MASARYK UNIVERSITY
FACULTY OF SCIENCES
DEPARTMENT OF BOTANY AND ZOOLOGY
Habilitation thesis
BRNO 2018 NATÁLIA MARTÍNKOVÁ
MASARYK UNIVERSITY
FACULTY OF SCIENCES
DEPARTMENT OF BOTANY AND ZOOLOGY
Modelling in phylogenetic
framework
Habilitation thesis
Natália Martínková
Brno 2018
Bibliographic Entry
Author: Mgr. Natália Martínková, Ph.D.
Faculty of Science, Masaryk University
Department of Botany and Zoology
Title of Thesis: Modelling in phylogenetic framework
Field of Study: Zoology
Academic Year: 2018/19
Number of Pages: x + 298
Keywords: Phylogenetic analysis; Maximum likelihood; Bayesian inferrence;
Molecular dating; Codivergence; Comparative phylogenetics;
Outlier analysis
Acknowledgement
Collaboration provides inspiration, drive, and increases achievements in scientific research.
This thesis contains research that would have been impossible without collaboration.
I thank all colleagues who worked with me in addressing numerous research
questions. Discussions with Jan Zima, Maarit Jaarola, Jeremy B. Searle, Gerald Heckel,
Jiří C. Moravec, Kamil S. Jaroň and Jiří Pikula have been very formative during progressing
stages of my career. My friends Alexandra Zahradníková, Jr., Danka Haruštiaková,
Nancy R. Irwin, Mabel Giménez and Pavel Škrabánek enriched not only my personal but
also my professional life. I thank Ladislav Dušek for his patience and support during my
pedagogical career.
Brno 9 Oct, 2018 . . . . . . . . . . . . . . . . . . . . . . . . . .
Natália Martínková
Contents
List of mathematical symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
Structure and goals of the habilitation thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Chapter 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1 Testing hypotheses using phylogenetic inference from genetic data . . . . . . 4
1.2 Modelling the time line of phylogenetic divergence . . . . . . . . . . . . . . . . . 11
1.3 Phylogenetic congruence in search of functional relationships . . . . . . . . . . 17
1.4 Predictive modelling of species traits in the context of genetic diversity . . . 21
1.5 Using outliers for identification of biologically relevant information in data
structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
Chapter 2. Author contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.1 Phylogenetic analyses papers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.2 Molecular dating papers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.3 Codivergence papers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.4 Comparative phylogenetics papers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.5 Outlier analyses papers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Paper 2.1.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Paper 2.1.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Paper 2.1.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Paper 2.1.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
Paper 2.1.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
Paper 2.1.6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
Paper 2.1.7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Paper 2.1.8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
– vii –
Paper 2.2.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
Paper 2.2.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
Paper 2.2.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
Paper 2.3.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
Paper 2.3.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
Paper 2.4.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
Paper 2.4.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
Paper 2.4.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
Paper 2.5.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
Paper 2.5.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
List of mathematical symbols
A matrix of associations between organisms
b vector of pairwise phylogenetic distances
b vector of pairwise phylogenetic distances of species that share a binary
trait
B phylogenetic distance matrix
C author contribution
c vector of the smallest phylogenetic distances for each species
c vector of the smallest phylogenetic distances of species that share a
binary trait
C phylogenetic distance matrix
D fourth-corner matrix
DIAS discrete interval accumulative score in the SigHunt analysis
f frequency
g generation
k number of categories partitioning research for a publication
K phylogenetic signal statistic
i index
M substitution model
M matrix
n number of units
N nearest neighbour interchange
NRI net relatedness index
NTI nearest taxon index
p probability
P population
ParaFitGlobal global statistic of phylogeny codivergence in the ParaFit analysis
Ri list of authors contributing to the i-th category
s selection coefficient
S sequence data
T phylogenetic tree
– ix –
x Přehled použitého značení
T matrix transposition
W global stastistic of matrix congruence from the CADM test
α significance level or parameter of the Γ distribution
β parameter of the Γ distribution
B bootstrap sampling
εεε vector of differences between the vector of trait values and the phylogenetic
mean of the trait value at the root of the tree
ˆεεε vector of trait values corrected for the variance-covariance of the tree
using the generalized least-squares
λ parameter of the exponential distribution
µ mean
ρ rate of evolution
σ standard deviation
φ function evaluating contribution of a specific author
Structure and goals of the habilitation
thesis
This thesis includes 17 publications in journals with an impact factor and one reviewed
conference paper. Included publications represent a selection of my work that is closely
related to the topic of statistical modelling in zoology. The publications are organized into
five topics, with each topic corresponding to a goal of the thesis.
1. Testing hypotheses using phylogenetic inference from genetic data to evaluate taxon
divergence and evolutionary history.
2. Molecular dating to model the time line of phylogenetic divergence at intra- and
inter-specific levels.
3. Testing phylogenetic congruence in search of functional relationships between groups
of taxa.
4. Predictive modelling of species traits in the context of genetic diversity.
5. Identifying biologically relevant information in data structure with detection of
outliers.
The habilitation thesis contains an Introduction that provides a wider knowledge base for
application of the methods and processes in my collated research. The Introduction updates
the current information in individual topics with respect to the accumulated progress in
research since the time of the publication. In some topics, the new advances explain
the justification for the practical applications of the analytical tools in my publication. I
demonstrate the concepts using diagnostic graphs, and I use simulation models to present
a streamlined picture of the described effects.
The diagnostic graphs originate directly from the archived analyses files, the results
of which are included in the respective publications. They explain and justify the applied
analytical procedure. The purpose of the diagnostic graphs is to guide a detailed commentary
into the problem of data analyses in phylogenetic framework and to critically evaluate
the obtained results.
Validating any result is a common practice in experimental biology, and the practice
needs to be applied to the data analysis as well. Simulations help to provide data for testing
the effect the analytical approach should reveal, and thus validate the results and help in
their interpretation. Statistical methods return a number if the user manages to input data
in the proper format. However, the more advanced methods challenge the user intuition in
– 1 –
2 Úvod
spotting incorrect results. A method might report a result with precision to several decimal
places that is several orders of magnitude incorrect. The diagnostics of the analysis run in
conjunction with simulations of the expected process enable the user to better differentiate
the correct results from analysis artefacts.
Chapter 1
Introduction
Genetic databases contain data from varied types of studies, where the respective authors
used DNA sequences and related information to test their specific hypothesis. As a
result, the databases accumulate information on model organisms and other taxa that were
considered relevant with respect to the multitude of different hypotheses in the scientific
endeavours of the last decades. Given time, the genetic databases accumulate DNA
sequences with great diversity both in terms of sampled taxa and genes. The amount of
the available genetic information enables the researchers to ask new questions that were
unfeasible at the time of the original research that generated the data. Using publically
available data is becoming routine not only as an addition to the original data, but datasets
from the public domain can now fully support new research questions on their own. Ease in
obtaining data increases demand for creativity in designing novel hypotheses and inspires
the ability to quickly and cheaply test the hypothesis in a pilot study.
The keystone hypotheses stemming from genetic information relate to differentiation of
biological diversity in form of reconstruction of phylogenetic relationships. Phylogenetic
relationships provide information on the succession of divergences and relatedness of
taxon groups. With help from additional information, a phylogeny can be transformed to
reflect time and rate of speciation events, providing a biological timeline corresponding to
geological or climate changes. In interactions between organisms, assessing codivergence
between phylogenetic trees addresses the character of associations between organisms
through time. A phylogeny facilitates comparing species traits, because similarity between
species is affected by their shared evolutionary history. To discover rare mechanisms that
may influence biological variability, the outlier analysis indicates directions for research
focus. A phylogeny provides a foundation for studying processes affecting biodiversity in
the web of interactions between organisms through time and space.
– 3 –
4 1. Introduction
1.1 Testing hypotheses using phylogenetic inference from
genetic data
Phylogenetic information in DNA sequences can be extracted using statistical methods
based on multiple sequence alignment or on alignment-free methods. Alignment-free
methods provide undisputable advantages in allowing the analysis without additional information
on gene content or location and appear to be ideal tools for the genomic era.
However, methods based on oligonucleotide composition, complexity measures or metrics
derived from information theory perform worse than statistical methods that use multiple
sequence alignment (Bogusz & Whelan, 2017; Chan, Bernard, Poirion, Hogan, & Ragan,
2014; Philippe et al., 2011). To construct a multiple sequence alignment means to find
positions in DNA (or amino-acid) sequences that share evolutionary history. An alignment
is then a matrix, where the rows represent individual sequences and columns correspond
to homologous positions in DNA sequences. When we construct an alignment from sequences
of multiple genes, the homology should be estimated per gene. In practice, we
first align gene sequences and then concatenate the individual alignments into a chimeric
supermatrix (Martínková & Moravec, 2012; Pečnerová & Martínková, 2012; Pečnerová,
Moravec, & Martínková, 2015).
Unlike numeric matrices, the cells in an alignment contain nucleotide bases. A transition
from one value in a matrix cell to another is a qualitative change reflecting a substitution
of one nucleotide for another. A challenge to quantify a change from one qualitative value
to the next requires a modelling approach. In phylogenetics, the substitution model quantifies
the probability of a specific substitution. For example, when a cytosine mutates to
a guanine, the probability of the change depends primarily on the frequency of guanines
in the sequence and the rate of cytosine to guanine mutations. The substitution models
can then be used to calculate genetic distances among sequences or to reconstruct their
phylogenetic relationships.
Current trends strongly prefer methods of phylogenetic reconstruction based on maximum
likelihood (ML) or Bayesian inference (BI) due to the reliability of the approaches in
estimating the correct phylogenetic relationships under variable conditions. Other methods
common in the past, such as parsimony or neighbour-joining clustering, do not scale well
with increasing sequence length and number of taxa in the dataset (Bogusz & Whelan,
2017). Although the ML and BI methods are also challenged with increasing dataset size,
they are sufficiently robust and allow flexibility in algorithm design that facilitated studies
of thousands of taxa with thousands of genes (Dobrin, Zwickl, & Sanderson, 2018).
Maximum likelihood searches the tree space using heuristics for a tree that will have
the highest probability p of observing the sequence data S under the condition of the
given phylogenetic tree T and the applied substitution model M. We call this conditional
probability p(S|T,M) a likelihood. A tree with the highest likelihood will best reflect the
evolutionary relationships between the sequences in the dataset.
Phylogenetic reconstruction using ML is computationally expensive in the attempts to
obtain node support using a bootstrap analysis. One bootstrap replicate imitates the analysis
of the full phylogenetic reconstruction, and thus the bootstrap analysis has linear time
complexity. First, the bootstrap analysis creates a pseudoreplicate of the DNA sequence
alignment, in which the columns are sampled from the original alignment with repetitions
1. Introduction 5
up to the total alignment length. The pseudoreplicate then contains identical traits (columns
of the alignment) but in different frequencies than those in the original data. In the bootstrap
analysis, each pseudoreplicate is subsequently analysed with ML and the analysis with 100
bootstrap replicates will take 100 times as long as the analysis with the original alignment.
The computational price lead to decrease in utility of the ML analysis with the increase in
available sequence data. Nowadays, the ML analysis has rejuvenated owing to increased
speed in bootstrap analysis (Stamatakis, 2014) and development of alternative measures
for node support estimation (Anisimova & Gascuel, 2006).
Bayesian inference evaluates the probability that we will observe a specific phylogeny
and its substitution model given the data in the alignment p(T,M|S). In BI, we need to
collate information about the model and specify its prior distribution before the analysis.
The advantage of the BI is that after the data are collected, the BI directly provides statistical
support for the hypothesis in form of the posterior probability (Nascimento, dos Reis, &
Yang, 2017). The posterior probability is the probability that the hypothesis (tree and
model) is correct given the data (alignment).
The BI uses Markov chain Monte Carlo (MCMC) to search the tree space. We can
imagine the tree space as a multidimensional space with all trees that are possible for a
given number of taxa (Fig. 1.1) (Hillis, Heath, & St John, 2005). The MCMC searches
the tree space with respect to the priors defined by the user by slightly modifying the
tree and evaluating whether to accept or reject the tree modification. Each tree and model
modification will be drawn from a prior for the substitution model parameters, tree topology
and branch lengths. The user chooses a substitution model and the distributions from which
the model parameters will be drawn. The tree topology can be influenced by priors that
assign a taxon or a group of monophyletic taxa as an outgroup, thereby specifying the
root of the ingroup and restricting the allowed topology changes during MCMC sampling.
Alternatively, monophyly of a group deep in the tree can be assigned as a prior, reducing
the tree space that will be searched by the MCMC. The last category of priors in BI is the
prior for branch lengths.
Length of branches in a phylogenetic tree is a linear combination of mutation rate and
time. Short branches represent evolutionary scenarios when the divergence occurred fast
with respect to the mutation rate of the respective locus. In BI, the mutation rate can freely
vary across the tree and the branch lengths are drawn randomly from a prior distribution.
Program MrBayes (Ronquist et al., 2012) uses an exponential distribution for the branch
length prior with the default value for its parameter λ = 10 (Fig. 1.2, red line). The default
prior is called uninformative, because it has similar probability density for a relatively wide
interval of possible branch lengths. In theory, an uninformative prior should have little
influence on the posterior and the posterior should be data-driven (Nelson, Andersen, &
Brown, 2015). In practice, the priors could in some cases override information in the data
in sampling the posterior.
The distribution of the posterior sample of branch lengths may mimic the prior distribution
in situations when the data are not informative (Fig. 1.3). In case of phylogeny
of voles from the subfamily Arvicolinae (Rodentia: Cricetidae) (Martínková & Moravec,
2012), the data were uninformative probably because of large amount of missing sequences
in the multilocus alignment. In several loci, relatively few species had homologous DNA
sequences resulting in more than 70% of missing data in the alignment. Matrices of aligned
6 1. Introduction
q
q
q
q
qq
q
q
q
q
q
qqq
q
q
q
q q
qq
qq
q
q
qq
qq
q
q
qqqqq
qq
qqqq
qqq qqqqqqqq
q
q qqq qqq
qqqq
q
qq
q
q
qqq q
qqqqqqq qq q
q
q
q
qqq qqq q
qqq
qq
q
q q
qq q
q
qq
q
q
q
qqqqqqqq
q
qqqqqq
qq
qqq
q qq
qqqqqqqq
q
qqq
qqqqqqqq
q
qqqq qq
q
q qq
q
qq
qqqqqq
q
q
q
q
q q
q
q
q
q q
qq
q
q
qq
q
qq qq qq
q
q qqq
q
qq
q
q
qqq q qqq
qq
q qq
q
qq qqq qqq
q
qq qqqq
qqqq
qqq
q
q q
q qqq
qq
q
q
qqq
qqq qq
qq
q
q
qq
q qqqqqqq
q
qq
qq q
qqqqqqq q
qq
qq qqqq q
q
q q
qq q
q q
q
q
qq
q
qq qqqq qqqq
qq qq
q q
q
q
qqq
qq
qqqqq q
qq
q
q
q
q
qq q
q
q
q
q
q
q
q
q
qq
qq q
q
qqq
qqq
q
q
qqqqqqq
qq
qq q
qq qq
q
q
q
q
q
qqq
q
qqq
qqqqqqqq
q
q
qq qq
q
qq
qqqqq q
q q
qqqq qq
qqqq q
q
qqqqq
qqq
qq
q
q q
q
qq qq
q
q
qqqq
qq q
q
qqq
q
qqqq
q
qqq
q
q
q q
q
qq
qqqq
q
qq qq
qq
qq
q
q
q
[1,600]
(600, 1200]
(1200, 1800]
(1800, 2400]
(2400, 3000]
(3000, 3600]
(3600, 4200]
(4200, 4800]
(4800, 5400]
(5400, 6000]
Figure 1.1: Markov chain Monte Carlo sampling of the tree space. Phylogenetic trees
occupy a multidmensional tree space, where their distances correspond to differences
in tree topology and branch lengths (Hillis, Heath, & St John, 2005). The tree space
here is reduced to two dimensions with multidimensional scaling for visualisation. Dots
show MCMC sampling of the tree space, with colours representing the progression of the
algorithm through the saved trees.
0.0 0.1 0.2 0.3 0.4
0
5
10
15
20
25
30
Branch length (substitutions bp−1
)
Probabilitydensity
λ = 10
λ = 20
λ = 30
λ = 40
λ = 50
Figure 1.2: Probability density of the exponential distribution. Exponential distribution
is used as a prior for branch lengths in Bayesian inference with default value of parameter
λ = 10. By increasing the parameter value, each generation of Markov chain Monte Carlo
modifies the phylogeny in Bayesian inference by drawing low values for branch lengths
with higher probabilities.
1. Introduction 7
0.0 0.1 0.2 0.3 0.4 0.5
0
5
10
15
Branch length (substitutions bp−1
)
Probabilitydensity
Prior
Posterior (internal branches)
Posterior (terminal branches)
a)
0.0 0.1 0.2 0.3 0.4 0.5
0
10
20
30
Branch length (substitutions bp−1
)
Probabilitydensity
b)
0.0 0.1 0.2 0.3 0.4 0.5
0
10
20
30
40
Branch length (substitutions bp−1
)
Probabilitydensity
c)
0.0 0.1 0.2 0.3 0.4 0.5
0
10
20
30
40
Branch length (substitutions bp−1
)
Probabilitydensity
d)
0.0 0.1 0.2 0.3 0.4 0.5
0
10
20
30
40
Branch length (substitutions bp−1
)
Probabilitydensity
e)
5 10 15
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Tree length (substitutions bp−1
)
Probabilitydensity
f)
λ = 10
λ = 20
λ = 30
λ = 40
λ = 50
Figure 1.3: Influence of priors on the posterior in Bayesian inference. Line - probability
density of the prior distribution, polygon - probability density of the posterior distribution
estimated from real data (Martínková & Moravec, 2012). Branch lengths are displayed
separately for internal branches and for terminal branches leading to tree tips. a) Prior
for branch lengths is given as an exponential distribution for λ = 10, b) λ = 20, c)
λ = 30, d) λ = 40, e) λ = 50. f) Prior for tree length is not explicitly defined, but it is
implicitly determined by the prior for the branch lengths as a Γ distribution with parameters
α = 2n−3 and β = λ, where n is the number of sequences in the dataset. Vertical black
line - maximum likelihood estimate of the tree length.
DNA sequences that contain missing data should not be confused with sparse matrices,
characterised by high frequency of zeroes. The large amount of missing data coupled
with their non-random distribution across the alignment matrix may produce inconsistent
results of phylogenetic reconstruction (Xi, Liu, & Davis, 2016). Reconstruction of phylogenies
from multilocus sequence data requires an efficient computational algorithm due
to high number of traits that are included in the analysis, and potential problems when
the result may contain statistically supported incorrect nodes. To analyse relationships
between arvicoline voles from the alignment with large amount of missing data, we chose
BI. Compared to ML, BI is more robust against missing data and the results contain fewer
artefacts (Dwivedi & Gadagkar, 2009; Simmons, 2012), although the BI could become
trapped in the tree space that contains unrealistically long trees (Ekman & Blaalid, 2011;
Marshall, 2010). Despite expected challenges, including multiple loci in phylogenetic reconstruction
improves estimation of relationships and better approximates true divergence
processes and their progression (Martínková & Moravec, 2012).
8 1. Introduction
My work in multilocus phylogenetics addressed the influence of branch length priors
on tree topology and the overall length of the tree in BI. The tree length is a sum of all
branch lengths in a phylogeny. If the posterior contains samples from a prior exponential
distribution, the posterior will reflect a convolution of exponential densities of all branches
(Nelson et al., 2015). A phylogeny contains 2n−3 branches for n sequences. Convolution
of 2n−3 exponential densities with parameters λ = 10 is a Γ distribution with parameters
α = 2n − 3 and β = λ. What follows is that there exists an implicit prior on the tree
length in BI that is sensitive to the number of sequences in the alignment and the branch
length prior (Nelson et al., 2015; Rannala, Zhu, & Yang, 2012). The branch length prior
strongly influences the theoretical tree lengths sampled in the posterior (Fig. 1.3f). Using
the example of the vole phylogeny (Martínková & Moravec, 2012), the implicit prior for
the tree length included the ML estimate of the tree length with the highest probability for
λ ∈ {30,40}.
Except the tree length, the branch length prior influences also the tree topology and
node support (Ekman & Blaalid, 2011; Kolaczkowski & Thornton, 2009; Martínková &
Moravec, 2012; Nelson et al., 2015). We observed a paradoxical situation in BI of the vole
phylogeny, when the node support decreased with increasing accuracy of the tree length
estimation (Fig. 3 in paper 2.1.2; Martínková & Moravec, 2012). Selection of the branch
length prior requires good justification, because the prior has an important influence on
the final phylogeny. Whenever possible, an uninformative prior with the parameter λ = 10
should be used.
We have shown that the BI analysis of alignments with a large amount of missing data
may result in unrealistically long phylogenies. An opposite situation can arise in an analysis
of divergent sequences, where the dataset contains information on insertions and deletions,
the indels (Dwivedi & Gadagkar, 2009; Kopecna et al., 2014; Wallace et al., 2012).
Indels are inheritable traits that appear in sequences during DNA replication when the
DNA polymerase slips along a DNA strand. As the offspring inherits the indels, they carry
phylogenetic information that could be informative for phylogeny reconstruction. However,
phylogenetic information in indels is lost in default uses of the phylogeny reconstruction
algorithms, and the indels are treated as missing data. The reason for the omission is
in the model of evolution that quantifies the qualitative change in the DNA sequence.
Common substitution models used in the ML and BI methods assume nucleotide changes,
and calculate the likelihood from DNA sequence nucleotide composition and rates of
substitutions. To include the information in indels, the dataset needs to be partitioned with
indels encoded as a binary morphological character. The evolution in the indels is then
modelled with an F81 model modified for binary data (McGuire, Denham, & Balding,
2001).
In phylogenetic reconstruction of aquaporins (Wallace et al., 2012), we utilised phylogenetic
information in indels and we obtained an unrealistically short tree given that the
dataset contained both protostomes and deuterostomes. The tree length estimated from
the BI was short compared to the ML estimate, but also with respect to the implicit prior
on the tree length (Fig. 1.4). The reason why the BI tree was too short is probably the
specifics of the model. The F81 model assumes an equal mutation rate at all sites. When
the alignment contains sites that mutate at a higher or slower rate, the BI tends to systematically
underestimate branch lengths (Nascimento et al., 2017). Different mutation rate at
1. Introduction 9
0.0 0.1 0.2 0.3 0.4 0.5
0
5
10
15
20
25
Branch length (substitutions bp−1
)
Probabilitydensity
Prior
Posterior (internal branches)
Posterior (terminal branches)
a)
5 10 15 20 25 30 35
0.0
0.5
1.0
1.5
Tree length (substitutions bp−1
)
Probabilitydensity
b)
Figure 1.4: Influence of a prior on the posterior in Bayesian inference with a mixed
model including DNA sequences and binary encoded indels. Line - probability density
of the prior distribution, polygon - probability density of the posterior estimated from real
data (Wallace et al., 2012). Branch lengths are displayed separately for internal branches
and for terminal branches leading to tree tips. a) Prior for branch lengths is given as an
exponential distribution for λ = 10, b) Prior for tree length is not explicitly defined, but it is
implicitly determined by the prior for the branch lengths as a Γ distribution with parameters
α = 2n−3 and β = λ, where n is the number of sequences in the dataset. Vertical black
line - maximum likelihood estimate of the tree length.
different alignment positions is biologically obvious. The most straightforward mechanism
generating different mutation rates for indels involves survival of carriers of mutations in
coding versus non-coding genomic regions. In non-coding sequences, indels will persist,
and thus appear to mutate, more often than in protein coding sequences, because insertions
or deletions of nucleotide bases in coding regions might lead to a damaged gene product
through frame shift or truncation of the coding sequence. Current implementations of the
evolutionary models for indels ignore variable mutation rates across sites. The caveat on
how to obtain a solution that can still utilise the robust BI is to estimate the tree topology
with BI, and then to calculate the branch lengths in the ML framework on a fixed topology
(Kopecna et al., 2014; Wallace et al., 2012).
We have demonstrated that multilocus alignments with missing data are informative
in reconstructing phylogenetic relationships (Martínková & Moravec, 2012; Pečnerová &
Martínková, 2012; Pečnerová et al., 2015). The amount of missing data present in the
sequence alignments would be prohibitive in other statistical analyses and the datasets will
need to be pruned or the missing data will be imputed to allow statistical evaluation. In
phylogenetics, we work with the temporal information inherent to the biological systems
and their continuity, and we are able to mine the information about taxon divergence also
from incomplete data matrices. Missing data nevertheless lead to loss of diversifying sensitivity
in additive elements of the tree likelihood calculations. In other words, likelihood
10 1. Introduction
will be identical for all possible scenarios of species divergence if the clade topology is
assessed from missing data (Sanderson, McMahon, & Steel, 2011). Trees with different
topology that have identical likelihood are called a terrace. When the alignment includes
complete data for all taxa and loci, a terrace will contain a single tree. That is; there will
be a single phylogeny with the given likelihood. In alignments with missing data, small
terraces can exist and they will ideally contain one tree (Dobrin et al., 2018; Martínková
& Moravec, 2012; Sanderson, McMahon, Stamatakis, Zwickl, & Steel, 2015; Sanderson
et al., 2011). The terrace size generally increases in datasets where the missing data affect
multiple loci or taxa (Dobrin et al., 2018); more so if the missing data are distributed in
the alignment non-randomly (Xi et al., 2016).
Occurrence of the terraces in the tree space based on alignments with missing data is
influential in interpretation of the result. First, the reported topology estimated with ML or
BI can belong to a terrace, meaning that there exist multiple trees that represent an equally
correct result, have the same likelihood, as the reported tree. In extreme published cases,
1023 − 10388 alternative tree topologies reside on the terrace (Dobrin et al., 2018). The
topological differences between trees from a terrace should be addressed in interpreting
the phylogeny (Martínková & Moravec, 2012; Pečnerová et al., 2015).
Except for the tree topology, terraces influence node support. A monophyletic group
can have high node support in bootstrap analysis in the ML framework if the clade is
present in trees on a terrace (Sanderson et al., 2015). In BI, the terraces affect node support
in a different way. While the trees on a terrace will have an identical likelihood, the priors
influence their posterior probability. The BI that uses exponential prior for branch lengths
(e.g. in Ronquist et al., 2012) will place a taxon with missing data as a sister to long
branches in the tree with higher probability (Kolaczkowski & Thornton, 2009; Sanderson
et al., 2015). Differences in posterior probability of trees from one terrace are given by a
complex interplay of priors, branch lengths and structure of missing data on topology of
the trees on the terrace (Sanderson et al., 2015).
There are no universal guidelines on how to address phylogenetic inference of large and
complex datasets. Each specific application of the phylogeny reconstruction methods on
concatenated data either from multiple loci or multiple traits requires sophisticated search
for balance between accuracy and precision of the solution. Fortunately, recent advances
in computational phylogenetics begin to suggest new algorithms that consider terraces.
Heuristics that search the tree space will soon be modified in such a way that they will
be robust against effects of terraces on phylogenetic reconstruction (Biczok et al., 2018;
Chernomor, von Haeseler, & Minh, 2016).
1. Introduction 11
1.2 Modelling the time line of phylogenetic divergence
In the last chapter, we addressed a problem of branch length estimation and the potential
inconsistencies under different scenarios. Correct estimation of the tree topology but also
branch lengths gains importance when we interpret the phylogeny in terms of evolutionary
change through time. Since branch lengths reflect change in data, they approximate the
biological phenomenon of mutation accumulation in the DNA sequences through time. We
can extract the time component inherent to the phylogenetic reconstruction and estimate the
divergence times by decomposing the branch lengths. If the branch lengths are a product of
mutation rate and time, calibrating mutation rate will enable transformation of the branch
lengths to time. We speak about molecular clocks (Kumar, 2005; Zuckerkandl & Pauling,
1962). Molecular dating analyses transform the node heights to time units and we will be
able to place the divergence events into relation to the history of Earth. The potential of
knowing when evolutionary scenarios happened is tantalising.
High variability between biological entities and processes complicates the elegant
molecular clock hypothesis. Mutation rate varies between different organisms, different
loci and when we observe mutation rate through time, it changes even within one lineage
(dos Reis, Donoghue, & Yang, 2016; Ho, Phillips, Cooper, & Drummond, 2005).
The differences in estimates of the mutation rate can be small with a small effect on
the interpretation of the results. For example, alternative molecular dating analyses can
differentiate whether Phoenicians or Vikings transported house mice along the European
coasts (Gabriel, Jóhannesdóttir, Jones, & Searle, 2010). Unfortunately, the application of
the molecular clock hypothesis and the time estimation from molecular data can seduce
even a single team into vastly different mutation rate estimates and skew the interpretations
of such results. As an example, two very close biological events were dated to times
that differed by more than five orders of magnitude using overlapping datasets within a
year of scientific advances. Specifically, the split of a pathogen from its sister taxon was
dated to tens of millions years ago (Palmer, Drees, Foster, & Lindner, 2018), whereas
the same pathogen intra-specific divergence was dated to thousands years ago (Drees et
al., 2017). Molecular dating is one of the most used and misused analyses in molecular
genetics. Availability of clickable software in combination with mathematical complexity
of the underlying statistics can lead and mislead biologists through parameter distributions,
hyperparameter distributions and transition kernels to sensational discoveries. Ascertaining
that the molecular dating discovery is not only sensational but also correct requires
additional information about the biological system that the analyst suitably incorporates
into the analyses (cf. Aghova et al., 2018).
The differences in mutation rates between evolutionary lineages will be notable on a
phylogeny as long branches that exceed the average root-to-tip distances. (In chapter 1.1,
we addressed the shortening or elongation of the whole tree, the tree length. Here, we
discuss length change in individual branches.) Unless the long branches are a sampling
or analysis artefact (Kolaczkowski & Thornton, 2009), they have a biological meaning. In
some organisms, such as viruses, an elevated mutation rate might be characteristic for the
whole clade (Smith et al., 2009). In others, long braches may indicate fast evolutionary
processes that might be adaptive (Fink, Excoffier, & Heckel, 2007).
Loci that encode proteins are limited in where the mutations can persist. In effect, we
12 1. Introduction
observe that the protein-coding loci mutate in a way that is compatible with survival. That
does not mean that mutations do not occur elsewhere in the gene. It means that carriers of
such mutations did not survive long enough to be sampled. Only a mutation that enables
the organism to survive to adulthood and to reproduce will have a chance to spread in a
population and to persist through time. On the other hand, mutations that are beneficial
to their carriers and increase the number of their offspring and their survival will have a
selective advantage. Beneficial mutations will thus have a higher probability in persisting
in a population through time. Looking at genetic diversity from the present situation back
through time, we can imagine that we observe mutations that had been sieved through
evolution in the past. Old and new beneficial mutations will remain in populations more
likely, and old deleterious mutations will not be visible in today’s genetic information due
to their carriers not passing their genes onto the offspring.
To better demonstrate the process of mutation loss with time, we can simulate a
model situation. Let’s have a population of individuals that carry mutations with selection
coefficient s, where s < 0 means the mutation is deleterious, s = 0 the mutation is neutral
and s > 0 represents beneficial mutations. Empirical research shows that about 30%
of mutations are deleterious, 42% neutral and only 18% of mutations bring the carrier
a selective advantage (Barrick & Lenski, 2013). Starting from these conditions, the
mutations become lost in time through genetic drift and selection (Fig. 1.5). We speak
about mutation rate decay (Barrick & Lenski, 2013; Ho et al., 2005).
We can use Markov chains to simulate the basic principle of the mutation rate decay,
when the relative decrease in mutation rate will develop from the starting conditions
through genetic drift and selection. Let’s assume that, for the purpose of simulating the
molecular rate decay, the frequency of appearance of a mutation with the specific selection
coefficient is equal to frequency of the mutation in the starting population. A population
P with n individuals is characterised as P = {p1, p2,..., pn}, where p = 1 in individuals
that carry the mutation and p = 0 in individuals without the mutation. The number of
individuals with a mutation with a given selection coefficient follows information in Box
1 in Barrick & Lenski (2013) that lists the frequency of mutations f for different selection
coefficients s. We start the simulation with a population P1 in the first generation g1.
Development of the frequency of the mutation in the populations can be simulated as a
Markov process:
fg+1 =
1
n ∑B Pg+1 , if fg (1+s) < 1
1, if fg (1+s) ≥ 1
, (1.1)
where B is bootstrap samplings of the population Pg+1. The population Pg+1 is created as a
permutation of fg (1+s)n individuals with the mutation and the remaining individuals
without the mutation up to the total of n individuals. We can see that no lethal mutations
occur in the simulated populations after the initial generations (Fig. 1.5, red line) and the
deleterious mutations also quickly disappear from the populations (Fig. 1.5, orange lines).
Selectively neutral and beneficial mutations persist in populations, although not universally
(Fig. 1.5, blue and green lines), and some mutations disappear from the simulated
populations through random drift. Multiple other factors such as demography, population
fragmentation or associations between mutations in the genome will influence persistence
and loss of mutations in a real biological system and in more advanced models (Ho et
1. Introduction 13
0 20 40 60 80 100
0.0
0.2
0.4
0.6
0.8
1.0
Generation
Frequencyofmutationinpopulations
a)
0 20 40 60 80 100
0.0
0.2
0.4
0.6
0.8
1.0
Generation
Frequencyofpopulationswiththemutation
b)
Highly beneficial, s = 0.3, f1 = 0.05
Slightly beneficial, s = 0.1, f1 = 0.13
Neutral, s = 0, f1 = 0.42
Slightly deleterious, s = −0.1, f1 = 0.17
Highly deleterious, s = −0.5, f1 = 0.13
Lethal, s = −1, f1 = 0.1
Figure 1.5: Simulation of the principle of the mutation rate decay. a) Persistance of
individual mutations modelled with Markov chains in 20 populations consisting of 100
individuals each. b) Frequency of populations in which the mutation persists. Black line
shows the total proportion of populations that have the starting mutation with a certain
selection coefficient s at the given time. Coloured lines demonstrate that the loss of
mutations occurs predominantly in selectively deleterious mutations, and that the decay is
fastest in early generations.
14 1. Introduction
al., 2005). Despite limitations, our simple simulation shows that when we estimate the
mutation rate from old divergences, we will evaluate fewer mutations than how many truly
appeared in the populations early after their split. When we categorised mutations into six
groups, from lethal with s = −1 to selectively beneficial with s = 0.3, we lost more than
half of all initial mutations in 100 generations (Fig. 1.5, black line).
Because of high variability of the mutation rate between lineages and through time,
extrapolating the mutation rate outside of the range of data used to estimate it is risky.
To avoid the errors, we need to find information that will provide a rough time frame for
the possible divergence in the group. Using the terminology of BI, we need to specify
the priors. Subsequent molecular dating analysis will be based on the most accurate and
reliable calibration that can be found.
There are two ways how to calibrate the molecular clock - by calibrating tree tips or by
calibrating the internal nodes of the phylogeny. Calibrating the molecular clock from the
tree tips assumes that the sequences originated at different times (Martínková et al., 2013;
Murray et al., 2016). Depth of the dated tree tips (difference between the oldest and the
youngest sequence in the dataset) should reach at least one tenth of the expected time frame
between the tips and the root of the tree. This means that we will not be able to use DNA
sequences obtained from museum collections several decades old for dating divergence of
taxa that could have occurred hundred thousands to millions of years ago. Instead, we
will have to find suitable fossil material several thousands to tens of thousands years old,
estimate the age of the fossils using radiocarbon dating or a similar method, isolate DNA
from the fossils, sequence it and use those sequences for the tip-dated calibration of the
molecular clock (Martínková et al., 2013).
Calibration of internal nodes is less demanding on technological applications than
the tip-dated calibration, but we should verify the suitability of the node height limits
for the specific divergence. Paleontological, geological or biogeographic information can
guide the node calibration. Nodes in a phylogeny characterise points of divergence of two
lineages. That is, a node represents a hypothetical most recent common ancestor of the
two lineages. A fossil organism known to be a common ancestor of a certain monophyletic
group can be used as a suitable calibration anchor, a so-called time constraint. Some
paleontological data may contradict one another with respect to the phylogeny (Aghova
et al., 2018; Near, Bolnick, & Wainwright, 2005). For example, a fossil considered an
ancestor for a genus would be dated as older than a fossil ancestor of a family. We need to
identify the conflicts using cross-validation (Near & Sanderson, 2004). In addition to fossil
calibration, we can use geological (e.g. island emergence above the sea level or ice-sheet
retreat (Martínková, McDonald, & Searle, 2007)) or biogeographic data (e.g. formation of
land bridges between continents (Pečnerová et al., 2015)). Geological and biogeographic
calibration enables the use the molecular clock hypothesis in groups and at times when the
fossil record is rare.
We can demonstrate the effect of additional information on molecular dating using
the example of the Irish population of stoats (Mustela erminea; Carnivora: Mustelidae)
(Martínková et al., 2007). Using additional information in molecular dating requires
accepting a set of assumptions about the biological system that should be firmly anchored
in reality. For example, the analysis of genetic diversity in Irish stoats indicates a single
colonisation of the island followed by a population expansion. That means that the number
1. Introduction 15
of first colonists was relatively small and the influence of the founder effect and genetic drift
resulted in low genetic diversity shortly after the colonisation. The subsequent population
expansion means that many offspring, and thus carriers of mutations, survived and thus
we observe an increase in genetic variability. The burst of diversification will look like a
star tree. In Irish stoats, we observe a star tree in the DNA sequence data (Martínková et
al., 2007). When we identified the pattern of colonisation, we can consider the time frame.
Arrival of stoats to Ireland was unlikely before the ice-sheets retreated 40 thousand years
ago. On the other hand, the oldest stoat fossils from Ireland are dated to 13 thousand years
ago. The two dates set the youngest and the oldest possible date for the colonisation. The
time limits represent a uniform prior for the tree root of mitochondrial DNA sequences
of stoats from Ireland (Fig. 1.6, extent of the blue field on the x axis). We chose the
prior for mutation rate as uninformative (Fig. 1.6, extent of the blue field on the y axis).
Using the Bayesian coalescence analysis, we estimate the posterior distribution of the tree
height and mutation rate based on the coalescence of the DNA sequences given the priors.
The root height thus reflects the age of the Irish stoat population. The obtained mutation
rate estimates the rate of evolution in stoats during late Pleistocene glaciations and can
be applied to other stoat populations in which we expect genetic changes at similar time
frames. The intraspecific mutation rate in stoats equals to about 14% change per million
years, which is an order of magnitude faster than mutation rates dating speciation events
(Ho et al., 2005; Pečnerová et al., 2015; Smith et al., 2009). We need to bear in mind
that the estimated mutation rate cannot be used in million years old divergences, as it will
decay with time (Fig. 1.5).
Mastering the molecular dating analyses requires good statistical knowledge of the
analyst, but equally imporant is the ability to cooperate with specialists in other fields.
Namely, biologists planning molecular dating analyses should work with paleontologists,
radiologists and geologists to obtain reliable data for the molecular clock calibration.
Avoiding the pitfalls of molecular dating is easiest with correctly identified fossil remains
that represent ancestors of the studied contemporary taxa, reliably estimated ages of the
fossils from radiocarbon dating and properly chosen geological and climate events for
constraining the nodes of the dated phylogeny.
16 1. Introduction
0.010 0.015 0.020 0.025 0.030 0.035 0.040
Root height (My)
Mutationrate(substitutionsbp−1
My−1
)
0.01
0.1
1
10
100
Figure 1.6: Relationship between the prior and the posterior in molecular dating using
Bayesian coalescence analysis. Blue shading - priors for root height and mutation rate.
The root height prior for the Irish population of stoats was set as a uniform distribution in
range [0.013,0.04] million years (My). The mutation rate prior is set as uninformative as
a uniform distribution in range [0,150] substitutions per base pair (bp) per My. Contour
graphs - the area of the highest density in the posterior sample (red to green colour).
1. Introduction 17
1.3 Phylogenetic congruence in search of functional rela-
tionships
During the course of evolution, organisms influence one another, sometimes to the extent,
where a change in one organism leads to a reciprocal change in another. We speak of
coevolution (Krasnovyd, Vetešník, Gettová, Civáňová, & Šimková, 2017; Simoes et al.,
2016). Similarity in topology of phylogenetic trees of two groups of associated organisms
is called codivergence. Where the coevolution assumes mutual influence of the pairs of
organisms, codivergence does not imply causality in change. Two trees codiverge, when we
observe a corresponding divergence in both of the associated clades in the two phylogenies.
The character of the association may differ, but we study codivergence usually between
parasites and their hosts. Following fundamental principles of parasitism, we would expect
that random selection of hosts might seldom occur in generalists, but would still be restricted
to a certain taxonomic group. For example, a virus causing rabies, RABV, may infect many
mammalian species, but plants are resistant to the infection (Marston et al., 2018). Most
parasite-host relationships will reflect variable specificity of the parasite and differences in
susceptibility of the host. Their interactions through time will modify not only the ability
of the parasite to infect the host, but also protective mechanisms of the host. Herein, we
will assume non-random associations. Phylogenetic codivergence between parasites and
hosts can be defined in situations, when we observe a speciation event in hosts co-occurring
with a speciation event in parasites. The direction of the interaction is not relevant for the
purposes of codivergence.
We can test associations between parasites and hosts in phylogenetic context using comparisons
of matrices of pairwise phylogenetic distances (Legendre et al., 2002; MartinezAquino,
2016). The strictest comparison in assessing codivergence is a test whether the
matrix of phylogenetic distances between parasites is congruent with a similar matrix of
the hosts. Two matrices B and C are congruent, when there exists a matrix M, for which
we can equate:
C = MT
BM, (1.2)
where T means transposition. Illustratively, we can understand the essence of matrix
congruence as if we compared two photographs of the same place taken at different
weather conditions.
Congruence between phylogenetic trees can be tested with a permutation test of congruence
between distance matrices (CADM; Campbell et al., 2011). The CADM test
calculates a global statistic W that is defined over the interval [0,1]. W = 0 means that two
matrices are fully incongruent, and W = 1 means that matrices are fully congruent. The
statistic W provides an intuitive guide to the extent to which two matrices are congruent.
Statistical significance of W can be tested with a permutation test that evaluates a null
hypothesis that the two matrices are incongruent (W = 0). A non-significant CADM test
means that we can expect at least partial congruence between matrices of phylogenetic
distances of parasites and hosts.
The CADM test can be used to test the global congruence between a group of parasites
and a group of hosts with the caveat that the associations must be strictly paired. This
18 1. Introduction
means that in the CADM test, each parasite infects exactly one host and each host has
exactly one parasite. In natural applications, we often observe that multiple parasites infect
one host or that one parasite infects multiple host species. In host-parasite systems with
multiple associations per taxon, ParaFit can test the global signal for codivergence. ParaFit
uses a permutation test to evaluate whether the associations between parasites and hosts
are random (Legendre et al., 2002).
The ParaFit method calculates a fourth-corner statistic using matrix algebra:
D = CAT
B, (1.3)
where B is matrix of phylogenetic distances between parasites, modified with principal
coordinate analysis, C is a similar matrix obtained for the host relationships and A is an
association matrix, where value 1 means association between the corresponding taxa and
0 means no association. Matrix D is thus a linear combination of phylogenetic information
in the parasite and host trees weighted through their associations (Legendre et al., 2002).
ParaFit calculates a statistic ParaFitGlobal, which is a sum of squares of the values in
matrix D.
Statistical significance of codivergence between the two phylogenies given their associations
does not fully suffice in explaining the biological phenomenon. Let’s use an
example with random simulated trees (Fig. 1.7). We tested the signal for codivergence
in a set of 100 random trees compared to the starting tree. We found several trees with
significantly non-random codivergence, where each tip has exactly one association in both
compared trees using ParaFit (Fig. 1.7c, orange colour). The non-random codivergence
is surprising in light of tree distance calculated as the weighted, normalised Robinson &
Foulds distance (RF; Robinson & Foulds, 1981). The RF distance between random trees
with non-random associations reached more than 50% of maximum RF distance weighted
to the total tree length.
Applying a biologically more realistic model, we can simulate the divergence of two
trees with evolving host-parasite interactions using a Markov chain:
Ti = N (N (Ti−1)), (1.4)
where T is a phylogenetic tree, i ∈ N+ up to the total length of the simulation, T0 is a
starting tree and N is the nearest neighbour interchange. We observed statistically nonrandom
codivergence with the starting tree in 13 simulated phylogenies from a series of 100
trees, where each tree differed from a previous tree by two nearest neighbour interchanges
(Fig. 1.7d).
In most simulated paired trees of parasites and hosts, in which we observed nonrandom
codivergence for p(ParaFitGlobal) ≤ 0.05, we also observed that the matrices of
phylogenetic distances were congruent at p(W) ≤ 0.05 (Fig. 1.7d, f). This is to be expected.
Our simulations showed a surprising result, where some randomly codiverging trees tested
with ParaFit can be significantly congruent as tested with CADM (Fig. 1.7c, e), indicating
a partial congruence in some parts of the two trees.
Our simulations demonstrate that we can observe higher than random congruence even
in relatively divergent trees. Phylogenetic codivergence and congruence thus merits a more
detailed study.
1. Introduction 19
p(ParaFitGlobal) = 0.052, p(W) = 0.023, RF = 0.63
a)
p(ParaFitGlobal) = 0.008, p(W) = 0.012, RF = 0.41
b)
q q
q
q
q qq q
q
q
q
q
q
q
q
q
q
q
q q
q
q
q
q
q
qq
q
q
qq qq
qq
q
q
q
q
q
q
q
qq
q
q
q q
q
q
q
q
q
q
q
q
q
q
q qq
q
q
q
q
q
q
q
q
q
q
q
q q
q
q
q
q q
q
qq
q
q
q
q
q
q
qqq
q
q
q
q
q
q q
q
q
0.45 0.55 0.65 0.75
100
200
300
400
500
600
Weighted, normalised RF distance
ParaFitGlobal
c)
q
q
q
q
q
q
q
q
qqqq
q
q
q
q
qqq
q
q
q
qq
qqqq
q
qq
qq
qq
qqqqqqq
qq
q
q
q
qqq
q
q
q
q
q
q qq
qq
qq
q
q
qq
q
qq
q
q
q
q
q
q
q
q
qq
q
q
qqq
q
qq
qqq
q
q
qq
q
q
q
qqq
0.1 0.2 0.3 0.4
100
200
300
400
500
Weighted, normalised RF distance
ParaFitGlobal
d)
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
qq q
q q
q
q
q
q
q
q
q
q
q
q
q
q
q
qq
q
q
q q
q
q
q q
q
q
q
q
q
q
qq
q
q
q
q
q
q
q
q
q
q
q
q q
q
qq
q
q
q
0.45 0.55 0.65 0.75
0.4
0.5
0.6
0.7
0.8
Weighted, normalised RF distance
W
e)
q
q
q
qq
qq
q
q
qqq
q
q
q
qq
qq
qqqq
q
qqq
q
q
q
q
qq
q
q
qq
qqqqq
q
q
q
qq
qqq
q
q
q
q
q
q q
q
qq
qq
q
q
qq
q
qq
q
q
q
q
qq
q
qq
q
q
qqqq
q
qq
qqqq
q
qq q
qq
qqq
0.1 0.2 0.3 0.4
0.4
0.5
0.6
0.7
0.8
0.9
Weighted, normalised RF distance
W
f)
Figure 1.7: Codivergence and congruence of simulated phylogenies. a) Tanglegram
between randomly generated trees shows that codivergence between them cannot be distinguished
from random. b) Example of codiverging phylogenetic trees. c-f) Weighted,
normalised Robinson & Foulds distances (RF; Robinson & Foulds, 1981) between random
(c, e) and step-wise diverging (d, f) phylogenies in relation to the statistic ParaFitGlobal
(Legendre, Desdevises, & Bazin, 2002) (c, d) and CADM test statistic W (Campbell,
Legendre, & Lapointe, 2011) (e, f). Orange colour - ParaFitGlobal is significantly higher
than in random associations between the trees (c, d), W is significantly higher than in fully
incongruent matrices (e, f); green colour - the test were not significant at α = 0.05; cross
- example of associations displayed in panels a) and b).
20 1. Introduction
We studied codivergence of New World arenaviruses with their rodent hosts from
subfamilies Sigmodontinae and Neotominae (Rodentia: Cricetidae) (Irwin, Bayerlová,
Missa, & Martínková, 2012). The relationship was historically believed to represent
codivergence based on the fact that arenaviruses in the New and Old World infect rodents
from different families and the infections may be asymptomatic (Bowen, Peters, & Nichol,
1997). Even though empirical data rejected the hypothesis of long-term codivergence
of arenaviruses with their hosts (Jackson & Charleston, 2004), the opinion lingered in
predominantly biomedical literature. Studying all known arenaviruses from South and
North America in the context of all available hosts from the respective rodent subfamilies
with DNA sequence data, we found signal for codivergence only in some groups of the
parasites. In these groups, phylogenetic clustering of arenaviral infection on the host
phylogeny (Chapter 1.4) in conjunction with geographical modelling showed that the
parasites utilise related hosts that are locally available (Irwin et al., 2012).
We found no codivergence when comparing a molecular phylogeny with a similarity tree
derived from skull morphological data in tree squirrels from the tribe Sciurini (Rodentia:
Sciuridae) (Pečnerová et al., 2015). Instead, the skull morphology corresponded to the
current taxonomy of tree squirrels at the genus level and was likely driven by adaptation
to different food sources. The molecular phylogeny indicated a step-wise divergence
associated with lineage colonisation of new regions.
Interpreting presence (Irwin et al., 2012) or absence (Pečnerová et al., 2015) of statistically
significant signal for codivergence of two phylogenies requires knowledge of the
studied system and search for covariates that may influence the respective evolutionary
histories. By applying the codivergence analysis, we were able to find signals indicating
interesting functional variations in the diverse groups of organisms.
1. Introduction 21
1.4 Predictive modelling of species traits in the context of
genetic diversity
Characteristics of organisms, their traits, have a heritable component. The concept of trait
heritability forms a firm foundation of current systematics. For example, we speak of a
synapomorphy if a monophyletic group shares a trait and we expect that the trait was present
also in their ancestor. Traits change over the course of evolution. Qualitative traits arise
or become lost and quantitative change their size or range. The biological concept of trait
inheritance poses a complication in statistical modelling, because we cannot assume that
the observations are independent. We need to quantify the extent to which phylogenetic
relatedness influences trait variability and to account for the relatedness in the analyses.
When studying binary traits in the phylogenetic context, we can use metrics derived in
community ecology that assess phylogenetic similarity of communities. Specifically, the
net relatedness index and the nearest taxon index are useful to model evolution of a binary
trait on the phylogeny.
Net relatedness index (NRI) shows a scaled and transformed difference between mean
phylogenetic distance of the species with the given trait and the overall mean phylogenetic
distance in the tree (Webb, Ackerly, McPeek, & Donoghue, 2002):
NRI = −1
µ (b )− µ (b)
σ (b)
, (1.5)
where b is a vector of pairwise phylogenetic distances, b is a vector of pairwise phylogenetic
distance of the species that share the trait of interest, µ is a mean and σ is a standard
deviation. The multiplicaiton by −1 in the equation 1.5 simplifies the interpreration
of NRI. Values greater than zero indicate that the trait is distributed on the tree in a
phylogenetically closely related group and the values less the zero mean that the trait is
overdispersed on the tree. Values of NRI ≈ 0 mean that the species with the trait are
randomly related with respect to the observed diversity of the studied system. We test the
statistical significance whether NRI = 0 using permutations. The model that permutes the
trait value across the whole phylogeny without topological limitations is the one applicable
for comparative phylogenetics.
The other community ecology index useful in comparative phylogenetics is the nearest
taxon index (NTI). We calculate the NTI similarly to the NRI (Webb et al., 2002). The
NTI also compares observed and expected relatedness scaled to the standard deviation of
the phylogenetic distances, but it does not work with all pairwise distances. Instead, for
each taxon, the NTI searches for the smallest phylogenetic distance that signifies its nearest
neighbour. We can define NTI as:
NTI = −1
µ (c )− µ (c)
σ (c)
, (1.6)
where c is a vector of the smallest phylogenetic distances for each species in the phylogeny
and c is a vector of the smallest phylogenetic distances of the species that share the trait
of interest.
The utility of the indices NRI and NTI in comparative phylogenetics is in their ability to
estimate the conservatism of a given trait (Fountain-Jones et al., 2017; Webb et al., 2002).
22 1. Introduction
When the trait is conservative across the whole studied diversity, both indices will show
high values. When the trait is present predominantly in closely related taxa, the NRI will
decrease, but the NTI will remain high. In community ecology, high values of NRI and
NTI mean selection of evolutionary conservative traits driven by environmental conditions
(Fountain-Jones et al., 2017). We used an analogous interpretation in studying the trait
distribution in a model of host-parasite interaction in fungi and viruses infecting mammals.
We found an opposite relationship in hibernating bats (Chiroptera: Vespertilionidae,
Miniopteridae, Rhinolophidae) infected with fungus Pseudogymnoascus destructans (Ascomycota:
Pseudeurotiaceae) than that reported by Fountain-Jones et al. (2017). Infected
bats were significantly clustered on the phylogenetic tree (NRI > 0), but the sister taxa
were randomly affected (NTI ≈ 0) (Zukal et al., 2014). We expect that the result reflects
non-representative sampling that concentrated on the representatives of the genus Myotis.
Myotis species are the most frequent bats in Holarctic hibernacula and due to their species
diversity, Myotis strongly influence the phylogenetic composition of bat communities in
hibernacula (Horáček, Bartonička, Lučan, & Czech Bat Conservation Trust, 2014).
Group B of New World arenaviruses are important human pathogens (Charrel & de
Lamballerie, 2003). Hosts of the group B arenaviruses are randomly distributed on the
phylogenetic tree of all possible rodent hosts from South and North America. Hosts of
other groups of arenaviruses, not pathogenic to humans, are closely related (Irwin et al.,
2012). Together with information on host-parasite codivergence (Chapter 1.3), the absence
of phylogenetic clustering on the host tree means that group B arenaviruses are capable of
frequent host switching.
For quantitative traits, we can study evolution of the trait using phylogenetic signal.
We consider a trait to show phylogenetic signal when two phylogenetically related taxa
will be more similar to one another than two taxa randomly selected from a phylogeny
(Blomberg, Garland jr., & Ives, 2003; Losos, 2008; Pyron, 2015). The phylogenetic signal
is thus covariance of the quantitative trait in different taxa that can be explained by their
shared evolutionary history (Revell, Harmon, & Collar, 2008). Whether we assess body
size, food preferences or susceptibility to infection, we assume that the mean of the given
trait in a species will be determined by heritable processes. Closely related species will
have similar mean values of the studied trait.
Phylogenetic signal as defined by Blomberg et al. (2003) is calculated as a ratio of
mean squared error of the trait given the phylogenetically corrected mean of the trait (trait
value at the root of the phylogeny) to the mean squared error of the trait transformed with
the variance-covariance matrix derived from the phylogeny:
µ εεε2
µ ˆεεε2
, (1.7)
where εεε is a vector of differences between the vector of trait values and the phylogenetic
mean of the trait value at the root of the tree and ˆεεε is a vector of trait values corrected
for the variance-covariance of the tree using the generalized least-squares. The values of
the phylogenetic signal calculated with equation 1.7 will be dependent on the tree size,
topology and length, making universal cross-comparisons difficult. Therefore, we now
express phylogenetic signal as a statistic K that is a ratio of the observed to expected ratios
1. Introduction 23
in equation 1.7 (Blomberg et al. 2003). The expected values are calculated assuming a
Brownian motion model of the trait evolution.
When K ≈ 1, the variable changes along the phylogeny as expected under the Brownian
motion model of trait evolution. Departures from one in either direction indicate
biologically meaningful interpretation in terms of adaptive evolution. For K < 1, closely
related species have trait values less similar than expected due to their relatedness and thus
they might be influenced by alternative adaptive processes. The adaptation to a similar
selection pressure affecting multiple taxa may drive trait similarity within a clade with
K > 1 (Blomberg et al., 2003), and a trait with K > 1 could be correlated with phylogenetic
niche conservatism (Losos, 2008).
We found statistically significant phylogenetic signal for K < 1 in pathogen load in
hibernating bats (Zukal et al., 2016). The fungus P. destructans causes disease white-nose
syndrome (WNS) in hibernating bats inhabiting Holarctic biogeographic ecozone, where
the Nearctic bats often experience drastic population declines following pathogen invasion
in the hibernacula, but Palearctic bats are able to tolerate the infection better (Lorch et al.,
2011; Wibbelt et al., 2010; Zukal et al., 2014; Zukal et al., 2016). The pattern with the
highly susceptible species in the Nearctic and the tolerant species in the Palearctic can be
observed in multiple affected genera with representatives in both regions, such as Myotis,
Eptesicus, or Pipistrellus/Perimyotis species group. However, species from one genus vary
in infection intensity also in the Palearctic (Pikula et al., 2017; Zukal et al., 2016).
We can demonstrate the scale of interspecific variation in fungal load by simulating the
expected fungal load at tree tips with a model of evolution following Brownian motion and
comparing the simulated data with the empirical data (Fig. 1.8). The Brownian motion
evolutionary model assumes that the rate of evolution ρ is stable through time and thus
the change in a trait value depends on the rate of evolution and varies with respect to
a stochastic component drawn from a normal distribution with mean equal to zero and
variance equal to ρ2 (Butler & King, 2004). With progressing time, the lineages will
accumulate more random changes in the trait values leading to their divergence from one
another. The density plots from simulated trait values in derived taxa (Myotis) will show
greater variance than taxa that branched from the common ancestor early (Rhinolophus)
on the phylogeny (Fig. 1.8, green colour).
The variable susceptibility of different species that cannot be attributed to the relatedness
of taxa might thus reflect other influences on infection intensity such as species
ecology, behaviour or environmental conditions during hibernation. Relationship between
quantitative traits in a phylogenetic context can be studied with the phylogenetic generalized
least-squares (PGLS; Ives, Midford, & Garland jr., 2007; Revell, 2010). In PGLS,
the inferred intercept and slope of the regression are corrected with respect to the fact
that the data are hierarchically structured and co-vary to a certain extent due to the relatedness
of the taxa (Pikula et al., 2017; Zukal et al., 2016). The ordinary least-squares
method is extended in the phylogenetic context by adding a variance-covariance structure
derived from a phylogeny to the regression (Symonds & Blomberg, 2014). The variancecovariance
structure reflects the non-independence of the samples in PGLS, where the
samples (trait values in species) will be influenced by their shared evolutionary history.
The variance-covariance matrix in the PGLS represents the expected covariance of residuals
in the regression (Symonds & Blomberg, 2014). That means that in using PGLS,
24 1. Introduction
Miniopterus schreibersii
Eptesicus nilssonii
Plecotus ognevi
Plecotus austriacus
Plecotus auritus
Barbastella barbastellus
Murina hilgendorfi
Myotis brandtii
Myotis gracilis
Myotis mystacinus
Myotis bechsteinii
Myotis myotis
Myotis bombinus
Myotis nattereri
Myotis petax
Myotis macrodactylus
Myotis daubentonii
Myotis dasycneme
Myotis alcathoe
Myotis emarginatus
Rhinolophus euryale
Rhinolophus hipposideros
Rhinolophus ferrumequinum
Fungal load (log10(ng cm−2
))
−8 −7 −6 −5 −4 −3 −2 −1
Figure 1.8: Trait evolution modelled with Brownian motion. The fungal load trait
values were modelled on a multilocus phylogeny of hibernating Palearctic bats infected
with Pseudogymnoascus destructans (Zukal et al., 2014). The rate of evolution ρ = 3.8 and
fungal load at the root equal to 3.62×10−6 ng cm−2 used in simulations were derived from
Zukal et al. (2016) and Pikula et al. (2017). Green - density plots from 100 simulations of
the expected fungal load at tree tips, orange - density of the empirical fungal load in each
species.
1. Introduction 25
we assume phylogenetic signal in the residuals from the regression, not necessarily the
variables themselves (Revell, 2010).
The current regression methods in the phylogenetic context take into account intraspecific
variability of the trait value (Ho, & Ané, 2014; Ives et al, 2010). This is an important
advancement, because including within species variation in a model improves the fit between
the model and the observations. We applied the PGLS to study the relationship
between infection intensity measures that characterize disease progression from a superficial
fungal infection to development of the WNS (Zukal et al., 2016). By correcting for
shared evolutionary history between taxa and trait variability within species, we showed
that the skin lesions diagnostic of WNS develop with increasing fungal load more slowly
than expected (Zukal et al., 2016), but severity of the disease increases faster if the bat had
more skin lesions (Pikula et al., 2017).
26 1. Introduction
1.5 Using outliers for identification of biologically relevant
information in data structure
Each biological system is variable and statistical methods assume that the values of a trait
follow a frequency distribution, often the normal distribution. Trends are then estimated
in relation to the parameters of the distributions such as their mean. Statistical analyses
transform data in such a way that we minimise observed variance, we explain the variance
or we categorise data into classes with minimum variance. In each of these applications, we
focus on the overall information content in the data and we consider outliers as undesirable.
The outliers would either be viewed as errors in measurement or unique aberrations likely
to be unsuccessful in the course of evolution. Since they may have a strong influence on
results of the statistical analyses, the outliers are deleted from the datasets upon identification.
While deleting measurement errors is always warranted, some outliers might carry
information about an alternative mechanism with a biological foundation. If we had a
sufficiently large sample size (hundreds to thousands of measurements), outliers generated
by an alternative mechanism would form a separate cluster in ordination analyses and we
would detect them during data explorations as relevant and worth of further investigation.
Datasets in biology are often smaller and the power of statistical tests cannot distinguish
difference in means of common and rare phenomena. Rare phenomena that generate outliers
in small datasets are long overlooked or they may appear as descriptions of individual
occurrences (case studies; cf. Pikula et al., 2017).
We used the outlier analysis in two applications to detect observations generated by
an unusual and rare mechanism. We designed a method SigHunt to search for genomic
islands in eukaryotic genomes. The rare mechanism that generates the genomic islands is
the horizontal gene transfer. The second method of outlier analysis is EEO that can be used
to identify exceptional behaviour in hibernating bats. We expect the exceptional behaviour
during hibernation in moribund bats with WNS, but lethal cases of WNS are rare in the
Palearctic (Pikula et al., 2017).
Genomic islands are regions in the genome that the organism obtained from unrelated
organisms through horizontal gene transfer. Horizontal gene transfer can equip organisms
with pre-evolved traits that may contain metabolic pathways enabling utilisation of new
resources. We hypothesised that the genome of the fungal pathogen P. destructans might
contain genomic islands based on the fact that the fungus belongs to a group of predominantly
saprophytic organisms. An ability to infect live tissues biochemically differs from
dead tissue decomposition. A parasite thus needs to produce compounds facilitating invasive
growth or protecting the parasite against the immune system of the host. Obtaining a
pre-evolved metabolic pathway for utilising new resources is advantageous, but the process
can occur only between organisms that are at least temporarily in close contact to expedite
exchange of genetic information. Bacteria use conjugation or transduction as common
mechanisms of horizontal gene transfer. In eukaryotes, transposons or retrotransposons
enable transfer of genetic material between organisms that do not normally produce viable
offspring (Schaack, Gilbert, & Feschotte, 2010).
We developed the method SigHunt that detects horizontally transferred genes in eukaryotic
genomes using composition of the DNA molecules (Jaron, Moravec, & Martínková,
2014). We built the method on the assumption that molecular mechanisms of DNA repli-
1. Introduction 27
cation and repair are heritable and in the long term lead to similarities in DNA composition
in related organisms (Blokzijl, Janssen, van Boxtel, & Cuppen, 2018). Related organisms
will have similar frequencies of nucleotide bases or short oligonucleotides in their genomes
(Karlin & Burge, 1995). We call a set of frequencies of oligonucleotides a genomic signature.
When a genomic region transfers from one organism to an unrelated organism, it will
initially retain the genomic signature of the original genome that could be distinguished
from the host genomic signature. Resolution of genomic signatures will be higher and they
will be more sensitive to differences between taxa for longer oligonucleotides, because
we will estimate oligonucleotide frequencies for more categories. The number of possible
oligonucleotides in a DNA sequence is 4n, where n is the oligonucleotide length.
We chose n = 4 for calculating genomic signatures in SigHunt and our vectors characterising
a genome contained frequencies of 256 tetranucleotides. Not all tetranucleotides are
informative for differentiating home genomic sequence from the horizontally transferred
genes. To find genomic islands in a eukaryotic genome, we selected those tetranucleotides
that had the smallest variance in the home genome and differed the most from genomes
of other organisms that could have been sources for the horizontal gene transfer. SigHunt
calculates genomic signatures for the selected tetranucleotides in sliding windows of the
DNA sequence. Based on a kernel density estimate of the frequencies in a chromosome,
SigHunt finds in which windows the tetranucleotide frequencies fall outside of the credibility
intervals (Fig. 1.9). Identification of the genomic island assumes that in horizontally
transferred regions, the genomic sequence will contain more tetranucleotides with extreme
frequencies (Jaron et al., 2014).
Research of hibernating bat behaviour tests the hypothesis that infection with fungus
P. destructans and development of WNS cause increased arousal frequency during the
hibernation period and increased flight activity. Moribund infected bats from the Nearctic
have been observed to dramatically change their behaviour during hibernation (Reeder et
al., 2012), but similar changes in activity were not observed in infected bats that survived
(Lilley et al., 2016). Since Palearctic bats with WNS survive better and we do not observe
an associated population decline in any species, their winter behaviour should mimic that
of Nearctic WNS survivors. We expect that the behaviour of most bats during hibernation
will show regular patterns of torpor-arousal as per Lilley et al. (2016). In rare cases, bats
with WNS die in the Palearctic (Pikula et al., 2017), and those should be detectable on
recordings of behaviour as increased arousal frequency as per Reeder et al. (2012).
We developed the method EEO (extraction of exceptional observations; Škrabánek &
Martínková, 2017) that identifies rare exceptional events based on their differentiation from
the common observations. The core principle of the EEO method is identification of a
point when we observe a sharp increase in sorted row sums of the distance matrix between
observations. We calculated the pairwise distances between observations as Mahalanobis
distances, because those are independent of the directionality of the change or units of the
observations. The distance matrix can be calculated from normalised data with Euclidean
distance for large datasets where computational expenses of Mahalanobis distances would
be prohibitive. The EEO method identifies outliers as those observations that have sum
of distances to all other observations higher than the threshold. Compared to other outlier
detection methods, the EEO is suitable for small datasets starting from 101 − 102 of
observations and it is capable to effectively seach for outliers in multidimensional space.
28 1. Introduction
0.000 0.004 0.008
0
50
100
150
200
250
300
350
AGCG, DIAS +3
0.0000.0020.004
0
100
200
300
400
500
GGGC, DIAS +0
0.000 0.002 0.004
0
100
200
300
400
500
600
GGCG, DIAS +3
0.000 0.006 0.012
0
50
100
150
200
AAGC, DIAS +3
0.000 0.004 0.008
0
50
100
150
200
250
300
GCGC, DIAS +0
0.0000.0040.0080.012
0
50
100
150
200
250
AGGC, DIAS +2
0.000 0.004
0
100
200
300
400
AGGG, DIAS +0
0.000 0.003 0.006
0
100
200
300
400
500
600
GAGC, DIAS +0
0.000 0.003 0.006
0
100
200
300
400
500
AGTG, DIAS +0
0.000 0.004 0.008
0
100
200
300
AGCA, DIAS +3
0.000 0.003 0.006
0
100
200
300
400
TGCT, DIAS +0
0.000 0.003 0.006
0
50
100
150
200
250
300
GGCC, DIAS +0
0.000 0.004 0.008
0
100
200
300
400
TGCC, DIAS +3
0.000 0.003 0.006
0
100
200
300
400
CATA, DIAS +0
0.000 0.004
0
100
200
300
400
CGGC, DIAS +0
0.000 0.004 0.008
0
50
100
150
200
250
AAGG, DIAS +3
Density
Tetranucleotide frequency
Figure 1.9: Searching for genomic islands in eukaryotic genomes using SigHunt.
SigHunt calculates frequencies of selected tetranucleotides in overlapping adjacent regions
of genomic sequence about 103 −104 bp long and estimates a kernel density for a longer
sliding window of the genomic sequence (about 105 −106 bp). The short window can be
considered a genomic island when its tetranucleotide freqencies (arrows) exceed marginal
values. In the worked example, the marginal values (dashed lines) are defined as kernel
density credibility intervals with probability α ∈ {0.05,0.025,0.01}. If the tetranucleotide
frequency is outside of the marginal values a discrete interval accumlative score (DIAS)
increases by 1, 2 and 3, respectively. The final value DIAS = 21 means that the genomic
signature of the tested DNA sequence considerably differs from its genomic context.
1. Introduction 29
In studying biodiversity, we are able to address exciting questions about the nature.
Phylogenetic and outlier analyses described in this thesis provide powerful tools to address
complex questions with genetic, ecological or behavioural data. We use phylogenetic
analysis to estimate the relationships between taxa and we are able to infer the age of their
divergence through the molecular dating analysis. Once we have a strong phylogenetic
hypothesis with good support, we are able to elucidate relationships between organisms
with a codivergence analysis and we can characterise functional relationships between traits
in light of the evolving systems. With outlier analyses, we can identify rare phenomena in
biological datasets and submit those for further testing.
Chapter 2
Author contributions
Research questions that investigate global biodiversity in light of ecological, behavioural
or epidemiological characteristics require an extensive access to data and expertise. Following
successful cooperation across regions and specialisations, research teams become
numerous in phylogenetic studies. The listed publications resulted predominantly from
collaborations with colleagues and students from the Czech and foreign universities and
research institutions. I derive estimation of individual contributions from the published
author contributions paragraphs. The author contributions list k categories, in which the
team members cooperated in the published research (e.g. conceptualisation, study design,
material collection, laboratory experiments, statistical analysis, writing, etc.). Where the
published papers did not contain the author contribution box, the number of categories and
individual contributions follow a documented input towards the research questions and the
publication. Each category i had equal weight for calculating the percent contribution of
individual authors, and in each category the contribution was equally divided between all
included authors Ri. Author contribution C was then the sum of all partial contributions in
all listed categories in the given publication for the given author:
C =
100
k
k
∑
i=1
φ(Ri)
|Ri|
, (2.1)
where Ri is a list of researchers contributing to the i-th category, |Ri| is the length of the
list of researchers, i.e. the number of authors contributing in the i-th category, and
φ(Ri) =
1, if ’NM’ ∈ Ri
0, otherwise
. (2.2)
For example, the paper 2.2.3 on phylogeography of Orkney voles (Martínková et al.,
2013) had 17 authors and the author box lists nine categories of contributions to the
research topic (Table 2.1). As each category has equal weight, it represents a fraction of
100
9 = 11.1%. During the course of the research, I collected material in the field together
with seven other authors, performed laboratory experiments with two others, analysed data
with two others and co-wrote the paper with all authors. My calculated contribution equals
to C = 11.1× 1
8 + 1
3 + 1
3 + 1
16 ≈ 9%.
The contribution discussions have an undeniable potential to sour professional relationships,
but they are required in this type of work. My intention was thus to reflect published
– 31 –
32 2. Author contributions
Table 2.1: Example of calculating my author contribution in paper 2.2.3 (Martínková et
al. 2013) using equation 2.1 for k = 9 categories. Ri - list of authors contributing to the
given category, |Ri| - number of authors contributing to the category, φ(Ri) - a function
evaluating whether I contributed to the category (equation 2.2).
Category i Contributing authors Ri |Ri| φRi
Project design KMD, JBS 2 0
Field material collection TC, MF, GH, NM, MP, MaP, JPQ, JBS 8 1
Museum material collection RB, TC, KMD 3 0
Laboratory experiments RB, TC, NM 3 1
Lab support KMD, ARH, GH, JBS, SB, TH 6 0
Data analyses RB, TC, NM 3 1
Analyses support RS, KMD, LE, POH, ARH, GH, SYH, JBS 7 0
Writing JBS 1 0
Writing support NM + 15 authors 16 1
or documented facts as objectively as possible. In the following section, I declare the
putative shift in interpreting the percentage of my contributions, where the strict adherence
to the equation 2.1 unjustly affected my students or colleagues.
2.1 Phylogenetic analyses papers
2.1.1 Jaarola M., Martínková N., Gündüz ˙I., Brunhoff C., Zima J., Nadachowski A.,
Amori G., Bulatova N. S., Chondropoulos B., Fraguedakis-Tsolis S., GonzálezEsteban
J., López-Fuster M. L., Kandaurov A. S., Kefelio˘glu H., da Luz Mathias
M., Villate I., Searle J. B. 2004. Molecular phylogeny of the speciose vole genus
Microtus (Arvicolinae, Rodentia) inferred from mitochondrial DNA sequences.
Molecular Phylogenetics and Evolution 33: 647-663.
Author contribution: 16%, I contributed to material collection, laboratory experiments,
data analysis and manuscript preparation.
Main innovation of the paper: The article reports phylogeny reconstruction of voles
from the genus Microtus based on a mitochondrial gene for cytochrom b. The analyses
compare results from neighbour-joining, maximum parsimony, ML and BI trees. Phylogeny
based on a single gene was unable to resolve deep divergencies, but it showed good
support for Microtus and Chionomys and subgenera Microtus and Terricola.
2.1.2 Martínková N., Moravec J. 2012. Multi-locus phylogeny of arvicoline voles
(Arvicolini, Rodentia) shows small tree terrace size. Folia Zoologica 61: 254-267.
Author contribution: 67%, I designed the study from the publically available DNA sequence
data and I contributed to the statistical analyses and writing.
Main innovation of the paper: A study inspired by the paper 2.1.1 (Jaarola et al., 2004)
that attempts to solve analytical problems in the older study. The paper resolves phylogenetic
relationships between voles of the tribe Arvicolini that were previously problematic.
2. Author contributions 33
We achieved the improvement by using both mitochondrial and nuclear sequences and we
tested the impact of missing data on results reliability and stability. Further, we tested
the impact of branch length priors on BI posterior node support with the alignment with
missing data. We found that the vole phylogeny from an alignment that contained over 70%
of missing data belongs to a small terrace; that is there exists a small number of alternative
trees that have the same likelihood as the consensus tree from the posterior sample of
the BI. In other statistical analyses, missing data need to be either imputed or samples or
variables containing missing data are omitted from the analyses, and thus practice calls for
less than 5% of missing data. Relative robustness of phylogenetic analyses from matrices
with missing data shows the remarkable information value of historical processes during
DNA sequence evolution.
2.1.3 Martínková N., Zima J., Jaarola M., Macholán M., Spitzenberger F. 2007. The
origin and phylogenetic relationships of Microtus bavaricus based on karyotype
and mitochondrial DNA sequences. Folia Zoologica 56: 39-49.
Author contribution: 37%, I co-designed the study together with J. Zima and I contributed
to the laboratory experiments, statistical analyses and writing.
Main innovation of the paper: Microtus bavaricus belongs among voles with small
distribution ranges that inhabit montane and submontane regions in Europe. We succeeded
in obtaining a karyotype as well as DNA sequences of the species. We used neighbourjoining,
maximum parsimony, ML and BI phylogenetic analyses to show that M. bavaricus
is a young, recently diverged species from the Microtus multiplex species group.
2.1.4 Pečnerová P., Martínková N. 2012. Evolutionary history of tree squirrels (Rodentia,
Sciurini) based on multilocus phylogeny reconstruction. Zoologica Scripta 41:
211-219.
Author contribution: 50%, although I contributed to all categories that constitute this work,
my real share was lower. The paper is in essence the Bachelor thesis of P. Pečnerová. I
designed the topic and I supervised the student throughout her work in data collection,
analyses and writing.
Main innovation of the paper: Tree squirrels of the tribe Sciurini are distributed in the
Holarctic and Neotropic biogeographic ecozones. Most species of tree squirrels and three
out of five genera belonging to the tribe can be found in the Neotropic. Using publically
available sequences of eight genes, we found well supported phylogenetic divergence
between genera Sciurus and Tamiasciurus based on the BI phylogeny. Other genera from
the tribe Sciurini were paraphyletic with respect to Sciurus and the relationships between
species were unresolved. Alternative approaches of phylogenetic reconstruction from
multilocus data with supermatrices or supertrees showed similar results. We concluded
that the unresolved relationships were a result of explosive radiation and incomplete lineage
sorting of tree squirrels in the Neotropic.
34 2. Author contributions
2.1.5 Kandemir ˙I., Sözen M., Matur F., Kankilic¸ T., Martínková N., C¸ olak F., Özkurt S.
Ö., C¸ olak E. 2012. Phylogeny of species and cytotypes of mole rats (Spalacidae) in
Turkey inferred from mitochondrial cytochrome b gene sequences. Folia Zoologica
61: 25-33.
Author contribution: 22%, I worked on the statistical analyses and manuscript preparation
and helped with the laboratory experiments and the study design.
Main innovation of the paper: Mole rats of the genus Nannospalax belong to a group of
underground rodents needing a taxonomic revision. We studied the relationships between
mole rat populations in Turkey using neighbour-joining, maximum parsimony, ML and BI
phylogenetic methods in conjunction with information on their karyotypes. Further, we altered
the prior for branch lengths in the BI analysis to ascertain that the posterior credibility
interval of the tree length included its ML estimate. We found that Nannospalax populations
with low number of chromosomes (2n ∈ {36,38,40}) usually form monophyletic
groups. Populations with high numbers of chromosomes (2n ≥ 52) were paraphyletic on a
phylogeny contructed from partial sequences of a single mitochondrial locus.
2.1.6 Vallo P., Benda P., Martínková N., Kaňuch P., Kalko E. K. V., Červený J., Koubek
P. 2011. Morphologically uniform bats Hipposideros aff. ruber (Hipposideridae)
exhibit high mitochondrial genetic diversity in southeastern Senegal. Acta
Chiropterologica 13: 79-88.
Author contribution: 14%, I contributed towards study design, statistical analyses and
writing.
Main innovation of the paper: Morphological and ecological differentiation sometimes
does not coincide with phylogenetic divergence. We identified genetically distinct groups in
Afrotropic bats from the Hipposideros ruber species complex (Chiroptera, Hipposideridae)
in the phylogenetic reconstructions with maximum parsimony, ML and BI. Using sample
classification based on the phylogeny, we were able to morphologically differentiate groups
that may belong to separate taxa. We then tested alternative phylogenetic hypotheses, and
the tests revealed that the group differentiation is not robust in the available data.
2.1.7 Kopecna O., Kubickova S., Cernohorska H., Cabelova K., Váhala J., Martínková
N., Rubes J. 2014. Tribe-specific satellite DNA in non-domestic Bovidae. Chromosome
Research 22: 277-291.
Author contribution: 12%, I reconstructed the phylogenetic relationships and wrote the
pertaining sections in the manuscript.
Main innovation of the paper: Satellite DNA in the centromers consists of repetitive
sequences, where individual repetitions are relatively long and may contain signal of past
speciation processes. We found high diversity in satellite DNA sequences in representatives
of Bovidae (Artiodactyla). The resulting alignment incorporated many indels. We mined
the phylogenetic information in indels by recoding the indels as a binary morphological
character in a partitioned phylogenetic analyses with ML and BI. The information in indels
revealed potential problem with homology of the satellite sequences in some taxa, but
including indels in the phylogenetic reconstruction helped to resolve relationships between
2. Author contributions 35
satellite DNA sequences of the subfamily Antilopinae.
2.1.8 Wallace I. S., Shakesby A. J., Hwang J. H., Choi W. G., Martínková N., Douglas
A. E., Roberts D. M. 2012. Acyrthosiphon pisum AQP2: A multifunctional insect
aquaglyceroporin. BBA Biomembranes 1818: 627-635.
Author contribution: 11%, I reconstructed phylogenetic relationships and wrote the pertaining
sections in the manuscript.
Main innovation of the paper: With help from phylogenetic analysis, we were able
to identify the newly discovered gene ApAQP2 as an aquaporin-encoding gene in insects.
We approached the task by designing the dataset representative of the diverse family of
aquaporins. We included aminoacid sequences of known aquaporins in arthropods and
chordates, and we analysed the data in a mixed model with indels coded as a binary
partition. The BI analysis may become trapped in the tree space with short branches
when analysing divergent sequences. The issue affected our analyses and the aquaporin
phylogeny was too short. We corrected the branch lengths by estimating them with ML on
the tree topology fixed to the consensus from the BI posterior trees.
2.2 Molecular dating papers
2.2.1 Martínková N., McDonald R. A., Searle J. B. 2007. Stoats (Mustela erminea)
provide evidence of natural overland colonisation of Ireland. Proceedings of the
Royal Society B-Biological Series 274: 1387-1393.
Author contribution: 63%, my contribution included material collections from European
museums and private collections, all laboratory experiments, data analyses and their interpretation,
manuscript preparation.
Main innovation of the paper: Unless molecular dating of recent divergencies addresses
mutation rate decay, the divergence times of events correlated with Pleistocene glaciations
could be overestimated by orders of magnitude. To avoid the pitfalls, we estimated the
mutation rate specifically for the model organism and expected time frame. We used
time constraint priors for root height of one stoat population and we set them based on
known paleontological data and geological events. Using Bayesian coalescence analysis,
we estimated mutation rate for the given population and we applied the rate to date stoat
populations in Europe. We inferred information on age and population size changes of
island and continental populations of stoats.
2.2.2 Seifertová M., Bryja J., Vyskočilová M., Martínková N., Šimková A. 2012. Multiple
Pleistocene refugia and post-glacial colonization in the European chub (Squalius
cephalus) revealed by combined use of nuclear and mitochondrial markers. Journal
of Biogeography 39: 1024-1040.
Author contribution: 12%, I joined the project in the phase of manuscript review, when a
reviewer requested additional analyses. I performed the molecular dating analysis of the
available data and I wrote the respective sections of the manuscript.
36 2. Author contributions
Main innovation of the paper: Similarly as in stoats in paper 2.2.1 (Martínková et al.,
2007), we expected the Pleistocene glaciations to influence distribution range changes and
thus genetic diversity in fish. We used paleontological and biogeographic information to
constrain the heights of the tree nodes. We constrained the tree root with the first known
fossil of the European chub from Germany and a fossil of its ancestor from Greece. Other
constrained nodes were ancestral to star-like phylogenies that characterise growing populations.
We expected the sudden population growth coincided with climate amelioration
after glaciations. The time constraints at different depth of the tree enabled us to estimate
the divergence times for the European chub lineages.
2.2.3 Martínková N., Barnett R., Cucchi T., Struchen R., Pascal M., Pascal M., Fischer
M. C., Higham T., Brace S., Ho S. Y. W., Quéré J.-P., O’Higgins P., Excoffier L.,
Heckel G., Hoelzel A. R., Dobney K. M., Searle, J. B. 2013. Divergent evolutionary
processes associated with colonization of offshore islands. Molecular Ecology 22:
5205-5220.
Author contribution: 9%, I collected all material from the Orkney Islands and participated
in material collection in France, I performed laboratory experiments on the recent material,
analysed and interpreted the data and co-wrote the paper.
Main innovation of the paper: Molecular dating that uses paleontological and geological
time constraints assumes that the chosen events influenced directly and strongly the
population of the model organism. We chose a direct, tip-dated molecular clock calibration
reflecting the observed genetic changes in time to date the origin of the Orkney voles and
the time of colonisation of the archipelago. We used fossil voles that were radiocarbondated
and sequenced to calibrate the mutation rate with the Bayesian coalescence analysis.
The precise and accurate mutation rate allowed us to test colonisation scenarios of islands
in the North Sea and we interpreted the results with respect to the archeological findings
from the Neolithic.
2.3 Codivergence papers
2.3.1 Irwin N. R., Bayerlová M., Missa O., Martínková N. 2012. Complex patterns of
host switching in New World arenaviruses. Molecular Ecology 21: 4137-4150.
Author contribution: 36%, I participated in study design, supervised the student M.
Bayerlová, contributed towards data analyses and manuscript preparation.
Main innovation of the paper: Using publically available data, we addressed a long
standing assumptions that arenaviruses coevolved with their hosts. We reconstructed
phylogenetic relationships between New World arenaviruses and all hosts and related taxa
with available DNA sequences using ML and BI phylogenetic analyses. We computed
contribution of individual associations to the overall codivergence between the parasite
and host phylogenies using the ParaFit analysis and we tested clustering of arenavirus hosts
on the rodent phylogeny with indices used in community ecology. We found that host
species of some groups of arenaviruses are randomly distributed across rodent diversity in
the Americas. Together with geographical modelling, our results suggest that medically-
2. Author contributions 37
important strains utilise host-switching between geographically, not phylogenetically close
hosts.
2.3.2 Pečnerová P., Moravec J. C., Martínková N. 2015. A skull might lie: Modeling
ancestral ranges and diet from genes and shape of tree squirrels. Systematic Biology
64: 1074-1088.
Author contribution: 23%, I designed the study, supervised the student P. Pečnerová,
contributed to analyses and manuscript preparation.
Main innovation of the paper: We expanded the research questions about speciation
explosion of tree squirrels in the Nearctic established in paper 2.1.4 (Pečnerová &
Martínková, 2012). Here, we applied an innovative approach of contrasting codivergence
of a molecular genetic phylogeny of tree squirrels with a tree characterising morphological
similarity between taxa and compared that to a tree of food preferrences of the tree squirrels.
We then modelled geographic spread of speciating lineages in the Nearctic. We found
striking dissimilarities between the trees, where the current taxonomy corresponds to the
morphological, not the phylogenetic relationships. Convergent evolution of skull shape in
tree squirrels is likely a result of specialisation and diversification in utilising specific food
sources, when the ancestors of contemporary tree squirrels crossed to South America after
formation of the Isthmus of Panama.
2.4 Comparative phylogenetics papers
2.4.1 Zukal J., Bandouchova H., Bartonicka T., Berkova H., Brack V., Brichta J., Dolinay
M., Jaron K. S., Kovacova V., Kovarik M., Martínková N., Ondracek K., Rehak Z.,
Turner G. G., Pikula J. 2014. White-nose syndrome fungus: a generalist pathogen
of hibernating bats. PLoS ONE 9: e97224.
Author contribution: 13%, I participated in study design, field research, data analyses and
writing, and I supervised the student M. Dolinay.
Main innovation of the paper: White-nose syndrome leads to increased mortality in
hibernating bats in the Nearctic, but Palearctic bats survive the infection better. We reconstructed
a multilocus phylogeny of vespertilionid and minioperid bats in the ML framework
from an alignment with 64% of missing data. We used comparative phylogenetics to study
the interaction between the parasite and its hosts. We modelled clustering of the infected
species on the bat phylogeny. We found that the probability that sister species will be
infected decreases with improved sampling and that the pathogenic fungus is not speciesspecific.
Ecological similarity of infected species varies as modelled with NRI and NTI
indices on a neighbour joining tree based on the matrix characterising ecological and
behavioural traits of hibernating bats. Using our models, we predicted P. destructans
infection in five bat species, previously expected to be uninfected and even not susceptible.
Research in subsequent years confirmed most of our predictions both in Palearctic (Zukal
et al., 2016) and in Nearctic bat species (Bernard et al., 2015).
38 2. Author contributions
2.4.2 Zukal J., Bandouchova H., Brichta J., Cmokova A., Jaron K. S., Kolarik M.,
Kovacova V., Kubátová A., Nováková A., Orlov O., Pikula J., Presetnik P., Šuba J.,
Zahradníková A. Jr., Martínková N. 2016. White-nose syndrome without borders:
Pseudogymnoascus destructans infection confirmed in Asia. Scientific Reports 6:
19829.
Author contribution: 17%, I co-designed the study and participated in material collection,
data analyses and writing.
Main innovation of the paper: We found high prevalence of the P. destructans infection
and the WNS in Palearctic hibernating bats. Fungal load at the wing surface of infected
animals is comparable to that in the Nearctic animals that die of the disease. Number of
UV fluorescent lesions diagnostic of WNS increases with increasing fungal load on the
wing surface with and without the correction for phylogenetic non-independence of the
data. Fungal load shows significant phylogenetic signal, but variance in number of UV
fluorescent lesions is not significantly explained by the phylogeny. Our results indicate that
in some (unrelated) bat species intensive fungal infection might not progress to development
of numerous lesions and thus to severe disease. In conjunction with histopathological data,
this means that the Palearctic bats are able to tolerate P. destructans infection.
2.4.3 Pikula J., Amelon S. K., Bandouchova H., Bartonička T., Berkova H., Brichta J.,
Hooper S., Kokurewicz T., Kolarik M., Köllner B., Kovacova V., Linhart P., Piacek
V., Turner G. G., Zukal J., Martínková N. 2017. White-nose syndrome pathology
grading in Nearctic and Palearctic bats. PLoS ONE 12: e0180435.
Author contribution: 33%, I contributed to all aspects of the study with the exception
of design of the method scoring disease severity of WNS from histopathology of the
wing punch biopsy. I worked on material collection, evaluation of the histopathologic
slides, preparation of materials for na¨ıve personnel, I analysed the data and co-wrote the
manuscript.
Main innovation of the paper: The only method that unambiguously diagnoses WNS
is a histopathologic evaluation of skin infected with P. destructans. We used UV light
transillumination to detect the skin lesions in the field and to navigate the biopsy sampling
of bat wings (Turner et al., 2014). We developed a methodology, how to use the biopsy
punch to assess the WNS severity. From the analytical perspective, we validated the
new method for use by experienced pathologists as well as by na¨ıve evaluators who passed
basic training. We analysed the relationships between the non-destructive methods of WNS
research and the histopathological findings and we evaluated the relationship between the
newly developed histopathology score to the quantitative measures of infection intensity
in the phylogenetic context.
2. Author contributions 39
2.5 Outlier analyses papers
2.5.1 Jaron K. S., Moravec J. C., Martínková N. 2014. SigHunt: Horizontal gene
transfer finder optimized for eukaryotic genomes. Bioinformatics 30: 1081-1086.
Author contribution: 17%, students K. S. Jaron and J. C. Moravec drove the development
of the method detecting genomic islands in eukaryotic genomes and its software implementation.
I tested the algorithms on model organisms, validated the results and wrote the
manuscript.
Main innovation of the paper: DNA sequence variability along the chromosome complicates
detection of horizontally transferred genes in eukaryotic genomes. We designed
SigHunt, a method that uses local genomic signatures from informative tetranucleotides,
i.e DNA composition from oligonucleotides that showed small variance in the host genome
and were different from genomes of putative source organisms. We classify a genomic
island as the DNA sequence region where multiple frequencies of tetranucleotides are
outliers with respect to the respective frequencies in the surrounding genomic context. We
quantify the outlier score as a discrete interval accumulative score from extreme values of
the cumulative distribution function of the tetranucleotide frequency. That is a recently
acquired genomic island will have markedly different DNA composition.
2.5.2 Škrabánek P., Martínková N. 2017. Extraction of outliers from imbalanced sets.
In: Martínez de Pisón F., Urraca R., Quintián H., Corchado E. (Eds.) Hybrid
Artificial Intelligent Systems. HAIS 2017. Lecture Notes in Computer Science, vol.
10334. pp. 402-412. Springer, Cham, DOI: 10.1007/978-3-319-59650-1 34.
Author contribution: 50%, the method is based on research by P. Škrabánek, with my
minor contribution to all aspects of the study, including method development, analyses and
writing.
Main innovation of the paper: We can identify exceptional observations using statistical
methods if we have sufficiently large dataset, the exceptional observations are sufficiently
similar to one another and we measured variables that characterise the exceptional observations
sufficiently well to differentiate them from the normal observations. In biology,
the requirements for statistical power to distinguish the exceptional observations pose a
requirement that we know the exceptional observation exists and we know how to search
for it. In other words, we test a hypothesis. The hypothesis-based research risks that
we will omit processes fundamentally different from the focus of the study that have the
potential to revolutionize the field. When we expect unknown processes to govern a biological
system, we can apply outlier analysis to detect the observations generated by a rare
mechanism. We designed a method, extraction of exceptional observations (EEO), that
can detect individual samples in small, imbalanced sets that markedly differ from other
samples. We use a threshold value where pairwise distance row sums rapidly increase
as an indicator for outliers in a multidimensional variable space. The EEO method can
single out observations that merit detailed study and evaluation of potential differences in
mechanisms that generated the outlier observations.
References
[1] Aghova, T., Kimura, Y., Bryja, J., Dobigny, G., Granjon, L., & Kergoat, G. J. (2018).
Fossils know it best: using a new set of fossil calibrations to improve the temporal
phylogenetic framework of murid rodents (Rodentia: Myomorpha: Muroidea:
Muridae). Mol Phylogenet Evol, 128, 98-111. doi: 10.1016/j.ympev.2018.07.017
[2] Anisimova, M., & Gascuel, O. (2006). Approximate likelihood-ratio test for branches:
A fast, accurate, and powerful alternative. Syst Biol, 55, 539-552.
[3] Barrick, Jeffrey E., & Lenski, Richard E. (2013). Genome dynamics during experimental
evolution. Nat Rev Genet, 14(12), 827-839. doi: 10.1038/nrg3564
[4] Biczok, R., Bozsoky, P., Eisenmann, P., Ernst, J., Ribizel, T., Scholz, F., . . . Stamatakis,
A. (2018). Two C++ Libraries for Counting Trees on a Phylogenetic Terrace.
Bioinformatics. doi: 10.1093/bioinformatics/bty384
[5] Bernard, R. F., Foster, J. T., Willcox, E. V., Parise, K. L., McCracken, G. F. (2015).
Molecular detection of the causative agent of white-nose syndrome on Rafinesque’s
big-eared bats (Corynorhinus rafinesquii) and two species of migratory bats in the
southeastern USA. J Wildl Dis, 51, 519-22. doi: 10.7589/2014-08-202
[6] Blokzijl, F., Janssen, R., van Boxtel, R., & Cuppen, E. (2018). MutationalPatterns:
comprehensive genome-wide analysis of mutational processes. Genome Med, 10, 33.
doi: 10.1186/s13073-018-0539-0
[7] Blomberg, S. P., Garland jr., T., & Ives, A.R. (2003). Testing for phylogenetic signal
in comparative data: behavioral traits are more labile. Evolution, 57, 717-745.
[8] Bogusz, M., & Whelan, S. (2017). Phylogenetic tree estimation with and without
alignment: New distance methods and benchmarking. Syst Biol, 66(2), 218-231. doi:
10.1093/sysbio/syw074
[9] Bowen, M. D., Peters, C. J., & Nichol, S. T. (1997). Phylogenetic analysis of
the Arenaviridae: patterns of virus evolution and evidence for cospeciation between
arenaviruses and their rodent hosts. Mol Phylogenet Evol, 8, 301-316. doi:
10.1006/mpev.1997.0436
[10] Butler, M. A., & King, A. A. (2004). Phylogenetic comparative analysis: A modeling
approach for adaptive evolution. The American Naturalist, 164, 683-695. doi:
10.1086/426002
– 41 –
42 References
[11] Campbell, V., Legendre, P., & Lapointe, F.-J. (2011). The performance of the congruence
among distance matrices (CADM) test in phylogenetic analyses. BMC Evol
Biol, 11, 64. doi: 10.1186/1471-2148-11-64
[12] Chan, C. X., Bernard, G., Poirion, O., Hogan, J. M., & Ragan, M. A. (2014). Inferring
phylogenies of evolving sequences without multiple sequence alignment. Scientific
Reports, 4, 6504. doi: 10.1038/srep06504
[13] Charrel, R. N., & de Lamballerie, X. (2003). Arenaviruses other than Lassa virus.
Antiviral Research, 57(1-2), 89-100. doi: 10.1016/s0166-3542(02)00202-4
[14] Chernomor, O., von Haeseler, A., & Minh, B. Q. (2016). Terrace aware data structure
for phylogenomic inference from supermatrices. Syst Biol, 65(6), 997-1008. doi:
10.1093/sysbio/syw037
[15] Dobrin, B. H., Zwickl, D. J., & Sanderson, M. J. (2018). The prevalence of terraced
treescapes in analyses of phylogenetic data sets. BMC Evol Biol, 18(1), 46. doi:
10.1186/s12862-018-1162-9
[16] dos Reis, M., Donoghue, P. C., & Yang, Z. (2016). Bayesian molecular clock dating
of species divergences in the genomics era. Nat Rev Genet, 17(2), 71-80. doi:
10.1038/nrg.2015.8
[17] Drees, K. P., Lorch, J. M., Puechmaille, S. J., Parise, K. L., Wibbelt, G., Hoyt, J. R., . . .
Foster, J. T. (2017). Phylogenetics of a fungal invasion: origins and widespread dispersal
of white-nose syndrome. mBio, 8, e01941-01917. doi: 10.1128/mBio.01941-17
[18] Dwivedi, B., & Gadagkar, S. R. (2009). Phylogenetic inference under varying proportions
of indel-induced alignment gaps. BMC Evol Biol, 9, 211. doi: 10.1186/1471-
2148-9-211
[19] Ekman, S., & Blaalid, R. (2011). The devil in the details: interactions between the
branch-length prior and likelihood model affect node support and branch lengths in
the phylogeny of the Psoraceae. Syst Biol, 60, 541-561. doi: 10.1093/sysbio/syr022
[20] Fink, S., Excoffier, L., & Heckel, G. (2007). High variability and non-neutral evolution
of the mammalian avpr1a gene. BMC Evol Biol, 7, 176. doi: 10.1186/1471-2148-7-
176
[21] Fountain-Jones, N. M., Pearse, W. D., Escobar, L. E., Alba-Casals, A., Carver, S.,
Davies, T. J., . . . Craft, M. E. (2017). Towards an eco-phylogenetic framework for
infectious disease ecology. Biol Rev Camb Philos Soc. doi: 10.1111/brv.12380
[22] Gabriel, S. I., Jóhannesdóttir, F., Jones, E. P., & Searle, J. B. (2010). Colonization,
mouse-style. BMC Biology, 8, 131. doi: 10.1186/1741-7007-8-131
[23] Hillis, D. M., Heath, T. A., & St John, K. (2005). Analysis and visualization of tree
space. Syst Biol, 54(3), 471-482. doi: 10.1080/10635150590946961
References 43
[24] Ho, S. Y., Phillips, M. J., Cooper, A., & Drummond, A. J. (2005). Time dependency
of molecular rate estimates and systematic overestimation of recent divergence times.
Mol Biol Evol, 22(7), 1561-1568. doi: 10.1093/molbev/msi145
[25] Ho, L. s. T., & Ané, C. (2014). A linear-time algorithm for Gaussian and non-Gaussian
trait evolution models. Syst Biol, 63, 397-408. doi: 10.1093/sysbio/syu005
[26] Horáček, I., Bartonička, T., Lučan, R. K., & Czech Bat Conservation Trust. (2014).
Macroecological characteristics of bat geomycosis in the Czech Republic: Results of
five years of monitoring. Vespertilio, 17, 65-77.
[27] Irwin, N. R., Bayerlová, M., Missa, O., & Martínková, N. (2012). Complex patterns
of host switching in New World arenaviruses. Mol Ecol, 21(16), 4137-4150. doi:
10.1111/j.1365-294X.2012.05663.x
[28] Ives, A.R., Midford, P. E., & Garland jr., T. (2007). Within-species measurement
error in phylogenetic comparative methods. Syst. Biol., 56, 252-270.
[29] Jackson, A. P., & Charleston, M. A. (2004). A cophylogenetic perspective of RNAvirus
evolution. Mol Biol Evol, 21(1), 45-57. doi: 10.1093/molbev/msg232
[30] Jaron, K. S., Moravec, J. C., & Martínková, N. (2014). SigHunt: Horizontal gene
transfer finder optimized for eukaryotic genomes. Bioinformatics, 30, 1081-1086.
[31] Karlin, S., & Burge, C. (1995). Dinucleotide relative abundance extremes: a genomic
signature. Trends Genet, 11, 283-290.
[32] Kolaczkowski, B., & Thornton, J. W. (2009). Long-branch attraction bias and inconsistency
in Bayesian phylogenetics. PLoS ONE, 4, e7891. doi: 10.1371/
[33] Kopecna, O., Kubickova, S., Cernohorska, H., Cabelova, K., Vahala, J., Martinkova,
N., & Rubes, J. (2014). Tribe-specific satellite DNA in non-domestic Bovidae. Chromosome
Res, 22(3), 277-291. doi: 10.1007/s10577-014-9401-4
[34] Krasnovyd, V, Vetešník, L, Gettová, L, Civáňová, K, & Šimková, A. (2017). Patterns
of parasite distribution in the hybrids of non-congeneric cyprinid fish species: Is
asymmetry in parasite infection the result of limited coadaptation? Int. J. Parasitol.,
47, 471-483.
[35] Kumar, S. (2005). Molecular clocks: four decades of evolution. Nat Rev Genet, 6(8),
654-662.
[36] Legendre, P., Desdevises, Y., & Bazin, E. (2002). A statistical test for host-parasite
coevolution. Syst Biol, 51, 217-234.
[37] Lilley, T. M., Johnson, J. S., Ruokolainen, L., Rogers, E. J., Wilson, C. A., Schell, S.
M., . . . Reeder, D. M. (2016). White-nose syndrome survivors do not exhibit frequent
arousals associated with Pseudogymnoascus destructans infection. Front Zool, 13,
12. doi: 10.1186/s12983-016-0143-3
44 References
[38] Lorch, J. M., Meteyer, C. U., Behr, M. J., Boyles, J. G., Cryan, P. M., & Hicks, A. C.
(2011). Experimental infection of bats with Geomyces destructans causes white-nose
syndrome. Nature, 480, 376-378.
[39] Losos, J. B. (2008). Phylogenetic niche conservatism, phylogenetic signal and the relationship
between phylogenetic relatedness and ecological similarity among species.
Ecol Lett, 11, 995-1003. doi: 10.1111/j.1461-0248.2008.01229.x
[40] Marshall, D. C. (2010). Cryptic failure of partitioned Bayesian phylogenetic analyses:
lost in the land of long trees. Syst Biol, 59(1), 108-117. doi: 10.1093/sysbio/syp080
[41] Marston, D. A., Banyard, A. C., McElhinney, L. M., Freuling, C. M., Finke,
S., de Lamballerie, X., . . . Fooks, A. R. (2018). The lyssavirus host-specificity
conundrum-rabies virus-the exception not the rule. Curr Opin Virol, 28, 68-73. doi:
10.1016/j.coviro.2017.11.007
[42] Martinez-Aquino, A. (2016). Phylogenetic framework for coevolutionary studies:
a compass for exploring jungles of tangled trees. Curr Zool, 62, 393-403. doi:
10.1093/cz/zow018
[43] Martínková, N., Barnett, R., Cucchi, T., Struchen, R., Pascal, M., Pascal, M., . . .
Searle, J. B. (2013). Divergent evolutionary processes associated with colonization
of offshore islands. Mol Ecol, 22(20), 5205-5220. doi: 10.1111/mec.12462
[44] Martínková, N., McDonald, R. A., & Searle, J. B. (2007). Stoats (Mustela erminea)
provide evidence of natural overland colonization of Ireland. Proc Biol Sci, 274(1616),
1387-1393. doi: 10.1098/rspb.2007.0334
[45] Martínková, N., & Moravec, J. C. (2012). Multilocus phylogeny of arvicoline voles
(Arvicolini, Rodentia) shows small tree terrace size. Folia Zool., 61, 254-267.
[46] McGuire, G., Denham, M. C., & Balding, D. J. (2001). Models of sequence evolution
for DNA sequences containing gaps. Mol Biol Evol, 18, 481-490. doi: 10.1093/ox-
fordjournals.molbev.a003827
[47] Murray, G. G., Wang, F., Harrison, E. M., Paterson, G. K., Mather, A. E., Harris, S.
R., . . . Welch, J. J. (2016). The effect of genetic structure on molecular dating and tests
for temporal signal. Methods Ecol Evol, 7(1), 80-89. doi: 10.1111/2041-210X.12466
[48] Nascimento, F. F., dos Reis, M., & Yang, Z. (2017). A biologist’s guide to Bayesian
phylogenetic analysis. Nat Ecol Evol, 1, 1446-1454. doi: 10.1038/s41559-017-0280-x
[49] Near, T. J., Bolnick, D. I., & Wainwright, P. C. (2005). Fossil calibrations and
molecular divergence time estimates in centrarchid fishes (Teleostei: Centrarchidae).
Evolution, 59, 1768-1782. doi: 10.1554/05-030.1.s1
[50] Near, T. J., & Sanderson, M. J. (2004). Assessing the quality of molecular divergence
time estimates by fossil calibrations and fossil-based model selection. Philos Trans R
Soc Lond B Biol Sci, 359, 1477-1483. doi: 10.1098/rstb.2004.1523
References 45
[51] Nelson, B. J., Andersen, J. J., & Brown, J. M. (2015). Deflating trees: improving
Bayesian branch-length estimates using informed priors. Syst Biol, 64, 441-447. doi:
10.1093/sysbio/syv003
[52] Palmer, J. M., Drees, K. P., Foster, J. T., & Lindner, D. L. (2018). Extreme sensitivity
to ultraviolet light in the fungal pathogen causing white-nose syndrome of bats. Nat
Commun, 9, 35. doi: 10.1038/s41467-017-02441-z
[53] Pečnerová, P., & Martínková, N. (2012). Evolutionary history of tree squirrels (Rodentia,
Sciurini) based on multilocus phylogeny reconstruction. Zoologica Scripta,
41, 211-219. doi: 10.1111/j.1463-6409.2011.00528.x
[54] Pečnerová, P., Moravec, J. C., & Martínková, N. (2015). A skull might lie: Modeling
ancestral ranges and diet from genes and shape of tree squirrels. Syst Biol, 64, 1074-
1088. doi: 10.1093/sysbio/syv054
[55] Philippe, H., Brinkmann, H., Lavrov, D. V., Littlewood, D. T. J., Manuel, M.,
Wörheide, G., & Baurain, D. (2011). Resolving difficult phylogenetic questions:
Why more sequences are not enough. PLoS Biol., 9, e1000602. doi: 10.1371/ jour-
nal.pbio.1000602
[56] Pikula, J., Amelon, S. K., Bandouchova, H., Bartonicka, T., Berkova, H., Brichta, J.,
. . . Martínková, N. (2017). White-nose syndrome pathology grading in Nearctic and
Palearctic bats. PLoS ONE, 12, e0180435. doi: 10.1371/journal.pone.0180435
[57] Pyron, R. A. (2015). Post-molecular systematics and the future of phylogenetics.
Trends Ecol Evol, 30, 384-389. doi: 10.1016/j.tree.2015.04.016
[58] Rannala, B., Zhu, T., & Yang, Z. (2012). Tail paradox, partial identifiability, and
influential priors in Bayesian branch length inference. Mol Biol Evol, 29, 325-335.
doi: 10.1093/molbev/msr210
[59] Reeder, D. M., Frank, C. L., Turner, G. G., Meteyer, C.U., Kurta, A., Britzke, E.
R., . . . Blehert, D.S. (2012). Frequent arousal from hibernation linked to severity
of infection and mortality in bats with white-nose syndrome. PLoS ONE, 7, e38920.
doi: 10.1371/journal.pone.0038920.g001
[60] Revell, L. J. (2010). Phylogenetic signal and linear regression on species data. Methods
Ecol Evol, 1, 319-329. doi: 10.1111/j.2041-210X.2010.00044.x
[61] Revell, L. J., Harmon, L. J., & Collar, D. C. (2008). Phylogenetic signal, evolutionary
process, and rate. Syst Biol, 57, 591-601. doi: 10.1080/10635150802302427
[62] Robinson, D. F., & Foulds, L. R. (1981). Comparison of phylogenetic trees. Mathematical
Biosciences, 53(1), 131-147. doi: 10.1016/0025-5564(81)90043-2
[63] Ronquist, F., Teslenko, M., van der Mark, P., Ayres, D. L., Darling, A., Hohna,
S., . . . Huelsenbeck, J. P. (2012). MrBayes 3.2: Efficient Bayesian phylogenetic
inference and model choice across a large model space. Syst Biol, 61, 539-542. doi:
10.1093/sysbio/sys029
46 References
[64] Sanderson, M. J., McMahon, M. M., Stamatakis, A., Zwickl, D. J., & Steel, M.
(2015). Impacts of terraces on phylogenetic inference. Syst Biol, 64, 709-726. doi:
10.1093/sysbio/syv024
[65] Sanderson, M. J., McMahon, M. M., & Steel, M. (2011). Terraces in phylogenetic
tree space. Science, 333, 448-450. doi: 10.1126/science.1206357
[66] Schaack, S., Gilbert, C., & Feschotte, C. (2010). Promiscuous DNA: Horizontal
transfer of transposable elements and why it matters for eukaryotic evolution. Trends
Ecol Evol, 25, 537-546. doi: 10.1016/j.tree.2010.06.001
[67] Simmons, Mark P. (2012). Misleading results of likelihood-based phylogenetic analyses
in the presence of missing data. Cladistics, 28, 208-222. doi: 10.1111/j.1096-
0031.2011.00375.x
[68] Simoes, M., Breitkreuz, L., Alvarado, M., Baca, S., Cooper, J. C., Heins, L., . . .
Lieberman, B. S. (2016). The evolving theory of evolutionary radiations. Trends Ecol
Evol, 31, 27-34. doi: 10.1016/j.tree.2015.10.007
[69] Škrabánek, P., & Martínková, N. (2017). Extraction of outliers from imbalanced sets.
In F. J. Martínez de Pisón, R. Urraca, H. Quintián & E. Corchado (Eds.), Lecture Notes
in Artificial Intelligence. Hybrid Artificial Intelligent Systems: 12th International
Conference, HAIS 2017, La Rioja, Spain, June 21-23, 2017, Proceedings. (Vol.
10334, pp. 402-412). Cham: Springer.
[70] Smith, G. J., Bahl, J., Vijaykrishna, D., Zhang, J., Poon, L. L., Chen, H., . . . Guan,
Y. (2009). Dating the emergence of pandemic influenza viruses. Proc Natl Acad Sci
U S A, 106, 11709-11712. doi: 10.1073/pnas.0904991106
[71] Stamatakis, A. (2014). RAxML version 8: a tool for phylogenetic analysis and postanalysis
of large phylogenies. Bioinformatics, 30, 1312-1313. doi: 10.1093/bioinfor-
matics/btu033
[72] Symonds, M. R. E., & Blomberg, S. P. (2014). A Primer on Phylogenetic Generalised
Least Squares. In L. Garamszegi (Ed.), Modern Phylogenetic Comparative Methods
and Their Application in Evolutionary Biology. Berlin, Heidelberg: Springer.
[73] Turner, G. G., Meteyer, C. U., Barton, H., Gumbs, J. F., Reeder, D. M., Overton,
B., . . . Blehert, D. S. (2014). Nonlethal screening of bat-wing skin with the use of
ultraviolet fluorescence to detect lesions indicative of white-nose syndrome. J Wildl
Dis, 50, 566-73. doi: 10.7589/2014-03-058
[74] Wallace, I. S., Shakesby, A. J., Hwang, J. H., Choi, W. G., Martínková, N.,
Douglas, A. E., & Roberts, D. M. (2012). Acyrthosiphon pisum AQP2: a multifunctional
insect aquaglyceroporin. Biochim Biophys Acta, 1818, 627-635. doi:
10.1016/j.bbamem.2011.11.032
[75] Webb, C. O., Ackerly, D. D., McPeek, M. A., & Donoghue, M. J. (2002). Phylogenies
and Community Ecology. Annual Review of Ecology and Systematics, 33, 475-505.
doi: 10.1146/annurev.ecolsys.33.010802.150448
References 47
[76] Wibbelt, G., Kurth, A., Hellmann, D., Weishaar, M., Barlow, A., & Veith, M. (2010).
White-nose syndrome fungus (Geomyces destructans) in bats, Europe. Emerg. Infect.
Dis., 16, 1237-1243.
[77] Xi, Z., Liu, L., & Davis, C. C. (2016). The impact of missing data on species tree
estimation. Mol Biol Evol, 33, 838-860. doi: 10.1093/molbev/msv266
[78] Zuckerkandl, E., & Pauling, L. (1962). Molecular disease, evolution, and genic
heterogeneity. In M. Kasha & B. Pullman (Eds.), Horizons in biochemistry (pp.
167-187). New York: Academic Press.
[79] Zukal, J., Bandouchova, H., Bartonička, T., Berková, H., Brack, V., Brichta, J., . . .
Pikula, J. (2014). White-nose syndrome fungus: A generalist pathogen of hibernating
bats. PLoS ONE, 9, e97224.
[80] Zukal, J., Bandouchova, H., Brichta, J., Cmokova, A., Jaron, K. S., Kolarik, M.,
. . . Martínková, N. (2016). White-nose syndrome without borders: Pseudogymnoascus
destructans infection confirmed in Asia. Scientific Reports, 6, 19829. doi:
10.1038/srep19829
Index
alignment, 4, 5, 9
missing data, 5, 8–10, 33, 37
aquaporins, 8, 9, 35
arenaviruses, 20, 22, 36
bats, 22–27, 34, 37, 38
bootstrap, 4, 10, 12
bovids, 9, 34
branch length, 5–11, 33–35
Brownian motion model, 23, 24
CADM, 17, 19
codivergence, 17–20, 22, 29, 36, 37
coevolution, 17
comparative phylogenetics, 21, 29, 37, 38
EEO, 26, 27, 39
fish, 36
genomic signature, 27, 28, 39
horizontal gene transfer, 26, 27, 39
host-parasite interaction, 17, 18, 20, 22–27,
36–38
indel, 8, 9, 34, 35
likelihood, 4, 7–10, 33
Markov chain, 5, 6, 12, 13, 18
matrix, 4, 9, 37
congruence, 17–19
distance, 4, 17, 18, 27, 39
sparse, 7
variance-covariance, 22, 23
mole rats, 34
molecular clock, 11, 14
calibration, 14, 16, 35, 36
molecular dating, 3, 11, 14, 15, 29, 35, 36
mutation rate, 5, 8, 11, 12, 14, 15, 36
decay, 12–15, 35
nearest taxon index, 21, 22, 36, 37
net relatedness index, 20–22, 36, 37
outlier, 26–29, 39
ParaFit, 18, 19, 36
partition, 8, 9, 34, 35
permutation, 12, 17, 18, 21
PGLS, 23, 25, 38
phylogenetic reconstruction, 3, 10, 11
Bayesian inference, 4–10, 32–36
maximum likelihood, 4, 8, 10, 32–37
maximum parsimony, 4, 32–34
neighbour joining, 4, 32–34, 37
phylogenetic signal, 22, 23, 25, 38
phylogeny, 5–11, 14, 17, 19–21, 23, 24, 29,
32, 37, 38
posterior, 5, 7–10, 15, 33–35
prior, 5–10, 14, 15, 34
selection, 12, 13, 23
SigHunt, 26–28, 39
simulation, 12, 13, 18, 19, 23, 24
stoats, 14, 35
substitution model, 4, 5, 8
trait, 21–25, 29
tree
length, 7–9, 11, 18, 34
space, 4–7, 10, 35
terrace, 10, 33
tree squirrels, 9, 10, 14, 15, 20, 33, 37
voles
arvicoline, 5, 7–10, 32, 33
Nearctic, 20, 22, 36
Orkney vole, 14, 36
white-nose syndrome, 22, 23, 25, 27, 37, 38
– 49 –
Paper 2.1.1
Jaarola M., Martínková N., Gündüz ˙I., Brunhoff C., Zima J., Nadachowski A., Amori
G., Bulatova N. S., Chondropoulos B., Fraguedakis-Tsolis S., González-Esteban J., LópezFuster
M. L., Kandaurov A. S., Kefelio˘glu H., da Luz Mathias M., Villate I., Searle J. B.
2004. Molecular phylogeny of the speciose vole genus Microtus (Arvicolinae, Rodentia)
inferred from mitochondrial DNA sequences. Molecular Phylogenetics and Evolution 33:
647-663.
– 51 –
Molecular phylogeny of the speciose vole genus
Microtus (Arvicolinae, Rodentia) inferred from mitochondrial
DNA sequences
Maarit Jaarolaa,*, Nata´lia Martı´nkova´b,c
, _Islam Gu¨ndu¨za,d
, Cecilia Brunhoffa
,
Jan Zimab
, Adam Nadachowskie
, Giovanni Amorif
, Nina S. Bulatovag
,
Basil Chondropoulosh
, Stella Fraguedakis-Tsolish
, Jorge Gonza´lez-Estebani
,
Marı´a Jose´ Lo´pez-Fusterj
, Andrei S. Kandaurovk
, Haluk Kefeliog˘lud
,
Maria da Luz Mathiasl
, Idoia Villatei
, Jeremy B. Searlem
a
Department of Cell and Organism Biology, Genetics Building, Lund University, So¨lvegatan 29, SE-223 62 Lund, Sweden
b
Institute of Vertebrate Biology, Academy of Science of the Czech Republic, Studenec 122, 675 02 Koneˇsˇı´n, Czech Republic
c
Biodiversity Research Group, Department of Zoology, Charles University, Vinicˇna´ 7, 128 44 Praha 2, Czech Republic
d
Department of Biology, Faculty of Arts and Sciences, Ondokuz Mayis University, TR-55139 Kurupelit, Samsun, Turkey
e
Institute of Systematics and Evolution of Animals, Polish Academy of Sciences, Sławkowska 17, 31-016 Krako´w, Poland
f
CNR—Institute of Ecosystem Studies, Via A. Borelli 50, 00161 Rome, Italy
g
Severtzov Institute of Ecology and Evolution, Russian Academy of Sciences, 33 Leninsky Prospect, 117071 Moscow, Russia
h
Section of Animal Biology, Department of Biology, University of Patra, GR-26 500 Rion, Patras, Greece
i
Sociedad de Ciencias Aranzadi, Alto de Zorroaga, E20014 San Sebastia´n, Spain
j
Department of Animal Biology, Faculty of Biology, University of Barcelona, Avinguda Diagonal 645, 08028-Barcelona, Spain
k
Institute of Zoology, Georgian Academy of Science, 31 Chavchavadze Av., Tbilisi, Georgia
l
Centre for Environmental Biology and Department of Animal Biology, Faculty of Science, University of Lisbon, Campo Grande,
1749-016 Lisbon, Portugal
m
Department of Biology, University of York, PO Box 373, York YO10 5YW, UK
Received 22 January 2004; revised 30 June 2004
Available online 16 September 2004
Abstract
Voles of the genus Microtus represent one of the most speciose mammalian genera in the Holarctic. We established a molecular
phylogeny for Microtus to resolve contentious issues of systematic relationships and evolutionary history in this genus. A total of 81
specimens representing ten Microtus species endemic to Europe as well as eight Eurasian, six Asian and one Holarctic species were
sequenced for the entire cytochrome b gene (1140 bp). A further 25 sequences were retrieved from GenBank, providing data on an
additional 23, mainly Nearctic, Microtus species. Phylogenetic analysis of these 48 species generated four well-supported monophyletic
lineages. The genus Chionomys, snow voles, formed a distinct and well-supported lineage separate from the genus Microtus. The
subgenus Microtus formed the strongest supported lineage with two sublineages displaying a close relationship between the arvalis
species group (common voles) and the socialis species group (social voles). Monophyly of the Palearctic pitymyid voles, subgenus
Terricola, was supported, and this subgenus was also subdivided into two monophyletic species groups. Together, these groupings
clarify long-standing taxonomic uncertainties in Microtus. In addition, the ‘‘Asian’’ and the Nearctic lineages reported previously
were identified although the latter group was not supported. However, relationships among the main Microtus branches were not
resolved, suggesting a rapid and potentially simultaneous radiation of a widespread ancestor early in the history of the genus. This
and subsequent radiations discernible in the cytochrome b phylogeny, show the considerable potential of Microtus for analysis of
1055-7903/$ - see front matter Ó 2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.ympev.2004.07.015
*
Corresponding author. Fax: +46 46 14 78 74.
E-mail address: Maarit.Jaarola@cob.lu.se (M. Jaarola).
Molecular Phylogenetics and Evolution 33 (2004) 647–663
MOLECULAR
PHYLOGENETICS
AND
EVOLUTION
www.elsevier.com/locate/ympev
52 Paper 2.1.1
historical and ecological determinants of speciation in small mammals. It is evident that speciation is an ongoing process in the
genus and that the molecular data provides a vital insight into current species limits as well as cladogenic events of the past.
Ó 2004 Elsevier Inc. All rights reserved.
Keywords: Voles; Microtus; Arvicolinae; Mitochondrial DNA; Cytochrome b; Holarctic; Phylogeny; Speciation
1. Introduction
Voles of the genus Microtus Schrank (1798) are ecologically
diverse and constitute the dominant herbivorous
small mammals in many habitats in the Northern
Hemisphere. Most species prefer open grasslands such
as meadows and pastures but some species also occupy
forests and highland habitats (Getz, 1985; Hoffmann
and Koeppl, 1985; Mitchell-Jones et al., 1999; Nowak,
1999). Microtus represents one of the most speciose
mammalian genera in the Holarctic, accounting for
nearly 50 percent of the species of Arvicoline rodents
(voles and lemmings) (e.g., Musser and Carleton,
1993). The genus represents one of the best known cases
of rapid mammalian radiation resulting in about 65 extant
species distributed throughout the Palearctic and
Nearctic regions (Musser and Carleton, 1993; Chaline
et al., 1999; Nowak, 1999). The shrew genus Sorex
(Soricidae, Insectivora) is the only other mammalian
genus that displays a comparable diversity across the
Holarctic region, but Sorex is much older than Microtus
(see Fumagalli et al., 1999).
The genus Microtus is apparently derived from the
fossil genus Allophaiomys, which itself appears to have
descended from Mimomys (Chaline and Graf, 1988;
Conroy and Cook, 1999; Garapich and Nadachowski,
1996). Palaeontological data suggest that Allophaiomys
radiated independently in northern Eurasia, central
Asia-Himalayas and North America (Brunet-Lecomte
and Chaline, 1991; Chaline et al., 1999). Until recently,
the appearance of Allophaiomys was dated to approximately
2 million years ago (Mya) (Chaline and Graf,
1988), but a new finding of Allophaiomys from China
traces the origin of the lineage back to 2.3–2.4 Mya
(Zheng and Zhang, 2000). The majority of extant Microtus
species, however, do not appear in the fossil record
until Middle Pleistocene about 0.7–0.5 Mya (Chaline et
al., 1999; Rabeder, 1986; Richmond, 1996) and it has
even been suggested that some species trace their origin
to the last glaciation (e.g., Brunet-Lecomte and Chaline,
1990; Chaline and Graf, 1988).
The genus Microtus displays a number of features
that makes it ideal for evolutionary studies of speciation
and the role of Quaternary glacial cycles on diversification.
However, the phylogenetic relationships within
Microtus and its closest relatives are uncertain and difficulties
remain both in delimiting species and defining
subgenera (e.g., Musser and Carleton, 1993; Nadachowski
and Zagorodnyuk, 1996; Zagorodnyuk, 1990).
Current speciesÕ boundaries and phylogenetic relationships
in Microtus rely mainly on morphology and
karyotypes but these taxonomic characters have not
been sufficient for solving all systematic questions in
Microtus. The treatment of this genus is consequently
characterized by inconsistency and lack of consensus
(Musser and Carleton, 1993). For example, studies of
dental and skull characters in Microtus demonstrate
great intra-specific variability, a high incidence of adaptive
convergence and many pairs of sibling species (Chaline
et al., 1999; Chaline and Graf, 1988; Nadachowski
and Zagorodnyuk, 1996; Zakrzewski, 1985). Karyotype
evolution on the other hand appears largely uncoupled
from morphological evolution (Baskevich, 1996;
Suchentrunk et al., 1998). The Microtus karyotype varies
between 2n = 17–62 (Zagorodnyuk, 1990; Zima and
Kra´l, 1984) and exhibits one of the highest rates of
karyotypic change in mammals (Maryama and Imai,
1981; Modi, 1987). Although some phylogenetic relationships
can be deduced, especially from G-banded
karyotypes (Mazurok et al., 2001; Meier et al., 1985,
1996; Modi, 1987; Orlov et al., 1983; Zagorodnyuk,
1990), the overall picture is that of extensive karyotypic
variation among closely related species with no apparent
phylogenetic trends (e.g., Akhverdyan et al., 1999; Chaline
et al., 1999; Machola´n et al., 2001; Modi, 1987). The
systematic uncertainties in Microtus are perhaps best
illustrated by the fact that the number of recognized species
varies widely in different accounts (e.g., Gromov
and Polyakov, 1992; Musser and Carleton, 1993; Nowak,
1999; Panteleyev, 1998; Zagorodnyuk, 1990) and
that several new species have been proposed over recent
years (Golenishchev et al., 2003; Kefeliog˘lu and Krysˇtufek,
1999; Krysˇtufek and Kefeliog˘lu, 2001; Yigit and
Colak, 2002). Other unresolved issues concern the delineation
and validity of higher-order relationships and
species groups such as Chionomys, Volemys, Lasiopodomys,
Blanfordimys, and Terricola that are alternately given
either generic or subgeneric rank (e.g., Gromov and
Polyakov, 1992; Musser and Carleton, 1993; Nowak,
1999).
Attempts to reconstruct the Microtus phylogeny
using molecular approaches have been made using allozymes
(e.g., Chaline and Graf, 1988; Gill et al., 1987;
Graf, 1982; Mezhzherin et al., 1993, 1995; Suchentrunk
et al., 1998), restriction enzyme analysis of mitochondrial
DNA (DeBry, 1992) and RAPD analysis (Potapov
et al., 1999) based on a limited number of Microtus
species. However, the most comprehensive molecular
648 M. Jaarola et al. / Molecular Phylogenetics and Evolution 33 (2004) 647–663
Paper 2.1.1 53
phylogenetic studies to date have been carried out by
Conroy and Cook (1999, 2000a) and Conroy et al.
(2001) using mitochondrial cytochrome b gene sequences.
These studies comprised the North American
endemics as well as some Asian and Eurasian Microtus
species. In addition, Mazurok et al. (2001) analyzed
cytochrome b sequences of four species of the M. arvalis
group and Haring et al. (2001) inferred a phylogeny for
species of the M. multiplex species group by analysis of
mitochondrial D-loop sequences. Phylogeographic surveys
of North American and Eurasian Microtus have
demonstrated extensive cytochrome b variation within
species (Brunhoff et al., 2003; submitted; Conroy and
Cook, 2000b; Haynes et al., 2003; Jaarola and Searle,
2002, 2004). The results even suggest the existence of
hitherto unidentified, cryptic species in Microtus (Jaarola
and Searle, 2002, 2004; Hellborg, 2004), thereby
pointing out the importance of within-species sampling
for phylogeny reconstruction in Microtus.
The aim of this study is to further the understanding
of phylogenetic relationships and evolutionary history in
Microtus voles. We establish a molecular phylogeny for
Palearctic Microtus including all the currently recognized
European species (Mitchell-Jones et al., 1999) except
M. bavaricus, as well as eight Eurasian and six
Asian species. For this purpose we use DNA sequence
analysis of the entire cytochrome b gene (1140 bp) since
it evolves rapidly over the expected divergence times and
because it enables us to incorporate previously published
data on Nearctic species (Conroy and Cook,
1999, 2000a; Conroy et al., 2001). The combined phylogeny
includes 48 out of 65 species of Microtus, with multiple
representatives of many of the species. This nearcomprehensive
phylogeny of the genus allows us not
only to resolve long-standing taxonomic controversies
in Microtus but also to provide a phylogenetic tool to
help understand speciation and speciesÕ radiations in
small mammals.
2. Materials and methods
2.1. Samples
A total of 81 specimens representing 25 Microtus species
were analyzed for variation in the mitochondrial
cytochrome b gene (Table 1). All 25 species except for
the Holarctic M. oeconomus are Palearctic; 10 are endemic
to Europe, eight occur in Eurasia and six are restricted
to Asia (cf. Gromov and Polyakov, 1992;
Mitchell-Jones et al., 1999; Musser and Carleton,
1993). The only European species not analyzed is M.
bavaricus that was considered extinct (Mitchell-Jones
et al., 1999), but recently rediscovered in the northern
Tyrol (Hutterer, 2001). The samples contain two species
from the genus Chionomys, until recently often classified
as Microtus, as well as six subgenera of Microtus (Table
1). Up to five individuals per taxon were surveyed to account
for intra-specific variation. When possible, conspecific
individuals were chosen from geographically
distant localities. A total of 17 sequences representing
M. agrestis, M. oeconomus, and M. arvalis have been
published previously (Brunhoff et al., 2003; Haynes
et al., 2003; Jaarola and Searle, 2002); their GenBank
accession numbers are given in Table 1. In addition,
24 sequences representing 19 North American endemics
and five Asian/Palearctic Microtus species (including
one species that we sequenced, M. gregalis) were retrieved
from GenBank (Conroy and Cook, 1999,
2000a; Conroy et al., 2001); for accession numbers see
Table 2. A second M. (Volemys) kikuchii sequence was
obtained from the whole mtDNA sequence in GenBank
(AF348082, Lin et al., 2002).
2.2. DNA extraction, PCR amplification and sequencing
Total genomic DNA was extracted from frozen or
ethanol preserved tail tips, ears, kidneys or liver. A
few samples consisted of skulls collected from owl and
raptor pellets. The Qiagen Dneasy Tissue kit was used
for all types of samples. Pure mtDNA was isolated from
two M. rossiaemeridionalis samples according to Jaarola
and Tegelstro¨m (1995).
The primers used in this study are given in Table 3.
For most specimens, the complete mitochondrial cytochrome
b gene (1140 bp) was amplified in a single
PCR reaction using the L14727-SP and H-15195-SP/
H-ISO-SP primers complementary to glutamate and
threonine/proline tRNA sequences, respectively. Alternatively,
two to four separate amplifications that produced
overlapping fragments were carried out. Due to
the presence of nuclear copies of the cytochrome b gene
in several Microtus species (DeWoody et al., 1999; Jaarola
and Searle, 2004; Jaarola et al., in prep.), the majority
of amplifications involved only Microtus- or speciesspecific
primers designed by us (Table 3).
PCR amplification was carried out using AmpliTaq
Gold DNA polymerase (Applied Biosystems). The
PCR protocol for tissue samples consisted of an initial
7 min denaturation step at 95 °C, 30–35 cycles of denaturation
at 94 °C for 1 min, annealing at 49 or 50 °C for
1 min and extension at 72 °C for 1–2 min, and a final 10min
extension step at 72 °C. PCR products were purified
using the Qiagen QIAquick kit. The PCR protocol for
skull samples is described in Jaarola and Searle (2004).
We sequenced each DNA fragment in both directions
using a combination of PCR primers and internal primers
(Table 3). Cycle sequencing reactions were carried
out using the BigDye Terminator cycle sequencing kit
(Applied Biosystems). Amplifications and sequencing
reactions were performed in a PTC-200 thermal cycler
(MJ Research). Sequencing products were purified using
M. Jaarola et al. / Molecular Phylogenetics and Evolution 33 (2004) 647–663 649
54 Paper 2.1.1
Table 1
Microtus species, specimens, locations, and GenBank accession numbers of cytochrome b sequences
Subgenus (species
group)
Species Common
name
No. Location Country Accession
Nos.
Agricola (agrestis) Microtus agrestis Field vole 1 Novosibirsk Russia AY167149
2 Bonn Germany AY167210
3 Sion, Valais Switzerland AY167160
4 Pyrenees Spain AY167187
Microtus (arvalis) M. arvalis arvalis Common vole 1 Mantet, Pyrenees Spain AY220789
2 Trento Italy AY220766
3 Nuijamaa Finland AY220770
4 Lauwersee Netherlands AY220778
Microtus (arvalis) M. arvalis obscurus (Altai vole) 1 Crimea Ukraine AY220762
2 Kavka River, Serov Russia AY220764
3 Neiva River Russia AY220765
4 Sisian Armenia AY220761
5 Ninotsminda Georgia AY220760
Agricola (agrestis) M. cabrerae CabreraÕs vole 1 Alandron Portugal AY513788
2 Idanha-a-Velha Portugal AY513789
Terricola M. daghestanicus Daghestan pine vole 1 Beniani Georgia AY513790
(subterraneus/majori) 2 Bag˘dasßan Turkey AY513791
3 Handere Turkey AY513792
Microtus (socialis) M. dogramacii 1 Ortako¨y-Aksaray Turkey AY513793
2 Amasya Turkey AY513794
3 Boyalı Ko¨yu¨-Amasya*
Turkey AY513795
Terricola M. duodecimcostatus Mediterranean pine vole 1 Setu`bal Portugal AY513796
(duodecimcostatus) 2 Algarve Portugal AY513797
Terricola (savii) M. felteni Balkan pine vole Mt. Pelister, Begova Cˇ esˇma Macedonia AY513798
Terricola (savii) M. gerbei Pyrenean pine vole 1 Arro´s, Vall dÕAran Spain AY513799
2 Riba Spain AY513800
3 Hecho Spain AY513801
4 Hecho Spain AY513802
Stenocranius M. gregalis Narrow-headed vole 1 Yamal Peninsula Russia AY513803
Microtus (socialis) M. guentheri GuentherÕs vole 1 Gravia Greece AY513804
2 Aqrabat Syria AY513805
3 Locality unknown Israel AY513806
4 Locality unknown Israel AY513807
Neodon M. juldaschi Juniper vole Mazarsay Kyrgyzstan AY513808
Microtus (arvalis) M. kirgisorum Tien Shan vole 1 Balkash Kyrgyzstan AY513809
2 Balkash Kyrgyzstan AY513810
Terricola (multiplex) M. liechtensteini LiechtensteinÕs pine vole Anhovo Slovenia AY513811
Terricola M. lusitanicus Lusitanian pine vole 1 Burgos Spain AY513812
(duodecimcostatus) 2 Melgar de Fernamental Spain AY513813
Terricola
(subterraneus/majori)
M. majori MajorÕs pine vole Damar Turkey AY513814
Terricola (multiplex) M. multiplex Alpine pine vole 1 Staffarda, Piedmont Italy AY513815
2 Trento Italy AY513816
3 Lillaz Italy AY513817
4 Me´ribel France AY513818
Pallasiinus (oeconomus) M. oeconomus Root vole 1 Ivvavik Nat. Park Canada AY220028
(Tundra vole) 2 Krasnoyarsk Russia AY220018
3 Hamningberg Norway AY219988
4 Texel Netherlands AY220006
Microtus (arvalis) M. rossiaemeridionalis Sibling vole 1 Kauhava Finland AY513819
2 Svalbard Norway AY513820
3 Gerede, Istanbul Turkey AY513821
650 M. Jaarola et al. / Molecular Phylogenetics and Evolution 33 (2004) 647–663
Paper 2.1.1 55
standard protocols and run in an ABI 310 or 3100 automated
DNA sequencer (Applied Biosystems).
2.3. Phylogenetic analysis
Sequences were aligned and ambiguous bases resolved
by eye using Sequencher v. 3.1.1 (Gene Codes
Corp.). Nucleotide and amino acid composition was
analyzed using MacClade v. 4.05 (Maddison and Maddison,
2000). Frequencies of transitions and transversions
were estimated from maximum parsimony trees using
MacClade.
The phylogenetic relationships among haplotypes
were reconstructed using neighbor-joining (NJ), maximum
parsimony (MP) and maximum likelihood (ML)
algorithms implemented in PAUP* v. 4.0b10 (Swofford,
2002) as well as the Bayesian approach (Huelsenbeck
et al., 2001) using the program MrBayes 3 (Ronquist
and Huelsenbeck, 2003). The parsimony analyses were
carried out ten times with the heuristic search approach
using the TBR swapping algorithm, steepest descent option,
random addition and 100–1000 replicates. Strict
and 50% majority consensus trees were constructed from
multiple equally parsimonious MP trees. The hierarchical
likelihood ratio test (hLRT) and the Akaike information
criterion (AIC) implemented in the computer
program MODELTESTODELTEST v. 3.06 (Posada and Crandall,
1998), were used to identify the most appropriate model
of DNA substitution for our data. The model selected
was the general time reversible model, GTR (Yang,
1994) with a gamma distributed shape parameter (a)
of 0.8722 and the proportion of invariable sites (I)
equaling 0.5320. This model, with parameters determined
by MODELTESTODELTEST, as well as several simpler substitution
models, was implemented in the NJ and ML
analyses. The ML tree search was conducted as described
for MP but with the ‘‘as is’’ addition replicate
(i.e. alphabetically by species). Relative stability of NJ
and MP trees was assessed with bootstrap analysis using
10 000 and 1000 replicates, respectively. Bootstrapping
of the ML tree could not be carried out due to the excessive
computer capacity required.
Table 1 (continued)
Subgenus (species
group)
Species Common name No. Location Country Accession
Nos.
4 Erciyes Mt., Kayseri Turkey AY513822
5 Kangal-Sivas Turkey AY513823
Terricola (savii) M. savii SaviÕs pine vole 1 Viterbo Italy AY513824
2 Torino, Piedmont Italy AY513825
3 Cerano, Piedmont Italy AY513826
4 Fiume Freddo Italy AY513827
5 Fiume Freddo Italy AY513828
Microtus (socialis) M. socialis Social vole 1 Iori River valley Georgia AY513829
2 Iori River valley Georgia AY513830
3 Reine Iran AY513831
Terricola M. subterraneus Common pine vole 1 Seli Greece AY513832
(subterraneus/majori) 2 Glocknerhaus Austria AY513833
3 C¸ ig˘likara Turkey AY513834
4 C¸ ig˘likara Turkey AY513835
5 Gu¨zyurdu Turkey AY513836
Terricola (multiplex) M. tatricus Tatra vole 1 Tretie Roha´cˇske pleso lake Slovakia AY513837
2 Smutna´ dolina valley Slovakia AY513838
3 Vel´ka´ studena´ dolina valley*
Slovakia AY513839
Terricola (duodecimcostatus) M. thomasi ThomasÕs vole 1 Agios Stefanos Greece AY513840
2 Ano Kastritsi Greece AY513841
3 Kyparissia Greece AY513842
4 Itea Greece AY513843
5 Trebinje, Herzegovina Bosnia AY513844
Chionomys nivalis Snow vole 1 Trento Italy AY513845
2 Trento Italy AY513846
3 Prve´ Roha´cˇske pleso lake Slovakia AY513847
4 Queralbs, Girona Spain AY513848
5 Saleh, As Suwayda Syria AY513849
C. roberti RobertÕs vole 1 Altındere Vadisi Turkey AY513850
2 Datvisi Georgia AY513851
Subgenus and species group designation follows Musser and Carleton (1993) whose classification is largely based on the reclassification of Zagorodnyuk
(1990).
*
Type localities.
M. Jaarola et al. / Molecular Phylogenetics and Evolution 33 (2004) 647–663 651
56 Paper 2.1.1
We conducted Bayesian phylogenetic analyses using
the GTR + I + G model with unequal base frequencies.
Model parameters were estimated as part of the analysis.
Altogether four independent runs, each with four
Markov chain Monte Carlo (MCMC), were performed.
All Bayesian analyses were initiated with random starting
trees and run for two million generations. Trees were
sampled every 10 generations. Log-likelihood scores of
trees were plotted against generation time to determine
the ‘‘burn-in’’ period and ensure that equilibrium loglikelihood
values for different runs approached similar
mean values (Huelsenbeck et al., 2002; Ronquist and
Huelsenbeck, 2003). After discarding burn-in trees, we
generated 50% majority rule consensus trees in PAUP
for each single run and compared posterior probabilities
for convergence among runs.
Since the correlation between and significance of tree
support values as estimated by standard, nonparametric
bootstrap values as opposed to posterior probabilities is
not well understood and intensely debated (e.g., Huelsenbeck
et al., 2002; Suzuki et al., 2002; Erixon et al.,
2003), we follow the conservative approach of Leache´
and Reeder (2002). Thus, only bootstrap values of
P70% (corresponding to 95% CI) and posterior probabilities
of P95% were considered significant.
Table 2
Microtus cytochrome b sequences retrieved from GenBank (Conroy
and Cook, 1999, 2000a; Conroy et al., 2001; Lin et al., 2002)
Species Accession Nos.
Microtus abbreviatus AF163890
M. californicus AF163891
M. canicaudus AF163892
M. chrotorrhinus AF163893
M. fortis AF163894
M. gregalis AF163895
M. guatemalensis AF410262
M. kikuchii AF163896
M. kikuchii AF348082
M. longicaudus AF187230
M. mexicanus AF163897
M. middendorffi AF163898
M. miurus AF163899
M. montanus AF119280
M. montebelli AF163900
M. oaxacensis AF410260
M. ochrogaster AF163901
M. oregoni AF163903
M. pennsylvanicus AF119279
M. pinetorum AF163904
M. quasiater AF410259
M. richardsoni AF163905
M. townsendii AF163906
M. umbrosus AF410261
M. xanthognathus AF163907
Table 3
Primers used for PCR amplification and sequencing of the cytochrome b gene in Microtus
Primer Sequence (50
–30
) Reference
L14724B CGAGATCTGAAAAACCATCGTTG Kocher et al. (1989)
L14727-SP GACAGGAAAAATCATCGTTG Jaarola and Searle (2002)
L14841M CCATCAAATATTTCATCATGATGAAA Jaarola and Searle (2002)
L15162M2 GCTACGTACTTCCATGAGGACAAATATC Jaarola and Searle (2002)
L15162Marv G(CT)TACGT(CT)CTTCCATGAGGCCAAATATC Haynes et al. (2003)
L15162MO CTTCCATGAGGCCAAATATC Brunhoff et al. (2003)
L15408M GCAGACAAAATCCCGTTCCA Jaarola and Searle (2002)
L15408Marv GCAGACAAAATCCCATTCCA Haynes et al. (2003)
L15408-SP GCAGACAAA AT(TC)CC(AG)TT(TC)CA Present study
H15177-SP2 AGGAGGTTTGT(AG)ATGACTG Present study
H15177-SP3 A(AG)GAGGTTTGT(AG) ATNACTG Present study
H15177Marv AAGAGATTTGTAAT(CT)ACTG Present study
H15177MO AGGAGGTTTGTGATTACTG Brunhoff et al. (2003)
H 15319Marv AAAGGTGGACTAATACGAGG Haynes et al. (2003)
H15348A-SP GTTGGA(CT)CCTGTTTCGTG Jaarola and Searle (2002)
H15408M TGGAACGGGATTTTGTCTGC Jaarola and Searle (2002)
H15408MO TGGAATGGGATTTTGTCTGT Brunhoff et al. (2003)
H15497-SP T(AG)TAATT(AG)TCNGGGTCTCC Present study
H15497-SP2 TGTAATT(AG)TCGGGGTCTCC Present study
H15549M AAGAGGAAATACCATTCTGGTTTAA Jaarola and Searle (2002)
H15576M GACCGTAAAATGGCGTAGG Jaarola and Searle (2002)
H15576MO GATCGTAGGATGGCGTAGG Brunhoff et al. (2003)
H15915 AACTGCAGTCATCTCCGGTTTACAAGAC Irwin et al. (1991)
H15915-SP TTCATTACTGGTTTACAAGAC Jaarola and Searle (2002)
H-ISO-M AAGTAGTTTAATTAGAATGTCAG Haynes et al. (2003)
H-PRO AAGTAGTTTAATTAGAATATCAG Brunhoff et al. (2003)
H-ISO-SP AGTAGTTTAATTAGAATGTCAGC Jaarola and Searle (2002)
SP: Microtus spp.; M: M. agrestis; Marv: M. arvalis; and MO: M. oeconomus.
652 M. Jaarola et al. / Molecular Phylogenetics and Evolution 33 (2004) 647–663
Paper 2.1.1 57
Various combinations of lemmings (Dicrostonyx,
Lemmus), Clethrionomys rutilus and C. glareolus and
Arvicola terrestris were tested as outgroups. Since the position
of Arvicola proved unstable and often tended to
occur within Microtus, we discarded this option. The
two Chionomys species analyzed, C. nivalis and C. roberti,
were also tested but were too closely related to Microtus
to function as outgroups. Our findings corroborate
the conclusion of Conroy and Cook (1999, 2000a) that
Clethrionomys is a sister taxon to Microtus, and we therefore
used C. glareolus and/or C. rutilus as outgroups.
2.4. Tests of sequence saturation and a cytochrome b
clock
To diagnose sequence saturation and homoplasy at
the third position, we constructed a scatter plot of
uncorrected pairwise transitions and transversion frequencies
versus corrected pairwise divergences. Sequence
divergence was corrected with the Jukes–
Cantor (JC, Jukes and Cantor, 1969) model as well as
a maximum likelihood model based on the
GTR + G + I model of substitution. Total and net
divergence (Dxy and Da) between species was estimated
according to Nei (1987). We also tested for a molecular
clock by comparing log likelihood scores of ML trees
constructed with and without a molecular clock constraint
(Felsenstein, 1988) in PAUP.
3. Results
GenBank accession numbers for the 64 new 1140 bp
cytochrome b sequences representing 22 species are given
in Table 1 (AY513788–AY513851) together with
accession numbers of 17 sequences, representing three
additional species, that we have published previously.
We are confident that the new sequences represent the
mitochondrial cytochrome b gene and not nuclear
pseudogenes since they closely match previously reported
Microtus sequences (Brunhoff et al., 2003; Conroy
and Cook, 2000a,b; Conroy et al., 2001; Haynes et
al., 2003; Jaarola and Searle, 2002) and because they
did not display any of the anomalies typical for nuclear
copies (cf. Bensasson et al., 2001; Mirol et al., 2000). We
did, however, also obtain pseudogenes for the cytochrome
b gene in a number of samples and taxa despite
using only Microtus-specific primers. These findings will
be reported elsewhere (Jaarola et al., unpublished).
3.1. Sequence composition and variation
The total, aligned data matrix included 100 cytochrome
b haplotypes derived from 106 sequences representing
48 currently recognized species. A total of 504
(44%) variable sites were observed and 456 of these were
informative for the parsimony analyses. More than one
type of nucleotide substitution was observed in 199
(17%) sites and 83 (7%) of these displayed all four nucleotides.
The majority of polymorphic sites were at third
positions (364, 72%), followed by first positions (115,
23%) and second positions (25, 5%). Most substitutions
were transitions (78%). The light strand nucleotide composition
was characterized by a deficit of guanines (13%)
similar to that described in North American Microtus
species (Conroy and Cook, 2000a,b) as well as other
mammals (e.g., Irwin et al., 1991). All but eight of the
86 variable amino acid residues and ten of the 136 amino
acid replacements were located within the variable matrix
and transmembrane region of cytochrome b (cf. Irwin
et al., 1991; McClellan and McCracken, 2001).
Inter-specific distances varied between 4.2% and
18.0% using the JC model, whereas maximum likelihood
distances estimated under the GTR + G + I model ranged
from 4.5% to 51.6%. Corresponding intra-specific
distances ranged up to 6.2% and 7.2% (M. agrestis).
M. arvalis, M. agrestis, M. daghestanicus, M. guentheri,
M. oeconomus, M. savii, M. subterraneus, and C. nivalis
showed intra-specific divergences of 4–7%, values similar
to net distance estimates between closely related species
such as M. duodecimcostatus–M. lusitanicus, M.
dogramacii–M. guentheri, and M. liechtensteini–M. multiplex.
Some saturation occurred at third position transitions
for divergences at the subgenus and genus
levels (not shown), similar to that described in other
studies of rodent cytochrome b (e.g., Yang and Yoder,
1999).
3.2. Phylogenetic results
The four phylogenetic methods used (MP, NJ, ML,
and Bayesian) displayed trees with very similar topologies
(Figs. 1–3). All methods discriminated two, wellsupported
major lineages corresponding to the two subgenera
Microtus and Terricola. Each of the two subgenera
was further divided into two well-supported
sublineages. The two Microtus lineages previously described
by Conroy and Cook (2000a) and Conroy et
al. (2001), the ‘‘Asian’’ and the Nearctic, were also observed
although the latter group was not supported.
The two species of the genus Chionomys formed a highly
supported branch outside the genus Microtus. Monophyly
of the genus Microtus was supported although
the bootstrap support depended much on the position
of M. gregalis.
We obtained 48 MP trees (3744 steps, CI = 0.214)
when using C. rutilus as outgroup, and 156 trees (3792
steps, CI = 0.212) when C. glareolus and C. rutilus were
used as outgroups. The strict consensus tree of the 48
trees is given in Fig. 1. The same tree topology was recovered
in the two consensus trees except for the position of
M. gregalis. The 48 MP trees rooted with C. rutilus only
M. Jaarola et al. / Molecular Phylogenetics and Evolution 33 (2004) 647–663 653
58 Paper 2.1.1
differed because of unresolved intra-specific branches of
M. multiplex, M. gerbei, M. socialis, and M. dogramacii.
However, the importance of intra-specific sampling was
clearly illustrated by the fact that removal of this intraspecific
variation for one or several of these taxa resulted
in fewer MP trees but also a decreased resolution of some
Fig. 1. Strict consensus tree of 48 maximum parsimony (MP) trees from cytochrome b haplotypes for 46 Microtus and two Chionomys species rooted
with Clethrionomys rutilus. Numbers above branches denote percentage bootstrap resampling support (>50%) from 1000 replications.
654 M. Jaarola et al. / Molecular Phylogenetics and Evolution 33 (2004) 647–663
Paper 2.1.1 59
of the deeper branches in the tree. The MP trees clearly
contained an excessive degree of homoplasy, but all major
lineages, except for the Nearctic, as well as many sublineages
exhibited high bootstrap values, and relatively
few MP trees were generated. Phylogenetic analyses
using only first and second positions did not provide enough
resolution (data not shown).
The NJ algorithm recovered the same topology independent
of substitution model used, except for statistically
unsupported differences in the positions of
Nearctic species and basal taxa such as M. agrestis, M.
cabrerae, M. gregalis, and M. juldaschi. The bootstrap
values increased with the simplicity of the substitution
model, the JC model generating the highest estimates.
The ML tree (Fig. 2) based on the GTR + G + I model
showed the same topology as the MP and NJ trees. The
ML tree constructed under a molecular clock constraint
differed significantly from the unconstrained ML tree
Fig. 2. Maximum likelihood (ML) tree based on GTR + G + I distances and rooted with Clethrionomys glareolus and C. rutilus. The tree shows the
inferred phylogenetic relationships among 100 cytochrome b haplotypes representing 46 Microtus and two Chionomys species. Bootstrap resampling
support from 10,000 iterations for a neighbor-joining (NJ) tree based on the JC model is listed above main branches. Estimates below branches show
corresponding bootstrap support values when M. gregalis, M. majori and M. cabrerae are removed from the phylogenetic analyses. Only values
greater than 50 percent are shown. *Percentage bootstrap support for Microtus monophyly in a NJ tree with M. gregalis included in Microtus.
Subgenus designation follows Musser and Carleton (1993) (see Table 1).
M. Jaarola et al. / Molecular Phylogenetics and Evolution 33 (2004) 647–663 655
60 Paper 2.1.1
when outgroups were included (v2
= 140.4, df = 100,
P < 0.001), but the difference was not significant when
outgroups were excluded (v2
= 120.5, df = 98, P > 0.05).
The Bayesian analyses reached stationarity around
100,000 generations (equaling 10,000 saved trees), so
that the last 190,000 trees were used to compute a
majority rule consensus tree for each run. The four independent
analyses converged on similar log-likelihood
values and the mean ln L score for the posterior distribution
of trees was À17674. Topology of consensus
trees, values of posterior probabilities and parameter
estimates were highly similar in all four analyses. The
Fig. 3. Fifty percent majority rule consensus tree of 190, 000 trees from a Bayesian analysis of cytochrome b haplotypes for 46 Microtus and two
Chionomys species rooted with Clethrionomys rutilus. Numbers above branches represent posterior probability values (>0.50).
656 M. Jaarola et al. / Molecular Phylogenetics and Evolution 33 (2004) 647–663
Paper 2.1.1 61
marginal probabilities of the nucleotide frequencies were
similar to those obtained with MODELTESTMODELTEST. The average
gamma distribution shape parameter (a) was 0.80 and
the proportion of invariable sites (I) averaged 0.48.
The lineages and sublineages also supported by standard,
nonparametric bootstrapping values P70%, calculated
using MP and/or NJ (Figs. 1 and 2), all
showed posterior probabilities of 100% (Fig. 3). In addition,
significance levels around 95% were obtained for a
few branches not supported by bootstrapping (Fig. 3).
The long branches of M. agrestis, M. juldaschi and,
especially, M. gregalis and M. cabrerae were basal and
only formed non-significant associations that were
dependent on the taxa present and the substitution model
used. The apparent relationship of the Iberian endemic
M. cabrerae with the Nearctic species represents an
example of this type of clustering (Figs. 1 and 3). The
Bayesian analyses supported the grouping of M. agrestis
with M. juldaschi with a significance level of 97% (Fig. 3),
but this was not supported by bootstrapping (Figs. 1 and
2). The position of M. gregalis was especially unstable.
Within the Terricola lineage, the position of the single
M. majori sequence could not be determined. Removal
of the unstable taxa M. gregalis, M. cabrerae, and M.
majori increased the bootstrap support for most major
lineages in the NJ tree so that all except the Nearctic
branch generated bootstrap values over 90% as well as
high (83%) support for Microtus monophyly (Fig. 1).
4. Discussion
4.1. The cytochrome b tree and Microtus taxonomy
Although there are many issues relating to details of
particular relationships, overall there is much concordance
between the Microtus cytochrome b tree and current
taxonomy based on morphology and karyotype.
Four monophyletic lineages with good bootstrap support
and 100% posterior probabilities were identified
(Figs. 1–3). First, the two species of snow voles, genus
Chionomys, formed a lineage separate from Microtus.
Second, species of the subgenus Microtus formed a
well-supported lineage with two highly supported sublineages
representing the arvalis species group (common
voles) and the socialis species group (social voles). Third,
monophyly of the Palearctic pitymyid voles, i.e. subgenus
Terricola was supported. Fourth, the ‘‘Asian’’ lineage
previously reported by Conroy and Cook (2000a)
and Conroy et al. (2001) was identified. All the Nearctic
species clustered as previously described by Conroy and
Cook (2000a) and Conroy et al. (2001) but this grouping
was not well supported. The following discussion will
deal with phylogenetic relationships, biogeographic scenarios
and species delimitations in Microtus. Data on
molecular evolution involving estimation of a molecular
clock and datings will be reported elsewhere (Jaarola et
al., unpublished). Our data support the analyses of Conroy
and Cook (2000b) suggesting rapid cytochrome b
evolution in Microtus.
4.2. Genus Chionomys (snow voles)
Chionomys nivalis and C. roberti form a distinct and
well-supported clade separate from Microtus (Figs. 1–
3). Thus, our cytochrome b results corroborate the ranking
of Chionomys as a genus separate from Microtus.
The genus Chionomys was previously included in Microtus
but recent analyses have suggested separation (reviewed
in Musser and Carleton, 1993; Nadachowski,
1991). According to palaeontological data, Chionomys
represents an early split, 1.3–1.5 Mya, from Allophaiomys
or, possibly, Mimomys, the ancestor of Allophaiomys
(Hora´cˇek, pers. comm.). The distant relationship
in cytochrome b between C. nivalis and C. roberti is also
in accord with fossil data (cf. Nadachowski, 1991).
4.3. Genus Microtus
The relationship among the main Microtus branches
was not resolved and the data indicate that Eurasian
species are not basal to North American endemics as
suggested by Conroy and Cook (2000a). This ‘‘hard’’
polytomy is likely to indicate an early burst of rapid
diversification of a widespread ancestor resulting in the
nearly simultaneous appearance and radiation of major
Microtus lineages in Eurasia, central Asia-Himalayas
and North America as advocated by palaeontologists
(Brunet-Lecomte and Chaline, 1991; Chaline et al.,
1999). We cannot, however, entirely dismiss the added
effects of a high degree of homoplasy at third positions
in the cytochrome b gene confounding the resolution
of basal relationships (cf. Conroy and Cook, 1999; Reed
and Sperling, 1999). Furthermore, some relationships
among subgenera were indicated by Bayesian inference
(Fig. 3) but not supported by bootstrapping of NJ or
MP trees (Figs. 1 and 2).
4.4. Subgenus Microtus
The Microtus subgenus represents the strongest supported
lineage in the cytochrome b tree, displaying a
close relationship between the arvalis species group
(common voles) and the socialis species group (social
voles) sensu Zagorodnyuk (1990). The close phylogenetic
relationships among the morphologically cryptic
species in the arvalis group are concordant with the
cytochrome b studies of Haynes et al. (2003) and Mazurok
et al. (2001) involving members of this group. Mazurok
et al. (2001) also included M. transcaspicus but
since the sequence was not submitted to GenBank, we
could not include it in our analyses.
M. Jaarola et al. / Molecular Phylogenetics and Evolution 33 (2004) 647–663 657
62 Paper 2.1.1
The systematics of social voles has proven much
more complex than previously thought (see Kefeliog˘lu
and Krysˇtufek, 1999; Krysˇtufek and Kefeliog˘lu, 2001).
Only recently a number of new species have been suggested
(Golenishchev et al., 2003; Kefeliog˘lu and Krysˇtufek,
1999; Krysˇtufek and Kefeliog˘lu, 2001; Yigit and
Colak, 2002). Besides the established species, M. socialis
and M. guentheri, we also included sequences of the
newly described M. dogramacii (Kefeliog˘lu and Krysˇtufek,
1999) in our analyses. The cytochrome b data
demonstrate a recent divergence of M. dogramacii
(2n = 48) from M. guentheri (2n = 54). The NJ, ML
and Bayesian trees indicated a paraphyletic relationship
(Figs. 2 and 3) but the MP analyses supported reciprocal
monophyly (Fig. 1). However, the situation is
more complex in that our M. guentheri specimens from
Syria and Israel may be more appropriately considered
M. irani (cf. Kefeliog˘lu and Krysˇtufek, 1999; Krysˇtufek
and Kefeliog˘lu, 2001). The distribution of M. irani remains,
however, uncertain (Krysˇtufek and Kefeliog˘lu,
2001; Mitchell-Jones et al., 1999). Overall, it is clear
that a much more detailed molecular investigation of
the phylogenetic relationships among social voles is
warranted.
4.5. Subgenus Terricola (ground voles)
The cytochrome b tree supports Brunet-Lecomte and
Chaline (1992), Chaline and Graf (1988), and Zagorodnyuk
(1989) in the separation of pitymyine forms
into Nearctic (Pitymys) and Palearctic (Terricola) components
(Figs. 1–3). The monophyly of the subgenus
Terricola was recently questioned by Krysˇtufek et al.
(1996) who claimed that it constituted an artificial group
of unrelated convergently evolved species with no shared
apomorphies. However, our molecular data demonstrate
that Terricola species do share a common ancestor.
The Bayesian analyses yielded strong supported
for Terricola (Fig. 3), but the bootstrap support was relatively
low (Figs. 1 and 2). Removal of M. majori from
the analyses, however, increased the bootstrap values
drastically for the whole group as well as the two subgroups
within Terricola (Fig. 2). Analysis of additional
M. majori sequences might stabilize the group, but the
results may also indicate that the Asian endemic M. majori
represents a separate evolutionary lineage (see
below).
The cytochrome b data strongly support two monophyletic
species groups within Terricola: one subgroup
with M. subterraneus and M. daghestanicus, species with
ranges extending to Asia Minor, and the other subgroup
consisting of the European endemics. The cytochrome b
phylogeny does not agree fully with the prevailing species
groups within Terricola (cf. Table 1). However,
extensive karyotypic and morphological polymorphism
in Terricola has spawned many alternative classifications
(e.g., Chaline et al., 1999; Kratochvı´l and Kra´l, 1974)
and our data imply that yet another systematic revision
is necessary.
Microtus majori, M. subterraneus, and M. daghestanicus
are believed to have diverged recently (reviewed in
Machola´n et al., 2001). Our data fully support a sister
relationship between M. subterraneus and M. daghestanicus
but the position of M. majori remains uncertain
(see above)––even in the Bayesian analyses. This result
is somewhat surprising since Machola´n et al. (2001)
found a close allozymic relationship between M. majori
and M. subterraneus. Our results are, however, in accordance
with Zagorodnyuk (1990) who considered M. majori
the sole member of its own species group.
SpeciesÕ relationships in the European subgroup of
Terricola are poorly resolved, although, again, the
Bayesian analyses support some groupings not recognized
by MP or NJ bootstrap analysis (Fig. 3). Since
these species are so closely related, the hard polytomy
observed cannot be ascribed to saturation in cytochrome
b. Instead, strong bootstrap support above
and below unresolved polytomies indicate a rapid radiation
involving nearly simultaneous diversification of
many lineages (cf. Conroy and Cook, 1999; Lessa and
Cook, 1998). Such a surge of speciation could have occurred
during a single or, more likely, a few consecutive
glacial periods by geographic isolation of small and
genetically differentiated populations in different glacial
refugia as suggested by Chaline (1987). The relative
importance of Mediterranean peninsulas as opposed to
more northern mountain areas as ‘‘speciation traps’’ deserves
further attention (cf. Chaline, 1987; Bilton et al.,
1998; Martı´nkova´ and Dudich, 2003). In this context it
is noteworthy that all taxa in this subgroup except M.
savii (see below) seem to harbor little intra-specific
variation.
4.6. Subgenera Pallasiinus, Alexandromys, and Volemys
(the ‘‘Asian’’ lineage)
The taxonomic validity of the subgenera Pallasiinus,
Alexandromys, and Volemys was not supported by our
cytochrome b analysis since their representatives formed
a strongly supported monophyletic group. This ‘‘Asian’’
group was previously described by Conroy and Cook
(2000a). Our addition of within-species sequences and
more species to the Microtus tree has significantly increased
the support for this lineage. The Japanese M.
montebelli and Taiwanese M. kikuchii are sister species
to the Holarctic M. oeconomus and constitute examples
of allopatric speciation on islands. The results are supported
by chromosome data. Thus, pairing of the X
and Y chromosome in Microtus meiosis seems to be a
rare lineage-specific phenomenon that can be used in
reconstructing systematic relationships in the genus
(Mekada et al., 2002; Megı´as-Nogales et al., 2002). To
658 M. Jaarola et al. / Molecular Phylogenetics and Evolution 33 (2004) 647–663
Paper 2.1.1 63
date, X–Y pairing is only reported for the three species
forming the Asian lineage (see Mekada et al., 2002) as
well as Chionomys nivalis (Megı´as-Nogales et al., 2002)
and two species representing the subgenera Lasiopodomys
and Neodon not included here (Gu et al., 1999;
Mekada et al., 2002).
4.7. Subgenera Agricola, Neodon, and Stenocranius
Microtus agrestis, M. cabrerae, M. juldaschi, and M.
gregalis were placed basal in the phylogenetic analyses
and did not form significant associations with other species.
For example, the association of M. cabrerae with
the Nearctic species was not supported by either bootstrapping
or posterior probabilities (Figs. 1–3) but most
probably due to long-branch attraction (e.g., Hendy and
Penny, 1989). These four, basal species represent three
subgenera––Agricola, Neodon, and Stenocranius––characterized
by few and ancestral species. Our data suggest
that M. agrestis and M. cabrerae should not both be
placed in the subgenus Agricola since they do not show
any sister relationship. Especially the M. cabrerae lineage
seems to be either very old or has undergone accelerated
evolution in cytochrome b. Actually, M. cabrerae
displays morphological characters that are archaic (Gromov
and Polyakov, 1992), and Chaline (1972) described
a new subgenus, Iberomys, for the fossil vole Microtus
(Iberomys) brecciensis, a direct ancestor of M. cabrerae.
Altogether, our data strongly support the classification
of M. cabrerae in the separate subgenus Iberomys.
Microtus juldaschi belongs to the subgenus Neodon
also containing M. irene and M. sikimensis (Musser
and Carleton, 1993; Zagorodnyuk, 1990). Neodon as
well as the subgenera Blanfordimys and Phaiomys (not
analyzed) are considered old Pleistocene relicts that
probably descended directly from the Allophaiomys
stock (Nadachowski and Garapich, 1998; Nadachowski
and Zagorodnyuk, 1996). Consequently, the position of
M. juldaschi in the phylogenetic tree is expected to be
basal, in line with our result.
Both our data and those of Conroy and Cook (2000a)
indicate that the subgenus Stenocranius is a polyphyletic
and artificial group as the Asian M. gregalis does not
cluster with the North American M. miurus and M.
abbreviatus. Nor does M. gregalis cluster with M. middendorffi
as suggested by the alternative classification
of Zagorodnyuk (1990). Thus, the morphological similarity
of these species is most probably due to adaptive
convergence as suggested by Chaline et al. (1999) and
Conroy and Cook (2000a). M. gregalis is by far the most
divergent Microtus species in our cytochrome b data set
and its position is unclear even in relation to Chionomys
(cf. Figs. 1–3). According to palaeontological data, M.
gregalis represents an early split from the Allophaiomys
stock (Rekovets and Nadachowski, 1995). The support
for Microtus as a bona fide taxonomic group is mainly
influenced by the instability of M. gregalis. Thus, in order
to fully evaluate monophyly of the genus Microtus,
the position of M. gregalis needs to be determined. In
addition, the position of the genus Arvicola needs to
be evaluated.
4.8. Species limits in Microtus?
Our data confirm that the genus Microtus contains
many closely related species as well as many species that
are characterized by extensive intra-specific variation
(Fig. 2). Consequently, there is an overlap between inter-
and intra-specific cytochrome b distances around
4–8%, similar to what has been described in other mammals
(e.g., Avise, 2000; Bradley and Baker, 2001). These
data corroborate the notion that genetic distances and/
or reciprocal monophyly in mtDNA are highly uncertain
criteria for delimiting closely related species
(Bradley and Baker, 2001; Hudson and Coyne, 2002;
Hudson and Turelli, 2003; Nichols, 2001; Rosenberg
and Nordborg, 2002). Moreover, the overlap in cytochrome
b divergence between and within currently
recognized species demonstrates that speciation is an
ongoing process in Microtus and that many taxa and
species groups offer a huge potential for molecular, evolutionary
research, particularly with regards to phylogeography
and speciation (cf. Barraclough and Nee, 2001).
Examples of sister species that show low cytochrome
b divergence, but do represent widely accepted species
include M. duodecimcostatus––lusitanicus and M. arvalis––rossiaemeridionalis
(Fig. 1). The cytochrome b distance
of 4–5% between M. duodecimcostatus and M.
lusitanicus is among the lowest recorded for Microtus
species, the exception being M. miurus––M. abbreviatus
that differ by only 1.5%. However, the latter two taxa
are probably conspecific (Conroy and Cook, 2000a).
The taxonomic rank of M. lusitanicus has varied in recent
years but it is now considered a species (reviewed
in Spitzenberger et al., 2000), its ranking validated by
sterility of male F1 hybrids between M. duodecimcostatus
and M. lusitanicus.
The sibling species M. arvalis (2n = 46) and M. rossiaemeridionalis
(2n = 54) are closely related, demonstrating
a divergence of only 6–8% in cytochrome b.
However, hybrids between the two taxa are sterile (Meier
et al., 1985). The status of the eastern ÔobscurusÕ taxon in
the arvalis group is unclear. While some authors regard
this taxon as a species, M. obscurus, separate from the
western M. arvalis, other authors describe the taxa as
two karyotypic forms, M. arvalis ÔobscurusÕ and M. arvalis
ÔarvalisÕ. We have used the latter classification since
hybridization between the two taxa appears to occur in
the wild (Bulatova, unpublished) and hybridization studies
have shown that F1, F2 and subsequent backcrosses
are fertile, albeit with lowered fertility (Malygin and
Panteleichuk, 2003). The cytochrome b data imply a reM.
Jaarola et al. / Molecular Phylogenetics and Evolution 33 (2004) 647–663 659
64 Paper 2.1.1
cent split between Ôarvalis Õ and ÔobscurusÕ; the divergence
is only 2–4%. For references and a recent discussion
based on cytochrome b data see Haynes et al. (2003).
Other species with unclear status accompanied by low
cytochrome b divergence include M. liechtensteini and
M. atticus. M. liechtensteini seems to represent a cytochrome
b lineage distinct from M. multiplex (Figs. 1–
3). Our observation is consistent with the results obtained
by Haring et al. (2001) who conducted a more
extensive study of the Alpine voles of the M. multiplex
complex using mitochondrial D-loop sequences. However,
additional data on hybridization and fertility of
offspring in the M. multiplex complex are needed to fully
evaluate the taxonomic ranking of M. liechtensteini. The
M. thomasi samples 1–3 represent the karyotype form
ÔatticusÕ previously ascribed to a separate species, M.
atticus. Our data corroborate the present ranking of
ÔatticusÕ as a mere karyotype form of M. thomasi (references
in Tsekoura et al., 2002).
Species with high intra-specific cytochrome b variation
include M. savii, M. agrestis, M. daghestanicus,
M. oeconomus, M. subterraneus, and C. nivalis (Fig. 1).
The subspecies M. savii savii in northern and central
Italy and M. s. brachycercus in southern Italy differ in
karyotype (Galleni et al., 1992) and because the male
F1 hybrids are sterile, Galleni et al. (1994) suggested that
the two subspecies should be elevated to species rank.
Our data set is limited, but indicates a recent mtDNA
split between southern and central-northern M. savii,
with cytochrome b divergences of 4–5%. Recent molecular
data on M. agrestis demonstrate a south-north split
in Europe, with a net divergence of 5.2% in cytochrome
b and 0.7% in the X and Y chromosome, all three genealogies
exhibiting reciprocal monophyly. These data
suggest that the southern M. agrestis represent a new,
morphologically and karyotypically cryptic species (Jaarola
and Searle, 2002, 2004; Hellborg, 2004).
Another species with high intra-specific variation is
M. daghestanicus that exhibits a highly divergent haplotype
from Georgia differing by 4% from the two Turkish
haplotypes. Georgian specimens are sometimes ascribed
to M. nasarovi, but M. daghestanicus and M. nasarovi
differ in karyotype (2n = 52/54 and 38, respectively),
habitat preference and distribution range (Bukhnikashvili
and Kandaurov, 2002). Since the Georgian M. daghestanicus
originated from the eastern part of the country,
i.e. outside the distribution range of M. nasarovi, and the
Turkish specimens both carried a 2n = 54 karyotype
(Machola´n, pers. comm.), the large cytochrome b divergence
observed between Georgian and Turkish M. daghestanicus
is likely to reflect intra-specific variation.
Microtus oeconomus is divided into four divergent
cytochrome b lineages differing by net distances up to
3.5% (Brunhoff et al., 2003). The largely allopatric distribution
of these lineages and inferences on their late Quaternary
history is presented in Brunhoff et al. (2003,
submitted). Yet another species that is characterized
by very high cytochrome b distances between haplotypes
is M. subterraneus; one of the haplotypes from Turkey
differed from the other by 6–7%. Finally, C. nivalis, a
species with fragmented distribution over Europe and
Middle East mountain ranges, exhibited intra-specific
distances up to 4%. Studies of morphological characters
(Krysˇtufek, 1999; Nadachowski, 1991) and allozymes
(Filippucci et al., 1991) have also demonstrated much
diversity in this species.
5. Conclusions
ÔSmall mammalsÕ are far more speciose than their larger
relatives, and it is interesting to speculate on the reasons
for this (Searle, 1996). However, before the present
study, there have been few detailed attempts to obtain
molecular phylogenies for particularly species-rich genera
of small mammals. Our study, through its coverage
of almost all European and North American species, as
well as many from Asia, provides a fascinating insight
into Microtus, the most speciose genus of Arvicoline rodents
and one of the most speciose genera of all mammals
(Nowak, 1999). The cytochrome b phylogeny
demonstrates speciesÕ radiations at a variety of temporal
and spatial scales. In a temporal sense, an early rapid
radiation about 2 Mya appears to have generated the
major subgenera, which then subsequently radiated further
to generate the variety of extant species. Moreover,
the currently recognized species are not static forms, but
are clearly differentiating and contain cryptic entities
that may in many cases best be considered species
(e.g., within M. agrestis: Jaarola and Searle, 2004). In
a spatial sense, the radiations of subgenera and equivalent
groupings of Microtus have occurred in different
geographic areas as evidenced by the current geographic
ranges of major cytochrome b lineages.
Now that a molecular phylogeny of Microtus is available,
there is an opportunity to use it to understand the
ecological and historical circumstances that lead to speciation
of small mammals (cf. Searle, 1996). The phylogeny
can also be used for a range of future comparative
studies in ecology, behaviour, physiology, parasitology
etc. It will, for instance, be possible to use the phylogeny
for coevolutionary studies (e.g., Microtus and their parasites;
Wickstro¨m, 2004) and for analyses of trait evolution
(e.g., mating systems).
A further, practical achievement of this study is to resolve
a variety of long-standing taxonomic uncertainties
in the genus Microtus since the molecular results are
clear-cut and provide an objective basis for taxonomic
revision. Our data show that there is a justification for
subdividing the genus into subgenera, as proposed by
previous workers, but there are also groupings of species
within subgenera (Figs. 1–3) and an appropriate nomen-
660 M. Jaarola et al. / Molecular Phylogenetics and Evolution 33 (2004) 647–663
Paper 2.1.1 65
clature will need to be developed, building on that previously
established.
Acknowledgments
We are grateful to M. Akhverdyan, P. Basset, P.
Benda, F. Catzeflis, V. Fedorov, D. Frynta, H. Hauffe,
S. Haynes, V. Haukisalmi, H. Henttonen, R.A. Ims,
B. Krysˇtufek, M. Machola´n, J. Michavx, P. Munclinger,
F. Poitevin, A. Polyakov, K. Sperling, C. Tez, N. Yoccoz,
and D. Ziak for providing samples. We also want
to thank C. Conroy, K. Fredga, F. Golenishchev, and
I. Hora´cˇek for valuable advice, discussions and comments
on a previous version of the manuscript and
R.S. Hoffmann and two anonymous reviewers for comments
on the manuscript. Financial support was received
from the Swedish Institute (Guest Scholarship
to I.G.), the Swedish Research Council, the NilssonEhle
Foundation, the Erik Philip-So¨rensenÕs Foundation,
the Grant Agency of the Czech Republic (GACˇ R
206/01/0562), and the Ministry of Education, Youth
and Sport of the Czech Republic (MSMT 311100004)
and from the Russian Foundation of Basic Research
and INTAS (01-2163).
References
Akhverdyan, M.R., Lyapunova, E.A., Vorontsov, N.N., Teslenko,
S.V., 1999. Intrapopulation autosomal polymorphism in the
common vole Microtus arvalis of the Transcaucasian region. Russ.
J. Genet. 35, 1452–1463.
Avise, J.C., 2000. Phylogeography: the History and Formation of
Species. Harvard University Press, Cambridge, MA.
Barraclough, T.G., Nee, S., 2001. Phylogenetics and speciation. Trends
Ecol. Evol. 16, 391–399.
Baskevich, M.I., 1996. On the karyological differentiation in Caucasian
populations of common vole (Rodentia, Cricetidae, Microtus).
Zool. Zh. 75, 297–308 (in Russian with English summary).
Bensasson, D., Zhang, D.-X., Hartl, D.L., Hewitt, G.M., 2001.
Mitochondrial pseudogenes: evolutionÕs misplaced witnesses.
Trends Ecol. Evol. 16, 314–321.
Bilton, D.T., Mirol, P.M., Mascheretti, S., Fredga, K., Zima, J.,
Searle, J.B., 1998. Mediterranean Europe as an area of endemism
for small mammals rather than a source for northwards postglacial
colonization. Proc. R. Soc. Lond. Ser. B 265, 1219–1226.
Bradley, R.D., Baker, R.J., 2001. A test of the genetic species concept:
cytochrome b sequences and mammals. J. Mammal. 82, 960–973.
Brunet-Lecomte, P., Chaline, J., 1990. Relations phyloge´ne´tiques et
e´volution des campagnols souterrains dÕEurope (Terricola, Arvicolidae,
Rodentia). C.R. Acad. Sci. Paris, Se´rie II 311, 745–750.
Brunet-Lecomte, P., Chaline, J., 1991. Morphological evolution and
phylogenetic relationships of the European ground voles (Arvicolinae,
Rodentia). Lethaia 24, 45–53.
Brunet-Lecomte, P., Chaline, J., 1992. Morphological convergences
versus biochemical divergences in the holarctic ground voles:
Terricola and Pitymys (Arvicolidae, Rodentia). N. Jb. Geol.
Pala¨ont. Mh. 12, 721–734.
Brunhoff, C., Galbreath, K.E., Fedorov, V.B., Cook, J.A., Jaarola,
M., 2003. Holarctic phylogeography of the root vole (Microtus
oeconomus): implications for late Quaternary biogeography of high
latitudes. Mol. Ecol. 12, 957–968.
Bukhnikashvili, A., Kandaurov, A., 2002. The annotated list of
mammals of Georgia. Proc. Inst. Zool. Acad. Sci. Georgia,
Metsniereba, Tbilisi 21, 319–340.
Chaline, J., 1972. Les rongeurs du Ple´istoce`ne moyen et supe´rieur de
France. Cahiers Pale´ont., CNRS, Paris, pp. 1–410.
Chaline, J., 1987. Arvicolid data (Arvicolidae, Rodentia) and evolutionary
concepts. Evol. Biol. 21, 237–310.
Chaline, J., Brunet-Lecomte, P., Montuire, S., Viriot, L., Courant, F.,
1999. Anatomy of the arvicoline radiation (Rodentia): palaeogeographical,
palaeoecological history and evolutionary data. Ann.
Zool. Fenn. 36, 239–267.
Chaline, J., Graf, J.-D., 1988. Phylogeny of the Arvicolidae (Rodentia):
biochemical and paleontological evidence. J. Mammal. 69, 22–
33.
Conroy, C.J., Cook, J.A., 1999. MtDNA evidence for repeated pulses
of speciation within arvicoline and murid rodents. J. Mamm. Evol.
6, 221–245.
Conroy, C.J., Cook, J.A., 2000a. Molecular systematics of a holarctic
rodent (Microtus: Muridae). J. Mammal. 81, 344–359.
Conroy, C.J., Cook, J.A., 2000b. Phylogeography of a post-glacial
colonizer: Microtus longicaudus (Rodentia: Muridae). Mol. Ecol. 9,
165–175.
Conroy, C.J., Hortelano, Y., Cervantes, F.A., Cook, J.A., 2001. The
phylogenetic position of southern relictual species of Microtus
(Muridae: Rodentia) in North America. Mamm. Biol. 66, 332–344.
DeBry, R.W., 1992. Biogeography of New World taiga-dwelling
Microtus (Mammalia: Arvicolidae): a hypothesis test that accounts
for phylogenetic uncertainty. Evolution 46, 1347–1357.
DeWoody, A.J., Chesser, R.K., Baker, R.J., 1999. A translocated
mitochondrial cytochrome b pseudogene in voles (Rodentia:
Microtus). J. Mol. Evol. 48, 380–382.
Erixon, P., Svennblad, B., Britton, T., Oxelman, B., 2003. Reliability
of Bayesian posterior probabilities and bootstrap frequencies in
phylogenetics. Syst. Biol. 52, 665–673.
Felsenstein, J., 1988. Phylogenies from molecular sequences—inference
and reliability. Annu. Rev. Genet. 22, 521–565.
Filippucci, M.G., Fadda, V., Krysˇtufek, B., Simson, S., Amori, G.,
1991. Allozyme variation and differentiation in Chionomys nivalis
(Martins, 1842). Acta Theriol. 36, 47–62.
Fumagalli, L., Taberlet, P., Stewart, D.T., Gielly, L., Hausser, J.,
Vogel, P., 1999. Molecular phylogeny and evolution of Sorex
shrews (Soricidae: Insectivora) inferred from mitochondrial DNA
sequence data. Mol. Phylogenet. Evol. 11, 222–235.
Galleni, L., Stanyon, R., Tellini, A., Giordano, G., Santini, L., 1992.
Karyology of the Savi pine vole, Microtus savii (De Se´lysLongchamps,
1838) (Rodentia, Arvicolidae): G-, C-, DA/DAPI-,
and AluI-bands. Cytogenet. Cell Genet. 59, 290–292.
Galleni, L., Tellini, A., Stanyon, R., Cicalo`, A., Santini, L., 1994.
Taxonomy of Microtus savii (Rodentia, Arvicolidae) in Italy:
cytogenetic and hybridization data. J. Mammal. 75, 1040–
1044.
Garapich, A., Nadachowski, A., 1996. A contribution to the origin of
Allophaiomys (Arvicolidae, Rodentia) in Central Europe: the
relationships between Mimomys and Allophaiomys from Kamyk
(Poland). Acta Zool. Cracov. 39, 179–184.
Getz, L.L., 1985. Habitats. In: Tamarin, R.H.(Ed.), Biology of New
World Microtus. American Society of Mammalogists. Special
Publication No. 8, pp. 286–309.
Gill, A., Petrov, B., Zivkovic, S., Rimsa, D., 1987. Biochemical
comparison in Yugoslavian rodents of the families Arvicolidae and
Muridae. Z. Sa¨ugetierk. 52, 247–256.
Golenishchev, F.N., Malikov, V.G., Nazari, F., Vaziri, A.S., Sablina,
O.V., Polyakov, A.V., 2003. New species of guentheri group
(Rodentia, Arvicolinae, Microtus) from Iran. Russ. J. Theriol. 1,
117–123.
M. Jaarola et al. / Molecular Phylogenetics and Evolution 33 (2004) 647–663 661
66 Paper 2.1.1
Graf, J.-D., 1982. Ge´ne´tique biochimique, zooge´ographie et taxonomie
des Arvicolidae (Mammalia, Rodentia). Rev. Suisse Zool. 89, 749–
787.
Gromov, I.M., Polyakov, I.Y., 1992. Fauna of the USSR, vol. 3, Voles
(Microtinae). Brill Publishing Company, Leiden.
Gu, W., Wang, T., Zhu, B., 1999. Study on the morphology of sex
chromosomes pairing of the synaptonemal complex in Mandarin
vole (Microtus mandarinus). Acta Theriol. Sin. 19, 150–154 (in
Chinese).
Haring, E., Herzig-Straschil, B., Spitzenberger, F., 2001. Phylogenetic
analysis of Alpine voles of the Microtus multiplex complex using the
mitochondrial control region. J. Zool. Syst. Evol. Res. 38, 231–238.
Haynes, S., Jaarola, M., Searle, J.B., 2003. Phylogeography of the
common vole (Microtus arvalis) with particular emphasis on the
colonization of the Orkney archipelago. Mol. Ecol. 12, 951–956.
Hellborg, L., 2004. Evolutionary studies of the mammalian Y
chromosome. Ph.D, thesis, Uppsala University.
Hendy, M.D., Penny, D., 1989. A framework for the quantitative
study of evolutionary trees. Syst. Zool. 38, 297–309.
Hoffmann, R.S., Koeppl, J.W., 1985. Zoogeography. In: Tamarin,
R.H. (Ed.), Biology of New World Microtus. American Society of
Mammalogists. Special Publication No. 8, pp. 84–0115.
Hudson, R.R., Coyne, J.A., 2002. Mathematical consequences of the
genealogical species concept. Evolution 56, 1557–1565.
Hudson, R.R., Turelli, M., 2003. Stochasticity overrules the threetimes
rule: genetic drift, genetic draft, and coalescence times for
nuclear loci versus mitochondrial DNA. Evolution 57, 182–190.
Huelsenbeck, J.P., Larget, B., Miller, R.E., Ronquist, F., 2002.
Potential applications and pitfalls of Bayesian inference of
phylogeny. Syst. Biol. 51, 673–688.
Huelsenbeck, J.P., Ronquist, F., Nielsen, R., Bollback, J.P., 2001.
Bayesian inference of phylogeny and its impact on evolutionary
biology. Science 294, 2310–2314.
Hutterer, R., 2001. Bavarian pine vole still alive. Oryx 35, 185.
Irwin, D.M., Kocher, T.D., Wilson, A.C., 1991. Evolution of the
cytochrome b gene of mammals. J. Mol. Evol. 32, 128–144.
Jaarola, M., Searle, J.B., 2002. Phylogeography of field voles (Microtus
agrestis) in Eurasia inferred from mitochondrial DNA sequences.
Mol. Ecol. 11, 2613–2621.
Jaarola, M., Searle, J.B., 2004. A highly divergent mitochondrial DNA
lineage of Microtus agrestis in southern Europe. Heredity 92, 229–
234.
Jaarola, M., Tegelstro¨m, H., 1995. Colonization history of north
European field voles (Microtus agrestis) revealed by mitochondrial
DNA. Mol. Ecol. 4, 299–310.
Jukes, T.H., Cantor, C.R., 1969. Evolution of protein molecules. In:
Munro, H.N. (Ed.), Mammalian Protein Metabolism. Academic
Press, New York, pp. 21–132.
Kefeliog˘lu, H., Krysˇtufek, B., 1999. The taxonomy of Microtus socialis
group (Rodentia: Microtinae) in Turkey, with the description of a
new species. J. Nat. Hist. 33, 289–303.
Kocher, T.D., Thomas, W.K., Meyer, S.A., Edwards, S.V., Pa¨a¨bo, S.,
Villablanca, F.X., Wilson, A.C., 1989. Dynamics of mitochondrial
DNA evolution in animals: Amplification and sequencing with
conserved primers. Proc. Natl. Acad. Sci. USA 86, 6196–6200.
Kratochvı´l, J., Kra´l, B., 1974. Karyotypes and relationships of
Palaearctic 54-chromosome Pitymys species (Microtidae, Rodentia).
Zool. Listy 23, 289–302.
Krysˇtufek, B., 1999. Snow voles, genus Chionomys, of Turkey.
Mammalia 63, 323–339.
Krysˇtufek, B., Griffiths, H.I., Vohralı´k, V., 1996. The status and use of
Terricola Fatio, 1867 in the taxonomy of Palaearctic pine volesÕ
(Pitymys) (Rodentia, Arvicolinae). Bull. Inst. R. Sci. Nat. Belg.
Biol. 66, 237–240.
Krysˇtufek, B., Kefeliog˘lu, H., 2001. Redescription of Microtus irani,
the species limits and a new social vole from Turkey. Bonn. Zool.
Beitr. 50, 1–14.
Leache´, A.D., Reeder, T.W., 2002. Molecular systematics of the
eastern fence lizard (Sceloporus undulatus): A comparison of
parsimony, likelihood and Bayesian approaches. Syst. Biol. 51,
44–68.
Lessa, E.P., Cook, J.A., 1998. The molecular phylogenetics of tucotucos
(genus Ctenomys, Rodentia: Octodontidae) suggests an early
burst of speciation. Mol. Phylogenet. Evol. 9, 88–99.
Lin, Y.-H., Waddell, P.J., Penny, D., 2002. Pika and vole mitochondrial
genomes increase support for both rodent monophyly and
Glires. Gene 294, 119–129.
Machola´n, M., Filippucci, M.G., Zima, J., 2001. Genetic variation and
zoogeography of pine voles of the Microtus subterraneus/majori
group in Europe and Asia Minor. J. Zool. 255, 31–42.
Maddison, D.R., Maddison, W.P., 2000. MacClade 4: Analysis of
phylogeny and character evolution, Version 4.0. Sinauer Associates,
Sunderland, MA.
Malygin, V.M., Panteleichuk, T.M.S.L., 2003. Efficiency of reproductive
isolation mechanisms in six species of common voles (Microtus,
Rodentia). In: Kryokov, A.P., Yakimenko, L.V. (Eds.), Problems
of Evolution, vol. V. Russian Academy of Sciences, Far East
Division, Vladivostok, Dalnauka, pp. 198–206 (in Russian with
English summary).
Martı´nkova´, N., Dudich, A., 2003. The fragmented distribution range
of Microtus tatricus and its evolutionary implications. Folia Zool.
52, 11–22.
Maryama, T., Imai, H.T., 1981. Evolutionary rate of the mammalian
karyotype. J. Theor. Biol. 90, 111–121.
Mazurok, N.A., Rubtsova, N.V., Isaenko, A.A., Pavlova, M.E.,
Slobodyanyuk, S.Y., Nesterova, T.B., Zakian, S.M., 2001. Comparative
chromosome and mitochondrial DNA analyses and
phylogenetic relationships within common voles (Microtus, Arvicolidae).
Chrom. Res. 9, 107–120.
Megı´as-Nogales, B., Marchal, J.A., Acosta, M.J., Bullejos, M., Dı´az
de la Guardia, R., Sa´nchez, A., 2002. Sex chromosome pairing in
two Arvicolidae species: Microtus nivalis and Arvicola sapidus.
Hereditas 138, 114–121.
McClellan, D.A., McCracken, K.G., 2001. Estimating the influence of
selection on the variable amino acid sites of the cytochrome b
protein functional domains. Mol. Biol. Evol. 18, 917–925.
Meier, M.N., Golenishchev, F.N., Radjabli, S.I., Sablina, O.V., 1996.
Voles (subgenus Microtus Schrank) of Russia and adjacent
territories. Proc. Zool. Inst. Russ. Acad. Sci. 232, 1–320 (in
Russian).
Meier, M.N., Radjabli, S.I., Bulatova, N. Sh., Golenishchev, F.N.,
1985. Karyological peculiarities and probable relations of common
voles of the group ÔarvalisÕ (Rodentia, Cricetidae, Microtus). Zool.
Zh. 64, 417–428 (in Russian with English summary).
Mekada, K., Koyasu, K., Harada, M., Narita, Y., Shrestha, K.C.,
Oda, S.-I., 2002. Karyotype and X–Y chromosome pairing in the
Sikkim vole (Microtus (Neodon) sikimensis). J. Zool. 257, 417–
423.
Mezhzherin, S.V., Morozov-Leonov, S.Yu., Kuznetsova, I.A., 1995.
Biochemical variation and genetic divergence in Palearctic voles
(Arvicolidae): subgenus Terricola, true lemmings Lemmus Link
1795, pied lemmings Dicrostonyx Gloger 1841, steppe lemmings
Lagurus Gloger 1842, mole voles Ellobius Fischer von Waldheim
1814. Genetika 31, 788–797 (in Russian with English summary).
Mezhzherin, S.V., Zykov, A.E., Morozov-Leonov, S.Y.u., 1993.
Biochemical variation and genetic divergence in Palearctic voles
(Arvicolidae): meadow voles Microtus Schrank 1798, snow voles
Chionomys Miller 1908, water voles Arvicola Lacepede 1799.
Genetika 29, 28–41 (in Russian with English summary).
Mirol, P.M., Mascheretti, S., Searle, J.B., 2000. Multiple nuclear
pseudogenes of mitochondrial cytochrome b in Ctenomys (Caviomorpha,
Rodentia) with either great similarity to or high divergence
from the true mitochondrial sequence. Heredity 84, 538–
547.
662 M. Jaarola et al. / Molecular Phylogenetics and Evolution 33 (2004) 647–663
Paper 2.1.1 67
Mitchell-Jones, A.J., Amori, G., Bogdanowicz, W., Krysˇtufek, B.,
Reijnders, P.J.H., Spitzenberger, F., Stubbe, M., Thissen, J.B.M.,
Vohralı´k, V., Zima, J., 1999. The Atlas of European Mammals.
Poyser, London.
Modi, W.S., 1987. Phylogenetic analyses of chromosomal banding
patterns among the Nearctic Arvicolidae (Mammalia: Rodentia).
Syst. Zool. 36, 109–136.
Musser, G.G., Carleton, M.D., 1993. Family Muridae. In: Wilson,
D.E., Reeder, D.M. (Eds.), Mammal Species of the World: A
Taxonomic and Geographic Reference, second ed. Smithsonian
Institution Press, Washington, pp. 510–756.
Nadachowski, A., 1991. Systematics, geographic variation, and
evolution of snow voles (Chionomys) based on dental characters.
Acta Theriol. 36, 1–45.
Nadachowski, A., Garapich, A., 1998. Allophaiomys evolutionary
stage in extant Microtus. Paludicola 2, 91–94.
Nadachowski, A., Zagorodnyuk, I., 1996. Recent Allophaiomys-like
species in the Palaearctic: Pleistocene relicts or a return to an initial
type? Acta Zool. Cracov. 39, 387–394.
Nei, M., 1987. Molecular Evolutionary Genetics. Columbia University
Press, New York.
Nichols, R., 2001. Gene trees and species trees are not the same.
Trends Ecol. Evol. 16, 358–364.
Nowak, R.M., 1999. WalkerÕs Mammals of the World, sixth ed. The
Johns Hopkins University Press, Baltimore, ML.
Orlov, V.N., Yatsenko, V.N., Malygin, V.M., 1983. Karyotype
homology and species phylogeny in a group of field mice
(Cricetidae, Rodentia). Doklady Biol. Sci. 269, 217–219.
Panteleyev, P.A., 1998. The Rodents of the Palaearctic Fauna:
Composition and Areas. Russian Academy of Sciences, Moscow.
Posada, D., Crandall, K.A., 1998. Modeltest: testing the model of
DNA substitution. Bioinformatics 14, 817–818.
Potapov, S.G., Orlov, V.N., Kovalskaya, Yu.M., Malygin, V., M,
Ryskov, A.P., 1999. Genetic differentiation in the voles of the tribe
Arvicolini (Cricetidae, Rodentia) using DNA taxonprint and
RAPD-PCR. Russ. J. Genetics 35, 403–410.
Rabeder, G., 1986. Herkunft und fru¨he Evolution der Gattung
Microtus (Arvicolidae, Rodentia). Z. Sa¨ugetierk. 51, 350–367.
Reed, R.D., Sperling, F.A.H., 1999. Interaction of process partitions
in phylogenetic analysis: an example from the swallowtail butterfly
genus Papilio. Mol. Biol. Evol. 16, 286–297.
Rekovets, L., Nadachowski, A., 1995. Pleistocene voles (Arvicolidae)
of the Ukraine. Paleont. Evolutio´ 28-29, 145–245.
Richmond, G.M., 1996. The INQUA-approved provisional LowerMiddle
Pleistocene boundary. In: Turner, C. (Ed.), The Early
Middle Pleistocene in Europe. Balkema, Rotterdam.
Ronquist, F., Huelsenbeck, J.P., 2003. MrBayes 3: Bayesian phylogenetic
inference under mixed models. Bioinformatics 19, 1572–1574.
Rosenberg, N.A., Nordborg, M., 2002. Genealogical trees, coalescent
theory and the analysis of genetic polymorphisms. Nat. Rev.
Genet. 3, 380–390.
Searle, J.B., 1996. Speciation in small mammals. Symp. Zool. Soc.
Lond. 69, 143–156.
Spitzenberger, F., Brunet-Lecomte, P., Nadachowski, A., Bauer, K.,
2000. Comparative morphometrics of the first lower molar in
Microtus (Terricola) cf. liechtensteini of the Eastern Alps. Acta
Theriol. 45, 471–483.
Suchentrunk, F., Markov, G., Haiden, A., 1998. On gene pool
divergence of the two karyotypically distinct sibling vole species
Microtus arvalis and M. rossiaemeridionalis (Arvicolidae, Rodentia).
Folia Zool. 47, 103–114.
Suzuki, Y., Glazko, G.V., Nei, M., 2002. Overcredibility of molecular
phylogenies obtained by Bayesian phylogenetics. Proc. Natl. Acad.
Sci. USA 99, 16138–16143.
Swofford, D.L., 2002. PAUP*. Phylogenetic Analysis Using Parsimony
(*and Other Methods), Version 4. Sinauer Associates,
Sunderland, MA.
Tsekoura, N., Fraguedakis-Tsolis, S., Chondropoulos, B., Markakis,
G., 2002. Morphometric and allozyme variation in central and
southern Greek populations of Microtus (Terricola) thomasi. Acta
Theriol. 47, 137–149.
Wickstro¨m, L.M., 2004. Phylogeny, phyletic coevolution and phylogeography
of anoplocephaline cestodes in mammals. Ph.D. thesis,
University of Helsinki, Finnish Forest Research Institute, Research
Papers 918.
Yang, Z., 1994. Estimating the pattern of nucleotide substitutions. J.
Mol. Evol. 39, 105–111.
Yang, Z., Yoder, A.D., 1999. Estimation of the transition/transversion
rate bias and species sampling. J. Mol. Evol. 48, 274–283.
Yigit, N., Colak, E., 2002. On the distribution and taxonomic status of
Microtus guentheri (Danford and Alston, 1880) and Microtus lydius
Blackler, 1916 (Mammalia: Rodentia) in Turkey. Turk. J. Zool. 26,
197–204.
Zagorodnyuk, I.V., 1989. Taksonomiya, rasprostranenie i morfologicheskaya
izmenchivostÕpolevok roda Terricola vostochnoi Evropy.
[Taxonomy, distribution and morphological variation of the
Terricola voles in East Europe]. Vestnik Zool. 5, 3–14 (in Russian
with English summary).
Zagorodnyuk, I.V., 1990. Kariotipicheskaya izmenchivostÕi sistematika
serykh polevok (Rodentia, Arvicolini). Soobshchenie1. Vidovoi
sostav i khromosomnye chisla. [Karyotypic variability and
systematics of the gray voles (Rodentia, Arvicolini). Communication
1. Species composition and chromosomal numbers]. Vestnik
Zool. 2, 26–37 (in Russian).
Zakrzewski, R.J., 1985. The fossil record. In: Tamarin, R.H.(Ed.),
Biology of New World Microtus. American Society of Mammalogists.
Special Publication No. 8, pp. 1–51.
Zheng, S.-H., Zhang, Z.-Q., 2000. Late Miocene–Early Pleistocene
micromammals from Wenwanggou of Lingtai, Gansu, China.
Vertebrata PalAsiatica 38, 58–71.
Zima, J., Kra´l, B., 1984. Karyotypes of European mammals II. Acta
Sc. Nat. Brno 18, 1–62.
M. Jaarola et al. / Molecular Phylogenetics and Evolution 33 (2004) 647–663 663
68 Paper 2.1.1
Paper 2.1.2
Martínková N., Moravec J. 2012. Multi-locus phylogeny of arvicoline voles (Arvicolini,
Rodentia) shows small tree terrace size. Folia Zoologica 61: 254-267.
– 69 –
254
Introduction
Arvicoline voles (Rodentia, Arvicolini) are a young
group of small rodents distributed on the northern
hemisphere. They started to diverge probably as
recently as two to three million years ago, but in the
short time frame, they speciated into one of the most
speciose mammalian groups (Wilson & Reeder 2005).
Today, the species show rapid temporal changes in
genetic composition of populations (Bryja et al. 2007,
Oliver et al. 2009, Rudá et al. 2010; but see Spaeth et
al. 2009) and fast karyotype reorganisation (Mazurok
et al. 2001, Mekada et al. 2002, Sitnikova et al. 2007,
Mitsainasetal.2010)coupledwithgenereorganisation
between mitochondrial and nucleotide genomes
(Triant & DeWoody 2007, 2008). Populations of
arvicoline voles diverge quickly when they become
fragmented in refugia, leading at times to speciation
in these areas, and refugia become speciation traps
(Martínková & Dudich 2003, Martínková et al. 2007,
Tougard et al. 2008, Kryštufek et al. 2009, Haring
et al. 2011). Multiple vole species co-occur in many
regions and habitat partitioning was found with
sympatric occurrence (Jurdíková et al. 2000, Santos
et al. 2011). Yet, they are morphologically similar
with small differences between distantly related taxa
(Fraguedakis-Tsolis et al. 2009).
Reconstruction of phylogenetic relationships in
arvicoline voles is complicated by morphological
similarities and rapid karyotype rearrangements,
making the group an ideal model for molecular genetic
studies. Both mitochondrial (mtDNA) and nuclear
(nucDNA) markers have been extensively used to
resolve relationships within the group and to study
evolutionary history of different taxa. The relative
merit of information value of mtDNA and nucDNA
markers overlaps in arvicoline rodents. MtDNA
sequences provide valuable information in phylogeny
reconstruction at specific and intrageneric level,
contributing to resolving phylogeographic histories
of taxa (e.g. Conroy & Cook 2000a, Jaarola & Searle
2002, Fink et al. 2004, Brunhoff et al. 2006, Fan et
al. 2011), identification of putative cryptic species
Folia Zool. – 61 (3–4): 254–267 (2012)
Multilocus phylogeny of arvicoline voles (Arvicolini,
Rodentia) shows small tree terrace size
Natália Martínková1,2
* and Jiří Moravec2
1
	Institute of Vertebrate Biology, Academy of Sciences of the Czech Republic, v.v.i., Květná 8, 603 65 Brno,
	 Czech Republic; e-mail: martinkova@ivb.cz
2
	Institute of Biostatistics and Analyses, Masaryk University, Kamenice 3, 625 00 Brno, Czech Republic;
	 e-mail: jirka678@seznam.cz
Received 4 January 2012; Accepted 25 June 2012
Abstract. We combined mitochondrial (cyb, control region, coi, nd4) and nuclear (irbp, ghr, sry, lcat) DNA
sequence data to infer phylogenetic relationships of arvicoline voles. The concatenated supermatrix contained
72.8 % of missing data. From this dataset, Bayesian inference showed close relationships of Arvicola and
Chionomys, Proedromys with Lasiopodomys and Microtus gregalis, Phaiomys with Neodon and M. clarkei.
Genus Microtus formed a supported group with Blanfordimys and N. juldaschi. The gene partition taxon sets
were explained in the multilocus phylogeny in such a way that the resulting Bayesian inference tree represented
a unique solution on a terrace in the tree space. This means that although the supermatrix contained a large
proportion of missing data, it was informative in retrieving a phylogeny with a unique optimality score, tree
likelihood.
Key words: divergence, evolutionary history, supertree, supermatrix, phylogenetic tree terrace, Microtus,
Arvicolinae
* Corresponding Author
70 Paper 2.1.2
255
(Kefelioğlu & Kryštufek 1999, Hellborg et al. 2005,
Castiglia et al. 2008, Conroy & Neuwald 2008, Weksler
et al. 2010) and phylogenetic placement of taxa with
unstable position based on other data (Macholán et
al. 2001, Jaarola et al. 2004, Martínková et al. 2007,
Kryštufek et al. 2009, Bannikova et al. 2010). The
phylogenetic signal of mtDNAand its ability to resolve
relationships decrease at higher level of phylogenies
that exhibits as a rapid burst of diversification (Jaarola
et al. 2004). However, also nuclear markers, either
DNA sequence data or AFLP markers (Galewski et al.
2006, Abramson et al. 2009, Fink et al. 2010), fail to
fully resolve the signal of the rapid diversification of
voles at the base of the tree. Incongruence of sampling
between studies further complicates interpretation of
relationships within the group (Bužan et al. 2008,
Haring et al. 2011).
Here, we combine available mitochondrial and nuclear
DNA sequence markers to reconstruct phylogenetic
relationships of arvicoline rodents and to assess
stability of the retrieved model. We use Bayesian
inference analysis of a concatenated supermatrix and
SuperTriplets supertree reconstruction to estimate
parent trees. These we then use to establish the size of
the terrace where trees will have the same likelihood
with the dataset with large content of missing data.
Material and Methods
Sequences were downloaded from GenBank for 74
species belonging to genera Arvicola, Blanfordimys,
Chionomys, Lasiopodomys, Microtus, Neodon,
Phaiomys and Proedromys (Table 1). Herewith,
we accept species designation of Wilson & Reeder
(2005) with the addition of recently established
species Microtus gromovi (Bannikova et al. 2010)
and Proedromys liangshanensis (Liu et al. 2007).
Alignments were constructed in Geneious 5.4
(Drummond et al. 2011) with sequences from
mitochondrial genes for cytochrome b (cyb), control
region (CR), cytochrome c oxidase subunit I (coi),
NADH dehydrogenase subunit 4 (nd4) and nuclear
genes for interphotoreceptor retinoid-binding protein,
exon 1 (irbp), growth hormone receptor, exon 10
(ghr), sex-determining region Y (sry), lecithin:
cholesterol acyl transferase, exons 2 through 5 (lcat).
The alignments were reduced to contain at least
three sequences at every base. In the sry gene, the
microsatellite (TC)n
(TG)n
(Acosta et al. 2010) could
not be aligned unambiguously, and the region was
deleted. The data and results are available through
TreeBASE (http://purl.org/phylo/treebase/phylows/
study/TB2:S12667).
Two additional loci have good taxonomic sampling.
However, both loci in the avpr1a gene, its upstream
region and exon 1, were previously documented to
be disparate with the species tree (Fink et al. 2007,
2010), and some sequences of the avpr1a exon 1 were
shared between distantly related species (Fink et al.
2007). To reduce conflict between gene trees in our
analyses, we chose to omit these loci.
Optimal substitution model was estimated in
MrModeltest 2.3 (Nylander 2004) with Akaike
Information Criterion (AIC) and applied to the
individual gene alignments and partitions of the
concatenated alignment. Where the parameters of
the selected model were extreme, a simpler model
was used. Bayesian Inference (BI) was conducted in
MrBayes 3.1 (Ronquist & Huelsenbeck 2003) with
Markov chain Monte Carlo (MCMC) parameters
set to 2-5 million steps sampled every 1000th
, five to
six chains in two runs with chain temperature 0.08-
0.12 and chain swapping attempted once every third
generation. The MCMC runs were optimised to mix
and ideally to finish with average standard deviation of
split frequencies below 0.01, potential scale reduction
factor for model parameters approaching 1.000 and
proportion of successful chain stage swaps between
0.4 and 0.7. BI is robust in recovering the correct tree
topology, but it might fail to establish appropriate
branch lengths with default branch length prior
(Marshall 2010). The 95 % credibility intervals of the
tree lengths from BI were compared to the tree length
obtained from maximum likelihood (ML) analysis.
TheMLtreewascalculatedinRAxML7.2(Stamatakis
2006). The trees were re-rooted to midpoint root to
allow for uncertainty in the phylogenetic position of
Arvicola (Galewski et al. 2006, Bužan et al. 2008,
Abramson et al. 2009, Bannikova et al. 2009).
Divergence events between all taxa were investigated
in two ways; from a combination of gene trees and
directly from the concatenated supermatrix. The gene
trees were combined into a SuperTriplets supertree
(Ranwez et al. 2010). The method breaks down the
gene trees to their smallest components containing three
taxa, where any two taxa are more closely related than
either is to the third. Supertree then contains medians of
relationships from the triplets as they were found in the
gene trees. Edge support in the SuperTriplets analysis
is the proportion of triplets that support a given edge.
The supermatrix was resolved with partitioned BI ran
for 6 million generations with five heated and one
cold MCMC, chain temperature set to 0.09 and one
chain swap attempted every third generation. Partition
rates were allowed to vary.
Paper 2.1.2 71
256
Table1.AccessionnumbersofDNAsequencesusedinthisstudytoreconstructphylogeneticrelationships.
SpeciesDistrib.mitochondrialnuclear
cybCRcoind4irbpghrsrylcat
Arvicola
A.amphibius(=terrestris)EuropeAF119269AY948543AY332681AF128938AY277407AM392380FN433499GQ267511
A.sapidusEuropeFJ539345FJ502319FN433498
Blanfordimys
B.afghanusAsiaEF599109
B.bucharensisAsiaAM392369AM392392
Chionomys
C.gudAsiaEU700087GQ267512
C.nivalisEuropeAY513845AF267284AY332687AM919424AM392378FN433493AH005248
C.robertiAsiaAY513850
Lasiopodomys
L.brandtiiAsiaGQ142008FN401371
L.mandarinusAsiaFJ986322AM919413AM392396GQ267514
Microtus
M.abbreviatusNAmericaAF163890
M.agrestisEuropeFJ619786AF267270AF128940AM919427AM910792FN433492
M.anatolicusAsiaFJ767740
M.arvalisEuropeAY220789AF267285AY332683AM919416AM392386FN433490GQ267517
M.bavaricusEuropeGQ243218AF267277
M.brachycercusEuropeAY513828
M.breweriNAmericaFN433501
M.cabreraeEuropeAY513788
M.californicusNAmericaAF163891FN433510
M.canicaudusNAmericaAF163892
M.chrotorrhinusNAmericaAF163893AM919403AM392383FN433503
M.clarkeiAsiaAY641526
M.daghestanicusAsiaAY513790GQ142009GQ267518
M.dogramaciiAsiaAY513795
M.duodecimcostatusEuropeAJ717744AM392400FN433496
M.evoronensisAsiaHM135862
M.felteniEuropeAY513798
M.fortisAsiaAF163894JF261174JF261174JF261174
M.gerbeiEuropeAY513799
M.gregalisAsiaAF163895GQ142007GQ267513
M.gromoviAsiaFJ986319HM135891
M.guatemalensisCAmericaAF410262
M.guentheriEurasiaAY513804AM919420AM392397FN433488
M.ilaeusAsiaAY513809
M.iraniAsiaFJ767748
72 Paper 2.1.2
257
M.kikuchiiAsiaAF348082AF348082AF348082AF348082AM919410AM392385
M.levisEurasiaDQ015676DQ015676DQ015676DQ015676FN433489
M.liechtensteiniEuropeAY513811AF267281
M.limnophilusAsiaFJ986323AM919426
M.longicaudusNAmericaAF119267AF128936AM919414AM392379
M.lusitanicusEuropeAY513812FN433497
M.majoriEuropeAY513814AM919409AM910796
M.maximowicziiAsiaFJ986311HM135863HM137737
M.mexicanusNAmericaAF163897AF251260
M.middendorffiiAsiaHM119493HM135899HM137752AM919419AM392390GQ267516
M.miurusNAmericaGU809171
M.mongolicusAsiaFJ986305HM137754
M.montanusNAmericaAF119280GU394082AF128946FN433513
M.montebelliAsiaAF163900AM919421AM910793
M.mujanensisAsiaHM135854
M.multiplexEuropeAY513815AF267276
M.oaxacensisCAmericaAF410260
M.ochrogasterNAmericaAF163901AM919423AM392389FN433506
M.oeconomusHolarcticAB372207AJ616853AM919418AM392388FN433491GQ267515
M.oregoniNAmericaAF163903FN433516
M.pennsylvanicusNAmericaAF119279AY369777JF443830AF128945AM919415AF540633
M.pinetorumNAmericaAF163904
M.quasiaterCAmericaAF410259
M.richardsoniNAmericaAF163905AM919404AM392387FN433517
M.sachalinensisAsiaFJ986318HM135900
M.saviiEuropeAY513824FN433495
M.schelkovnikoviAsiaAM910619AM919408AM910794
M.socialisAsiaAY513829FM162055FM162073
M.subterraneusEuropeAY513832AF267271AY332685
M.tatricusEuropeAY513837AF267267
M.thomasiEuropeAY513840AY560558AM919422AM910797FN433494
M.townsendiiNAmericaAF163906
M.transcaspicusAsiaAM919405AM910795
M.umbrosusCAmericaAF410261
M.xanthognathusNAmericaAF163907
Neodon
N.ireneAsiaAM392370AM919412AY294924
N.juldaschiAsiaAY513808
N.sikimensisAsiaAY163593
Phaiomys
Ph.leucurusAsiaAM392371AM919400AM392394
Proedromys
Pr.liangshanensisAsiaFJ463038FJ463038FJ463038FJ463038
Paper 2.1.2 73
258
Given the fractional nature of the supermatrix,
multiple distinct trees were likely to display the same
set of subtrees representing taxa sampled per gene.
Such trees will have the same log-likelihood for a
partitioned analysis (Sanderson et al. 2011). A set of
trees that display the same set of subtrees for sampled
loci and have the same log-likelihood is called a
terrace. The trees from a terrace are derived from
each other by nearest-neighbour interchange (NNI)
rearrangement, and, in a dataset with considerable
missing data content, the terraces might contain many
trees. The size of terraces for our dataset was assessed
with perl scripts from the PhyloTerraces package
(Sanderson et al. 2011). SuperTriplets supertree
and BI tree based on the supermatrix were used as
parent trees. Terrace identification requires binary
(fully resolved) trees. The BI tree was resolved by
accepting all relationships resolved in the posterior
sample. The SuperTriplets supertree was resolved
using relationships from the cyb tree. In terrace
identification analysis, the parent tree was broken to
subtrees, where each subtree represented relationships
of taxa sampled for the respective gene as resolved
in the parent tree. All relationships in the subtrees
were further characterized by triplets. From these, all
alternative parent trees that contain the subtrees were
constructed.
Results
The dataset contained 1143 base-pairs (bp) long
alignment with 68 taxa for cyb, CR alignment was
1025 bp long and contained 25 taxa, coi was 1545
bp long with 12 taxa, nd4 was 1378 bp long with 9
taxa, irbp was 1181 bp long with 24 taxa, ghr was
911 bp long with 27 taxa, sry was 908 bp long with
19 taxa and lcat was 590 bp long with 10 taxa. The
concatenated supermatrix had 72.8 % missing data
composed of missing sequences of individual genes,
alignment gaps and unknown nucleotides.
Substitution models selected by AIC for each gene
were GTR + Γ + I for cyb, HKY + Γ + I for CR, GTR
+ Γ + I for coi, GTR + Γ for nd4, HKY + Γ for irbp,
GTR + Γ + I for ghr, HKY + I for sry and HKY + I for
lcat. As the GTR model requires estimation of the rate
matrix, a simpler HKY model was tested for coi, nd4
and ghr genes. The difference between log-likelihood
based on selected and tested model was 10.8 for coi,
1.8 for nd4 and 1.1 for ghr and the HKY model was
used for nd4 and ghr genes. The resulting trees were
similar, and MCMC convergence was faster with the
simpler model. The results from the simpler models
are reported (Table 2, Figs. 1-2).
The 95 % credibility interval (CI) of the BI tree
length based on the partitioned concatenated dataset
was 7.89-15.12 with default rate parameter of the
exponential branch length prior (λ = 10.0). This CI of
the BI tree length did not contain the tree length 3.86
estimated from the ML analysis. Increasing the rate by
increments of 10.0 to the final value of 50.0, the tree
length decreased to 4.0-4.59, but we did not further
test the branch length prior because of decrease in
node support for high λ. Node support improved with
optimisation of the branch length prior when λ = 20.0,
and subsequently decreased (Fig. 3).We further analyse
the BI tree with the highest average node support.
The BI supermatrix phylogeny re-rooted with
midpoint root showed two initial groups (Fig. 4). The
root separated genera Arvicola and Chionomys from
Microtus, Blanfordimys, Phaiomys, Proedromys and
Lasiopodomys. Proedromys liangshanensis was a
sister species to a well-supported group containing
Lasiopodomys and Microtus gregalis. Similarly,
Phaiomys leucurus was a sister taxon to a supported
group with unresolved internal relationships
containing Neodon irene, N. sikimensis and M.
clarkei. Remaining taxa formed a monophyletic
group with high Bayesian posterior probability (BPP).
It contained species currently attributed to genus
Microtus not mentioned above, genus Blanfordimys
and N. juldaschi. The first group that diverged at this
level was the subgenus Microtus (Alexandromys)
predominantly distributed in the eastern Palaearctic.
Microtus fortis group was well differentiated from
the basal species of Alexandromys, M. kikuchii,
M. montebelli and M. oeconomus. We confirmed
position of M. gromovi as a distinct taxon rather than
a subspecies of M. maximowiczii.
Further notable group consisted of N. juldaschi
with Blanfordimys afghanus and B. bucharensis.
M. agrestis was a sister species to this group, but
the relationship was unsupported. Nearctic species
formed an unsupported group with M. cabrerae.
Within the Nearctic group, three pairs of sister taxa
had significant node support. Subgenera Microtus
(Terricola) and Microtus (Microtus) were sister
groups that were most derived in the BI phylogeny
(Fig. 4). In the subgenus Microtus, M. arvalis group
and M. socialis/guentheri group were separated, but
classification of M. schelkovnikovi to the M. socialis/
guentheri group had low support (BPP = 0.85).
Subgenus Terricola was separated to the eastern and
western clade, where the eastern clade containing M.
majori, M. subterraneus and M. daghestanicus had
BPP = 0.94.
74 Paper 2.1.2
259
The SuperTriplets supertree agreed with the BI tree in
distinguishing groups Arvicola, Microtus (Terricola),
Microtus (Microtus), Microtus (Pitymys) with M.
guatemalensis, Blanfordimys with N. juldaschi and
a separate group including taxa from the subgenus
Microtus (Alexandromys) (data available at TreeBASE).
The other groups were distorted due to different
position of Chionomys nivalis, N. sikimensis and
Lasiopodomys brandtii. In the supertree, C. nivalis
was placed as a basal taxon after diversification of
Arvicola. Neodon sikimensis formed a polyphyly with
C. gud and C. roberti. Lasiopodomys brandtii was
placed within the group containing Nearctic species.
Phylogenetic terraces where the trees belonged to
were small. The terrace with the BI tree used as the
parent tree consisted of a single tree, and the terrace
with the supertree consisted of 15 trees.
Discussion
Tree space of the concatenated dataset
We found that by optimising branch length prior of the
Bayesianinferenceanalysisofthemultilocusphylogeny
of arvicoline rodents with missing data comprising
nearly 73 % of the concatenated supermatrix we were
able to retrieve a phylogeny that is unique on a terrace.
This means that there are no trees with alternative
topology that would explain sets of taxa from
individual gene partitions that would be present in the
BI phylogeny. For the supertree approach, the terrace
size was also small, and it contained 15 trees with
alternative topology that explained the relationships of
gene tree datasets as depicted on the supertree.
The topology of our phylogeny that well explained
gene sets in the final trees was not reflected in
similar congruence in branch length estimations.
Our Bayesian phylogeny with highest node support
was longer than the ML tree as is known to occur in
partitioned datasets (Marshall et al. 2006, Brown et al.
2010, Marshall 2010). The cause of this discrepancy
was recently identified to be a branch length prior
that places too much probability density on large tree
lengths (Rannala et al. 2012). The default on branch
lengths in MrBayes assigns independent and identical
exponential priors for individual branch lengths. As
the default initial value for branch lengths is 0.1, the
MCMC starts from very long trees for large datasets
with many taxa, which is often unrealistic. The prior
then places too much influence on the posterior. The
effect is exacerbated for large datasets where there
are partitions with low variability or correlations in
rate variation or substitution models in the posterior
(Rannala et al. 2012). This seems to be the case in
our dataset. We optimised the branch length prior in
our partitioned analyses, assuming that by setting the
branch length exponential prior mean closer to mean
branch length estimated from the ML tree length, the
posterior tree length would be more similar to the ML
tree length (Zhang et al. 2012). This is not a suitable
approach in the Bayesian framework (c.f. Zhang et
al. 2012). Our data showed that decreasing the tree
length in this way lead to decrease of node support
for high values of rate parameter of the exponential
distribution of the uncorrelated branch length prior.
Using compound Dirichlet priors on branch lengths in
Table 2. Substitution models used for separate gene tree analyses and partitions of the supermatrix. Model
parameters are means estimated from sampled posterior distribution of the gene trees after burn-in. κ – transition/
transversion rate ratio, r – substitution rate, f – base frequency, α – shape parameter of the Γ distribution, I –
proportion of invariable sites, n/a – not available.
Parameter cyb CR coi nd4 irbp ghr sry lcat
GTR + Γ + I HKY + Γ + I GTR + Γ + I HKY + Γ HKY + Γ HKY + I HKY + I HKY + I
κ n/a 3.0452 n/a 14.1872 4.0786 4.8725 3.3133 6.5095
r(A ↔ C) 0.0176 n/a 0.0236 n/a n/a n/a n/a n/a
r(A ↔ G) 0.4088 n/a 0.4862 n/a n/a n/a n/a n/a
r(A ↔ T) 0.0402 n/a 0.0515 n/a n/a n/a n/a n/a
r(C ↔ G) 0.0065 n/a 0.0070 n/a n/a n/a n/a n/a
r(C ↔ T) 0.4714 n/a 0.4152 n/a n/a n/a n/a n/a
r(G ↔ T) 0.0555 n/a 0.0165 n/a n/a n/a n/a n/a
f(A) 0.3713 0.3262 0.3006 0.3421 0.2205 0.2611 0.2909 0.2015
f(C) 0.3596 0.2586 0.2736 0.3032 0.2809 0.2877 0.2640 0.2728
f(G) 0.0774 0.1141 0.1478 0.0976 0.2909 0.2291 0.2166 0.2766
f(T) 0.1918 0.3012 0.2780 0.2572 0.2077 0.2221 0.2285 0.2491
α 0.6046 1.1420 8.9089 0.2172 0.4326 n/a n/a n/a
I 0.4968 0.3503 0.6647 n/a n/a 0.7059 0.3133 0.6735
Paper 2.1.2 75
260
Fig. 1. Bayesian inference phylogenetic trees based on complete mitochondrial sequences for cyb (A), and
partial sequences for control region (B), coi (C) and nd4 (D) genes. All relationships are shown that were
compatible with the consensus tree from the posterior sample of trees after burn-in. Scale bars indicate 0.1
substitutions per site, asterisk denotes nodes with Bayesian posterior probability (BPP) ≥ 0.95.
76 Paper 2.1.2
261
modified MrBayes 3.1 (Zhang et al. 2012), we were
not able to obtain a tree with CI of tree length that
would include the ML estimate and the average BPP
remained lower than in our optimal tree.
Arvicoline phylogeny
Phylogenetic position of Arvicola within subfamily
Arvicolinae is unstable in studies utilising mtDNA
(Bužanetal.2008,Bannikovaetal.2009)incomparison
with studies that use nucDNA (Galewski et al. 2006,
Abramson et al. 2009). We also observed this in gene
trees. Our supermatrix results show that by rooting the
tree of the Arvicolini tribe sensu Galewski et al. (2006)
with midpoint root, Arvicola forms a supported group
with Chionomys at the base of the tree.
Microtus gregalis represented a phylogenetic enigma.
In early molecular phylogenies, it was placed distantly
from other supposedly related species at the base of
the phylogeny of Microtus, but its basal position was
unsupported (Conroy & Cook 2000b, Conroy et al.
2001, Jaarola et al. 2004). It was later retrieved as a
sister taxon to Chionomys based on mtDNA (Bužan &
Fig. 2. Bayesian inference phylogenetic trees based on nuclear sequences for partial irbp (A), ghr (B), sry
(C) and lcat (D) genes. All relationships are shown that were compatible with the consensus tree from the
posterior sample of trees after burn-in. Scale bars indicate 0.01 substitutions per site, asterisk denotes nodes
with BPP ≥ 0.95.
Paper 2.1.2 77
262
Kryštufek2008),butnucDNAgroupedM.gregaliswith
Lasiopodomys (Abramson et al. 2009). This grouping
elucidates towards rapid karyotype rearrangements
between species, as M. gregalis has 36 chromosomes
(Martínková et al. 2004), whereas L. mandarinus
chromosome number varies between 47 and 52 (Liu et
al. 2010). If the ancestral karyotype of the group was
2n = 54, the rearrangements leading to M. gregalis that
branches close to the base of the tree were extensive
(c.f. Lemskaya et al. 2010). The phylogenetic position
of M. gregalis in the vicinity of Lasiopodomys was
confirmed once Lasiopodomys was sequenced for
the mtDNA (Bannikova et al. 2010). Our combined
phylogeny placed M. gregalis close to the base of the
trees in a well-supported group with L. mandarinus
and L. brandtii in accordance with recent studies
(Abramson et al. 2009, Bannikova et al. 2010). The
sister taxon of the group is Proedromys liangshanensis
that was described recently (Liu et al. 2007). Based on
phylogeny of complete mtDNA, Pr. liangshanensis
is a sister species to Microtus (Hao et al. 2011). Our
analysis included more comprehensive sampling, and
we found that the species forms a supported sister
relationship with Lasiopodomys and M. gregalis.
The genus Neodon was polyphyletic with N. juldaschi
grouping with Blanfordimys deeper in the phylogeny
than other species attributed to Neodon, N. irene and
N. sikimensis. The latter two species consistently
belonged to the Phaiomys/Neodon lineage (Galewski
et al. 2006, Robovský et al. 2008, Bannikova et al.
2009, 2010) that included also M. clarkei in our
analyses. Its relationship with Neodon was not
resolved, forming a strongly supported trichotomy,
where BPP for the monophyly of Neodon within this
lineage was as low as 0.41.
The phylogenetic position of M. agrestis was unstable
in the gene trees, and it was placed as an unsupported
sister taxon to the N. juldaschi/Blanfordimys group
in our multilocus phylogeny. Microtus agrestis split
from the common ancestors of Microtus early in the
radiation of the genus, but its closest relatives might
not be identifiable today similarly as in the case of M.
cabrerae. Interestingly, M. agrestis from the Iberian
peninsula, where M. cabrerae is also distributed,
forms a phylogenetic lineage distinct from other M.
agrestis populations. This divergence is present both
in mitochondrial and nuclear phylogenies and might
represent a cryptic species that was not formally
described to date (Jaarola & Searle 2004, Hellborg
et al. 2005). The erratic placement of M. agrestis
in different gene trees and unsupported position
of M. cabrerae with North American Microtus
indicates that these species represent relicts of a
very early colonization of Arvicolini to Western
Europe. Phylogenies of Arvicolidae improve with
more comprehensive sampling (Bužan et al. 2008),
and based on the fact that we analysed majority of
species in the tribe Arvicolini, we are confident to
state that the close relatives of M. agrestis and M.
cabrerae are extinct today and their phylogenetic
position is influenced more by stochastic processes
in DNA sequence evolution such as saturation or
Fig. 3. Changes of the tree length (black diamonds, primary axis) and average Bayesian posterior probability of
node support (empty circles, secondary axis) with increasing shape parameter of the exponential distribution
of uncorrelated branch lengths prior. Dashed line indicates tree length estimated from maximum likelihood
analysis with GTR + Γ substitution model for each partition. Tree length is given with 95 % credibility interval.
78 Paper 2.1.2
263
Fig. 4. Bayesian inference phylogenetic tree based on concatenated sequence of the genes depicted in Figs.
1–2. All relationships are shown that were compatible with the consensus tree from the posterior sample of
trees after burn-in. Scale bar is in substitutions per site, BPP ≥ 0.95 is shown.
Paper 2.1.2 79
264
Literature
Abramson N.I., Lebedev V.S., Tesakov A.S. & Bannikova A.A. 2009: Supraspecies relationships in the
subfamily Arvicolinae (Rodentia, Cricetidae): an unexpected result of nuclear gene analysis. Mol. Biol.
43: 834–846. (in Russian)
Acosta M.J., Marchal J.A., Romero-Fernández I., Megías-Nogales B., Modi W.S. & Sánchez Baca A. 2010:
Sequence analysis and mapping of the sry gene in species of the subfamily Arvicolinae (Rodentia). Sex.
Dev. 4: 336­–347.
Bannikova A.A., Lebedev V.S. & Golenishchev F.N. 2009: Taxonomic position of Afghan vole (subgenus
Blanfordimys) by the sequence of the mitochondrial cytb gene. Russ. J. Genet. 45: 91–97.
Bannikova A.A., Lebedev V.S., Lissovsky A.A., Matrosova V., Abramson N.I., Obolenskaya E.V. & Tesakov
A.S. 2010: Molecular phylogeny and evolution of the Asian lineage of vole genus Microtus (Rodentia:
Arvicolinae) inferred from mitochondrial cytochrome b sequence. Biol. J. Linn. Soc. 99: 595–613.
Brown J.M., Hedtke S.M., Lemmon A.R. & Lemmon E.M. 2010: When trees grow too long: investigating the
causes of highly inaccurate Bayesian branch-length estimates. Syst. Biol. 59: 145–161.
Brunhoff C., Yoccoz N.G., Ims R.A. & Jaarola M. 2006: Glacial survival or late glacial colonization?
Phylogeography of the root vole (Microtus oeconomus) in north-west Norway. J. Biogeogr. 33: 2136–2144.
Bryja J., Charbonnel N., Berthier K., Galan M. & Cosson J.-F. 2007: Density-related changes in selection
pattern for major histocompatibility complex genes in fluctuating populations of voles. Mol. Ecol. 16:
5084–5097.
Bužan E.V. & Kryštufek B. 2008: Phylogenetic position of Chionomys gud assessed from a complete cytochrome
b gene. Folia Zool. 57: 274–282.
Bužan E.V., Kryštufek B., Hänfling B. & Hutchinson W.F. 2008: Mitochondrial phylogeny of Arvicolinae using
comprehensive taxonomic sampling yields new insights. Biol. J. Linn. Soc. 94: 825–835.
Castiglia R., Annesi F., Aloise G. & Amori G. 2008: Systematics of the Microtus savii complex (Rodentia,
Cricetidae) via mitochondrial DNA analyses: paraphyly and pattern of sex chromosome evolution. Mol.
Phylogenet. Evol. 46: 1157–1164.
convergence and by computational artefacts in phylogeny
reconstruction.
Within Microtus, we retrieved three groups that
represent subgenera recognised by Wilson & Reeder
(2005) with minor changes. Subgenus Alexandromys
was supported without M. clarkei as per Wilson
& Reeder (2005), Microtus without M. cabrerae
and Terricola with M. tatricus. The species within
subgenera showed relationships established in
previous studies (Jaarola et al. 2004, Martínková et
al. 2007, Kryštufek et al. 2009, Bannikova et al. 2010,
Haring et al. 2000, 2011).
Nearctic Microtus consistently suffer from lack of
resolution of many relationships (Conroy & Cook
2000b, Conroy et al. 2001, Jaarola et al. 2004, Fink et
al. 2010), although recent studies show that these taxa
are also strongly influenced by rapid diversification
in speciation traps. Weksler et al. (2010) found M.
abbreviatus from Wrangell Mts. to be divergent
and potentially merit species status, and Conroy &
Neuwald (2008) distinguished two species within M.
californicus. In our multilocus study, phylogenetic
position of M. breweri was particularly unstable. This
was probably in lieu of the fact that only the sry gene
sequence was available for this species.
Species with small ranges were often part of rapidly
differentiating groups. This leads to an assumption
that geographic isolation in small refugia triggers
diversification on both molecular and morphological
levels. In Microtus, the results of phylogeography
couplewithspeciesphylogenieswherephylogeography
nowadays indicates regions and populations that might
give rise to new species in the future.
Acknowledgements
We appreciate M. Macholán’s invitation to participate
in the issue of Folia Zoologica in tribute to Jan Zima’s
anniversary. N. Martínková is thankful to J. Zima
for his guidance and support at the beginning of her
career. The bioinformatic analyses were conducted at
the computational cluster of the Institute of Vertebrate
Biology and at the MetaCentrum. The access to the
MetaCentrum computing facilities was provided under
the programme “Projects of Large Infrastructure for
Research, Development, and Innovations” LM2010005
funded by the Ministry of Education, Youth, and Sports
of the Czech Republic. This study was supported by
the Grant Agency of the Academy of Sciences of the
Czech Republic, grant no. IAA600930609 and with
institutional support RVO: 68081766.
80 Paper 2.1.2
265
Conroy C.J & Cook J.A. 2000a: Phylogeography of a post-glacial colonizer: Microtus longicaudus (Rodentia:
Muridae). Mol. Ecol. 9: 165–175.
Conroy C.J. & Cook J.A. 2000b: Molecular systematics of a holarctic rodent (Microtus: Muridae). J. Mammal.
81: 344–359.
Conroy C.J., Horteland Y., Cervantes F.A. & Cook J.A. 2001: The phylogenetic position of southern relictual
species of Microtus (Muridae: Rodentia) in North America. Mamm. Biol. 66: 332–344.
Conroy C.J. & Neuwald J.L. 2008: Phylogeographic study of the California vole, Microtus californicus. J.
Mammal. 89: 755–767.
Drummond A.J., Ashton B., Buxton S., Cheung M., Cooper A., Duran C., Field M., Heled J., Kearse M.,
Markowitz S., Moir R., Stones-Havas S., Sturrock S., Thierer T. & Wilson A. 2011: Geneious v5.4.
Available from http://www.geneious.com/
Fan Z., Liu S., Liu Y., Zhang X. & Yue B. 2011: How Quaternary geologic and climatic events in the
southeastern margin of the Tibetan Plateau influence the genetic structure of small mammals: inferences
from phylogeography of two rodents, Neodon irene and Apodemus latronum. Genetica 139: 339–351.
Fink S., Excoffier L. & Heckel G. 2004: Mitochondrial gene diversity in the common vole Microtus arvalis
shaped by historical divergence and local adaptations. Mol. Ecol. 13: 3501–3514.
Fink S., Excoffier L. & Heckel G. 2007: High variability and non-neutral evolution of the mammalian avpr1a
gene. BMC Evol. Biol. 7, 176.
Fink S., Fischer M.C., Excoffier L. & Heckel G. 2010: Genomic scans support repetitive continental colonization
events during the rapid radiation of voles (Rodentia: Microtus): the utility of AFLPs versus mitochondrial
and nuclear sequence markers. Syst. Biol. 59: 548–572.
Fraguedakis-Tsolis S.E., Chondropoulos B.P., Stamatopoulos C.V. & Giokas S. 2009: Morphological variation
of the five vole species of the genus Microtus (Mammalia, Rodentia, Arvicolinae) occurring in Greece.
Acta Zool. 90: 254–264.
Galewski T., Tilak M., Sanchez S., Chevret P., Paradis E. & Douzery E.J.P. 2006: The evolutionary radiation
of Arvicolinae rodents (voles and lemmings): relative contribution of nuclear and mitochondrial DNA
phylogenies. BMC Evol. Biol. 6, 80.
Hao H., Liu S., Zhang X., Chen W., Song Z., Peng H., Liu Y. & Yue B. 2011: Complete mitochondrial genome
of a new vole Proedromys liangshanensis (Rodentia: Cricetidae) and phylogenetic analysis with related
species: are there implications for the validity of the genus Proedromys? Mitochondr. DNA 22: 28–34.
Haring E., Herzig-Straschil B. & Spitzenberger F. 2000: Phylogenetic analysis of Alpine voles of the Microtus
multiplex complex using the mitochondrial control region. J. Zool. Syst. Evol. Res. 38: 231–238.
Haring E., Sheremetyeva I.N. & Kryukov A.P. 2011: Phylogeny of Palearctic vole species (genus Microtus,
Rodentia) based on mitochondrial sequences. Mamm. Biol. 76: 258–267.
Hellborg L., Gündüz İ. & Jaarola M. 2005: Analysis of sex-linked sequences supports a new mammal species
in Europe. Mol. Ecol. 14: 2025–2031.
Jaarola M., Martínková N., Gündüz İ., Brunhoff C., Zima J., Nadachowski A., Amori G., Bulatova N.S.,
ChondropoulosB.,Fraguedakis-TsolisS.,González-EstebanJ.,López-FusterM.J.,KandaurovA.S.,Kefelioğlu
H., da Luz Mathias M., Villate I. & Searle J.B. 2004: Molecular phylogeny of the speciose vole genus Microtus
(Arvicolinae, Rodentia) inferred from mitochondrial DNA sequences. Mol. Phylogenet. Evol. 33: 647–663.
Jaarola M. & Searle J.B. 2002: Phylogeography of field voles (Microtus agrestis) in Eurasia inferred from
mitochondrial DNA sequences. Mol. Ecol. 11: 2613–2621.
Jaarola M. & Searle J.B. 2004: A highly divergent mitochondrial DNA lineage of Microtus agrestis in southern
Europe. Heredity 92: 228–234.
Jurdíková N., Žiak D. & Kocian Ľ. 2000: Habitat requirements of Microtus tatricus: macrohabitat and
microhabitat. In: Urban P. (ed.), Výskum a ochrana cicavcov na Slovensku 4. ŠOP COPK, Banská Bystrica.
Kefelioğlu H. & Kryštufek B. 1999: The taxonomy of Microtus socialis group (Rodentia: Microtinae) in Turkey,
with the description of a new species. J. Nat. Hist. 33: 289–303.
Kryštufek B., Bužan E.V., Vohralík V., Zareie R. & Özkan B. 2009: Mitochondrial cytochrome b sequence
yields new insight into the speciation of social voles in south-west Asia. Biol. J. Linn. Soc. 98: 121–128.
Lemskaya N., Romanenko S.A., Golenishchev F.N., Rubtsova N.V., Sablina O.V., Serdukova N.A., O’Brian
P.C.M., Fu B.,Yiğit N., Ferguson-Smith M.A.,Yang F. & GraphodatskyA.S. 2010: Chromosomal evolution
Paper 2.1.2 81
266
of Arvicolinae (Cricetidae, Rodentia). III. Karyotype relationships of ten Microtus species. Chromosome
Res. 18: 459–471.
Liu S., Sun Z., Zeng Z. & Zhao E. 2007: A new vole (Cricetidae: Arvicolinae: Proedromys) from the Liangshan
Mountains of Sichuan Province, China. J. Mammal. 88: 1170–1178.
Liu H., Yan N. & Zhu B. 2010: Two new karyotypes and bandings in Microtus mandarinus faeceus (Rodentia).
Hereditas 147: 123–126.
Macholán M., Filippucci M.G. & Zima J. 2001: Genetic variation and zoogeography of pine voles of the
Microtus subterraneus/majori group in Europe and Asia Minor. J. Zool. 255: 31–42.
Martínková N. & Dudich A. 2003: The fragmented distribution range of Microtus tatricus and its evolutionary
implications. Folia Zool. 52: 11–22.
Martínková N., Nová P., Sablina O.V., Graphodatsky A.S. & Zima J. 2004: Karyotypic relationships of the
Tatra vole (Microtus tatricus). Folia Zool. 53: 279–284.
Martínková N., Zima J., Jaarola M., Macholán M. & Spitzenberger F. 2007: The origin and phylogenetic
relationships of Microtus bavaricus based on karyotype and mitochondrial DNA sequences. Folia Zool.
56: 39–49.
Marshall D.C. 2010: Cryptic failure of partitioned Bayesian phylogenetic analyses: lost in the land of long
trees. Syst. Biol. 59: 108–117.
Marshall D.C., Simon C. & Buckley T.R. 2006: Accurate branch length estimation in partitioned Bayesian
analyses requires accommodation of among-partition rate variation and attention to branch length priors.
Syst. Biol. 55: 993–1003.
Mazurok N.A., Rubtsova N.V., Isaenko A.A., Pavlova M.E., Slobodyanyuk S.Y., Nesterova T.B. & Zakian
S.M. 2001: Comparative chromosome and mitochondrial DNA analyses and phylogenetic relationships
within common voles (Microtus, Arvicolidae). Chromosome Res. 9: 107–120.
Mekada K., Koyasu K., Harada M., NaritaY., Shrestha K.C. & Oda S.-I. 2002: Karyotype and X-Ychromosome
pairing in the Sikkim vole (Microtus (Neodon) sikimensis). J. Zool. 257: 417–423.
Mitsainas G.P., Rovatsos M.T. & Giagia-Athanasopoulou E.B. 2010: Heterochromatin study and geographical
distribution of Microtus species (Rodentia, Arvicolinae) from Greece. Mamm. Biol. 75: 261–269.
Nylander J.A.A. 2004: MrModeltest v2. Program distributed by the author. Evolutionary Biology Centre,
Uppsala University.
Oliver M.K., Lambin X., Cornulier T. & Piertney S.B. 2009: Spatio-temporal variation in the strength and
mode of selection acting on major histocompatibility complex diversity in water vole (Arvicola terrestris)
metapopulations. Mol. Ecol. 18: 80–92.
Rannala B., Zhu T. & Yang Z. 2012: Tail paradox, partial identifiability, and influential priors in Bayesian
branch length inference. Mol. Biol. Evol. 29: 325–335.
Ranwez V., Criscuolo A. & Douzery E.J.P. 2010: SuperTriplets: a triplet-based supertree approach to
phylogenomics. Bioinformatics 26: i115–i123.
Robovský J., Řičánková V. & Zrzavý J. 2008: Phylogeny of Arvicolinae (Mammalia, Cricetidae): utility of
morphological and molecular data sets in a recently radiating clade. Zool. Scr. 37: 571–590.
Ronquist F. & Huelsenbeck J.P. 2003: MrBayes 3: Bayesian phylogenetic inference under mixed models.
Bioinformatics 19: 1572–1574.
Rudá M., Žiak D., Kocian Ľ. & Martínková N. 2010: Low genetic variability in a mountain rodent, the Tatra
vole. J. Zool. 281: 118–124.
Sanderson M.J., McMahon M.M. & Steel M. 2011: Terraces in phylogenetic tree space. Science 333: 448–450.
Santos S.M., da Luz Mathias M. & Mira A.P. 2011: The influence of local, landscape and spatial factors on the
distribution of the Lusitanian and the Mediterranean pine voles in a Mediterranean landscape. Mamm. Biol.
76: 133–142.
Sitnikova N.A., Romanenko S.A., O’Brien P.C.M., Perelman P.L., Fu B., Rubtsova N.V., Serdukova N.A.,
GolenishchevF.N.,TrifonovV.A.,Ferguson-SmithM.A.,YangF.&GraphodatskyA.S.2007:Chromosomal
evolution of Arvicolinae (Cricetidae, Rodentia). I. The genome homology of tundra vole, field vole, mouse
and golden hamster revealed by comparative chromosome painting. Chromosome Res. 15: 447–456.
Spaeth P.A., van Tuinen M., Chan Y.L., Terca C. & Hadly E.A. 2009: Phylogeography of Microtus longicaudus
in the tectonically and glacially dynamic Central Rocky Mountains. J. Mamm. 90: 571–584.
82 Paper 2.1.2
267
Stamatakis A. 2006: RAxML-VI-HPC: maximum likelihood-based phylogenetic analyses with thousands of
taxa and mixed models. Bioinformatics 22: 2688–2690.
Tougard C., Brunet-Lecomte P., Fabre M. & Montuire S. 2008: Evolutionary history of two allopatric Terricola
species (Arvicolinae, Rodentia) from molecular, morphological, and palaeontological data. Biol. J. Linn.
Soc. 93: 309–323.
Triant D.A. & Dewoody J.A. 2007: Extensive mitochondrial DNA transfer in a rapidly evolving rodent has
been mediated by independent insertion events and by duplications. Gene 401: 61–70.
Triant D.A. & Dewoody J.A. 2008: Molecular analyses of mitochondrial pseudogenes within the nuclear
genome of arvicoline rodents. Genetica 132: 21–33.
Weksler M., Lanier H.C & Olson L.E. 2010: Eastern Beringian biogeography: historical and spatial genetic
structure of singing voles in Alaska. J. Biogeogr. 37: 1414–1431.
Wilson D.E. & Reeder D.M. 2005: Mammal species of the world. A taxonomic and geographic reference (3rd
ed.). John Hopkins University Press, Baltimore.
Zhang C., Rannala B. & Yang Z. 2012: Robustness of compound Dirichlet priors for Bayesian inference of
branch lengths. Syst. Biol. 61: 779–784.
Paper 2.1.2 83
Paper 2.1.3
Martínková N., Zima J., Jaarola M., Macholán M., Spitzenberger F. 2007. The origin and
phylogenetic relationships of Microtus bavaricus based on karyotype and mitochondrial
DNA sequences. Folia Zoologica 56: 39-49.
– 85 –
39
Folia Zool. – 56(1): 39–49 (2007)
The origin and phylogenetic relationships of Microtus bavaricus based
on karyotype and mitochondrial DNA sequences
Natália Martínková1,2
, Jan Zima1
, Maarit Jaarola3,
*, Miloš Macholán4
and Friederike
Spitzenberger5
1	
Department of Population Biology, Institute of Vertebrate Biology of the ASCR, v.v.i., Studenec 122, 675 02
	 Koněšín, Czech Republic; e-mail: martinkova@brno.cas.cz, jzima@ivb.cz
2 	
Department of Biology, University of York, PO Box 373, YO10 5YW York, United Kingdom
3 	
Department of Cell and Organism Biology, Genetics Building, Lund University, Sölvegatan 29, SE-223 62
	 Lund, Sweden; e-mail: m_jaarola@zbs.bialowieza.pl
4
	Laboratory of Mammalian Evolutionary Genetics, Institute of Animal Physiology and Genetics of the
	 ASCR, v.v.i., Veveří 97, 602 00 Brno, Czech Republic; e-mail: macholan@iach.cz
5 	
Natural History Museum, Burgring 7, A-1010 Vienna, Austria; e-mail: friederike.spitzenberger@nhm-wien.ac.at
*	Present address: Mammal Research Institute, Polish Academy of Sciences, ul. Waszkiewicza 1c, 17-230
	 Białowieża, Poland
Received 2 August 2006; Accepted 2 March 2007
A b s t r a c t . Geographic isolation of small populations in refugia during late Pleistocene
glaciations resulted in population differentiation that in some cases lead to speciation. We report
the karyotype of Microtus bavaricus, an evolutionary young and threatened rodent endemic
to the Alps. Our results show that the karyotype of M. bavaricus is almost identical to that of
M. liechtensteini (2N = 46, NF = 54). A close relationship between the two species was
also supported by phylogenetic analysis of complete mitochondrial DNA sequences for
the cytochrome b gene. The cytochrome b divergence between Microtus bavaricus and
M. liechtensteini was 1.7 %, the lowest estimate observed among the 14 currently recognised
species of Eurasian pine voles (subgenus Terricola).
Key words: Terricola, molecular divergence, glaciation
Introduction
Vole species of the genus Microtus (Arvicolinae, Rodentia) differ considerably in age and
various evolutionary stages of speciation can be observed within the genus. The Eurasian pine
voles, subgenus Terricola, include species groups that are especially suitable for analysis of
recent divergence events (J a a r o l a et al. 2004). Geographic isolation of small populations
in refugia during the late Pleistocene glaciations could have served as “speciation traps” for
several of these young taxa, thus promoting speciation (C h a l i n e 1987, M a r t í n k o v á
& D u d i c h 2003).
The Bavarian pine vole, Microtus bavaricus (König, 1962), is an endemic species of the
Alps with an extremely restricted range and rather enigmatic phylogenetic relationships.
In fact, its distribution area covers only six known localities in the Innsbruck Alps in
Bavaria, Germany (terra typica at Garmisch-Partenkirchen), and northern Tyrol in Austria
(K ö n i g 1982, S p i t z e n b e r g e r 2002, C a r l e t o n & M u s s e r 2005). Because
of the species’ distributional pattern, it was suggested that M. bavaricus survived the
last glacial period in a refugium situated in the northern Alps (K r a t o c h v í l 1970,
86 Paper 2.1.3
40
S p i t z e n b e r g e r 2002). Since the original morphological description by K ö n i g
(1962), affinities of M. bavaricus to the two other lineages of pine voles endemic to the Alps
and some nearby mountain ranges, M. multiplex and M. liechtensteini, have been indicated
(K r a t o c h v í l 1970, S p i t z e n b e r g e r 2002) and confirmed by both morphological
(S p i t z e n b e r g e r et al. 2000) and molecular genetic analysis (H a r i n g et al. 2000).
This group, the M. multiplex complex, including M. multiplex, M. liechtensteini and M.
bavaricus, is characterised by low morphological divergence and M. multiplex and M.
liechtensteini were occasionally considered to be conspecific (K r a p p 1982). This opinion
found particular support in a finding of a single natural hybrid in the area of parapatric
contact between the two taxa (S t o r c h & W i n k i n g 1977). The F1-hybrid from
Calliano, Trento province in Italy showed karyotype characteristics of both parental species
and had an intermediate diploid number of chromosomes (2N = 47).
M. multiplex and M. liechtensteini can be distinguished by their parapatric distribution
patterns and differences in karyotype: M. multiplex is distributed in the western parts of the
Alps and certain adjacent mountain ranges (eastern margins of Massif Central, northern
Apennines), whereas M. liechtensteini occurs in the eastern and north-eastern Alps and
the western Dinaric mountains (M i t c h e l l – J o n e s et al. 1999). The diploid number
of chromosomes differs between the two taxa (2N = 48 in M. multiplex, 2N = 46 in
M. liechtensteini) and the two karyotypes can also be distinguished also by other details
in the morphology of individual chromosomes (Z i m a & K r á l 1984). However, the
molecular divergence between M. multiplex and M. liechtensteini falls within the 4–8 %
cytochrome b range that includes both inter- and intra-specific divergence in Microtus
(J a a r o l a et al. 2004).
The aim of the present paper is to report on the hitherto unknown karyotype of
M. bavaricus. We have also examined sequences of the mitochondrial cytochrome b gene
for this species, in order to estimate the taxonomic position and phylogenetic relationships
of M. bavaricus to other species of pine voles (subgenus Terricola). Analysis of cytochrome
b sequences enabled us to utilize the extensive data set of publicly available sequences of
closely related Terricola species.
Material and Methods
C h r o m o s o m e s
The karyotype was studied in a male of M. bavaricus collected in the Rofangebirge in
northern Tyrol, Austria. It was collected on 3 August, 2004 by Simon Engelberger northwest
of Steinberg/Rofan in an open spruce forest near a small brook, a tributary to the Ampelsbach
River (47o
32’N, 11o
45’E, 1100 m a.s.l.). The voucher skin and skull (NMW65362) are stored
in the Mammal Collection of the Natural History Museum in Vienna. Mitotic chromosomes
were prepared from bone marrow cells obtained from short-term culture, using the standard
technique with hypotonic treatment and fixation in a mixture of ethanol and acetic acid. The
karyotype was then analyzed by conventional Giemsa staining.
C y t o c h r o m e b s e q u e n c e s
Complete or partial sequences of the mitochondrial gene for cytochrome b were obtained
for 14 individuals of Microtus bavaricus, M. tatricus, M. majori and M. liechtensteini
Paper 2.1.3 87
41
Table1.Individualsvouchernumbers,samplelocalities,cytochromebhaplotypesandGenBankaccessionnumbersforsequencedspecimensofMicrotus,subgenus
Terricola.
SpeciesVoucherNo.LocalityHaplotypeAccessionNo.
M.bavaricusNMW65362SteinbergamRofan,NorthernTyrol,Austriabavaricus1DQ841693
NMW8072Garmisch-Partenkirchen,Bavaria,Germanybavaricus2DQ841694
NMW26592SteinbergamRofan,NorthernTyrol,Austriabavaricus3DQ841695
M.tatricusNM-179PrvéRoháčskeplesoLake,WesternTatraMts,Slovakiatatricus4DQ841696
NM-546PrvéRoháčskeplesoLake,WesternTatraMts,Slovakiatatricus5DQ841697
NM-182PrvéRoháčskeplesoLake,WesternTatraMts,Slovakiatatricus6DQ841698
NM-194RakytovskádolinaValley,VeľkáFatraMts,Slovakiatatricus7DQ841699
NM-195RakytovskádolinaValley,VeľkáFatraMts,Slovakiatatricus7DQ841699
NM-202DolnýHarmanec,VeľkáFatraMts,Slovakiatatricus8DQ841700
NM-76TretieRoháčskeplesoLake,WesternTatraMts,Slovakiatatricus9DQ841701
NM-84SmutnádolinaValley,WesternTatraMts,Slovakiatatricus10DQ841702
M.majoriTU601Damar,Turkeymajori2DQ841703
MM388Hopa,Turkeymajori3DQ841704
M.liechtensteiniCR43Croatialiechtensteini2EF379100
NMW–NationalMuseum,Vienna,Austria;NM–collectionofN.Martínková;TU,MM–collectionofM.Macholán;CR–collectionofHeikkiHenttonen.
88 Paper 2.1.3
42
(Table 1). Additional sequences were downloaded from GenBank (Accession Numbers
in Fig. 2; J a a r o l a et al. 2004, G a l e w s k i et al. 2006, T o u g a r d et al., in press).
The haplotype names used here are identical to the original references except for
brachycercus1–3(cf.C a r l e t o n & M u s s e r 2005)thatwerenamedsavii1–3inJ a a r o l a
et al. (2004).
Primers used for amplification and sequencing of the cytochrome b gene, PCR and sequencing
conditions are described in detail in J a a r o l a et al. (2004). Sequences were completed in
Sequencher 4.2 (GeneCodes) and manually aligned in BioEdit 5.0 (H a l l 1999).
Sequence composition and nucleotide diversity (π) were calculated in DnaSP 4.1
(R o z a s et al. 2003). Total (Dxy) and net (Da) divergence was estimated in MEGA
3.1 (K u m a r et al. 2004) using Kimura 2-parameter distances (K i m u r a 1980).
Substitution model was assessed in ModelTest 3.7 (P o s a d a & C r a n d a l l 1998)
using the Akaike Information Criterion. The model selected, GRT+I+Γ (T a v a r é 1986;
I = 0.59, α = 1.5008), was used to estimate a maximum likelihood (ML) phylogenetic tree in
PAUP* 4.0b10 (S w o f f o r d 2003). Bootstrap support of taxonomic units was calculated
from 10,000 neighbour-joining (NJ) and 100 maximum parsimony (MP) parametric
replicates. Bayesian posterior probabilities were estimated in MrBayes 3.1 (R o n q u i s t
& H u e l s e n b e c k 2003) from 2,000,000 generations sampled every 1000th
generation
excluding a burn-in of 200,000 steps.
Results
C h r o m o s o m e s
The male diploid complement of M. bavaricus contains 46 chromosomes (Fig. 1). The largest
pair consists of two subtelocentric chromosomes with tiny, but clearly visible short arms. The
second largest chromosome is an odd submetacentric. There are three meta- or submetacentric
chromosomes of medium size. The other chromosomes are acrocentric, and their size
decreases gradually. The largest acrocentric pair approximately equals in size to the larger
arm of the odd submetacentric chromosome. The number of chromosomal arms is therefore
54, however, additional small short arms are apparent in at least two pairs of acrocentric
chromosomes. The odd large submetacentric element can be identified as the X chromosome.
The Y chromosome is one of the three elements with the meta- or submetacentric position of
the centromere. In most metaphases, one of these chromosomes was slightly larger than the
other two, and this could be considered as the Y chromosome.
C y t o c h r o m e b
Altogether 11 complete cytochrome b haplotypes (1143 base pairs; bp) were submitted
to public databases (GenBank Accession Numbers: DQ841693-DQ841704, EF379100).
Additionally, two M. bavaricus individuals yielded partial cytochrome b sequences, of 1052
(DQ841695) and 650 bp (DQ841694). Phylogenetic analyses were based on the complete
cytochrome b alignment of 38 haplotypes of European endemic Terricola species, including
one available complete M. bavaricus sequence, and 11 haplotypes of Terricola species with
distribution areas covering Europe and Asia Minor. Within- and between-species divergences
were estimated from all available Terricola sequences using a 650 bp alignment of 54
Paper 2.1.3 89
43
sequences. Phylogenetic analysis of this alignment, including three M. bavaricus and two
M. liechtensteini, shows that M. bavaricus and M. liechtensteini are reciprocally monophyletic
(data not shown).
For the 1143 bp cytochrome b alignment, a total of 371 (32 %) polymorphic sites were
observed and 325 (28 %) of those were parsimony informative. Phylogenies inferred with
ML, MP, NJ and Bayesian methods had similar topologies. The phylogenetic analysis shows
that M. bavaricus is most closely related to M. liechtensteini and the result is supported
by high bootstrap support and posterior probability values (Fig. 2). Monophyly of the
M. multiplex complex had bootstrap support of 71 % in the NJ tree, 75 % in the MP trees
and a Bayesian posterior probability of 1.0. Total and net divergence between M. bavaricus
and M. liechtensteini was estimated at 2.3 % and 1.7 %, respectively, which is the lowest
between-species divergence reported within the subgenus Terricola (Table 2). Despite the
limited sample size, nucleotide diversity of the M. bavaricus/liechtensteini group was within
the range of intraspecific nucleotide diversity in other Terricola species (Fig. 3).
Discussion
The karyotype of the M. bavaricus male studied is almost identical with those reported
from various parts of the range of M. liechtensteini (e.g., P e t r o v & Ž i v k o v i ć
1971, S t o r c h & W i n k i n g 1977). The autosome sets are apparently quite similar,
and the only difference between the karyotypes reported for individual populations of M.
liechtensteini and M. bavaricus can be found in the size and the centromeric position of
Fig. 1 . Karyotype of a male Microtus bavaricus from Rofangebirge, northern Tyrol, Austria analysed by
conventional Giemsa staining.
90 Paper 2.1.3
44
Table2.Total(Dxy;belowthediagonal)andnet(Da;abovethediagonal)divergence(Kimura2-parameterdistances)betweenspeciesoftheTerricolasubgenusbasedon
partialcytochromebgenesequences(650bp).Numberofsequencesinparenthesis.NumbersinboldrepresenttotalandnetdivergencebetweenMicrotusbavaricusandM.
liechtensteini.
Microtus
Dxy/Dabav.brach.dagh.duod.felt.gerbeiliecht.lusit.majorimultipl.saviisubt.tatr.thom.
M.bavaricus(3)0.0900.0760.0740.0700.0580.0170.0780.0890.0580.0980.0670.0760.074
M.brachycercus(3)0.0990.1090.0940.1080.0920.0770.0940.1140.1000.0440.0980.1020.094
M.daghestanicus(3)0.0920.1290.0970.0780.0730.0800.1000.0900.0780.1150.0520.0880.077
M.duodecimcostatus(4)0.0880.1120.1230.0830.0660.0660.0300.0820.0720.1010.0910.0830.080
M.felteni(1)0.0720.1140.0920.0950.0670.0760.0830.1010.0700.1060.0780.0770.061
M.gerbei(5)0.0650.1040.0920.0830.0720.0530.0690.0750.0670.0900.0690.0680.059
M.liechtensteini(2)0.0230.0870.0980.0820.0800.0630.0690.0770.0560.0870.0660.0740.070
M.lusitanicus(3)0.0840.1040.1170.0450.0860.0780.0770.0850.0750.1060.0940.0830.079
M.majori(3)0.0920.1210.1060.0960.1030.0820.0830.0900.0840.1250.0850.0930.099
M.multiplex(5)0.0650.1110.0970.0880.0750.0770.0650.0830.0900.1070.0750.0680.066
M.savii(2)0.1010.0510.1290.1140.1070.0960.0920.1110.1270.1130.0980.1090.092
M.subterraneus(6)0.0890.1240.0850.1230.0980.0940.0900.1170.1060.0990.1180.0860.073
M.tatricus(10)0.0810.1120.1050.0980.0800.0770.0810.0890.0980.0760.1130.1090.075
M.thomasi(5)0.0880.1130.1030.1040.0730.0770.0860.0950.1120.0830.1040.1050.090
Paper 2.1.3 91
45
Fig. 2. Maximum likelihood phylogenetic tree (-ln L = 7584.58) based on the GTR+I+Γ substitution model
showing the inferred phylogenetic relationships among 49 Terricola cytochrome b haplotypes (1143 bp). The
tree was rooted with representatives of Microtus sensu stricto subgenus. Numbers above branches represent
bootstrap support based on neighbour-joining, maximum parsimony analysis and Bayesian posterior probability,
respectively. Only values for major branches greater than 70 % and 0.95, respectively, are shown.
92 Paper 2.1.3
46
the Y chromosome. The Y chromosome observed in M. bavaricus is almost identical to
that of a population of M. liechtensteini from northern Velebit Mts in Croatia studied by
P e t r o v & Ž i v k o v i ć (1974). On the other hand, the Y chromosome of males from M.
liechtensteini populations from southern Austria (Defereggen-Gebirge, Karnische Alpen) was
subtelocentric and distinctly larger than a similar metacentric pair of small autosomes (K r á l
et al. 1978). Submetacentric and subtelocentric morphs of the Y chromosome were found
also in populations of M. liechtensteini from northern Italy (Trento and Belluno provinces;
S t o r c h & W i n k i n g 1977). The distribution of individual morphs of the Y chromosome
within the range of M. liechtensteini has no distinct geographic pattern, and it probably yields
no phylogenetic information. Similar variation in the sex chromosomes was reported also
among populations of M. multiplex (G r a f & M e y l a n 1980, B r u n e t – L e c o m t e
& Vo l o b o u e v 1994). The karyotype structure in the studied male of M. bavaricus thus
indicates its close relatedness to populations of M. liechtensteini, and no distinct specific
features were observed.
The phylogenetic inference derived from complete cytochrome b sequences confirms
the close relationship of M. bavaricus and M. liechtensteini reported by H a r i n g et al.
(2000) based on mitochondrial control region sequences. Similarly, in accordance with
previous phylogenetic analyses (H a r i n g et al. 2000, J a a r o l a et al. 2004), our data
demonstrate that M. multiplex, M. liechtensteini and M. bavaricus represent a well supported
monophyletic lineage.
Fig. 3. Comparison of variation of within-species nucleotide diversity (π ± SD) in the Terricola subgenus and
between-species divergence of the Microtus bavaricus/liechtensteini group based on partial cytochrome b gene
sequences (650 bp).
Terricola species
Paper 2.1.3 93
47
The primary divergence within the M. multiplex complex occurs between M. multiplex
sensu stricto and the sister species M. liechtensteini and M. bavaricus. The total and net
divergence between M. bavaricus and M. liechtensteini is the lowest observed between any
pine vole species, indicating a very recent origin of the two taxa.
The branching pattern within the M. multiplex complex, namely the sister relationship
of M. multiplex to closely related but reciprocally monophyletic M. bavaricus and M.
liechtensteini, suggests that the ancestral population of the complex survived the last
glaciations at the rims of the ice sheet covering the Alps and/or in the unglaciated
mountainous areas. The ancestral population became divided into two glacial refugia,
one situated probably in the southwestern and western Alps, while the second refugium
occurred in the south, east and north of the Alpine main ridge. This geographic isolation
caused speciation of M. multiplex and subsequently lead to a split between M. liechtensteini
and M. bavaricus. Additionally, the range of M. liechtensteini occurring north of the main
alpine ridge became fragmented leading to the apparent isolation of the contemporary
populations of M. liechtensteini in Niedere Tauern, Salzburg and Totes Gebirge, Styria
(S p i t z e n b e r g e r 2002). A similar scenario can also be proposed for the origin of the
current M. bavaricus populations.
Our cytochrome b results and the control region sequence analyses reported by
H a r i n g et al. (2000) are congruent with respect to the phylogenetic relationships of the
M. multiplex complex as well as other species of pine voles. Specifically, the data show that
M. tatricus and M. subterraneus are not closely related to the M. multiplex complex and
that they belong to other lineages of pine voles (cf. K r a t o c h v í l 1970, J a a r o l a et al.
2004). The topology of the phylogenetic tree presented here indicates an apparent discord
between molecular and chromosomal data (see Z i m a & K r á l 1984 for review) since
monophyletic lineages contain species with distinctly divergent karyotypes. For example,
the sister groups of the M. multiplex complex with 2N = 46-48 are M. tatricus (2N = 32),
M. felteni (2N = 54), and M. thomasi (2N = 40-44), whereas M. gerbei (2N = 54) is a sister
group of the Iberian species, M. duodecimcostatus and M. lusitanicus (2N = 62). Altogether,
the data strongly suggest that morphological, chromosomal and molecular evolution has
proceeded independently during pine vole evolution, and that each evolutionary process
has had its own specific rate. This model may be applied for divergence between species as
well as between populations within single species. The low mitochondrial DNA variability
observed in many pine vole species today suggests that chromosomal evolution in this
subgenus could have been facilitated by small historical population sizes.
We conclude that both our chromosomal and mitochondrial DNA findings demonstrate
a close affinity between M. bavaricus and M. liechtensteini. The pattern of evolutionary
divergence between populations of this group must have been rather complex in the past,
and some populations probably survived the last glaciation period in refugia situated in the
northern Alps.
A c k n o w l e d g e m e n t
We are grateful to Heikki H e n t t o n e n for providing a sample of Microtus liechtensteini and to two anonymous
reviewers for their helpful comments on the earlier version of the manuscript. This study was supported by a grant
from the Ministry of Education of the Czech Republic (no. LC06073) and the Carl Tryggers Foundation.
94 Paper 2.1.3
48
L I T E R A T U R E
Brunet-Lecomte P. & Volobouev V. 1994: Comparative morphometry and cytogenetics of Microtus (Terricola)
multiplex (Arvicolidae, Rodentia) of the western French Alps. Z. Säugetierkd. 59: 116–125.
Carleton M.D. & Musser G.G. 2005: Superfamily Muroidea. In: Wilson D.E. & Reeder D.M. (eds), Mammal
species of the world. A taxonomic and geographic reference. 3rd
ed. The Johns Hopkins University Press,
Baltimore: 894–1531.
Chaline J. 1987: Arvicolid data (Arvicolidae, Rodentia) and evolutionary concepts. Evol. Biol. 21: 237–310.
Galewski T., Tilak M.-K., Sanchez S., Chevret P., Paradis E. & Douzery E. J. P. 2006: The evolutionary radiation
of Arvicolinae rodents (voles and lemmings): relative contribution of nuclear and mitochondrial DNA
phylogenies. BMC Evol. Biol. 6: 80.
Graf J.-D. & MeylanA. 1980: Polymorphisme chromosomique et biochemique chez Pitymys multiplex (Mammalia,
Rodentia). Z. Säugetierkd. 45: 133–148.
Hall T.A. 1999: BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows
95/98/NT. Nucl. Acids. Symp. Ser. 41: 95–98.
Haring E., Herzig-Straschil B. & Spitzenberger F. 2000: Phylogenetic analysis of Alpine voles of the Microtus
multiplex complex using the mitochondrial control region. J. Zool. Syst. Evol. Research 38: 231–238.
Jaarola M., Martínková N., Gündüz İ., Brunhoff C., Zima J., Nadachowski A., Amori G., Bulatova N.,
Chondropoulos B., Fraguedakis-Tsolis S., González-Esteban J., Lopez-Fuster M.J., Kandaurov A., Kefelioğlu
H., Mathias M.L., Villate I. & Searle J.B. 2004: Molecular phylogeny of the speciose vole genus Microtus
(Arvicolinae, Rodentia) inferred from mitochondrial DNA sequences. Mol. Phylogenet. Evol. 33: 647–663.
Kimura M. 1980: A simple method for estimating evolutionary rate of base substitutions through comparative
studies of nucleotide sequences. J. Mol. Evol. 16: 111–120.
König C. 1962: Eine neue Wühlmaus aus der Umgebung Garmisch-Partenkirchen (Oberbayern): Pitymys
bavaricus. Senckenbergiana biol. 43: 1–10.
König C. 1982: Microtus bavaricus (König, 1962) – Bayerische Kurzohrmaus. In: Niethammer J. & Krapp F. (eds),
Handbuch der Säugetiere Europas 2/1. Akademische Verlagsges., Wiesbaden: 447–451.
Král B., Zima J. & Herzig-Straschil B. 1978: Karyotype analysis of voles of the genus Pitymys from southern
Austria. Folia Zool. 27: 129–133.
Krapp F. 1982: Microtus multiplex (Fatio, 1905) – Alpen-Kleinwühlmaus. In: Niethammer J. & Krapp F. (eds),
Handbuch der Säugetiere Europas 2/1. Akademische Verlagsges., Wiesbaden: 419–428.
Kratochvíl J. 1970: Pitymys-Arten aus der Hohen Tatra (Mamm., Rodentia). Acta Sc. Nat. Brno 4(12): 1–63.
Kumar S., Tamura K. & Nei M. 2004: MEGA3: Integrated Software for Molecular Evolutionary Genetics Analysis
and Sequence Alignment. Briefings in Bioinformatics 5: 150–163.
Martínková N. & Dudich A. 2003: The fragmented distribution range of Microtus tatricus and its evolutionary
implications. Folia Zool. 52: 11–22.
Mitchell-Jones A.J., Amori G., Bogdanowicz W., Kryštufek B., Reijnders P.J.H., Spitzenberger F., Stubbe M.,
Thissen J.B.M., Vohralík V. & Zima J. 1999: The atlas of European mammals. Academic Press, London.
Petrov B. & Živković S. 1971: Zur Kenntnis der Pitymys liechtensteini Wettstein, 1972 (Rodentia, Mammalia) in
Jugoslawien. Arhiv bioloških nauka, Beograd 23: 31–32.
Petrov B. & Živković S. 1974: Der taxonomische Status einiger Vertreter der Untergattung Pitymys (Rodentia,
Mammalia) in Jugoslawien im Lichte der Daten über ihren Karyotyp. In: Kratochvíl J. & Obrtel R. (eds),
Symposium Theriologicum II. Academia, Praha: 283–290.
Posada D. & Crandall K.A. 1998: Modeltest: testing the model of DNA substitution. Bioinformatics 14: 817–
818.
Ronquist F. & Huelsenbeck J. P. 2003: MrBayes 3: Bayesian phylogenetic inference under mixed models.
Bioinformatics 19: 1572–1574.
Rozas J., Sánchez-DelBarrio J.C., Messeguer X. & Rozas R. 2003: DnaSP, DNA polymorphism analysis by the
coalescent and other methods. Bioinformatics 19: 2496–2497.
Spitzenberger F. 2002: Bayerische Kurzohrmaus Microtus bavaricus (König, 1962); Illyrische Kurzohrmaus
Microtus liechtensteini (Wettstein, 1927). In: Spitzenberger F. (ed.), Die Säugetierfauna Österreichs. Grüne
Reihe des Bundesministeriums für Land- und Forstwirtschaft, Umwelt und Wasserwirtschaft, Band 13, Graz:
441–449.
Spitzenberger F., Brunet-Lecomte P., Nadachowski A. & Bauer K. 2000: Comparative morphometrics of the first
lower molar in Microtus (Terricola) cf. liechtensteini of the Eastern Alps. Acta Theriol. 45: 471–483.
Paper 2.1.3 95
49
Storch G. & Winking H. 1977: Zur Systematik der Pitymys multiplex-Pitymys liechtensteini-Gruppe (Mammalia,
Rodentia). Z. Säugetierkd. 42: 78–88.
Swofford D.L. 2003: PAUP*. Phylogenetic Analysis Using Parsimony (*and Other Methods). Version 4. Sinauer
Associates, Sunderland, Massachusetts.
Tavaré S. 1986: Some probabilistic and statistical problems on the analysis of DNA sequences. Lec. Math. Life
Sci. 17: 57–86.
Tougard C., Brunet-Lecomte P., Fabre M. & Montuire S. 2007: Evolutionary history of two allopatric Terricola
species (Arvicolinae, Rodentia) from molecular, morphological and palaeontological data. Biol. J. Linn. Soc.,
in press.
Zima J. & Král B. 1984: Karyotypes of European mammals II. Acta Sc. Nat. Brno 18(8): 1–62.
96 Paper 2.1.3
Paper 2.1.4
Pečnerová P., Martínková N. 2012. Evolutionary history of tree squirrels (Rodentia,
Sciurini) based on multilocus phylogeny reconstruction. Zoologica Scripta 41: 211-
219.
– 97 –
Evolutionary history of tree squirrels (Rodentia, Sciurini)
based on multilocus phylogeny reconstruction
PATRI´CIA PECˇ NEROVA´ & NATA´ LIA MARTI´NKOVA´
Submitted: 6 October 2011
Accepted: 13 December 2011
doi:10.1111/j.1463-6409.2011.00528.x
Pecˇnerova´, P. & Martı´nkova´, N. (2012). Evolutionary history of tree squirrels (Rodentia,
Sciurini) based on multilocus phylogeny reconstruction. —Zoologica Scripta, 41, 211–219.
Tree squirrels of the tribe Sciurini represent a group with unresolved phylogenetic relationships
in gene trees. We used partial sequences of mitochondrial genes for 12S rRNA,
16S rRNA, cytochrome b and d-loop, and nuclear irbp, c-myc exon 2 and 3 and rag1 genes
to reconstruct phylogenetic relationships within the tribe, maximizing the number of analysed
species. Bayesian inference analysis of the concatenated sequences revealed common
trends that were similar to those retrieved with supertree reconstruction. We confirmed
congruence between phylogeny and zoogeography. The first group that diverged from a
common ancestor was genus Tamiasciurus, followed by Palaearctic Sciurus and Indomalayan
Rheithrosciurus macrotis. Nearctic and Neotropical Sciurus species formed a monophyletic
group that included Microsciurus and Syntheosciurus. Neotropical Sciurini were monophyletic
with a putative exception of Syntheosciurus brochus that was included in a polychotomy
with Nearctic Sciurus in supertree analyses. Our data indicate that Sciurini tree squirrels
originated in the northern hemisphere and ancestors of contemporary taxa attained their
current distribution through overland colonization from the nearest continent rather than
through trans-Pacific dispersal.
Corresponding author: Nata´lia Martı´nkova´, Institute of Vertebrate Biology, Academy of Sciences
of the Czech Republic, v.v.i., Kveˇtna´ 8, 603 65 Brno, Czech Republic. E-mail: martinkova@ivb.cz
and Institute of Biostatistics and Analyses, Masaryk University, Kamenice 3, 625 00 Brno, Czech
Republic. E-mail: martinkova@ivb.cz
Patrı´cia Pecˇnerova´, Department of Botany and Zoology, Masaryk University, Kotla´rˇska´ 2, 602 00
Brno, Czech Republic. E-mail: pata.pecnerova@gmail.com
Introduction
Tree squirrels of the tribe Sciurini are sciuromorph
rodents that favour arboreal habitat. They are common in
temperate forests of the northern hemisphere, where they
feed on seeds, often storing them in underground food
caches. In the tropical forests, Sciurini tree squirrels are
found in Borneo and in Central and South America
(Nowak 1999).
Five genera belong to the tribe, Microsciurus, Sciurus,
Syntheosciurus, Rheithrosciurus and Tamiasciurus. Most of the
37 currently recognized species of the tribe are distributed
in Central America and northern part of South America,
and only four species inhabit the large area of Eurasia, one
of which is found in Borneo (Nowak 1999; Wilson &
Reeder 2005).
Distribution of Sciurini squirrels exhibits latitudinal gradient
in species richness (Wiens et al. 2006; Mittelbach
et al. 2007; Fuhrman et al. 2008) with higher number of
species occurring in the tropics than in more northern latitudes.
In Sciuridae, the high species richness in equatorial
regions on different continents is probably a result of
higher diversification rate in the tropics based on higher
lineage ‘birth’ and ‘death’ rates (Roth & Mercer 2008).
Other hypotheses that explain the species richness gradient
from tropics to the poles assume that lineages in the tropics
are older or occupy larger areas facilitating lineage
diversification through historical and spatial processes that
influence populations (Stebbins 1974; Rosenzweig 1995;
Chown & Gaston 2000; Fine 2001; Stephens & Wiens
2003; Wiens & Donoghue 2004; Wiens et al. 2006).
The earliest known well-preserved fossil of the family
Sciuridae is Douglassciurus jeffersoni from North America
that lived about 36 million years ago (Mya; Emry & Thorington
1982; Mercer & Roth 2003). Diversification of the
tribe Sciurini began later with divergence of the North
American genus Tamiasciurus about 13 Mya (Mercer &
ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters, 41, 3, May 2012, pp 211–219 211
Zoologica Scripta
98 Paper 2.1.4
Roth 2003). This date, based on multilocus molecular dating
(Mercer & Roth 2003), coincides with the oldest
Sciurus fossil, S. olsoni, which was dated to the Clarendonian
period (approximately 10–13 Mya) in Nevada (Emry
et al. 2005). However, Oshida et al. (2009) suggested that
Eurasia might be the most probable place of early divergence
of the genus Sciurus with subsequent dispersal
through Beringia to North America. If the tribe Sciurini
originated in North America, this scenario would assume
that common ancestors of Sciurus dispersed to Eurasia
after diversification of Tamiasciurus, diverged, and subsequently,
as Sciurus, returned to North America conditioned
by opening of the Bering Strait. Alternatively, as
Eurasian Sciurini squirrels are not monophyletic based on
mitochondrial DNA phylogeny (Oshida et al. 2009), their
ancestors could have crossed from North America to Eurasia
multiple times. In both cases, the extension of the
genus Sciurus proceeded to Central America and, after formation
of the Isthmus of Panama about three Mya, also to
South America (Fig. 1A). In tropical parts of Central and
South America, the Sciurini squirrels reach the highest
levels of species richness.
Support for a second scenario assuming Eurasian origin
of the tribe Sciurini can be found again in the fossil
record. The earliest fossil with evident traits of modern
squirrels was Palaeosciurus from the Oligocene period in
France that was most likely a ground dweller (Thorington
& Ferrell 2006; Michaux et al. 2008). It is considered a
putative ancestor of Sciurini and Pteromyini (Thorington
& Santana 2007). In this scenario, the group originated in
Eurasia and colonized the Americas twice. First, the ancestors
of contemporary Tamiasciurus and the second time the
ancestors of Sciurus sensu lato crossed Beringia and further
diversified in the Americas (Fig. 1B).
This explains the evolutionary history of the genera with
known fossil record, Sciurus and Tamiasciurus, but not the
other currently recognized genera belonging to Sciurini.
Microsciurus, Rheithrosciurus and Syntheosciurus are distributed
in Central and South America, and in south-east Asia
(Wilson & Reeder 2005), and their direct fossil ancestors
are not known from the area of their current distribution.
The tropical regions are generally unsuitable for fossil
preservation (Tappen 1994); hence, the timing of colonization
is open to speculation. The lineage of ancestors of
Rheithrosciurus occupying Eurasia is either extinct without
known fossil remains or the genus supposedly colonized
Borneo in an independent long-distance colonization event
(Mercer & Roth 2003). Genera Microsciurus and Syntheosciurus
from South America are considered late colonizers
together with the extension of the genus Sciurus following
the opening of the Isthmus of Panama (Mercer & Roth
2003; Roth & Mercer 2008).
Our intention was to establish the phylogenetic relationships
between species of the tribe Sciurini in a comprehensive
way, using supertree reconstruction algorithms and
Bayesian inference analysis of a supermatrix of multiple
gene alignments. We analysed all species and loci that had
sufficient number and length of available DNA sequences
to increase robustness of the analyses. We confirmed the
origin of the group in the northern hemisphere with two
colonization events across Beringia. Sciurini squirrels from
the Neotropic ecozone including species from Sciurus,
Microsciurus and Syntheosciurus formed a monophyletic
group, indicating recent rapid diversification in South and
Central America, but Rheithrosciurus diverged close to the
base of Sciurini phylogeny indicating an overland colonization
of Borneo from Asia.
Materials and methods
We analysed partial sequences of eight loci (12S rRNA,
16S rRNA, mt-cyb, d-loop, irbp, c-myc exon 2, c-myc exon 3,
rag1), both mitochondrial and nuclear, that were publicly
available (Wettstein et al. 1995; Oshida et al. 1996; Matthee
& Robinson 1997; Barratt et al. 1999; Bentz & Montgelard
1999; Arbogast et al. 2001; Montgelard et al. 2002; Harrison
et al. 2003; Lance et al. 2003; Mercer & Roth 2003;
<13 Mya<13 Mya
<3 Mya
<13 Mya<13 Mya
<3 Mya
3. Sciurus
expansion
1. Tamiasciurus
divergence
2. Sciurus
divergence
4. Sciurus
expansion
4. Sciurus
expansion
2. Tamiasciurus
divergence
3. Sciurus
divergence
4. Sciurus
expansion
1. Origin of
Sciurini
A
B
Fig. 1 Colonization and diversification scenarios of the tribe
Sciurini. —A. Origin of Sciurini in North America. —B. Origin
of Sciurini in Eurasia. The indicated dates are based on molecular
dating analysis by Mercer & Roth (2003) and opening of the
Isthmus of Panama for passage to South America.
Phylogeny of tree squirrels d P. Pecˇnerova´ & N. Martı´nkova´
212 ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters, 41, 3, May 2012, pp 211–219
Paper 2.1.4 99
Steppan et al. 2004; Lee et al. 2008; Blanga-Kanfi et al.
2009; Oshida et al. 2009; Chavez et al. 2011; B. R. Barber,
unpubl. data; R. D. Bradley, G. Ceballos, P. Manzano,
F. M. Mendez-Harclerode, M. L. Haynie & D. H. Walker,
unpubl. data; A. A. Neverov & D. V. Volokhov, unpubl.
data; P. D. Sudman & M. S. Hafner, unpubl. data; L. J.
Uimaniemi, M. I. Orell & P. O. Reunanen unpubl. data;
N. Yaekashiwa & H. B. Tamate, unpubl. data). We used
data sets with partially overlapping species content, comprising
in total of 19 species of the tribe Sciurini and two
outgroup taxa. Samples of all recently recognized genera of
the tribe (Microsciurus, Rheithrosciurus, Sciurus, Syntheosciurus,
Tamiasciurus) were included. Gene data sets included multiple
sequences per species (data not shown). For final analyses,
the gene data sets were purged to include one sequence
per species (see below for details). Accession numbers of
sequences in the final data set are available in Table S1.
We aligned the DNA sequences in GENEIOUS 4.7
(Biomatters Ltd., Auckland, New Zealand; Drummond
et al. 2009). To estimate phylogenetic relationships for each
locus separately, we calculated Bayesian inference analysis
for every data set in MRBAYES 3.1.2 (Huelsenbeck & Ronquist
2001), utilizing substitution models selected by the
Bayesian Information Criterion in MODELTEST 3.7 (Table 1;
Posada & Crandall 1998). To optimize chain convergence,
we ran five Markov chains Monte Carlo (MCMC) for
2 million generations, sampling trees every 1000th generation.
Chain heating parameter was 0.1, and 1 chain swap
was attempted every 3rd generation. The burn-in fraction
was set to 30%. The outgroup taxa included Glaucomys volans
and Pteromys volans, which were previously identified as
sister taxa to Sciurini (e.g. Montgelard et al. 2002). The
gene trees were assessed visually for intraspecific variability,
and single representatives of each species were chosen. For
the supertrees approach, the gene trees were purged to
include one tree leaf per species, and for the supermatrix
approach, we used the respective sequence to concatenate
the chimeric sequence to represent the species.
To estimate phylogenetic relationships between all sampled
taxa, we used both supermatrix and supertree
approaches. The supermatrix contained concatenated
sequences of the eight loci, and we inferred the Bayesian
phylogenetic tree using partitions corresponding to the
loci. The substitution models for each partition were
the same as were used for the gene tree inferences, where
the model parameters were unlinked. The MCMC ran for
4 million generations, and the chain temperature was 0.08.
We used the gene trees to reconstruct supertrees that
would include all taxa with available DNA sequence data.
We applied six methods: SuperTriplets (Ranwez et al.
2010), MinCut (Semple & Steel 2000), modified MinCut
supertrees (Page 2002), standard matrix representation
with parsimony (MRP; Baum 1992; Ragan 1992), PurvisMRP
(Purvis-MRP; Purvis 1995) and veto supertree
reconstruction (Scornavacca et al. 2008). The supertree
reconstruction methods combine information from gene
trees where each individual gene tree contains a subsample
of taxa.
The SuperTriplets method searches for a supertree that
is based on decomposition of the source trees to the simplest
trees, the triplets (Ranwez et al. 2010). The resulting
supertree contains medians of triplet relationships as
resolved in the source trees. We conducted the analysis in
the program SuperTriplets (Ranwez et al. 2010).
The MinCut (Semple & Steel 2000) and modified MinCut
(Page 2002) are robust to contradictions in the source
trees. The algorithms convert the source trees into a graph
of edges with allocated weights. The weights are assigned
based on the number of trees where an edge occurs in one
cluster. The supertree is constructed from the graph after
execution of a minimum cut that discards relationships
with the lower than threshold weight. The MinCut
method does not include information uncontradicted in
the source trees, which modified MinCut algorithm
accounts for (Page 2002). We worked with these algorithms
in the Supertree software (Page 2002).
Table 1 Proportion of missing data in gene alignments and substitution models for each analysed locus with model parameter means
estimated from their Bayesian posterior probability distribution
Locus Species Sequences Missing data (%) Model j a I
12S rRNA 10 16 25.6 GTR + C + I N ⁄ A 0.529 0.448
16S rRNA 9 13 34.6 GTR + C N ⁄ A 0.203 N ⁄ A
mt-cyb 14 365 25.8 GTR + C + I N ⁄ A 0.243 0.329
d-loop 8 378 66.6 GTR + C + I N ⁄ A 86.6 0.423
irbp 15 16 5.3 GTR + C N ⁄ A 0.149 N ⁄ A
c-myc exon 2 5 5 5.7 HKY 4.475 N ⁄ A N ⁄ A
c-myc exon 3 6 6 1.3 HKY 5.389 N ⁄ A N ⁄ A
rag1 6 7 34.9 HKY + I 6.049 N ⁄ A 0.634
j – transition ⁄ transversion rate ratio, a – shape parameter of the C distribution, I – proportion of invariable sites, N ⁄ A – not available.
P. Pecˇnerova´ & N. Martı´nkova´ d Phylogeny of tree squirrels
ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters, 41, 3, May 2012, pp 211–219 213
100 Paper 2.1.4
Matrix representation with parsimony and Purvis-MRP
are consensus methods, where taxa relationships defined in
the phylogenetic trees are translated into a binary matrix.
We constructed the matrix representation in r8s 1.70
(Sanderson 2003) and analysed it using maximum parsimony
in PAUP* 4b10 (Sinauer Associates, Inc., Sunderland,
MA) with 10 heuristic search replicates and TBR branch
swapping algorithm. Maximum number of the swapped
trees was limited to 10 000. The final tree was constructed
as a 50% majority rule consensus.
In the veto method, as implemented in PhySIC_IST,
the resulting supertree is a combination of the relationships
agreed upon by all source trees (Ranwez et al. 2007;
Scornavacca et al. 2008) as opposed to the relationships
most favoured by the source trees in the other supertree
reconstruction algorithms. We used the veto supertree
method without the source tree correction preprocess and
with the correction of the source trees where less frequent
triplets are dropped if they contradict more frequent ones.
Results
Gene trees
The eight analysed loci represented 9065 base pairs (bp)
of genomic sequence. T. hudsonicus was the only ingroup
species represented in all gene data sets. The alignment of
partial 12S rRNA gene was 851 bp long and contained ten
taxa, 16S rRNA 1058 bp for nine taxa, mt-cyb 1140 bp for
14 taxa, d-loop 1254 bp for eight taxa, irbp 1180 bp for 15
taxa, c-myc exon 2 674 bp for five taxa, c-myc exon 3
956 bp for six taxa, and the alignment for the partial rag1
gene was 2141 bp long for six taxa. Six gene alignments
contained multiple sequences per species, including
sequences that were shorter than the overall gene alignment.
This resulted in missing data content (alignment
gaps) ranging from 1.3% in the c-myc exon 3 alignment to
66.6% in the d-loop alignment (Table 1).
Presented gene trees were purged of intraspecific variability
to include one representative from within each species
clade. Six of eight phylogenetic trees yielded by
Bayesian inference analyses of gene alignments revealed a
similar basic pattern in the distribution of taxa with successive
divergence of Tamiasciurus, followed by Palaearctic,
Nearctic and Neotropical species of Sciurini (Fig. 2). The
gene tree for 12S rRNA showed Tamiasciurus included in
a poorly resolved group of New World taxa (Fig. 2A) and
the d-loop tree depicted a basal polychotomy of Sciurini
tree squirrels (Fig. 2D). Unresolved polychotomies were
present in all data sets, but several relationships were supported
[Bayesian posterior probability (BPP) ‡ 0.95] in
individual gene trees.
Sciurus lis and S. vulgaris were confirmed as sister taxa
based on the 12S rRNA, mt-cyb and d-loop gene trees.
According to the mt-cyb gene tree, S. anomalus diverged
after split of the ancestor of S. lis and S. vulgaris. The
other Palaearctic genus, represented by Rheithrosciurus macrotis
(12S rRNA, 16S rRNA and irbp), diverged early but
without significant support of its relationship to Palaearctic
Sciurus. New World taxa formed supported monophyletic
groups in five data sets (16S rRNA, mt-cyb, c-myc
exon 2, c-myc exon 3 and rag1). In the irbp gene tree, the
support for monophyly of the New World group was low
(BPP = 0.69). In addition to the differentiation of the New
World clade, Neotropical taxa were supported as a monophylum
in mt-cyb and c-myc exon 3 trees and occurred as a
distinct, unsupported lineage in irbp (BPP = 0.45) and rag1
(BPP = 0.77) trees. Relationships among the Nearctic and
Neotropical species were unresolved (Fig. 2).
To create the concatenated data set, one sequence per
species per gene was chosen. The selected sequence was as
long as possible for the given gene and its position with
respect to other species is shown on the gene trees
(Fig. 2). The concatenated data set had 64.4% of missing
data. Bayesian phylogenetic tree based on the concatenated
data set distinguished two main groups, Tamiasciurus and a
lineage including Sciurus, Rheithrosciurus, Microsciurus and
Syntheosciurus (Fig. 3A). In Tamiasciurus, the phylogeny
showed significant relationships, where T. hudsonicus was a
sister species to a lineage including T. douglasii and T. mearnsi.
Within the latter group, the New World tree squirrels
formed a significantly supported monophyletic group
with the Nearctic taxa branching first. The analysis
retrieved a monophyletic lineage containing all Neotropical
taxa, but relationships within the group were mostly
unresolved. Of five pairs of sister taxa apparent in the tree,
only S. vulgaris with S. lis and S. aestuans with S. ignitus
were significantly supported (Fig. 3).
The diverse methods of supertree reconstruction partially
coincided in the branching pattern with the Bayesian
tree inferred from the concatenated sequence data set
(Figs 3B and S1). In particular, SuperTriplets, modified
MinCut, MRP and veto supertree without source tree correction
preprocessing showed successive diversification of
Tamiasciurus, Palaearctic ⁄ Indomalayan, Nearctic and Neotropical
species with Sy. brochus included in early radiation
of the New World taxa (Figs 3B and S1B,C,E). The MinCut
supertree differed in showing successive diversification
of Tamiasciurus species at the base of the Sciurini clade
(Fig. S1A) and veto supertree with source tree correction
preprocessing included Tamiasciurus in a polychotomy with
S. lis and S. vulgaris (Fig. S1F). The Purvis-MRP supertree
showed heretic relationships of the ingroup taxa that were
not present in the other trees (Fig. S1D).
The veto supertrees eliminated several taxa because of
conflict in source trees. In the veto supertree without
Phylogeny of tree squirrels d P. Pecˇnerova´ & N. Martı´nkova´
214 ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters, 41, 3, May 2012, pp 211–219
Paper 2.1.4 101
preprocessing, M. flaviventer and S. carolinensis were
excluded (Fig. S1E). The veto supertree with preprocessing
did not include five taxa, S. aestuans, S. carolinensis,
S. niger, Sy. brochus and R. macrotis (Fig. S1F).
Discussion
Most trees showed one common pattern in Sciurini phylogeny,
where Tamiasciurus diverged first in the radiation
of Sciurini and the other ingroup species were organized
according to their geographic distribution rather than taxonomic
status. In wider phylogenetic context, Tamiasciurus
is consistently placed at the root of the tribe Sciurini
(Hafner et al. 1994; Mercer & Roth 2003; Herron et al.
2004; Steppan et al. 2004; Villalobos & Cervantes-Reza
2007; Roth & Mercer 2008). We included three species
belonging to the genus, and we confirmed close relationships
of T. douglasii and T. mearnsi with T. hudsonicus
(Arbogast et al. 2001; Herron et al. 2004; Chavez et al. 2011).
Three Palaearctic Sciurus taxa (S. anomalus, S. lis and
S. vulgaris) and R. macrotis from Borneo branched close to
the base of the trees. The New World squirrels formed a
monophyletic group. Within the New World clade, there
was a gradual transition from the Nearctic to the Neotropical
species. The Nearctic species (S. aberti, S. carolinensis,
S. griseus and S. niger) were paraphyletic with respect
to the monophyletic clade comprising the Neotropical
species (M. alfari, M. flaviventer, S. aestuans, S. granatensis,
S. ignitus, S. stramineus and S. variegatoides). Syntheosciurus
A
0.02
Syntheosciurus brochus
Glaucomys volans
S. lis
Pteromys volans
Tamiasciurus hudsonicus
S. niger
S. vulgaris
Sciurus aestuans
Microsciurus flaviventer
Rheithrosciurus macrotis
*
*
*
* B
0.02
T. hudsonicus
S. vulgaris
M. flaviventer
P. volans
R. macrotis
G. volans
S. niger
S. carolinensis
Sy. brochus
*
*
*
*
C
1.0
M. flaviventer
S. aberti
T. hudsonicus
G. volans
S. stramineus
S. vulgaris
S. aestuans
S. niger
P. volans
S. lis
S. carolinensis
S. anomalus
T. douglasii
T. mearnsi
*
*
*
*
*
*
*
*
*
D
0.05
S. niger
G. volans
S. carolinensis
S. lis
T. hudsonicus
S. vulgaris
T. douglasii
P. volans
*
*
E
0.01
S. variegatoides
S. griseus
G. volans
T. hudsonicus
S. stramineus
S. granatensis
M. flaviventer
Sy. brochus
S. aestuans
S. vulgaris
M. alfari
S. aberti
S. ignitus
P. volans
R. macrotis
*
*
*
F
0.01
M. flaviventer
S. stramineus
T. hudsonicus
G. volans
S. carolinensis*
G
0.01
S. stramineus
M. flaviventer
S. carolinensis
S. ignitus
T. hudsonicus
G. volans
*
*
0.01
S. ignitus
M. flaviventer
G. volans
S. carolinensis
T. hudsonicus
S. stramineus
H
*
*
Fig. 2 Bayesian phylogenetic trees of Sciurini based on the mitochondrial and nuclear gene sequences. —A. 12S rRNA. —B. 16S rRNA.
—C. mt-cyb. —D. d-loop. —E. irbp. —F. c-myc exon 2. —G. c-myc exon 3. —H. rag1. The trees were purged from phylogenies that
included multiple sequences per species. See text for details. The stars above edges indicate significantly supported relationships
(BPP ‡ 0.95). All scale bars show substitutions site)1
.
P. Pecˇnerova´ & N. Martı´nkova´ d Phylogeny of tree squirrels
ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters, 41, 3, May 2012, pp 211–219 215
102 Paper 2.1.4
brochus was included in the Neotropical group in the
Bayesian analysis of the concatenated data set, but it was
placed among Nearctic taxa in the supertrees. This discrepancy
between supermatrix and supertrees approaches
was probably a consequence of variable placement of
Sy. brochus with respect to S. aestuans in the 12S rRNA
and the irbp gene trees.
Rheithrosciurus is an island species from Borneo, geographically
isolated from the nearest representatives of
Sciurini that occur in Japan and north-east China (Nowak
1999; Wilson & Reeder 2005). In our study, R. macrotis
branched early in the trees and diverged after the split of
ancestors of Tamiasciurus, possibly together with initial
diversification of Sciurus in Eurasia (see also Mercer &
Roth 2003). We believe that R. macrotis diverged from a
common ancestor as early as the Palaearctic species of
the genus Sciurus and colonized Borneo overland from
south-east Asia. During the evolution of Sciuridae in the
last 36 million years, Borneo fauna benefited from multiple
land connections to continental Asia (Hall 2001), and
given historical biogeography of other arboreal animals
from Borneo (examples in Metcalfe et al. 2001), the area
was forested at least during some of those periods. In
light of our reconstruction of the phylogenetic position
of R. macrotis, the overland colonization hypothesis seems
more plausible than trans-Pacific colonization from the
Americas.
Two species of Microsciurus, which we included in this
study, form a close relationship with Central and South
American species of the genus Sciurus. M. flaviventer was
included in a supported group with several Neotropical
taxa in the mt-cyb, irbp and c-myc exon 3 gene trees. Both
M. flaviventer and M. alfari were sampled in the irbp gene
tree, but this tree supported monophyly of Sciurini and a
group including M. flaviventer, S. aestuans and S. ignitus.
Other relationships were not significant. The Bayesian
phylogeny inferred from the concatenated data set and the
supertrees showed M. flaviventer close to the base of
the Neotropical clade and M. alfari deeper in the group.
The irbp gene tree showed an unsupported sister relationship
of M. alfari with S. granatensis that was also present
in the analysis by Villalobos & Cervantes-Reza (2007). We
treat this pattern with caution because M. alfari was sampled
for the irbp gene only, and this gene tree was overall
poorly resolved.
Phylogenetic position of the genus Syntheosciurus varies
in trees obtained with different methods. In the Bayesian
tree based on the concatenated data set, Sy. brochus
belonged to a basal polychotomy of the Neotropical clade,
and in four supertrees, it formed a polychotomy with different
Nearctic species. The position of Sy. brochus was
based on the 12S, 16S rRNA and irbp gene trees, where
the 12S rRNA gene tree showed its supported sister relationship
with S. aestuans, but in the irbp gene tree, the two
A
0.05
S. lis
T. mearnsi
S. ignitus
Tamiasciurus douglasii
S. aberti
S. niger
S. variegatoides
Syntheosciurus brochus
M. flaviventer
Sciurus aestuans
S. granatensis
Rheithrosciurus macrotis
S. stramineus
S .griseus
T. hudsonicus
S. vulgaris
Glaucomys volans
Microsciurus alfari
S. anomalus
S. carolinensis
0.96
0.6
1.0
1.0
1.0
0.95
0.67
1.0
1.0
0.85
1.0
1.0
0.85
0.96
Pteromys volans Outgroup
B
S. aestuans
S. aberti
T. douglasii
M. flaviventer
Sy. brochus
T. mearnsi
G. volans
M. alfari
R. macrotis
S. granatensis
S. niger
S. variegatoides
S. lis
S. carolinensis
S. anomalus
T. hudsonicus
S. ignitus
S. griseus
S. vulgaris
S. stramineus
P. volans
85
67
83
100
100
100
100
100
100
100
100
100
100
67
100
Fig. 3 Multilocus phylogenies of the Sciurini tribe based on information from four mitochondrial and four nuclear loci as depicted in
Fig. 2. —A. Bayesian inference phylogenetic tree based on the concatenated supermatrix. The numbers above edges indicate Bayesian
posterior probability and the scale bar represents substitutions site)1
. —B. SuperTriples supertree based on gene trees in Fig. 2. The
numbers above edges indicate percentage of triplets that support the edge in the SuperTriplets analysis. The dashed vertical bar indicates
a Neotropical species that grouped with the Nearctic species. Maps of the biogeographic ecozones are not to scale.
Phylogeny of tree squirrels d P. Pecˇnerova´ & N. Martı´nkova´
216 ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters, 41, 3, May 2012, pp 211–219
Paper 2.1.4 103
species were separate and the tree indicated a more basal
position of Sy. brochus in a polychotomy with North American
S. aberti and S. griseus. In the 16S rRNA gene tree,
Sy. brochus belonged to a polychotomy that included also
M. flaviventer, S. carolinensis and S. niger. Previously suggested
sister relationship with S. variegatoides (Villalobos &
Cervantes-Reza 2007) was not confirmed in any of our
trees.
Evolutionary history
The colonization pathways of Sciurini tree squirrels
started most likely in the northern hemisphere. Tamiasciurus
was a basal group of Sciurini in all our analyses with
the exception of the 12S rRNA gene tree. The genus Tamiasciurus
is distributed in North America, and its basal
position in the phylogeny indicates origin of the group in
the northern hemisphere. This is supported by the fossil
record as known fossils attributed to the tribe were found
in North America and Eurasia, but not in tropical regions
(Emry & Thorington 1982; Emry et al. 2005). Our results
do not conclusively distinguish whether the tribe originated
in Eurasia or North America. Both continents could
be the regions where early diversification of the tribe
occurred, the alternative scenarios differing in the directions
of the following colonization events across Beringia
(Fig. 1).
First, phylogeny of Sciurini tree squirrels indicated that
the ancestor of the tribe first diverged in North America
and subsequently colonized Eurasia (Oshida et al. 2009).
After further differentiation in Eurasia, during which time
the ancestors of Rheithrosciurus entered Sundaland, the
Sciurus squirrels returned to North America and spread
southwards, which manifests in our multilocus phylogeny
as a supported monophyletic group of New World Sciurus
with Microsciurus and Syntheosciurus.
Second, broader analyses of Sciuridae showed that the
sister group of Sciurini was Pteromyini, and members of
this tribe are primarily distributed in the Palaearctic
region (Montgelard et al. 2002; Mercer & Roth 2003;
Herron et al. 2004; Steppan et al. 2004; Michaux et al.
2008; Roth & Mercer 2008). If an ancestor of Sciurini was
distributed in Eurasia, two eastward colonization events
across Beringia would be needed to explain the phylogeny
we present here; Tamiasciurus first, followed by ancestors
of New World Sciurus.
Fossil record indicates that Sciurus originated in North
America, supporting the latter colonization scenario. The
earliest Sciurus is known from late Miocene in the US, but
a representative of the genus appeared as late as Pliocene
in Europe (Emry et al. 2005), making the oldest Nearctic
tree squirrel fossil up to 10 million years older than the
oldest Palaearctic one. If the known fossils represented the
true diversification history of the group, we should observe
a similar branching pattern in the Old World as we see in
the Neotropical squirrels. A more recent colonization of
Eurasia should appear as a monophyletic lineage with
short branches that split deep in the phylogenetic tree.
Instead, the Old World taxa are basal. They do not form
a monophyletic group, and therefore, we are reluctant to
accept a single westward colonization event of a common
ancestor of Old World Sciurini squirrels from North
America. Multiple independent lineages could explain the
observed trees. If multiple lineages entered Eurasia at
approximately the same time, they would diversify into the
four species, R. macrotis, S. anomalus and a lineage that differentiated
into S. vulgaris and S. lis. But such a scenario is
speculative, as the timing of the diversification would need
to occur after the split of the ancestors of Tamiasciurus,
but before Sciurus began to diversify in North America.
Rather, we assume that the genus Sciurus first diverged in
Eurasia. The squirrels then spread eastwards to North
America and colonized the continent as far south as Central
America. Expansion to South America occurred last,
in time with the formation of the Isthmus of Panama (as
dated in Mercer & Roth 2003). The species of Microsciurus
and possibly also Syntheosciurus originated in tropical
South America after that, and the tribe attained its highest
diversification in the region.
The fact that Sciurini tree squirrels from South and
Central America form a monophyletic group of closely
related taxa in our analyses shows that the latitudinal species
richness gradient of the tree squirrels formed recently,
probably due to the remarkably high diversification rate
suggested by Roth & Mercer (2008). Phylogeography of
species from the tribe Sciurini from higher latitudes illustrates
that periods of geographic isolation exacerbate intraspecific
differentiation. For example, S. vulgaris lineage
endemic to Calabria in southern Italy differs from other
S. vulgaris populations in Europe by about 2% in mt-cyb
gene sequence data, whereas diversity elsewhere in Europe
is about 0.3% (Grill et al. 2009). Grill et al. (2009)
explained the difference as a result of an isolation of the
Calabrian population during Pleistocene glaciations. Phylogeographic
structure of S. niger indicates a rapid and
recent population expansion, co-occurring with size and
coat diversification attributed to postglacial range expansion
after the last glaciation (Moncrief et al. 2010). Pleistocene
glacial cycles probably played a critical role in
differentiation of T. hudsonicus and T. douglasii, where their
genetic admixture at present is limited because of different
habitat utilization (Arbogast et al. 2001; Chavez et al.
2011). At the same time, population viability simulations
show that a very small number of tree squirrels can establish
a new population (Wood et al. 2007). We propose that
P. Pecˇnerova´ & N. Martı´nkova´ d Phylogeny of tree squirrels
ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters, 41, 3, May 2012, pp 211–219 217
104 Paper 2.1.4
localized foci established from a limited number of founders
that rapidly adapted to specific environmental conditions
could have facilitated explosive diversification of
Sciurini tree squirrels in Central and South America.
Acknowledgements
We thank Peter Vallo and two anonymous reviewers for
commenting on the earlier draft of this manuscript. The
analyses were conducted on a computational cluster of the
Institute of Vertebrate Biology. This study was funded
from grant number AV0Z60930519 provided by the Academy
of Sciences of the Czech Republic.
References
Arbogast, B. S., Browne, R. A. & Weigl, P. D. (2001).
Evolutionary genetics and Pleistocene biogeography of North
American tree squirrels (Tamiasciurus). Journal of Mammalogy,
82, 302–319.
Barratt, E. M., Gurnell, J., Malarky, G., Deaville, R. & Bruford,
M. W. (1999). Genetic structure of fragmented populations of
red squirrel (Sciurus vulgaris) in the UK. Molecular Ecology, 8,
S55–S63.
Baum, B. R. (1992). Combining trees as a way of combining data
sets for phylogenetic inference, and the desirability of
combining gene trees. Taxon, 41, 3–10.
Bentz, S. & Montgelard, C. (1999). Systematic position of the
African dormouse Graphiurus (Rodentia, Gliridae) assessed from
cytochrome b and 12S rRNA mitochondrial genes. Journal of
Mammalian Evolution, 6, 67–83.
Blanga-Kanfi, S., Miranda, H., Penn, O., Pupko, T., DeBry, R.
W. & Huchon, D. (2009). Rodent phylogeny revised: analysis
of six nuclear genes from all major rodent clades. BMC
Evolutionary Biology, 9, 71.
Chavez, A. S., Saltzberg, C. J. & Kenagy, G. J. (2011). Genetic
and phenotypic variation across a hybrid zone between
ecologically divergent tree squirrels (Tamiasciurus). Molecular
Ecology, 20, 3350–3366.
Chown, S. L. & Gaston, K. J. (2000). Areas, cradles and
museums: the latitudinal gradient in species richness. Trends in
Ecology and Evolution, 15, 311–315.
Drummond, A. J., Ashton, B., Cheung, M., Heled, J., Kearse, M.,
Moir, R., Stones-Havas, S., Thierer, T. & Wilson, A. (2009).
Geneious v4.7. Available via http://www.geneious.com/.
Emry, R. J. & Thorington, R. W., Jr (1982). Descriptive and
comparative osteology of the oldest fossil squirrel, Protosciurus
(Rodentia, Sciuridae). Smithsonian Contributions to Paleobiology,
47, 1–35.
Emry, R. J., Korth, W. W. & Bell, M. A. (2005). A tree squirrel
(Rodentia, Sciuridae, Sciurini) from the Late Miocene
(Clarendonian) of Nevada. Journal of Vertebrate Paleontology, 25,
228–235.
Fine, P. V. A. (2001). An evaluation of the geographic area
hypothesis using the latitudinal gradient in North American
tree diversity. Evolutionary Ecology Research, 3, 413–428.
Fuhrman, J. A., Steele, J. A., Hewson, I., Schwalbach, M. S.,
Brown, M. V., Green, J. L. & Brown, J. H. (2008). A
latitudinal diversity gradient in planktonic marine bacteria.
Proceedings of the National Academy of Sciences of the United States
of America, 105, 7774–7778.
Grill, A., Amori, G., Aloise, G., Lisi, I., Tosi, G., Wauters, L. A.
& Randi, E. (2009). Molecular phylogeography of European
Sciurus vulgaris: refuge within refugia? Molecular Ecology, 18,
2687–2699.
Hafner, M. S., Barkley, L. J. & Chupasko, J. M. (1994).
Evolutionary genetics of New World tree squirrels (tribe
Sciurini). Journal of Mammalogy, 75, 102–109.
Hall, R. (2001). Cenozoic reconstructions of SE Asia and the SW
Pacific: changing patterns of land and sea. In I. Metcalfe, J. M.
B. Smith, M. Morwood & I. D. Davidson (Eds) Faunal and
Floral Migrations and Evolution in SE Asia – Australasia (pp. 35–
56). Lisse: Balkema.
Harrison, R. G., Bogdanowicz, S. M., Hoffmann, R. S., Yensen,
E. & Sherman, P. W. (2003). Phylogeny and evolutionary
history of the ground squirrels (Rodentia, Marmotinae). Journal
of Mammalian Evolution, 10, 249–276.
Herron, M. D., Castoe, T. A. & Parkinson, C. L. (2004). Sciurid
phylogeny and the paraphyly of Holarctic ground squirrels
(Spermophilus). Molecular Phylogenetics and Evolution, 31, 1015–
1030.
Huelsenbeck, J. P. & Ronquist, F. (2001). MrBayes: Bayesian
inference of phylogeny. Bioinformatics, 17, 754–755.
Lance, S. L., Maldonado, J. E., Bocetti, C. I., Pattee, O. H.,
Ballou, J. D. & Fleischer, R. C. (2003). Genetic variation in
natural and translocated populations of the endangered
Delmarva fox squirrel (Sciurus niger cinereus). Conservation
Genetics, 4, 707–718.
Lee, M.-Y., Park, S.-K., Hong, Y.-J., Kim, Y.-J., Voloshina, I.,
Myslenkov, A., Saveljev, A. P., Choi, T.-Y., Piao, R.-Z., An, J.H.,
Lee, M.-H., Lee, H. & Min, M.-S. (2008). Mitochondrial
genetic diversity and phylogenetic relationships of Siberian
flying squirrel (Pteromys volans) populations. Animal Cells and
Systems, 12, 269–277.
Matthee, C. A. & Robinson, T. J. (1997). Molecular phylogeny of
the springhare, Pedetes capensis, based on mitochondrial DNA
sequences. Molecular Biology and Evolution, 14, 20–29.
Mercer, J. M. & Roth, V. L. (2003). The effects of Cenozoic
global change on squirrel phylogeny. Science, 299, 1568–
1572.
Metcalfe, I., Smith, J. M. B., Morwood, M. & Davidson, I.
(2001). Faunal and Floral Migration and Evolution in SE AsiaAustralasia.
Lisse: Balkema.
Michaux, J., Hautier, L., Simonin, T. & Vianey-Liaud, M. (2008).
Phylogeny, adaptation and mandible shape in Sciuridae
(Rodentia, Mammalia). Mammalia, 72, 286–296.
Mittelbach, G. G., Schemske, D. W., Cornell, H. V., Allen, A. P.,
Brown, J. M., Bush, M. B., Harrison, S. P., Hurlbert, A. H.,
Knowlton, N., Lessios, H. A., McCain, C. M., McCune, A. R.,
McDade, L. A., McPeek, M. A., Near, T. J., Price, T. D.,
Ricklefs, R. E., Roy, K., Sax, D. F., Schluter, D., Sobel, J. M.
& Turelli, M. (2007). Evolution and the latitudinal diversity
gradient: speciation, extinction and biogeography. Ecology
Letters, 10, 315–331.
Moncrief, N. D., Lack, J. B. & Van den Bussche, R. A. (2010).
Eastern fox squirrel (Sciurus niger) lacks phylogeographic
structure: recent range expansion and phenotypic
differentiation. Journal of Mammalogy, 91, 1112–1123.
Phylogeny of tree squirrels d P. Pecˇnerova´ & N. Martı´nkova´
218 ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters, 41, 3, May 2012, pp 211–219
Paper 2.1.4 105
Montgelard, C., Bentz, S., Tirard, C., Verneau, O. & Catzeflis, F.
M. (2002). Molecular systematics of Sciurognathi (Rodentia):
the mitochondrial cytochrome b and 12S rRNA genes support
the Anomaluroidea (Pedetidae and Anomaluridae). Molecular
Phylogenetics and Evolution, 22, 220–233.
Nowak, R. M. (1999). Walker’s Mammals of the World, 6th edn.
Baltimore: The Johns Hopkins University Press.
Oshida, T., Masuda, R. & Yoshida, M. C. (1996). Phylogenetic
relationships among Japanese species of the family Sciuridae
(Mammalia, Rodentia), inferred from nucleotide sequences of
mitochondrial 12S ribosomal RNA genes. Zoological Science, 13,
615–620.
Oshida, T., Arslan, A. & Noda, M. (2009). Phylogenetic
relationships among the Old World Sciurus squirrels. Folia
Zoologica, 58, 14–25.
Page, R. D. M. (2002). Modified mincut supertrees. In: R. Guigo´
& D. Gusfield (Eds) Proceedings of the Second International
Workshop on Algorithms in Bioinformatics. Lecture Notes in
Computer Science, Vol. 2452 (pp. 537–551). London: Springer.
Posada, D. & Crandall, K. A. (1998). Modeltest: testing the
model of DNA substitution. Bioinformatics, 9, 817–818.
Purvis, A. (1995). A modification to Baum and Ragan’s method for
combining phylogenetic trees. Systematic Biology, 44, 251–255.
Ragan, M. A. (1992). Matrix representation in reconstructiong
phylogenetic-relationships among the Eukaryotes. Biosystems,
28, 47–55.
Ranwez, V., Berry, V., Criscuolo, A., Fabre, P. H., Guillemot, S.,
Scornavacca, C. & Douzery, E. J. P. (2007). PhySIC: a veto
supertree method with desirable properties. Systematic Biology,
56, 798–817.
Ranwez, V., Criscuolo, A. & Douzery, E. J. P. (2010).
Supertriplets: a triplet-based supertree approach to
phylogenomics. Bioinformatics, 26, i115–i123.
Rosenzweig, M. L. (1995). Species Diversity in Space and Time.
New York: Cambridge University Press.
Roth, V. L. & Mercer, J. M. (2008). Differing rates of
macroevolutionary diversification in arboreal squirrels. Current
Science, 95, 857–861.
Sanderson, M. J. (2003). r8s: inferring absolute rates of molecular
evolution and divergence times in the absence of a molecular
clock. Bioinformatics, 19, 301–302.
Scornavacca, C., Berry, V., Lefort, V., Douzery, E. J. P. &
Ranwez, V. (2008). PhySIC_IST: cleaning source trees to infer
more informative supertrees. BMC Bioinformatics, 9, 413.
Semple, C. & Steel, M. (2000). A supertree method for rooted
trees. Discrete Applied Mathematics, 105, 147–158.
Stebbins, G. L. (1974). Flowering Plants: Evolution Above the Species
Level. Cambridge: The Belknap Press of Harvard University Press.
Stephens, P. R. & Wiens, J. J. (2003). Explaining species
richness from continents to communities: the time-forspeciation
effect in emydid turtles. American Naturalist, 161,
112–128.
Steppan, S. J., Storz, B. L. & Hoffmann, R. S. (2004). Nuclear
DNA phylogeny of the squirrels (Mammalia: Rodentia) and the
evolution of arboreality from c-myc and rag1. Molecular
Phylogenetics and Evolution, 30, 703–719.
Tappen, M. (1994). Bone weathering in the tropical rain forest.
Journal of Archaeological Science, 21, 667–673.
Thorington, R. W., Jr & Ferrell, K. (2006). Squirrels: The Animal
Answer Guide. Baltimore: Johns Hopkins University Press.
Thorington, R. W., Jr & Santana, E. (2007). How to make a
flying squirrel: Glaucomys anatomy in phylogenetic perspective.
Journal of Mammalogy, 88, 882–896.
Villalobos, F. & Cervantes-Reza, F. (2007). Phylogenetic
relationships of Mesoamerican species of the genus Sciurus
(Rodentia: Sciuridae). Zootaxa, 1525, 31–40.
Wettstein, P. J., Strausbauch, M., Lamb, T., States, J.,
Chakraborty, R., Jin, L. & Riblet, R. (1995). Phylogeny of six
Sciurus aberti subspecies based on nucleotide sequences of
cytochrome b. Molecular Phylogenetics and Evolution, 4, 150–162.
Wiens, J. J. & Donoghue, M. J. (2004). Historical biogeography,
ecology, and species richness. Trends in Ecology and Evolution,
19, 639–644.
Wiens, J. J., Graham, C. H., Moen, D. S., Smith, S. A. & Reeder,
T. W. (2006). Evolutionary and ecological causes of the
latitudinal diversity gradient in hylid frogs: treefrog trees
unearth the roots of high tropical diversity. American Naturalist,
168, 579–596.
Wilson, D. E. & Reeder, D. M. (Eds) (2005). Mammal species of
the world. A Taxonomic and Geographic Reference, 3rd edn.
Baltimore: John Hopkins University Press.
Wood, D. J. A., Koprowski, J. L. & Lurz, P. W. W. (2007). Tree
squirrel introduction: a theoretical approach with population
viability analysis. Journal of Mammalogy, 88, 1271–1279.
Supporting Information
Additional Supporting Information may be found in the
online version of this article:
Fig. S1. Supertrees based on gene trees obtained from
four mitochondrial and four nuclear loci.
Table S1. Accession numbers of sequences of Sciurini
tree squirrels and the outgroup taxa used for the reconstruction
of phylogenetic relationships.
Please note: Wiley-Blackwell are not responsible for the
content or functionality of any supporting materials supplied
by the authors. Any queries (other than missing
material) should be directed to the corresponding author
for the article.
P. Pecˇnerova´ & N. Martı´nkova´ d Phylogeny of tree squirrels
ª 2012 The Authors d Zoologica Scripta ª 2012 The Norwegian Academy of Science and Letters, 41, 3, May 2012, pp 211–219 219
106 Paper 2.1.4
Paper 2.1.5
Kandemir ˙I., Sözen M., Matur F., Kankilic¸ T., Martínková N., C¸ olak F., Özkurt S.
Ö., C¸ olak E. 2012. Phylogeny of species and cytotypes of mole rats (Spalacidae) in
Turkey inferred from mitochondrial cytochrome b gene sequences. Folia Zoologica 61:
25-33.
– 107 –
25
Folia Zool. – 61 (1): 25–33 (2012)
Introduction
The mole rats are adapted for subterranean life. They
are distributed in the Palaearctic region, throughout
eastern and southeastern Europe, Anatolia, the
Caucasus, and the Middle East up to northeastern
Africa (Topachevskii 1969, Wilson & Reeder 1993).
Their evolutionary history and taxonomic status are
difficult to ascertain and molecular, karyological and
morphological studies are needed to determine the
phylogenetic relationships of mole rats in Turkey
(Nevo et al. 1995, Suziki et al. 1996, Sözen et al. 2000,
Kankılıç et al. 2005, Ivanitskaya et al. 2008). For
severaldecades,scientistsagreethattaxonomyofmole
rats (Spalacinae, Rodentia) needs a modern revision
based on chromosome and molecular genetic data
coupled with morphology, physiology and behavior
Phylogeny of species and cytotypes of mole rats
(Spalacidae) in Turkey inferred from mitochondrial
cytochrome b gene sequences
İrfan Kandemİr1
, Mustafa Sözen2
, Ferhat Matur2
*, Teoman Kankılıç3
,
Natália MartínkovÁ4
, Faruk Çolak2
, Sakir Ö. Özkurt5
and Ercument Çolak1
1	
Department of Biology, Faculty of Science, Ankara University, 06100-Tandoğan, Ankara, Turkey;
	e-mail: ikandemir@gmail.com, allactaga@hotmail.com
2	
Department of Biology, Faculty of Arts and Sciences, Zonguldak Karaelmas University, 67100-
	Zonguldak, Turkey; e-mail: ferhat.matur@gmail.com, spalaxtr@hotmail.com, farukcolak@gmail.com
3	
Department of Biology, Faculty of Arts and Sciences, Nigde University, Niğde, Turkey;
	e-mail: spalax06@hotmail.com
4
	
Institute of Vertebrate Biology, Academy of Sciences of the Czech Republic, v.v.i., Květná 8, 603 65
	Brno, Czech Republic; e-mail: martinkova@ivb.cz
5	
Department of Biology Education, Faculty of Education, Ahi Evran University, Kırşehir, Turkey;
	e-mail: onderÖzkurt64@gmail.com
Received 4 October 2010; Accepted 26 August 2011
Abstract. We described the genetic variation of cytochrome b gene sequences of blind mole rats in Turkey.
We examined 47 individuals belonging to nine cytotypes of three superspecies Nannospalax leucodon, N.
xanthodon and N. ehrenbergi in the 402bp gene sequence of cytochrome b. Phylogenetic analyses showed
that relationships between cytotypes were well supported, but deeper divergence between species showed insignificant
relationships. Cytotypes of N. xanthodon with low diploid number of chromosomes from western
Turkey formed a monophyletic group distinct from the populations with higher number of chromosomes (2n =
56-60). The monophyly of N. xanthodon was supported with respect to N. leucodon (2n = 56) in the Bayesian
and maximum likelihood phylogenies. The divergence between two analyzed cytotypes of N. ehrenbergi (2n
= 52, 2n = 56) was 9.4 %, and the Kilis cytotype (2n = 52) appeared as the basal branch of the whole analysed
dataset. N. ehrenbergi cytotypes were paraphyletic and they formed unsupported relationships with previously
described N. galili (2n = 52), N. golani (2n = 54), N. carmeli (2n = 58) and N. judaei (2n = 60) from Israel. The
results of this study showed that the Nannospalax species complex most likely represents more species than
currently recognized, especially in N. xanthodon. We suggest that cytotypes of N. xanthodon and N. ehrenbergi
from Turkey should be investigated in detail as possible candidates for being separate species.
Key words: Nannospalax, molecular phylogeny, chromosomal form, Anatolia, Thrace
* Corresponding Author
108 Paper 2.1.5
26
(Savić & Nevo 1990, Kryštufek & Vohralík 2009).
The genus Spalax differs from representatives of the
genus Nannospalax in having a less variable diploid
number (2n = 60, 62 in Spalax versus 2n = 36-60 in
Nannospalax) and a higher number of subtelocentric
chromosomes (NF = 116-124 in Spalax versus NF =
72-98 in Nannospalax cytotypes) (Lyapunova et al.
1971, Savić & Nevo 1990, Németh et al. 2009), and
shows slow chromosomal evolution rate (Kryštufek
& Vohralík 2009). Kryštufek & Vohralík (2009) and
Németh et al. (2009) ranked Nannospalax species
as superspecies. Here we use superspecies order for
Turkish mole rats instead of species.
According to Wilson & Reeder (1993) and Yiğit et
al. (2006), three species, Nannospalax leucodon, N.
xanthodon (senior synonym of N. nehringi; Kryštufek
& Vohralík 2009) and N. ehrenbergi, occur in Turkey.
N. leucodon is found in the Turkish Thrace, and N.
ehrenbergi in southeastern Turkey. N. xanthodon has
a wide distribution area extending over Anatolia.
Karyological studies revealed 17 cytotypes in Turkey:
one cytotype (2n = 56) in N. leucodon, 11 (2n = 36,
38, 40, 48, 50, 52, 54, 56, 58, 60 and 60R) in N.
xanthodon and five (2n = 48, 52, 54, 56, 58) in N.
ehrenbergi (see reviews in Sözen et al. 1999, 2006a,
2011, Coşkun et al. 2006, Kankılıç et al. 2007, 2010).
Additionally, Nevo et al. (1994, 1995) recorded the
2n = 62 form, but Ivanitskaya et al. (2008) reported
that the 2n = 62 forms should be eliminated from
the list of Turkish mole rats. Later Arslan et al.
(2011) studied C- and AgNOR banding patterns of
three cytotypes (2n = 40, 58 and 60) from southern
Anatolia. These karyological results show that one of
the most complex chromosomal diversity within the
distribution range of the genus Nannospalax is found
in Turkey. These results complicate the taxonomical
status of the cytotypes in Nannospalax in Turkey.
While taxonomic status of cytotypes of mole rats in
Turkey has been under discussion, Nevo et al. (2001)
raised four cytoypes (2n = 52, 54, 58 and 60) in Israel
to species level; Nannospalax galili (2n = 52), N.
golani (2n = 54), N. carmeli (2n = 58) and N. judaei
(2n = 60). Arslan et al. (2010) studied mitochondrial
DNA (mtDNA) variation between three cytotypes
(2n = 40, 58 and 60) in southern Anatolia, and they
showed well-supported lineages in the phylogenetic
tree. But there is no detailed study based on mtDNA
sequences on the genetic structure of cytotypes of
mole rats in Western Turkey. This study focused on
phylogenetic relationships among cytotypes of N.
xanthodon in Western Turkey, and N. leucodon in
the Turkish Thrace and N. ehrenbergi in southeast
of Turkey inferred from mtDNA cytochrome b
gene sequences. We aimed to highlight the genetic
relationships between and among the cytotypes of
the species in Turkey, and also the recently described
species of Nannospalax from Israel.
Material and Methods
Sampling for molecular studies
A total of 47 mole rat individuals were used in the
molecular studies. These samples came from three
previously recognized species, namely N. leucodon
(N = 3), N. xanthodon (N = 40; cytotypes 2n = 36, 38,
40, 50, 56, 60) and N. ehrenbergi (N = 4; cytotypes
2n = 52 and 56) (Fig. 1). Karyotypes were prepared
according to Ford & Hamerton (1956). Skins and
skulls were deposited to the Department of Biology
of Zonguldak Karaelmas and Ankara University,
Turkey.
DNA isolation, amplification, and sequencing
We sequenced all 47 samples for a part of the
cytochrome b locus. Total genomic DNA was isolated
from muscle tissues following the techniques reported
by Doyle & Doyle (1987) known as CTAB method.
Fragment of the cytochrome b gene was amplified
using polymerase chain reaction in 25 μl reaction
volume and each reaction included 1 μl of each primer
(20 pmol) (L14724 and H15154; Irwin et al. 1991,
Smith & Patton 1993), 4 μl of dNTPs, 2.5 μl of 10× Taq
Buffer with (NH4
)2
SO4
, 1.5 μl of MgCl2
(25 mM) and
0.25 μl of Taq DNAPolymerase (5 units/μl, Fermentas,
Ontario, Canada). The PCR cycling conditions first
included four cycles of 1 min of denaturation at 95
°C, 1 min of annealing at 40 °C, and 1 min extension
at 72 °C, followed by 33 cycles using annealing
temperature at 50 °C. Before the sequencing reaction
Fig. 1. Sampled sites in Turkey and respective mole
rat cytotypes.
Paper 2.1.5 109
27
the PCR fragments were cleaned with Nucleospin
extract kit (Macherey-Nagel, Düren, Germany) and
later sequencing reactions of both DNA strands were
commercially performed using Big Dye Terminator
v. 3.1 sequencing chemistry on an ABI 3100 Genetic
Analyzer (Applied Biosystems, California, USA). All
specimens were deposited to the GenBank under the
accession numbers FJ656259-FJ656305 (Table 1).
Table 1. Samples used in the present study.
Species Map code 2n Locality
Genbank
accession
number
Voucher name Coordinates
N. leucodon
1 56Tr Kırklareli FJ656299 TR-KIRKLARELİ 5097 41°25′34.05′′ N 27°7′9.09′′ E
1 56 Kırklareli FJ656300 TR-KIRKLARELİ 5094 41°25′34.05′′ N 27°7′9.09′′ E
1 56 Kırklareli FJ656301 TR-KIRKLARELİ 5095 41°25′34.05′′ N 27°7′9.09′′ E
N. xanthodon
2 36 Aydın FJ656275 TR-AYDIN 6218 37°51′46.19′′ N 27°49′27.67′′ E
2 36 Aydın FJ656276 TR-AYDIN 6256 37°51′46.19′′ N 27°49′27.67′′ E
2 36 Aydın FJ656277 TR-AYDIN 6222 37°51′46.19′′ N 27°49′27.67′′ E
2 36 Aydın FJ656278 TR-AYDIN 6207 37°51′46.19′′ N 27°49′27.67′′ E
2 36 Aydın FJ656279 TR-AYDIN 6250 37°51′46.19′′ N 27°49′27.67′′ E
2 36 Aydın FJ656280 TR-AYDIN 6532 37°51′46.19′′ N 27°49′27.67′′ E
2 36 Aydın FJ656281 TR-AYDIN 5258 37°51′46.19′′ N 27°49′27.67′′ E
3 38 Çanakkale FJ656259 TR-GOKCEADA 4959 40°9′30.61′′ N 25°50′30.05′′ E
3 38 Çanakkale FJ656260 TR-GOKCEADA 4957 40°9′30.61′′ N 25°50′30.05′′ E
3 38 Çanakkale FJ656261 TR-GOKCEADA 4955 40°1′18.10′′ N 26°25′26.44′′ E
3 38 Çanakkale FJ656262 TR-GOKCEADA 4954 40°1′18.10′′ N 26°25′26.44′′ E
3 38 Çanakkale FJ656263 TR-GOKCEADA 4958 40°1′18.10′′ N 26°25′26.44′′ E
3 38 Manisa FJ656282 TR-MANISA 5214 39°6′41.11′′ N 27°40′14.72′′ E
3 38 Manisa FJ656283 TR-MANISA 5211 39°6′41.11′′ N 27°40′14.72′′ E
3 38 Balıkesir FJ656284 TR-BALIKESIR 6120 39°22′14.33′′ N 27°59′25.34′′ E
3 38 Balıkesir FJ656285 TR-BALIKESIR 6161 39°22′14.33′′ N 27°59′25.34′′ E
3 38 Manisa FJ656286 TR-MANISA 5210 39°10′24.88′′ N 27°51′18.40′′ E
3 38 Manisa FJ656287 TR-MANISA 6131 39°10′24.88′′ N 27°51′18.40′′ E
3 38 Manisa FJ656288 TR-MANISA 6153 39°10′24.88′′ N 27°51′18.40′′ E
3 38 Izmir FJ656289 TR-IZMIR 6138 38°27′40.06′′ N 27°12′51.77′′ E
4 40 Isparta FJ656264 TR-ISPARTA 6202 37°43′7.76′′ N 30°56′55.64′′ E
4 40 Isparta FJ656265 TR-ISPARTA 6225 37°43′7.76′′ N 30°56′55.64′′ E
4 40 Isparta FJ656266 TR-ISPARTA 6265 37°43′7.76′′ N 30°56′55.64′′ E
4 40 Isparta FJ656267 TR-ISPARTA 6266 37°43′7.76′′ N 30°56′55.64′′ E
4 40 Konya FJ656268 TR-KONYA 6268 37°30′42.12′′ N 31°24′3.10′′ E
4 40 Konya FJ656269 TR-KONYA 5346 37°30′42.12′′ N 31°24′3.10′′ E
4 40 Konya FJ656270 TR-KONYA 6203 37°30′42.12′′ N 31°24′3.10′′ E
4 40 Konya FJ656271 TR-KONYA 4734 37°30′42.12′′ N 31°24′3.10′′ E
4 40 Konya FJ656272 TR-KONYA 6267 37°30′42.12′′ N 31°24′3.10′′ E
5 50 Aydın FJ656273 TR-AYDIN 5255 38°15′15.61′′ N 28°23′15.30′′ E
5 50 Aydın FJ656274 TR-AYDIN 5256 38°15′15.61′′ N 28°23′15.30′′ E
6 56 Uşak FJ656290 TR-USAK 6224 38°40′28.91′′ N 29°14′15.28′′ E
6 56 Manisa FJ656291 TR-MANISA 6141 38°28′1.76′′ N 28°38′36.98′′ E
6 56 Isparta FJ656292 TR-ISPARTA 4784 37°47′46.80′′ N 30°54′23.74′′ E
6 56 Isparta FJ656294 TR-ISPARTA 6257 37°47′46.80′′ N 30°54′23.74′′ E
6 56 Karabük FJ656297 TR-KARABUK 4858 41°13′21.30′′ N 32°43′16.30′′ E
7 60 Manisa FJ656293 TR-MANISA 6143 38°43′17.87′′ N 28°52′13.76′′ E
7 60 Kütahya FJ656295 TR-KUTAHYA 3683 39°24′5.99′′ N 29°16′14.63′′ E
7 60 Kütahya FJ656296 TR-KUTAHYA 6119 39°24′5.99′′ N 29°16′14.63′′ E
7 60R Kastamonu FJ656298 TR-KASTAMONU 5607 41°42′25.42′′ N 33°35′24.63′′ E
N. ehrenbergi
8 56 Osmaniye FJ656304 TR-OSMANIYE 5301 37°4′34.75′′ N 36°14′22.08′′ E
8 56 Osmaniye FJ656305 TR-OSMANIYE 5302 37°4′34.75′′ N 36°14′22.08′′ E
9 52 Kilis FJ656302 TR-KILIS 5120 36°47′34.56′′ N 37°15′27.46′′ E
9 52 Kilis FJ656303 TR-KILIS 5121 36°47′34.56′′ N 37°15′27.46′′ E
110 Paper 2.1.5
28
Phylogenetic analyses
Previously published cytochrome b sequences
of Nannospalax were obtained from GenBank
(AF140523-AF140534; Nevo et al. 1999), and two
other sequences for Mus musculus (NC010339) and
Rattus norvegicus (EU273707) were added as an
outgroup. The dataset was aligned in Clustal X 2.0.9
(Larkin et al. 2007) and had the total length 402 base-
pairs.
Genetic divergence was estimated in MEGA 4 (Tamura
et al. 2007). Neighbour-joining (NJ) and maximum
parsimony (MP) phylogenetic analyses were executed
in PAUP 4.0b10 (Swofford 1999), using HKY +
G substitution model for the NJ analysis selected
by Bayesian Information Criterion in Modeltest 3.7
(Posada & Crandall 1998), which was the simplest
suitable model, and 10000 bootstrap replicates.
Maximum likelihood (ML) tree was obtained from
RAxML 7.4 (Stamatakis 2006) with GTR + G model
and 10000 bootstrap replicates. Bayesian Inference
(BI) analysis was run in MrBayes 3.1.2 (Ronquist &
Huelsenbeck 2003) with 12 Markov chains Monte
Fig. 2. Bayesian inference phylogenetic tree based on Nannospalax partial sequences of the mitochondrial
cytochrome b gene (402 bp). Node support from Bayesian posterior probabilities and NJ, ML and MP
bootstrapping is shown for main groups. ‘-’ indicates non-significant support, < 70 % for bootstrap analyses
and < 0.95 for Bayesian posterior probability.
Paper 2.1.5 111
29
Carlo for two million generations in two independent
runs. The chain swapping was successful in 68-72
% of attempts indicating effective chain mixing.
We used default prior settings with the exception
of the branch length. That was limited to brlenspr =
unconstrained: exp(20) to ensure that the posterior
95 % confidence interval of the tree length included
the tree length obtained from the maximum likelihood
analysis (cf. Marshall 2010). The prior setting did
not influence tree topology compared to the default
setting (data not shown). A median-joining network
(Bandelt et al. 1999) was also reconstructed for the
sequenced mtDNA region. Finally, tree topologies
were compared by using the Shimodaira-Hasegawa
test in PAUP 4.0b10 (Swofford 1999).
Results
The 47 sequences represented 42 unique haplotypes,
which were used in the phylogenetic analyses.
The 402 bp long dataset contained 120 parsimony
informative sites and 31 singleton mutations. Support
for monophyly of the studied taxa based on partial
cytochrome b sequences was different for specific
analyses (Fig. 2). Total tree length in the MP analysis
was 364 steps and the groupings reflected the genetic
distances between cytotypes and species.
Nucleotide diversity within cytotypes ranged from
0 to 0.03 substitutions per site (Table 2). Within N.
xanthodon, the greatest diversity was found in the 2n
= 38 cytotype that was also represented by the highest
number of individuals in our dataset.
The genetic distances were calculated between all
taxa used in the present study using uncorrected
p-distance (Table 3). The distance between four Israeli
Nannospalax species ranged between 0.019 and 0.058
substitutions per site. The cytotypes attributed to N.
xanthodonshoweddeeperdivergencerangingbetween
0.024and0.105.Thegreatestgeneticdistancebetween
studied cytotypes was between N. carmeli and N.
xanthodon 2n = 40. Smaller genetic distances tended
to be found between the chromosomal cytotypes with
the highest diploid chromosome numbers. The genetic
distance between the two cytotypes of N. ehrenbergi
in Turkey was 0.087, which is greater than the genetic
distance between the two most distantly related Israeli
species (N. golani and N. carmeli) (Table 3).
Table 2. Within group nucleotide diversity for partial
sequences of the mitochondrial cytochrome b gene
(402 bp) for identified Nannospalax cytotypes.
Cytotypes
N Nucleotide
diversity
Standard
Error
N. galili 2n = 521
2 0.0017 0.0016
N. golani 2n = 541
2 0.0017 0.0016
N. judaei 2n = 601
3 0.0216 0.0058
N. carmeli 2n = 581
2 0.0265 0.0066
N. ehrenbergi 2n = 522
2 0.000 0.000
N. ehrenbergi 2n = 562
2 0.000 0.000
N. xanthodon 2n = 362
7 0.0021 0.0012
N. xanthodon 2n = 382
13 0.0192 0.0044
N. xanthodon 2n = 402
9 0.0072 0.0028
N. xanthodon 2n = 502
2 0.000 0.000
N. xanthodon 2n = 562
5 0.0033 0.0023
N. xanthodon 2n = 602
4 0.0108 0.0038
N. leucodon 2n = 562
3 0.0030 0.0058
1
Nevo et al. 1999,2
this study.
Table 3. Pairwise uncorrected p-distance for partial sequences of the mitochondrial cytochrome b gene
(402 bp) for identified Nannospalax cytotypes (below diagonal) and their standard errors (above diagonal).
1 2 3 4 5 6 7 8 9 10 11 12 13
N. galili 2n = 521
0.007 0.010 0.010 0.013 0.012 0.014 0.014 0.015 0.014 0.014 0.015 0014
N. golani 2n = 541
0.019 0.011 0.010 0.013 0.013 0.015 0.015 0.015 0.014 0.015 0.015 0.014
N. judaei 2n = 601
0.052 0.056 0.005 0.012 0.014 0.014 0.014 0.015 0.014 0.014 0.014 0.014
N. carmeli 2n = 581
0.056 0.058 0.022 0.012 0.014 0.015 0.014 0.015 0.014 0.014 0.014 0.014
N. ehrenbergi 2n = 522
0.080 0.078 0.076 0.076 0.013 0.014 0.015 0.015 0.015 0.015 0.015 0.015
N. ehrenbergi 2n = 562
0.068 0.066 0.096 0.098 0.087 0.014 0.015 0.015 0.014 0.015 0.015 0.015
N. xanthodon 2n = 402
0.119 0.118 0.120 0.125 0.104 0.108 0.010 0.007 0.011 0.012 0.014 0.013
N. xanthodon 2n = 382
0.119 0.118 0.117 0.123 0.113 0.117 0.059 0.010 0.010 0.012 0.014 0.013
N. xanthodon 2n = 502
0.119 0.118 0.117 0.121 0.104 0.112 0.029 0.053 0.012 0.012 0.014 0.013
N. xanthodon 2n = 362
0.100 0.099 0.107 0.107 0.102 0.097 0.062 0.052 0.065 0.013 0.014 0.013
N. leucodon 2n = 562
0.101 0.107 0.106 0.114 0.111 0.116 0.080 0.080 0.076 0.089 0.014 0.013
N. xanthodon 2n = 562
0.123 0.113 0.117 0.120 0.121 0.124 0.104 0.105 0.100 0.102 0.099 0.005
N. xanthodon 2n = 602
0.122 0.116 0.115 0.119 0.121 0.121 0.102 0.104 0.098 0.099 0.098 0.024
1
Nevo et al. 1999,2
this study.
112 Paper 2.1.5
30
We conducted four separate sets of analyses. The data
were analyzed using Bayesian inference, neighbour-
joining,maximumlikelihoodandmaximumparsimony
analyses. The trees showed similar topologies (Fig. 2).
The result of the Shimodaira-Hasegawa test showed
that only NJ tree topology was significantly different
from the other trees (Table 4). This result was caused
by the placement of the N. ehrenbergi clade. In
all trees, cytotypes with low diploid chromosome
numbers (2n ≤ 50) grouped as one supported clade.
The other N. xanthodon cytotypes (2n = 56 and 60)
occurred as a separate clade, but monophyly of N.
xanthodon was supported only in BI and ML analyses
(Fig. 2). N. leucodon formed a monophyletic group
with N. xanthodon. The other taxa showed a basal
polytomy in most of our analyses. Both cytotypes
of N. ehrenbergi were paraphyletic with respect to
the Israeli taxa. N. carmeli and N. judaei formed a
monophyletic group, but the taxa were not reciprocally
monophyletic within this group.
We found similar topology in the network analysis
(Fig. 3). The N. xanthodon cytotypes formed two
putative groups with low (2n ≤ 50) and high (2n ≥ 56)
chromosome numbers. The former group was more
closely related to N. leucodon as well as to the group
including N. ehrenbergi and the Israeli taxa (Fig. 3).
Discussion
Our phylogenetic analysis showed monophyly of
N. xanthodon and N. leucodon, but paraphyletic
Table 4. Results of Shimodaira-Hasegawa (SH) tests
for all trees. The NJ tree topology was found different
from ML (* p < 0.05).
Tree –ln L Diff –ln L P
Bayesian inference 227.988 0.434 0.52
Maximum likelihood 227.553 (best)
Maximum parsimony 228.626 1.073 0.29
Neighbour-joining 231.095 3.542 0.01*
Fig. 3. The 42 sequences appear as three different species and five clades on median joining network congruent
with Fig. 2. The samples from identical localities are indicated with small circles. The node size is not
proportional to the haplotype frequency. The numbers on branch show differences between two neighboring
haplotypes.
Paper 2.1.5 113
31
relationships of the two N. ehrenbergi sequences
with Israeli taxa. N. xanthodon sequences formed
two reciprocally monophyletic groups that included
cytotypes with low (2n ≤ 50) and high (2n ≥ 56)
chromosome numbers. Their sister species was
N. leucodon. These results are similar to recent
findings of Arslan et al. (2010) on a smaller dataset
including three cytotypes and Kryštufek et al. (2012)
that were based on a longer alignment. The taxa N.
ehrenbergi, N. galili, N. golani, N. carmeli and N.
judaei diverged in a basal polytomy wherein only a
single group including N. carmeli and N. judaei was
consistently retrieved. While Reyes et al. (2003) also
showed an undifferentiated clade of N. carmeli and
N. judaei samples, our analyses did not support their
clear relationship of N. galili and N. golani. This was
distorted by position of N. ehrenbergi 2n = 56 as a
sister taxon with unresolved relationship to the Israeli
species.
Cytotypes of N. xanthodon with low diploid number
(2n = 36, 38, 40 and 50) formed a monophyletic group.
This may be a result of specific evolutionary pathways.
For example, Matur & Sözen (2005) stated that the
River Sakarya separated 2n = 52 and 60 cytotypes of
N. xanthodon in Bilecik province, and the river might
act as a barrier between these cytotypes. But in our
analyses, 2n = 52 N. xanthodon was not analysed
and we cannot confirm this scenario. According to
our results, 2n = 38 from Gökçeada (Aegean island)
differentiated from the 2n = 38 populations in the
mainland indicating mole rat diversification in island
isolation.
Monophyly of specific cytotypes was present in all
cytotypesbelongingtotheN.xanthodon2n≤50group,
but the cytotypes were not likewise differentiated in
the 2n ≥ 56 group. Rather, the northern populations
of N. xanthodon, Karabük (2n = 56N) and Kastamonu
(2n = 60R), were differentiated from the other samples
of 2n = 56 and 60. This was suggested by Sözen
(2004) and Sözen et al. (2006b), who claimed that the
cytotypes from Northern Anatolia (2n = 54N, 56N,
58N and 60R) differentiated from other cytotypes.
Ivanitskaya et al. (2008) showed that the 2n = 60
population in Kastamonu is different from the 2n =
60 population in central Anatolia using classical and
molecular cytogenetic techniques. Matur et al. (2010)
studied chromosomal evolution of Turkish mole rats
inferred from G- and C- banding of 11 cytotypes. They
found that cytotypes from Kastamonu (2n = 60R) and
Karabük (2n = 56R) formed a clade together with 2n
= 54 cytotype not included in our study. This northern
clade has independent evolutionary pathway, and 2n
= 60R might be considered an ancestral karyotype
(Matur et al. 2010).
There is a conflict in the ancestral population of
Thracian mole rats (N. leucodon 2n = 56). They
grouped with N. xanthodon in our trees. Recently,
Matur et al. (2011) proposed the 2n = 60 cytotype as
an ancestral for all Anatolian cytotypes. Nevertheless,
this process needs more detailed studies in order to
verify the ancestral karyotype of the N. xanthodon/
leucodon group.
In our phylogenetic analyses, N. carmeli (2n = 58) and
N. judaei (2n = 60) did not appear in separate clusters.
Reyes et al. (2003) found similar results from the
sequence studies of the mitochondrial control region.
In their study, N. golani and N. galili were well
separated, whereas N. judaei (2n = 60) and N. carmeli
(2n = 58) formed a single cluster. Nevo et al. (1999)
argued that this is due to the fact that they represent
young species. The paraphyly of the two cytotypes
of N. ehrenbergi (2n = 52 and 2n = 56) from Kilis
and Osmaniye further complicate this in our study.
Both cytotypes formed unsupported relationships
and polychotomies. Tentatively, N. ehrenbergi from
Osmaniye grouped with N. galili and N. golani, and
N. ehrenbergi from Kilis was a basal taxon in the tree
(Fig. 2).
Further comprehensive and detailed multidisciplinary
research combining morphology, karyology, physiology,
behaviour, mtDNA and nuclear DNA phylogeny
should be applied to clarify the taxonomic status
and phylogenetic relationships of cytotypes of N.
leucodon, N. xanthodon, and N. ehrenbergi in Turkey.
In addition, research on intra-population variation
should also be considered for better understanding of
mole rats in Turkey.
Acknowledgements
ThisstudywassupportedbytheScientificandTechnical
Research Council of Turkey (TUBITAK 101T084 and
106T225), Zonguldak Karaelmas University (2004-
13-06-08, 2008-13-06-01) and Academy of Sciences
of the Czech Republic (AV0Z60930519). The authors
thank Ahmet Turkyılmaz from REFGEN Biotechnology
Inc. for performing the sequencing. Authors thank to
Zeynep Yuksel Sezen and Joan E. Johnston for careful
editing the manuscript and two anonymous reviewers
and the editor for helpful and detailed comments.
114 Paper 2.1.5
32
Literature
Arslan E., Gulbahce E., Arıkoglu H., Arslan A., Bužan E.V. & Kryštufek B. 2010: Mitochondrial divergence
between three cytotypes of the Anatolian mole rat Nannospalax xanthodon (Mammalia: Rodentia). Zool.
Middle East. 50: 27–34.
Arslan A., Akan Ş. & Zima J. 2011: Variation in C-heterochromatin and NOR distribution among chromosomal
races of mole rats (Spalacidae) from Central Anatolia,Turkey. Mamm. Biol. 76: 28–35.
Bandelt H.J., Forster P. & Röhl A. 1999: Median-joining networks for inferring intraspecific phylogenies. Mol.
Biol. Evol. 16: 37–48.
Coşkun Y., Ulutürk S. & Yürümez G. 2006: Chromosomal, diversity in mote-rats of the species Nannospalax
ehrenbergi (Rodentia: Spalacidae) from South Anatolia, Turkey. Mamm. Biol. 71: 244–250.
Doyle J. & Doyle L. 1987:Arapid DNAisolation procedure for small quantities of fresh leaf tissue. Phytochem.
Bull. 19: 11–15.
Ford C.E. & Hamerton J.L. 1956:Acolchicine hypotonic citrate, squash sequence for mammalian chromosomes.
Stain Technol. 31: 247–251.
Irwin D., Kocher T. & WilsonA. 1991: Evolution of the cytochrome b gene of mammals. J. Mol. Evol. 32: 128–144.
Ivanitskaya E., Sözen M., Rashkovetsky L., Matur F. & Nevo E. 2008: Discrimination of 2n = 60 Spalax
leucodon cytotypes (Spalacidae, Rodentia) in Turkey by means of classical and molecular cytogenetic
techniques. Cytogenet. Genome Res. 122: 139–149.
Kankılıç T., Çolak E., Çolak R. & Yiğit N. 2005: Allozyme variation in Spalax leucodon Nordmann, 1840
(Rodentia: Spalacidae) in the area between Ankara and Beyşehir. Turk. J. Zool. 29: 377–384.
Kankılıç T., Kankılıç T., Çolak R., Çolak E. & Karataş A. 2007: Karyological comparison of populations of the
Spalax leucodon Nordmann, 1840 superspecies (Rodentia: Spalacidae) in Turkey. Zool. Middle East. 42: 15–24.
Kankılıç T., Kankılıç T., Seker P.S., Çolak R., Selvi E. & Çolak E. 2010: Contributions to the karyology and
distribution areas of cytotypes of Nannospalax leucodon (Rodentia: Spalacidae) in Western Anatolia. Acta
Zool. Bulg. 62: 161–167.
Kryštufek B. &VohralíkV. 2009: Mammals ofTurkey and Cyprus. Rodentia II: Cricetinae, Muridae, Spalacidae,
Calomyscidae, Capromyida, Hystricidae, Castoridae. Annales Majora, Koper.
Kryštufek B., Ivanitskaya E., Arslan A., Arslan E. & Bužan E.V. 2012: Evolutionary history of mole rats (genus
Nannospalax) inferred from mitochondrial cytochrome b sequence. Biol. J. Linn. Soc. 105: 446–455.
Larkin M.A., Blakshields G., Brown N.P., Chenna R., McGettigan P.A., McWilliam H., Valentin. F., Wallace
I.M., Wilm A., Lopez R., Thompson J.D., Gibson T.J. & Higgins D.G. 2007: Clustal W and clustal X
version 2.0. Bioinformatics 23: 2947–2948.
Lyapunova E.A., Vorontsov N.N. & Martynova L. 1971: Cytological differentiation of burrowing mammals in
the Palaeartic. In Syposium Theriologicum II. Prague. 203–205.
Marshall D.C. 2010: Cryptic failure of partitioned Bayesian phylogenetic analyses: lost in the land of long
trees. Syst. Biol. 59: 108–117.
Matur F. & Sözen M. 2005: A karyological study on subterrranean mole rats of the Spalax leucodon Nordmann,
1840 (Mammalia: Rodentia) superspecies around Bilecik province in Turkey. Zool. Middle East 36: 5–10.
Matur F., Çolak F., Sevindik M. & Sözen M. 2010: Chromosomal evolution of blind mole rats. 12th
International
Conference on Rodent Biology. Rodens et Spatium. Zonguldak, Turkey, 19-23 July 2010, Abstract Book: 46.
Matur F., Çolak F., Sevindik M. & Sözen M. 2011: Chromosome differentiation of four 2n = 50 chromosomal
forms of Turkish mole rat, Nannospalax nehringi. Zool. Sci. 28: 61–67.
Németh A., Révay T., Hegyeli Z., Farkas J., Czabán D., Rózsás A. & Csorba G. 2009: Chromosomal forms and
risk assessment of Nannospalax (superspecies leucodon) (Mammalia: Rodentia) in the Carpathian Basin.
Folia Zool. 58: 349–361.
Nevo E., Filippucci M.G., Redi C., Korol A. & Beiles A. 1994: Chromosomal speciation and adaptive radiation
of mole-rats in Asia Minor correlated with increased ecological stress. Proc. Natl. Acad. Sci. USA 91:
8160–8164.
Nevo E., Filippucci M.G., Redi C., Simson S., Heth G. & Beiles A. 1995: Karyotype and genetic evolution in
speciation of subterranean mole rats of the genus Spalax in Turkey. Biol. J. Linn. Soc. 54: 203–229.
Nevo E., Beiles A. & Spradling T. 1999: Molecular evolution of cytochrome b of subterranean mole rats,
Spalax ehrenbergi superspecies, in Israel. J. Mol. Evol. 49: 215–226.
Paper 2.1.5 115
33
Nevo E., Ivanitskaya E. & Beiles A. 2001: Adaptive radiation of blind subterranean mole rats: naming and
revisiting the four sibling species of the Spalax ehrenbergi superspecies in Israel: Spalax galili (2n =
52), S. golani (2n = 54), S. carmeli (2n = 58) and S. judaei (2n = 60). Bachkhuys Publishers, Leiden, The
Netherlands.
Posada D. & Crandall K.A. 1998: Modeltest: testing the model of DNA substitution. Bioinformatics 14: 817–
818.
Reyes A., Nevo E. & Saccone C. 2003: DNA sequence variation in the mitochondrial control region of
subterranean mole rats, Spalax ehrenbergi superspecies, in Israel. Mol. Biol. Evol. 20: 622–632.
Ronquist F. & Huelsenbeck J.P. 2003: MrBayes 3: Bayesian phylogenetic inference under mixed models.
Bioinformatics 191: 572–1574.
Savić I. & Nevo E. 1990: The Spalacidae: evolutionary history, speciation, and population biology. In: Nevo
E. & Reig O.A. (eds.), Evolution of subterranean mammals at the organismal and molecular levels. Alan
R. Liss, Inc., New York: 129–153.
Smith M.F. & Patton J.L. 1993: The diversification of South American murid rodents: evidence from
mitochondrial DNA sequence data for the akodontine tribe. Biol. J. Linn. Soc. 50: 149–177.
Sözen M. 2004: A karyological study on subterranean mole rats of the Spalax leucodon Nordmann, 1840
superspecies in Turkey. Mamm. Biol. 69: 420–429.
Sözen M., Çolak E., Yiğit N., Özkurt Ş. & Verimli R. 1999: Contributions to the karyology and taxonomy of the
genus Spalax Güldenstaedt, 1770 (Mammalia: Rodentia) in Turkey. Z. Säugetierkd. 64: 210–219.
Sözen M., Yiğit N. & Çolak E.  2000: A study on karyotypic evolution of the genus Spalax Güldenstaedt, 1770
(Mammalia: Rodentia) in Turkey.  Isr. J. Zool. 46: 239–242.
Sözen M., Matur F., Çolak E., Özkurt Ş. & KarataşA. 2006a: Some karyological records and a new chromosomal
form for Spalax (Mammalia: Rodentia) in Turkey. Folia Zool. 55: 247–256.
Sözen M., Sevindik M. & Matur F. 2006b: Karyological and some morphological characteristics of Spalax
leucodon Nordmann, 1840 (Mammalia: Rodentia) superspecies around Kastamonu province, Turkey.
Turk. J. Zool. 30: 205–219.
Sözen M., Çataklı K., Eroğlu F., Matur F. & Sevindik M. 2011: Distribution of chromosomal forms of
Nannospalax nehringi (Satunin, 1898) (Rodentia: Spalacidae) in Çankırı and Çorum provinces, Turkey.
Turk. J. Zool. 35: 367–374.
Stamatakis A. 2006: RAxML-VI-HPC: Maximum likelihood-based phylogenetic analyses with thousands of
taxa and mixed models. Bioinformatics 22: 2688–2690.
Suzuki H., Wakana S., Yonekawa H., Moriwaki K., Sakurai S. & Nevo E. 1996: Variations in ribosomal DNA
and mitochondrial DNA among chromosomal species of subterranean mole rats. Mol. Biol. Evol. 13: 85–92.
SwoffordD.L.1999:PAUP*.PhylogeneticAnalysisUsingParsimony(*andothermethods).SinauerAssociates,
Sunderland, Massachusetts.
Tamura K., Dudley J., Nei M. & Kumar S. 2007: MEGA4: Molecular evolutionary genetics analysis (MEGA)
software version 4.0. Mol. Biol. Evol. 24: 1596–1599.
Topachevskii W. 1969: Fauna USSR: mammals, Spalacidae. Vol. 3, No. 3. Nauka, Leningrad. (in Russian)
Wilson D.E. & Reeder D.M. 1993: Mammal species of the world: a taxonomic and geographic reference. 2nd
edition. Smithsonian Institution Press, Washington and London.
Yiğit N., Çolak E., Sözen M. & Karataş A. 2006: Rodents of Turkey. Meteksan, Ankara.
116 Paper 2.1.5
Paper 2.1.6
Vallo P., Benda P., Martínková N., Kaňuch P., Kalko E. K. V., Červený J., Koubek P.
2011. Morphologically uniform bats Hipposideros aff. ruber (Hipposideridae) exhibit
high mitochondrial genetic diversity in southeastern Senegal. Acta Chiropterologica 13:
79-88.
– 117 –
INTRODUCTION
Genetic differences in living organisms may precede
phenotypic differences, making genetically
distinct forms difficult or even impossible to detect
by traditional morphological means (Yoder et al.,
2000; Jacobs et al., 2006). Such forms are believed
to belong to the same species until additional
evidence shows that they represent independent
evolutionary units. The existence of these so-called
cryptic species (Mayr, 1996; Lincoln et al., 1998)
has been revealed for many taxonomic groups
(Avise, 2004; Bickford et al., 2007). Although careful
examination of morphology, ecology or behaviour
helps to discover cryptic species, molecular
genetics has contributed enormously in the last
two decades to discovering cryptic forms within
traditionally recognised taxa (Avise, 2004; Bickford
et al., 2007).
The presence of cryptic species is a rather common
phenomenon in bats, and molecular data play
an important role in their recognition and formal
systematic acknowledgement (Jones, 1997; Mayer
and von Helversen, 2001; Baker and Bradley, 2006;
Ibáñez et al., 2006). Differences in call characteristics
of echolocation signals have been likewise useful
to recognize distinct forms deserving taxonomic
recognition. The European vespertilionid Pipistrellus
pipistrellus/P. pygmaeus complex is a classic
example where differences in peak frequencies of
search calls led to discovery of two distinct sonotypes,
which have been subsequently confirmed by
molecular methods to represent two species (Barratt
et al., 1997).
Acta Chiropterologica, 13(1): 79–88, 2011
PL ISSN 1508-1109 © Museum and Institute of Zoology PAS
doi: 10.3161/150811011X578633
Morphologically uniform bats Hipposideros aff. ruber (Hipposideridae) exhibit
high mitochondrial genetic diversity in southeastern Senegal
PETER VALLO1, 9
, PETR BENDA2, 3
, NATÁLIA MARTÍNKOVÁ1, 4
, PETER KAŇUCH1, 5
, ELISABETH K. V. KALKO6, 7
,
JAROSLAV ČERVENÝ1, 8
, and PETR KOUBEK1, 8
1Institute of Vertebrate Biology, Academy of Sciences of the Czech Republic, v.v.i., Květná 8, 603 65 Brno, Czech Republic
2
Department of Zoology, National Museum (Natural History), Václavské náměstí 68, 115 79 Praha 1, Czech Republic
3
Department of Zoology, Charles University, Viničná 7, 128 44 Praha 2, Czech Republic
4
Institute of Biostatistics and Analyses, Masaryk University, Kamenice 3, 625 00 Brno, Czech Republic
5
Institute of Forest Ecology, Slovak Academy of Sciences, Štúrova 2, 960 53 Zvolen, Slovak Republic
6Institute of Experimental Ecology, University of Ulm, Albert-Einstein Allee 11, 89069 Ulm, Germany
7Smithsonian Tropical Research Institute, Balboa, Republic of Panama
8Department of Forest Protection and Game Management, Faculty of Forestry and Wood Sciences, Czech University of Life
Sciences, Kamýcká 129, 165 21 Praha 6, Czech Republic
9
Corresponding author: E-mail: vallo@ivb.cz
Two mitochondrial lineages of bats that are morphologically attributed to Hipposideros ruber have been shown to occur
sympatrically in southeastern Senegal. We studied genetic diversity in these bats in the Niokolo Koba National Park using sequences
of mitochondrial cytochrome b gene to determine the taxonomic status of the two genetic forms, and included skull morphology for
comparison. Detailed multidimensional analysis of skull measurements indicated slight morphological differences between the two
genetic forms. Exploration of peak frequency of the constant-frequency echolocation signals in a local population of Hipposideros
aff. ruber was not available for both groups. Phylogenetic comparison with other available West African representatives of H. aff.
ruber revealed paraphyletic relationship of the two Senegalese forms, with the less abundant form from Senegal forming
a monophyletic group with that from Benin. Based on genetic divergence and sympatric occurrence, the two forms from Senegal
might represent cryptic species. However, absence of nuclear gene flow between them is yet to be investigated to demonstrate their
reproductive isolation.
Key words: cytochrome b, Hipposideros caffer complex, cryptic species, phylogeny
118 Paper 2.1.6
The genus Hipposideros Gray, 1831 represents
the most speciose group within the Palaeotropical
family of Old World leaf-nosed bats (Hipposideridae).
It contains a high number of phonic types,
some of which have been subsequently confirmed
as new cryptic species by molecular analyses, e.g.,
H. khakhouayensis (Guillén-Servent and Francis,
2006) and H. khasiana (Thabah et al., 2006) from
Southeast Asia. Distinct phonic types exist also
within African species of Hipposideros, although
none or only limited molecular justification appeared
to date to confirm their distinct taxonomic
statuses. The first is the case of phonic types of
African H. commersoni (E. Geoffroy, 1813), which
were revealed by Pye (1972) and which have been
recently raised to species rank as H. gigas (Wagner,
1845) and H. vittatus (Peters, 1852) (Simmons,
2005). The second example is the complex of forms
pertaining to H. caffer (Sundevall, 1846). Currently
recognised species of this complex, H. caffer and
H. ruber (Noack, 1893), are generally assumed to
differ in peak frequencies of their echolocation calls
(Pye, 1972; Fenton, 1986; Heller, 1992; Jones et al.,
1993). However, the first study of the H. caffer
complex using molecular genetic tools (Vallo et al.,
2008) showed that this group is composed of more
than two evolutionary units that might represent separate
species. As the metric characters traditionally
used to distinguish between H. caffer and H. ruber
overlap, importance of echolocation frequencies for
taxonomy should be re-evaluated in relation to molecular
phylogeny. Differences in echolocation parameters
can be good indicators of cryptic species.
Particularly with respect to social communication
between conspecific individuals, differences in
echolocation appear to be more valid for closely
related Hipposideros bats than the concept of adaptation
to feeding on hearing insects (Barratt et al.,
1997; Jones et al., 1997; Bogdanowicz et al., 1999;
Guillén-Servent et al., 2000; Sedlock and Weyandt,
2009).
Our field work in the Niokolo Koba National
Park (NKNP), Senegal, yielded a large number of
leaf-nosed bats tentatively identified as H. ruber.
These bats represent a distinct phylogenetic lineage
restricted to West Africa, ranging from Senegal to
Benin (‘lineage D’ of Vallo et al., 2008), and are
probably not closely related to H. ruber s. str., which
was described from Tanzania. Although samples of
Hipposderos aff. ruber bats from NKNP showed no
obvious morphological differences, in the subsample
used by Vallo et al. (2008), one haplotype significantly
differed from the main mitochondrial
lineage of the Senegalese population. This interesting
intrapopulational pattern inspired us to investigate
phylogenetic relationships in H. aff. ruber from
Senegal using sequences of the mitochondrial gene
for cytochrome b (cytb). We studied genetic diversity
of populations in the NKNP to evaluate the relevance
of two sympatric mitochondrial lineages for
the systematics of the H. caffer complex. We further
analysed sequences of male-specific nuclear zinc
finger region on the Y chromosome (zfy — Page et
al., 1987) in the two mitochondrial lineages to obtain
independent evidence for reproductive separation.
Additionally, we used skull morphometrics in
relation to patterns revealed on the genetic level.
As the divergent haplotype within Senegalese H. aff.
ruber came from the village of Dar Salam at the
northwestern border of the NKNP, we also explored
echolocation calls in the local population to determine
whether calls differ in correspondence to genetic
lineages. We hypothesised that possible differences
in echolocation frequencies and morphology
would support presence of cryptic species suggested
by genetic divergence.
MATERIALS AND METHODS
Sampling
Bats were netted at six localities in the NKNP, SE Senegal,
between 2004 and 2007 (Fig. 1 and Appendix). Captured specimens
were weighted and external measurements were recorded.
A subset of 52 specimens was collected and preserved in ethanol
for further study. Tissue samples (spleen) were taken from collected
specimens. Skulls were extracted from ethanol-preserved
vouchers for morphological analysis. All biological material
including voucher specimens was deposited at the Institute of
Vertebrate Biology of the Academy of Sciences, Brno, Czech
Republic. For comparison, we included additional samples and
GenBank sequences of West African H. aff. ruber from Benin,
Ghana, and Ivory Coast (Lim et al., 2007; Vallo et al., 2008),
and another hipposiderid bat, Asellia tridens, as an outgroup
(Benda and Vallo, 2009 — see Appendix).
Morphological Analysis
One external and 15 skull dimensions of collected specimens
were measured following Benda and Vallo (2009) and
using mechanical callipers with a precision of 0.02 mm: LAt =
forearm length, LCr = greatest length of skull including premaxillae,
LOc = occipito-canine length, LCc = condylo-canine
length, LaZ = zygomatic width, LaI = width of interorbital constriction,
LaInf = rostral width between foramina infraorbitalia,
LaN = neurocranium width, LaM = mastoid width, ANc =
height of neurocranium, LBT = largest horizontal length of
tympanic bulla, CC = width across upper canines at crowns,
M3M3 = width across third upper molars, CM3 = length of upper
tooth-row from front of canine to back of third molar, LMd
= condylar length of mandible, ACo = height of coronoid
80 P. Vallo, P. Benda, N. Martínková, P. Kaňuch, E. K. V. Kalko, et al.
Paper 2.1.6 119
process, CM3 = length of lower tooth-row from front of canine
to back of third molar. Analysis of variance and canonical discriminant
analysis using raw data values were employed to
reveal morphological differences in our dataset. Statistical
analyses were performed using Statistica 6.0 software (StatSoft,
Tulsa, OK, USA).
Analysis of Echolocation Calls
Echolocation calls of individuals captured at Dar Salam
were recorded in time expansion (10×) and heterodyne mode
with a bat-detector Pettersson D240x (Pettersson Elektronik
AB, Uppsala, Sweden; sampling rate 307 kHz) with built-in
microphone linked to a Sony MiniDisc MZ. Sound recordings
were saved in uncompressed digital format (.wav). Animals
were flown individually in a volary made of mosquito net
(2 × 2 × 2 m). Echolocation calls were recorded when bats were
flying towards the microphone only, thus compensation for the
Doppler effect should have the same rate in the small constantsized
volary. Peak frequency of the constant frequency (CF)
component of calls with maximum energy derived from power
spectra was extracted from time-expanded (10×) sequences of
approx. 1 s (FFT size 512 samples, Hanning window) with BatSound
3.31 software (Pettersson Elektronik AB). We measured
peak frequencies of four consecutive, randomly selected calls
from the sonograms of each individual and took their average as
the final value.
DNA Processing
Total genomic DNA was extracted from ethanol-preserved
tissue with DNeasy Tissue Kit (Qiagen, Halden, Germany)
Genetic diversity in Hipposideros aff. ruber in Senegal 81
FIG. 1. Distribution of localities sampled in Niokolo Koba NP,
southeastern Senegal, and of all West African populations
sampled; Senegal (circle; S), and Ivory Coast, Ghana, and Benin
(squares; IC, G, B, respectively)
according to the manufacturer’s protocol. Complete cytb
gene was amplified using universal primers L14724 and
H15915 (Irwin et al., 1991). Each PCR reaction contained
0.8 μM of each primer, 0.2 mM dNTP, 1U of HotMaster
Taq DNA polymerase with 5 μl of corresponding 10× buffer
(Eppendorf, Hamburg, Germany), and 2–5 μl of extracted
DNA in 50 μl volume. Initial denaturation at 94°C for 3 min
was followed by 35 cycles of denaturation for 40 s at 94°C,
annealing for 40 s at 50°C, and extension for 90 s at 65°C,
with final extension at 65°C for 5 min. Partial sequences of
zfy gene were amplified using primers 33X5YF and LGL
331 (Trujillo et al., 2009) and the same PCR protocol.
The resulting PCR products were purified with QIAquick
PCR Purification Kit (Qiagen) and sequenced commercially
(Macrogen, Seoul, Korea) with the same primers using
Big-Dye Terminator sequencing chemistry (Applied Biosystems,
Foster City, CA, USA) on ABI 3730xl sequencer. Sequences
were assembled and edited in Sequencher 4.6 (Gene
Codes, Ann Arbor, MI, USA). Sequences were submitted to
GenBank with accession numbers HQ343240–HQ343266
(Appendix).
Phylogenetic Analysis
Sequences of Hipposideros aff. ruber from Senegal were
aligned in BioEdit 7.0 (Hall, 1999). Polymorphism within the
Senegalese sequence dataset was assessed using DnaSP 4.0
(Rozas et al., 2003) and a median-joining network of cytb sequences
was constructed in Network 4.2 (Fluxus Technology,
Clare, UK). Based on this initial analysis, the cytb dataset
was reduced to haplotypes representing main haplogroups to
facilitate reconstruction of phylogenetic trees and additional
sequences including the outgroup were added for subsequent
analyses (Appendix).
Phylogenetic trees were reconstructed in PAUP* 4.10b
(Sinauer Associates, Sunderland, MA, USA) using maximum
parsimony (MP) and maximum likelihood (ML) methods. In
both methods, tree space was heuristically searched with tree
bisection-reconnection swapping algorithm on 100 random
sequence additions. Hasegawa-Kishino-Yano evolutionary
model with gamma-distributed among-site rate variation
(HKY85 + Γ — Hasegawa et al., 1985; Yang, 1996) was used
in ML analysis. This five-parameter model was suggested by the
program Modeltest 3.7 (Posada and Crandall, 1998) as the 3rd
best model under the Akaike Information Criterion (AIC),
and was chosen in preference to two more complex models
with eight parameters in order to reduce variance in parameter
estimates. Reliability of branching pattern was assessed by
bootstrapping using 1000 replicates in both analyses. Phylogeny
was further estimated using Bayesian inference in MrBayes
3.1.2 (Ronquist and Huelsenbeck, 2003) with the same model.
We used two independent simultaneous Metropolis-coupled
MCMC runs of four chains running for 106 generations, sampled
every 100th generation, starting from random trees. The
first 2,500 sampled trees were discarded as burn-in. A 50%
majority rule consensus tree was constructed from remaining
trees with posterior probabilities representing confidence
estimates of topology. Templeton test (Templeton, 1983) and
Shimodaira-Hasegawa test (SH test — Shimodaira and
Hasegawa, 1999) with RELL re-sampling algorithm and 1,000
bootstrap replicates were used to compare tree topologies.
Sequence divergences were based on pairwise Kimura twoparameter
genetic distances (K2P — Kimura, 1980).
120 Paper 2.1.6
RESULTS
In the Senegalese dataset, 29 haplotypes of cytb
(1,140 bp) were identified among 52 analysed
individuals, resulting in high haplotype diversity
(H = 0.966, SD = 0.01). As most haplotypes
were closely related, nucleotide diversity was low
(π = 0.009, SD = 0.002). Median-joining network
showed diversification into two groups within Senegalese
H. aff. ruber, denoted here with respect to the
original lineage D by Vallo et al. (2008) as D1 and
D2 (Fig. 2). Group D1 comprised 27 haplotypes representing
47 specimens. Group D2 was formed by
two haplotypes from five specimens. Groups D1 and
D2 differed by 2.4–3.6% K2P-distance. Genetic
structure was not related to distribution of sampling
sites, and bats of groups D1 and D2 occurred syntopically
at Dar Salam, Simenti and Lingue Kountou
(Appendix). Sequences of zfy (1,316 bp) from four
specimens of group D1 and from two specimens of
group D2 were identical.
We recorded echolocation calls of 16 specimens
from Dar Salam. Variation in peak frequencies within
the four measured consecutive calls of individual
bats was very low (around 0.1 kHz). Given the low
intra-individual variation in peak frequency and
similar recording conditions in the constant-sized
volary leading to similar recording bias (i.e., equal
compensation for the Doppler effect), we compared
average peak frequencies among individuals. Peak
frequencies ranged from 130 kHz to 139 kHz (Appendix).
Genetic results showed that all of the
recorded specimens belonged to group D1, hence
precluding a comparison with group D2 as originally
expected. Frequencies were related to sex,
as males called at significantly higher frequencies
(134–139 kHz, n = 8) than females (130–135 kHz,
n = 8 — Mann-Whitney test, U = 4, P < 0.01).
Univariate analysis of forearm length and skull
morphometrics did not reveal distinct morphotypes.
Measurements of the two haplogroups D1 and D2
mostly overlapped (Table 1); they differed only
in CM3 (ANOVA, F = 5.81, d.f. = 48, P < 0.05)
and slightly in ANc (ANOVA, F = 3.78, d.f. = 48,
P = 0.058). Males and females of groups D1 and
D2 did not differ in size. Canonical discriminant
analysis of skull dimension revealed LCo, LCc,
LaZ, LaI, CC, ANc, CM3 and CM3 as the most important
variables distinguishing between groups D1
and D2 along the 1st canonical axis (CV1), which
explained 71.0% of variance (Fig. 3).
Phylogenetic relationships were reconstructed
among 11 cytb sequences of H. aff. ruber: six from
Senegal (four from haplogroup D1 and two from
haplogroup D2), two from Ivory Coast, two from
Ghana, and one from Benin. Heuristic search under
MP criterion yielded a single best tree 269 steps long
(Fig. 4) showing five well-supported (bootstrap support
≥ 70%, Bayesian posterior probability ≥ 0.95)
phylogenetic lineages, which corresponded to the
respective geographic origin of samples, and included
the two distinct haplogroups from Senegal:
Senegal D1, Senegal D2, Benin, Ghana, and Ivory
Coast. Genetic divergences among the five lineages
ranged 2.0–5.4% (Table 2). Divergences within the
lineages except those from Senegal were small (up
to 0.1%). Monophyletic relationship was supported
between lineages Senegal D2 and Benin, and between
lineages Ivory Coast and Ghana, respectively.
Senegal D1 was placed as sister lineage to a clade
82 P. Vallo, P. Benda, N. Martínková, P. Kaňuch, E. K. V. Kalko, et al.
FIG. 2. Median-joining network of 29 haplotypes found in the population of NKNP. Size of nodes is proportional to frequency
of particular haplotypes. Black and white circles denote haplogroups D1 and D2, respectively, and this scheme is consistent with
Figs. 3 and 4
Paper 2.1.6 121
Senegal D2 + Benin, but the statistical support for
this group was low. ML tree (-lnL = 2741.72205)
and Bayesian consensus tree showed the same supported
groups as the MP tree: Senegal D2 + Benin
and Ivory Coast + Ghana, and their supported sister
relationship. However, they also revealed a rather
unclear pattern of paraphyletic haplotypes belonging
to the group Senegal D1 placed basally in the
tree. Thus, both monophyly of Senegal D1 and its
phylogenetic position considering the monophyly
were tested using Templeton and SH tests. Constrained
ML tree with forced monophyly of Senegal
D1 (-lnL = 2743.04876) did not differ significantly
from the original ML tree (Templeton test: diff.
length = 4, z = -1.1547, P = 0.2482; SH test: diff.
-lnL = 1.32671, P = 0.168) and monophyly of Senegal
D1 thus could not be rejected. Position of
other lineages in the constrained tree was otherwise
Genetic diversity in Hipposideros aff. ruber in Senegal 83
FIG. 3. Plot of main canonical variables CV1 and CV2 from canonical discriminant analysis of 15 skull measurements. Symbols as
in Fig. 2
Measurements
Haplogroup D1 Haplogroup D2
n 0 Min Max SD n 0 Min Max SD
LAt 47 47.88 44.20 49.80 1.100 4 47.15 46.30 48.30 0.780
LCr 45 18.92 18.27 19.62 0.298 5 18.77 18.28 19.11 0.338
LOc 45 18.68 17.78 19.23 0.294 5 18.56 18.14 18.86 0.264
LCc 45 16.34 15.83 16.83 0.238 5 16.23 16.03 16.37 0.149
LaZ 45 10.51 10.02 10.86 0.222 5 10.57 10.28 10.75 0.195
LaI 45 2.88 2.59 3.09 0.120 5 2.82 2.60 2.98 0.143
LaInf 45 5.07 4.81 5.42 0.114 5 5.06 4.98 5.18 0.088
LaN 45 8.31 7.75 8.68 0.176 5 8.27 7.85 8.73 0.370
LaM 45 9.89 9.52 10.28 0.154 5 9.80 9.69 9.97 0.121
ANc 45 5.85 5.29 6.38 0.223 5 6.07 5.80 6.67 0.353
CC 45 4.98 4.73 5.24 0.134 5 4.93 4.75 5.06 0.138
M3
M3
45 7.15 6.68 7.42 0.159 5 7.11 6.82 7.25 0.168
CM3
45 7.08 6.78 7.29 0.123 5 6.95 6.84 7.05 0.076
LMd 45 12.38 11.82 12.74 0.197 5 12.41 12.31 12.51 0.091
ACo 45 3.04 2.74 3.31 0.142 5 3.03 2.93 3.23 0.117
CM3 45 7.65 7.37 7.93 0.138 5 7.62 7.55 7.68 0.048
LBT 45 3.36 3.07 3.55 0.092 5 3.34 3.28 3.48 0.080
TABLE 1. Forearm length and skull measurements (in mm) of the examined specimens. See Materials and Methods for abbreviations
of measurements
CV1
CV2
122 Paper 2.1.6
identical to the original ML tree. Although ML
analysis favoured sister relationship between Senegal
D2 + Benin and Ivory Coast + Ghana (73% bootstrap
support) in contrast to MP analysis that pointed
towards sister relationship of Senegal D1 to Senegal
D2 + Benin (67% bootstrap support), these two
phylogenetic hypotheses did not differ from each
other significantly (Templeton test: diff. length = 3,
z = -0.9045, P = 0.37; SH test: diff. -lnL = 0.71613,
P = 0.29). Consequently, mutual positions of the lineages
Senegal D1, Senegal D2 + Benin and Ivory
Coast + Ghana remained unresolved.
DISCUSSION
Analysis of mitochondrial DNA sequences of
H. aff. ruber revealed two distinct genetic lineages
with sequence divergence of 2.4–3.6% co-occurring
in the Niokolo Koba NP, southeastern Senegal. Only
five specimens from our collection belonged to lineage
D2. Haplotypes of both lineages were present
at three localities in the NKNP. Absence of the lineage
D2 at two other localities may have been due to
low sample size, although habitat preferences or
competition between the two lineages may also be
a relevant explanation. As previous results indicated
84 P. Vallo, P. Benda, N. Martínková, P. Kaňuch, E. K. V. Kalko, et al.
FIG. 4. Maximum parsimony tree depicting phylogenetic relationships in West African H. aff. ruber. Bootstrap support for MP and
ML ≥ 70% (above branches) and posterior probabilities of BA ≥ 0.95 (below braches) are considered significant. Symbols as in Fig. 2
existence of both lineages at Dar Salam (Vallo et al.,
2008), we compared echolocation calls in the local
population. The differences in call frequencies
might indicate presence of cryptic species (Jones,
1997; Mayer and von Helversen, 2001). Our results
showed peak echolocation frequencies of CF calls
to range 130–139 kHz, which is fairly similar to the
combined range of cryptic forms of H. bicolor
(128.0–144.5 kHz — Kingston et al., 2001) and
H. ridley (65–72 kHz — Francis et al., 1999). However,
genetic analysis revealed that all 16 individuals,
in which we successfully recorded and analysed
echolocation frequencies, belonged to the more
abundant lineage D1. Therefore, the original hypothesis
of congruence between echolocation and
genetic diversity in Senegalese H. aff. ruber could
K2P Senegal D2 Senegal D1 Benin Ivory Coast
Senegal D1 2.6–3.6
Benin 2.0–2.1 3.2–4.1
Ivory Coast 5.1–5.2 4.0–4.3 5.3–5.4
Ghana 4.4–4.7 4.5–5.1 5.2–5.3 2.8–3.0
TABLE 2. Pairwise genetic divergences among phylogenetic
lineages of West African H. aff. ruber. K2P — Kimura twoparameter
distance (%)
Senegal D1
Senegal D2
Benin
Ivory Coast
Ghana
Paper 2.1.6 123
not be tested. Distribution of echolocation frequencies
within lineage D1 was related to sex, with
males calling at significantly higher frequencies
than females. Similar relationship of echolocation
frequency and sex has been shown in H. ruber from
Equatorial Guinea (Guillén-Servent et al., 2000).
The frequency range of 130–139 kHz found in the
bat population from Dar Salam should be regarded
as intraspecific variation.
Morphological analysis showed only slight differences
in skull dimensions. Overall, bats of the
two lineages remain indistinguishable by morphological
traits that are commonly used for species
identification. Although the canonical discriminant
analysis showed rather sufficient differences for
separation of the lineages in some (mainly rostral)
dimensions, the ranges of measured values overlap
to such an extent that competent determination of
taxa is not possible.
Given the values of sequence divergence and
sympatric occurrence, the two lineages from Senegal
could be considered separate species based on
known limits of divergence between cryptic species
(Baker and Bradley, 2006). Similar values of interspecific
divergence (2.4–3.9%) have been found
between cryptic forms of Scotophilus dinganii
(A. Smith, 1833) from South Africa (Jacobs et al.,
2006). Furthermore, similar divergence (3.9–4.1%)
was documented between two Hipposideros species
from Southeast Asia, H. khakhouayensis and H. rotalis
(Guillén-Servent and Francis, 2006). Sequence
divergence values alone, however, have to be interpreted
with caution when deciding on an appropriate
taxonomic rank of genetically separate populations.
Lausen et al. (2008) clearly showed that divergence
values in mitochondrial sequences led to false classification
of lineages within North American Myotis
lucifugus as distinct species, as they discovered substantial
nuclear gene flow among these lineages although
mitochondrial sequence divergence ranged
up to 5%. Similar mitochondrial differentiation not
reflected in nuclear markers was found in Myotis
capaccinii (Bilgin et al., 2008). It is therefore important
to document reproductive isolation of the
two genetically distinct forms in Senegal to confirm
their status as two species. Our use of a paternally
inherited zfy gene as an independent phylogenetic
marker to maternally inherited cytb gene (Lim et al.,
2008; Trujillo et al., 2009) did not provide a conclusive
answer because all obtained zfy sequences
were identical. On one hand, this uniformity may
indicate extensive, male-mediated gene flow between
haplogroups D1 and D2. On the other hand, it
could also result from low mutation rate of this
nuclear gene and a relatively recent split of both lineages.
More informative estimation of relationships
between the two Senegalese mitochondrial forms
could be achieved using analysis of microsatellites,
as this would detect gene flow between both lineages
and thus provide necessary data for a taxonomic
conclusion.
The basal node of phylogenetic tree remained
unresolved suggesting a rapid radiation of the three
basal lineages Senegal D1, Senegal D2 + Benin, and
Ghana + Ivory Coast. Moreover, a sister relationship
of the two Senegalese lineages was not supported
and the lineage D2 turned out to be closely related to
the lineage from Benin. According to this phylogenetic
structure and considering geographical distribution
of the lineages, the sympatric occurrence of
the two lineages from Senegal, D1 and D2, may be
explained as a secondary contact of two formerly
isolated populations. The genetic distance, reaching
over 5% among the sampled West African populations,
further suggests that even lineages from
Ghana and Ivory Coast might be regarded as cryptic
species. Unlike the Senegalese lineages D1 and D2,
these populations do not occur in sympatry and the
genetic differences could be explained by the isolation
by distance. On the other hand, conflict between
geographic distance between sampled localities in
Ghana and Benin (ca. 300 km), where the corresponding
divergence exceeds 5%, and Senegal and
Benin (ca. 2,000 km), with divergence only 2%,
indicate a rather more complicated relationship.
A broader sampling throughout West Africa would
thus be needed to better understand the indicated
phylogeographic pattern and resolve the question of
cryptic diversity in H. aff. ruber.
ACKNOWLEDGEMENTS
We thank Adam Konečný, Josef Bryja and other colleagues
from the Institute of Vertebrate Biology AS CR, v.v.i., Brno, for
their assistance in capturing of bats in Senegal. Field work in the
NKNP and collecting of bat specimens was approved and supervised
by the Direction des Parcs Nationaux du Sénégal, Dakar,
and we thank the director Col. Mame Balla Gueye for his kind
support. Further we thank Burton Lim (Royal Ontario Museum,
Toronto, Canada) for providing tissue samples of bats from
Ivory Coast, and Christian Drosten (University of Bonn Medical
Centre, Germany) and his team for acquiring samples from
Ghana within the project 01KI0701 by the German Federal
Ministry of Education and Research. This study was supported
by the grant No. IAA6093404 of the Grant Agency of the
Academy of Sciences of the Czech Republic and grant No.
MK00002327201 of the Ministry of Culture of the Czech Republic.
We also thank Jakob Fahr (University of Ulm, Germany)
for critical comments and suggestions on the manuscript.
Genetic diversity in Hipposideros aff. ruber in Senegal 85
124 Paper 2.1.6
LTERATURE CITED
AVISE, J. C. 2004. Molecular markers, natural history, and evolution,
2nd edition. Sinauer Associates, Sunderland, Massachusetts,
684 pp.
BAKER, R. J., and R. D. BRADLEY. 2006. Speciation in mammals
and the genetic species concept. Journal of Mammalogy, 87:
643–662.
BARRATT, E. M., R. DEAVILLE, T. M. BURLAND, M. W. BRUFORD,
G. JONES, P. A. RACEY, and R. K. WAYNE. 1997.
DNA answers the call of pipistrelle bat species. Nature, 387:
138–139.
BENDA, P., and P. VALLO. 2009. Taxonomic revision of the genus
Triaenops (Mammalia: Chiroptera: Hipposideridae) with
description of a new species from southern Arabia and definitions
of new genus and tribe. Folia Zoologica, 53 (Monograph
1): 1–45.
BICKFORD, D., D. J. LOHMAN, N. S. SODHI, P. K. L. NG, R.
MEIER, K. WINKER, K. K. INGRAM, and I. DAS. 2007. Cryptic
species as a window on diversity and conservation. Trends
in Ecology and Evolution, 22: 148–155.
BILGIN, R., A. KARATAŞ, E. ÇORAMAN, and J. MORALES. 2008.
The mitochondrial and nuclear genetic structure of Myotis
capaccinii (Chiroptera: Vespertilionidae) in the Eurasian
transition, and its taxonomic implications. Zoologica Scripta,
37: 253–262.
BOGDANOWICZ, W., M. B. FENTON, and K. DALESZCZYK. 1999.
The relationships between echolocation calls, morphology
and diet in insectivorous bats. Journal of Zoology (London),
247: 381–393.
FENTON, M. B. 1986. Hipposideros caffer (Chiroptera: Hipposideridae)
in Zimbabwe: morphology and echolocation calls.
Journal of Zoology (London), 210: 347–353.
FRANCIS, C. M., D. KOCK, and J. HABERSETZER. 1999. Sibling
species of Hipposideros ridleyi (Mammalia, Chiroptera,
Hipposideridae). Senckenbergiana Biologica, 79: 255–270.
GUILLÉN-SERVENT, A., and C. M. FRANCIS. 2006. A new species
of bat of the Hipposideros bicolor group (Chiroptera: Hipposideridae)
from Central Laos, with evidence of convergent
evolution with Sundaic taxa. Acta Chiropterologica, 8:
39–61.
GUILLÉN-SERVENT, A., C. IBÁÑEZ, and J. JUSTE. 2000. Variation
in the echolocation calls of Hipposideros ruber in the Gulf
of Guinea: An exploration of the adaptive meaning of the
constant frequency value in rhinolophoid bats. Journal of
Evolutionary Biology, 13: 70–80.
HALL, T. A. 1999. BioEdit: a user-friendly biological sequence
alignment editor and analysis program for Windows 95/98/
NT. Nucleic Acids Symposium Series, 41: 95–98.
HASEGAWA, M., H. KISHINO, and T. YANO. 1985. Dating of the
human-ape splitting by a molecular clock of mitochondrial
DNA. Journal of Molecular Evolution, 22: 160–174.
HELLER, K.-G. 1992. The echolocation calls of Hipposideros
ruber and Hipposideros caffer. Pp. 75–77, in Prague studies
in mammalogy (I. HORÁČEK and V. VOHRALÍK, eds.). Charles
University Press, Prague, xxi + 245 pp.
IBÁÑEZ, C., J. L. GARCÍA-MUDARRA, M. RUEDI, B. STADELMANN,
and J. JUSTE. 2006. The Iberian contribution to cryptic diversity
in European bats. Acta Chiropterologica, 8: 277–297.
IRWIN, D. M., T. D. KOCHER, and A. C. WILSON. 1991. Evolution
of the cytochrome b gene of mammals. Journal of Molecular
Evolution, 32: 128–144.
JACOBS, D. S., G. N. EICK, M. C. SCHOEMAN, and C. A. MATHEE.
2006. Cryptic species in an insectivorous bat, Scotophilus
dinganii. Journal of Mammalogy, 87: 161–170.
JONES, G. 1997. Acoustic signals and speciation: the roles of
natural and sexual selection in the evolution of cryptic species.
Advanced Study in Behaviour, 26: 317–354.
JONES, G., M. MORTON, P. M. HUGHES, and R. M. BUDDEN.
1993. Echolocation, flight morphology and foraging strategies
of some West African hipposiderid bats. Journal of Zoology
(London), 230: 385–400.
KIMURA, M. 1980. A simple method for estimating evolutionary
rates of base substitutions through comparative studies of
nucleotide sequences. Journal of Molecular Evolution, 16:
111–120.
KINGSTON, T., M. C. LARA, G. JONES, Z. AKBAR, T. H. KUNZ, and
C. J. SCHNEIDER. 2001. Acoustic divergence in two cryptic
Hipposideros species: a role for social selection? Proceedings
of the Royal Society of London, 268: 1381–1386.
LAUSEN, C. L., I. DELISLE, R. M. R. BARCLAY, and C. STROBECK.
2008. Beyond mtDNA: nuclear gene flow suggests taxonomic
over-splitting in the little brown bat (Myotis lucifugus).
Canadian Journal of Zoology, 86: 700–713.
LIM, B. K., M. D. ENGSTROM, J. W. BICKHAM, and J. C. PATTON.
2007. Molecular phylogeny of New World sheath-tailed
bats (Emballonuridae: Diclidurini) based on loci from the
four genetic transmission systems in mammals. Biological
Journal of the Linnean Society, 93: 189–209.
LINCOLN, R., G. BOXSHALL, and P. CLARK. 1998. A dictionary of
ecology, evolution and systematics, 2nd edition. Cambridge
University Press, Cambridge, United Kingdom, 371 pp.
MAYER, F., and O. VON HELVERSEN. 2001. Cryptic diversity in
European bats. Proceedings of the Royal Society of London,
268B: 1825–1832.
MAYR, E. 1996. What is a species, and what is not? Philosophy
of Science, 63: 262–277.
PAGE, D. C., R. MOSHER, E. M. SIMPSON, E. M. C. FISHER, G.
MARDON, J. POLLACK, B. MCGILLIVRAY, A. DE LA CHAPELLE,
and L. G. BROWN. 1987. The sex-determining region of the
human Y chromosome encodes a finger protein. Cell, 51:
1091–1104.
POSADA, D., and K. A. CRANDALL. 1998. Modeltest: testing the
model of DNA substitution. Bioinformatics, 14: 817–818.
PYE, J. D. 1972. Bimodal distribution of constant frequencies in
some hipposiderid bats (Mammalia: Hipposideridae). Journal
of Zoology (London), 166: 323–335.
RONQUIST, J., and J. J. HUELSENBECK. 2003. MrBayes 3: Bayesian
phylogenetic inference under mixed models. Bioinformatics,
19: 1572–1574.
ROZAS, J., J. C. SÁNCHEZ-DELBARRIO, X. MESSEGUER, and
R. ROZAS. 2003. DnaSP, DNA polymorphism analyses
by the coalescent and other methods. Bioinformatics, 19:
2496–2497.
SEDLOCK, J. L., and S. E. WEYANDT, 2009. Genetic divergence
between morphologically and acoustically cryptic bats: novel
niche partitioning or recent contact? Journal of Zoology
(London), 279: 388–395.
SHIMODAIRA, H., and M. HASEGAWA. 1999. Multiple comparisons
of log-likelihoods with applications to phylogenetic inference.
Molecular Biology and Evolution, 16: 1114–1116.
SIMMONS, N. 2005. Order Chiroptera. Pp. 312–529, in Mammal
species of the World: a taxonomic and geographic reference
(D. E. WILSON and D. M. REEDER, eds.). The Johns
Hopkins University Press, Baltimore, 2142 pp.
TEMPLETON, A. R. 1983. Phylogenetic inference from restriction
86 P. Vallo, P. Benda, N. Martínková, P. Kaňuch, E. K. V. Kalko, et al.
Paper 2.1.6 125
Genetic diversity in Hipposideros aff. ruber in Senegal 87
endonuclease cleavage site maps with particular reference
to the humans and apes. Evolution, 37: 221–244.
THABAH, A., S. J. ROSSITER, T. KINGSTON, S. ZHANG, S. PARSONS,
K. MYA MYA, A. ZUBAID, and G. JONES. 2006. Genetic
divergence and echolocation call frequency in cryptic species
of Hipposideros larvatus s.l. (Chiroptera: Hipposideridae)
from the Indo-Malayan region. Biological Journal of
the Linnean Society, 88: 119–130.
TRUJILLO, R. G., J. C. PATTON, D. A. SCHLITTER, and J. W. BICKHAM.
2009. Molecular phylogenetics of the bat genus Scotophilus
(Chiroptera: Vespertilionidae): perspectives from paternally
and maternally inherited genomes. Journal of Mammalogy,
90: 548–560.
VALLO, P., A. GUILLÉN-SERVENT, P. BENDA, D. B. PIRES, and
P. KOUBEK. 2008. Variation of mitochondrial DNA in the
Hipposideros caffer complex (Chiroptera: Hipposideridae)
and its taxonomic implications. Acta Chiropterologica, 10:
193–206.
YANG, Z. 1996. Among-site rate variation and its impact on
phylogenetic analyses. Trends in Ecology and Evolution,
11: 367–372.
YODER, A. D., R. M. RASOLOARISON, S. M. GOODMAN, J. A.
IRWIN, S. ATSALIS, M. J. RAVOSA, and J. H. GANZHORN.
2000. Remarkable species diversity in Malagasy mouse lemurs
(Primates, Microcebus). Proceedings of the National
Academy of Sciences of the USA, 97: 1325–1330.
Received 25 February 2010, accepted 10 February 2011
APPENDIX
List of specimens included in this paper. Echo — average peak frequency of CF echolocation call, Acc. number — accession
number in the GenBank database
Sample Haplotype Haplogroup Echo (kHz) Country Locality Acc. number Source of sequence
IVB S8 Hap_1 D1 – Senegal Mt. Assirik HQ343240 This study
IVB S95 Hap_9 D1 – Senegal Simenti – This study
IVB S112 Hap_2 D1 – Senegal Lingué Kountou HQ343241 This study
IVB S119 Hap_3 D1 – Senegal Lingué Kountou EU934478 Vallo et al. (2008)
IVB S139 Hap_6 D1 – Senegal Lingué Kountou HQ343243 This study
IVB S218 Hap_8 D1 – Senegal Simenti HQ343244 This study
IVB S253 Hap_6 D1 – Senegal Simenti HQ343245 This study
IVB S272 Hap_9 D1 – Senegal Simenti EU934481 Vallo et al. (2008)
IVB S273 Hap_10 D1 – Senegal Simenti EU934482 Vallo et al. (2008)
IVB S275 Hap_11 D1 – Senegal Simenti EU934483 Vallo et al. (2008)
IVB S278 Hap_5 D1 – Senegal Simenti HQ343246 This study
IVB S280 Hap_12 D1 – Senegal Simenti HQ343247 This study
IVB S283 Hap_13 D1 – Senegal Simenti HQ343249 This study
IVB S285 Hap_14 D1 – Senegal Simenti EU934484 Vallo et al. (2008)
IVB S290 Hap_15 D1 – Senegal Simenti HQ343250 This study
IVB S291 Hap_11 D1 – Senegal Simenti – This study
IVB S341 Hap_16 D1 – Senegal Simenti HQ343251 This study
IVB S342 Hap_9 D1 – Senegal Simenti – This study
IVB S362 Hap_17 D1 – Senegal Simenti HQ343252 This study
IVB S695 Hap_18 D1 – Senegal Dar Salam HQ343253 This study
IVB S701 Hap_14 D1 – Senegal Dar Salam – This study
IVB S702 Hap_19 D1 – Senegal Dar Salam HQ343254 This study
IVB S803 Hap_5 D1 – Senegal Lingué Kountou – This study
IVB S819 Hap_3 D1 – Senegal Dindéfélou – This study
IVB S820 Hap_20 D1 – Senegal Dindéfélou EU934485 Vallo et al. (2008)
IVB S821 Hap_21 D1 – Senegal Dindéfélou HQ343255 This study
IVB S825 Hap_20 D1 – Senegal Dindéfélou – This study
IVB S899 Hap_5 D1 – Senegal Dar Salam – This study
IVB S900 Hap_22 D1 – Senegal Dar Salam HQ343256 This study
IVB S1374 Hap_5 D1 – Senegal Dar Salam EU934479 Vallo et al. (2008)
IVB S1377 Hap_27 D1 – Senegal Dar Salam HQ343260 This study
IVB S1538 Hap_28 D1 138 Senegal Dar Salam HQ343261 This study
IVB S1539 Hap_11 D1 137 Senegal Dar Salam – This study
IVB S1540 Hap_9 D1 134 Senegal Dar Salam – This study
IVB S1541 Hap_14 D1 139 Senegal Dar Salam – This study
IVB S1551 Hap_23 D1 132 Senegal Dar Salam HQ343262 This study
126 Paper 2.1.6
88 P. Vallo, P. Benda, N. Martínková, P. Kaňuch, E. K. V. Kalko, et al.
APPENDIX. Continued
Sample Haplotype Haplogroup Echo (kHz) Country Locality Acc. number Source of sequence
IVB S1554 Hap_23 D1 136 Senegal Dar Salam – This study
IVB S1555 Hap_24 D1 137 Senegal Dar Salam HQ343257 This study
IVB S1561 Hap_29 D1 133 Senegal Dar Salam HQ343263 This study
IVB S1654 Hap_24 D1 133 Senegal Dar Salam – This study
IVB S1655 Hap_24 D1 136 Senegal Dar Salam – This study
IVB S1657 Hap_25 D1 134 Senegal Dar Salam HQ343258 This study
IVB S1660 Hap_11 D1 131 Senegal Dar Salam – This study
IVB S1662 Hap_5 D1 131 Senegal Dar Salam – This study
IVB S1663 Hap_14 D1 135 Senegal Dar Salam – This study
IVB S1664 Hap_14 D1 135 Senegal Dar Salam – This study
IVB S1665 Hap_26 D1 130 Senegal Dar Salam HQ343259 This study
IVB S132 Hap_4 D2 – Senegal Lingué Kountou HQ343242 This study
IVB S281 Hap_7 D2 – Senegal Simenti HQ343248 This study
IVB S286 Hap_7 D2 – Senegal Simenti – This study
IVB S403 Hap_7 D2 – Senegal Dar Salam – This study
IVB S1400 Hap_7 D2 – Senegal Dar Salam EU934480 Vallo et al. (2008)
NMP 91879 Hap_B1 – – Benin Tagayé EU934476 Vallo et al. (2008)
ROM 100516 Hap_IC1 – – Ivory Coast Sibabli EF584226 Lim et al. (2008)
ROM 100518 Hap_IC2 – – Ivory Coast Sibabli HQ343264 This study
IVB PV59 Hap_G1 – – Ghana Buoyem HQ343265 This study
IVB PV56 Hap_G2 – – Ghana Buoyem HQ343266 This study
Asellia tridens – – Egypt – FJ457617 Benda and Vallo (2009)
Paper 2.1.6 127
Paper 2.1.7
Kopecna O., Kubickova S., Cernohorska H., Cabelova K., Váhala J., Martínková N.,
Rubes J. 2014. Tribe-specific satellite DNA in non-domestic Bovidae. Chromosome
Research 22: 277-291.
– 129 –
Tribe-specific satellite DNA in non-domestic Bovidae
Olga Kopecna & Svatava Kubickova &
Halina Cernohorska & Katerina Cabelova &
Jiri Vahala & Natalia Martinkova & Jiri Rubes
Received: 4 November 2013 /Revised: 2 January 2014 /Accepted: 10 January 2014
# Springer Science+Business Media Dordrecht 2014
Abstract Satellite sequences present in the centromeric
and pericentric regions of chromosomes represent useful
source of information. Changes in satellite DNA composition
may coincide with the speciation and serve as
valuable markers of phylogenetic relationships. Here,
we examined satellite DNA clones isolated by laser
microdissection of centromeric regions of 38 bovid
species and categorized them into three types. Sat I
sequences from members of Bovini/Tragelaphini/
Boselaphini are similar to the well-documented 1.715
sat I DNA family. Sat I DNA from Caprini/Alcelaphini/
Hippotragini/Reduncini/Aepycerotini/Cephalophini/
Antilopini/Neotragini/Oreotragini form the second
group homologous to the common 1.714 sat I DNA.
The analysis of sat II DNAs isolated in our study
confirmed conservativeness of these sequences within
Bovidae. Newly described centromeric clones from
Madoqua kirkii and Strepsiceros strepsiceros were similar
in length and repetitive tandem arrangement but
showed no similarity to any other satellite DNA in the
GenBank database. Phylogenetic analysis of sat I sequences
isolated in our study from 38 bovid species
enabled the description of relationships at the subfamily
and tribal levels. The maximum likelihood and
Bayesian inference analyses showed a basal position
of sequences from Oreotragini in the subfamily
Antilopinae. According to the Bayesian inference analysis
based on the indels in a partitioned mixed model,
Antilopinae satellite DNA split into two groups with
those from Neotragini as a basal tribe, followed by a
stepwise, successive branching of Cephalophini,
Aepycerotini and Antilopini sequences. In the second
group, Reduncini sequences were basal followed by
Caprini, Alcelaphini and Hippotragini.
Keywords Bovidae . satellite DNA . centromeric
repeats . phylogeny. FISH . laser microdissection
Abbreviations
BAC Bacterial artificial chromosome
BI Bayesian inference
BRU-
PCR
Basic repeat unit obtained by PCR
DOP-
PCR
Degenerate oligonucleotide primed polymerase
chain reaction
FISH Fluorescence in situ hybridization
LINE Long interspersed nuclear element
Chromosome Res
DOI 10.1007/s10577-014-9401-4
Responsible editor: Walther Traut
Electronic supplementary material The online version of this
article (doi:10.1007/s10577-014-9401-4) contains supplementary
material, which is available to authorized users.
O. Kopecna (*) :S. Kubickova :H. Cernohorska :
K. Cabelova :J. Rubes
Department of Genetics and Reproduction, Veterinary
Research Institute,
Hudcova 70, 621 00 Brno, Czech Republic
e-mail: kopecna@vri.cz
J. Vahala
Dvur Kralove Zoo,
544 01 Dvur Kralove n. L., Czech Republic
N. Martinkova
Institute of Biostatistics and Analyses, Masaryk University,
Brno, Czech Republic
130 Paper 2.1.7
LTR Long terminal repeat
MCMC Markov chains Monte Carlo
ML Maximum likelihood
NCBI National Center for Biotechnology
Information
NFA Numbers of autosomal arms
RFLP Restriction fragment length polymorphism
sat Satellite
SINE Short interspersed nuclear element
SNP Single nucleotide polymorphism
Introduction
The family Bovidae (order Artiodactyla, i.e. the eventoed
ungulates) comprises approximately 140 species
(Nowak 1999). Many of these are of economic importance
(e.g. cattle, sheep and goats), while several species
are endangered or threatened with extinction (http://
www.iucnredlist.org). The chromosomal complement
of domestic cattle (2n=60) is considered to reflect the
ancestral condition forming the basis from which all the
recent bovid karyotypes have been derived (Wurster and
Benirschke 1968; Buckland and Evans 1978; Gallagher
and Womack 1992). Although diploid chromosome
numbers vary from 30 to 60 among species, the autosomal
arm number remains relatively constant (NFA=56–
58). This reflects the dominance of Robertsonian fusions
(i.e. centric fusions of two acrocentric chromosomes)
in the karyotype evolution of Bovidae (reviewed
in Robinson and Ropiquet 2011).
A substantial proportion of the eukaryote genome
consists of constitutive heterochromatin, a genomic
fraction that includes LINEs, SINEs, satellite DNAs
and other repetitive sequences. Satellite sequences, organized
in tandem, usually reside in the centromeric and
pericentric regions of chromosomes (D’Aiuto et al.
1997). DNA in the centromeric region appears quite
complex (Lee et al. 1997), and no sequence conservation
can be found among a wide range of species. Only
closely related species share homologous satellite sequences
(Jobse et al. 1995), but even among the most
closely related species, satellite DNAs can differ in
nucleotide sequence, copy number and the composition
of satellite families (Ugarković and Plohl 2002). In the
process of satellite DNA evolution, newer satellite sequences
may be derived from preexisting satellite DNA
sequences that replace or coexist with the old satellites
via ‘three-phase’ evolutionary processes (Nijman and
Lenstra 2001). Changes in satellite DNA composition
may coincide with speciation and serve as valuable
markers of phylogenetic relationships (Jobse et al.
1995; Saffery et al. 1999; Chaves et al. 2000, 2005), or
can occur subsequent to speciation permitting the retrieval
of species-specific profiles (Ugarković and Plohl
2002).
Bovine satellites constitute ∼25 % of the total nuclear
DNA content (Vaiman et al. 1999) but are quite heterogeneous.
Macaya et al. (1978) isolated eight major
satellite DNA fractions from the bovine genome DNA,
some of which are related to each other. Moreover,
certain sequence motifs have been found in different
bovid tribes. For example, Chaves et al. (2000, 2005)
hybridized satellite I sequences from sheep (1.714) and
cattle (1.715) to the metaphase chromosomes of 15
species (representative of seven tribes). They document
the existence of two major satellite clades in Bovidae—
the Bovinae and a composite comprising the Caprinae/
Alcelaphinae/Hippotraginae. Their findings of evolutionarily
older variants of cattle satellite I being preserved
in sheep satellite I provide support for an evolutionary
mechanism as proposed by Nijman and Lenstra
(2001). Kopecna et al. (2012) analysed fluorescence in
situ hybridization (FISH) patterns of the bovine satellite
I probe (1.715 DNA family) in different bovid tribes.
Their results were in most cases in accordance with the
results of other authors with the exception of three
tribes. The authors concluded that the determination of
phylogenetic relationships based only on FISH analysis
could be misleading and that a more rigorous comparison
that entailed the analysis of the main motif sequence
of the repetitive DNA could be more informative. On
the other hand, the evolutionarily young satellite IV
sequence represents 4.3 % of the bovine genome and
has no resemblance to other satellite DNAs
(Skowronski et al. 1984). Two studies that relied on this
fraction have been reported—Modi et al. (2004) and
Adega et al. (2006). Both sets of authors analysed
satellite IV DNA by Southern blotting and FISH in
species of the Bovini and Tragelaphini. In contrast to
the work done by Modi et al. (2004), Adega et al. (2006)
found bovine satellite IV DNA in Tragelaphini. Modi
et al. (1996) analysed the presence of six highly repeated
DNA families isolated from cattle in a comparative
FISH study comprising 46 species of the order
Artiodactyla. These authors found two of the repeated
families (Pst family and the 1.715 satellite I family)
O. Kopecna et al.
Paper 2.1.7 131
present in all pecoran ruminants. Different restriction
patterns of the 1.715 sequence were observed in different
taxonomic families suggesting that independent concerted
evolution events have homogenized different
motifs in different lineages.
Most of the investigations referred above were conducted
using caprine or bovine satellite DNAs separated
from genomic DNA by density-gradient centrifugation.
This approach ignores their chromosomal localization.
In sharp contrast, the isolation of satellite DNAs by laser
microdissection of centromeric regions has the advantage
of chromosomal specificity (Pauciullo et al.
2006; Rubes et al. 2008; Louzada et al. 2008;
Cernohorska et al. 2011, 2012). Moreover, it represents
an easy and rapid method for the isolation of satellite
DNAs of a larger number of species required for extensive
studies.
This study extends our preliminary work based on
the FISH analysis and distribution of tribe-specific
repeats representing a pool of different satellite
DNA families that coexist in centromeres
(Kopecna et al. 2012). The aim of the study was
to isolate various types of satellite DNAs and determine
their co-localization in centromeric regions by FISH.
Sequences representing the main motifs of the tribespecific
sat I were used for hybridizations across various
bovid tribes and provided data for the phylogenetic
analysis.
Materials and methods
Chromosome preparation
Material from 38 species representing 12 bovid tribes
was used in this study. Blood samples from 33 taxa were
taken from captive-born animals held in the zoological
gardens in Dvur Kralove, Plzen, Liberec and Olomouc
(Czech Republic) and cultured according to the standard
protocols. Fibroblast cell cultures were established from
a Tetracerus quadricornis female specimen held in the
Menagerie du Jardin des Plantes, Paris. Fibroblast cell
cultures from Alcelaphus lichtensteinii and Oreotragus
oreotragus and cell suspensions of Neotragus
moschatus and Raphicerus sharpei were obtained from
Evolutionary Genomics Group, Department of Botany
and Zoology, University of Stellenbosch, South
Africa. Cell cultures were grown and harvested
using conventional procedures (Gallagher and
Womack 1992). Classification of Bovidae was carried
out following the newest taxonomy by Groves
and Grubb (2011). In case the name of species was
changed, we present the former name in brackets
(Wilson and Reeder 2005).
Isolation of tribe-specific satellite DNAs
The centromeric regions of autosomes of species
representing each bovid tribe were microdissected by
the MicroLaser system (Carl Zeiss MicroImaging
GmbH, Munich, Germany). The pooled DNA was amplified
by DOP-PCR without pretreatment as described
by Kubickova et al. (2002). Amplification products
were cloned into a pDrive vector (Qiagen, Hilden,
Germany). Species-specific clones were selected by
DOT BLOT hybridization (Pauciullo et al. 2006), fluorescently
labelled and checked for specificity by FISH.
Plasmid DNAwas subsequently isolated and sequenced.
Sequences comprised satellite DNA but were not long
enough to represent the whole basic repeat unit.
Therefore, primers amplifying the 5′- and 3′-flanking
regions were designed. A simplified version of the inverse
PCR was performed on isolated untreated genomic
DNA. The primers were chosen with the emphasis
on the distance between primers being as short as possible.
This permitted the retrieval of almost full-length
satellite DNA basic repeat units (Cernohorska et al.
2012). A simplified version of inverse PCR generates
amplification product from untreated genomic DNA
only in the case of repetitive sequences organized in
tandem. In the case of Hippotragini, Alcelaphini and
Tragelaphini, primers selected from one representative
species of each tribe were used for the inverse PCR in
other members of the same tribe. The amplification
products representing the basic repeat unit obtained by
PCR (BRU-PCR) were cloned and plasmids isolated.
We chose a subset of clones on their Hae III RFLP
patterns (most common and most dissimilar) for sequencing
using BigDye terminator chemistry on an
automated sequencer. The sequences were compared
to those in the GenBank database using BLASTN
searches. BLAST2 was used to assess sequence homologies.
DNA sequences were screened for interspersed
repeats by RepeatMasker. One species of each tribe was
selected, and its most frequent clone was labelled with
Orange-dUTP (Vysis, Richmond, UK) and used in the
FISH analysis.
Satellite DNA in Bovidae
132 Paper 2.1.7
Preparation of sat II probe
The sat II probe for FISH on Antilopinae species was
prepared from ovine genomic DNA according to the
sequence from the GenBank (NCBI accession number
AF245169). The primers were designed to amplify the
region 24–385 of this sequence and generated a 362-bp
PCR product. The sat II probe for FISH on Bovinae
species was constructed from bovine genomic DNA
using primers designed to amplify the region 241–584
of the bovine sequence (NCBI accession number
M36668). The PCR products were cloned into pDrive
Cloning Vectors (Qiagen) and recombinant plasmids
were labelled by Green-dUTP (Vysis, Richmond, UK)
using Nick Translation Reagent Kit (Vysis, Richmond,
UK). Labelled clones were hybridized in double-colour
FISH together with tribe-specific sat I to chromosome
preparations of the selected species. The bovine satellite
II primers were also used for the isolation and sequencing
of sat II DNA from Nyala angasii (formerly
Tragelaphus angasii).
Moreover, Nanger dama, Gazella leptoceros,
Antidorcas marsupialis and Antilope cervicapra sat II
sequences were obtained from each species genomic
DNA using microdissection, cloning and inverse PCR
(described above). These four Antilopini species were
chosen for the isolation of sat II DNA on the basis of
high intratribe variability of sat I sequences found in our
study. The isolated sat II sequences were compared with
sat II sequences available in the GenBank database
using BLASTN searches.
Fluorescence in situ hybridization
Metaphase spreads for FISH analysis were prepared
from lymphocytes or fibroblast cell cultures using
standard methanol/acetic acid (3:1) fixation.
Centromeric probes were hybridized as described
by Kopecna et al. (2012). Slides were washed in
0.4xSSC/0.3 % Igepal at 73 °C, counterstained and
mounted with 4',6-diamidino-2-phenylindole
(DAPI)/Antifade. The FISH preparations were
evaluated using Olympus BX 51 and BX 60 microscopes
that were equipped with the necessary
fluorescence filters and automated pad shifts. Good
quality mitoses were scanned by CCD camera and
evaluated by image analysis (ISIS 3, MetaSystems,
Altlussheim, Germany).
Phylogenetic analysis
DNA sequences were pretreated to include the tandem
repeat sequence. Two sets of analyses were performed.
First, one sequence representing the most frequent clone
per species that originated from this study was included.
The dataset contained 38 sequences. Second, intraspecific
variation was included with the most common and
the most dissimilar clones per species (1–5 sequences)
for a total of 100 sequences. Three species were added
from the GenBank database. The datasets were analysed
separately.
The sequences were aligned in Clustal Omega
(Sievers et al. 2011; Goujon et al. 2010) using two
combined iterations of the guide tree and the hidden
Markov model algorithms.
The optimal substitution model was estimated from
the Bayesian information criterion comparison of tree
likelihoods of alternative substitution models with
a fixed tree topology in jModeltest 2.1 (Darriba
et al. 2012; Guindon and Gascuel 2003). The rate
heterogeneity specific to alignment sites was
modelled with the Γ distribution to avoid co-estimation
of the α parameter of the Γ distribution and the proportion
of invariable sites.
The maximum likelihood (ML) phylogeny was constructed
in RAxML 7.4 (Stamatakis 2006) with rapid
bootstrapping, followed by the full-likelihood search.
Optimal number of bootstrap replicates was estimated
with the automatic majority-rule bootstopping procedure
(Pattengale et al. 2010).
Bayesian inference (BI) was used to reconstruct the
phylogenetic trees in MrBayes 3.2 (Ronquist et al.
2012). To ensure convergence, two runs were employed
with one cold and five heated Markov chains Monte
Carlo (MCMC) for 2 million generations, sampled every
1,000th step and MCMC temperature set to 0.09.
Discarded burn-in fraction represented the initial 700
trees. The hypothesis of monophyly of the subfamily
Antilopinae was tested on the first dataset with Bayes
factors from marginal likelihoods in Tracer 1.5
(Rambaut and Drummond 2007). The phylogenetic signal
of insertions and deletions (indels) was investigated
by coding the alignment gaps as binary characters and
analysed as a partitioned mixed model together with
nucleotide sequence information in BI.
Midpoint root was used. A node was considered
supported when its bootstrap support was ≥70 and
Bayesian posterior probability ≥0.95. Interpretation of
O. Kopecna et al.
Paper 2.1.7 133
nodes with lesser reliability was considered speculative,
and nodes with support <50 % were collapsed.
Results and discussion
Sequence analysis
Satellite DNAs were isolated from 38 species representative
of the 12 bovid tribes (9 tribes in Antilopinae and
3 in Bovinae). A minimum of two clones were sequenced
from each specimen, and the sequence data
representing the respective BRU-PCR were deposited
to GenBank under accession numbers KF787894–
KF787988. The Cephalophini, Neotragini and
Oreotragini were represented by a single species; at least
two species were examined for the remainder:
Alcelaphini (3), Antilopini (7), Boselaphini (2), Bovini
(3), Caprini (3), Hippotragini (6), Reduncini (4) and
Tragelaphini (6). The monotypic tribe Aepycerotini
was represented by Aepyceros melampus. Sequences
from three specimens were compared.
The isolated centromeric sequences were compared
to those available in GenBank and on the basis of
ascertained sequence homologies were categorized into
three types: sat I, sat II and unclassified.
1. Sat I sequences
All sat I sequences isolated in our study
(GenBank: KF787894–KF787981) showed sequence
similarity to the 1.714 or 1.715 sat DNA
family. Sequence homologies of sat I clones within
a tribe were ascertained using BLAST2. Sat I clones
showed sequence similarities of >80 % in most of the
tribes, but significant variations were found in
Reduncini and Antilopini (Table 1). Madoqua kirkii
(tribe Antilopini) showed low sequence similarity to
other Antilopini species (70 %) and is therefore
assessed separately below. The variability of the
BRU-PCR among species within a tribe exceeded
moderately the variability detected in the repeat units
of a single species or for that matter, in a single
specimen. Examples of intraspecific and intratribe
variation are given in Table 4 for Hippotragini and
Table 5 for Antilopini (see Supplementary), both
tribes with a high number of analysed species.
Hippotragini showed the highest average values of
sequence similarity contrary to Antilopini with the
lowest values. The most common BRU-PCR was
∼800 bp in length and was found in Caprini,
Alcelaphini, Hippotragini, Reduncini, Aepycerotini,
Cephalophini, Antilopini and Neotragini. By contrast,
that of the Bovini and Tragelaphini was
∼1,400 bp. BRU-PCR of unique length occurred in
Boselaphini (630 bp in T. quadricornis, 2,200 bp in
Boselaphus tragocamelus) and Oreotragini
(1,300 bp). These lengths dissimilar to the common
form were found in all clones isolated from the
distinct species. In addition to the ∼1,300 bp long
clone, representing the main satellite DNA of
O. oreotragus, a minor BRU-PCR of ∼800 bp and
properties of sat I was detected. The atypical
∼2,200 bp BRU-PCR found in B. tragocamelus is
the result of an interspersed block of LTR elements
(400 bp) that were revealed by Repeat Masker. The
sequences of the most frequently encountered BRUPCR
from a specific tribal representative were chosen
to determine intertribe orthologies (Table 3).
It is possible to group the sat I sequences obtained
in this investigation on the basis of their length and
mutual homology and the degree of similarity shared
with sequences of species archived in GenBank. The
first group included those sequences isolated from
members of the subfamily Bovinae: Bovini/
Tragelaphini (1,400 bp) and Boselaphini (630 bp in
T. quadricornis, 2,200 bp in B. tragocamelus). These
showed sequence similarity 67–95 % to sequences of
1.715 sat I subfamily that were documented by
Table 1 Sequence similarity of all sat I clones from pairwise
comparisons within a distinct tribe. Minimum, maximum and
average values
Sequence similarity (%)
Tribe Min Max Average
Caprini 82 95 87
Alcelaphini 91 98 94
Hippotragini 88 99 93
Reduncini 71 96 83
Aepycerotini 90 94 92
Cephalophini 83 96 84
Antilopini 67 96 76
Neotragini 87 90 88
Oreotragini 98 99 98
Tragelaphini 83 93 88
Bovini 81 97 84
Boselaphini 85 96 90
Satellite DNA in Bovidae
134 Paper 2.1.7
several authors (Modi et al. 1996; Chaves et al. 2005;
Kopecna et al. 2012). On the other hand, the satellite
sequences isolated from the subfamily Antilopinae:
Caprini/Alcelaphini/Hippotragini/Reduncini/
Aepycerotini/Cephalophini/Antilopini/Neotragini
are ∼800 bp long and together with Oreotragini
(1,300 bp) form the second group that are similar
(sequence similarity 68–96 %) to the common 1.714
sat I DNA (Jobse et al. 1995; Chaves et al. 2005).
The sat I sequences of both groups were characterized
by high similarity which was exclusive to members
within each group. The last finding corresponds
with the dissimilarity between 1.714 and 1.715 sat I.
We found no significant sequence similarity between
sequences of 1.714 and 1.715 sat I subfamilies available
in the GenBank using MEGABLAST, only less
strict BLASTN searches revealed a partial similarity.
The affiliation of our sat I sequences isolated
across the whole family Bovidae either to 1.174 or
1.715 DNA subfamilies supports the wellestablished
division of Bovidae into two subfamilies,
Bovinae and Antilopinae (Hassanin and Douzery
1999; Matthee and Davis 2001; Hassanin and
Douzery 2003; Ropiquet et al. 2009; Groves and
Grubb 2011).
2. Sat II sequences
Satellite DNA II was isolated from four species
of the Antilopini (N. dama, A. marsupialis,
G. leptoceros and A. cervicapra), and the sequence
data representing the respective BRU-PCR were
deposited to GenBank under accession numbers
KF787982–KF787985. BRU-PCR length was
∼700 bp. We designated these sequences as sat II
DNA on the basis of high sequence similarity (72–
76 %) with ovine sat II DNA published by
Buckland in 1985 (GenBank: X03117) and the absence
of any internal repetition. We supplemented
the published bovine sat II sequences with newly
isolated orthologous sequences from N. angasii
(GenBank: KF787986). This in turn demonstrated
high sequence similarity to species within the subfamily
Bovinae—82 % similarity with Bos taurus
and Bubalus bubalis. In sharp contrast to highly
conserved sat II DNA in Bovinae, sat II isolated
from the four Antilopini species included in our
investigation varied considerably (from 70 to
97 %) and showed sequence similarity to ovine sat
II DNA (72–76 %). A rather high level of sequence
similarity (73 %) exists between sat II DNAs of both
subfamilies that consequently provide little phylogenetic
signal. The absence of any internal repetition
and similar length of repeat units described in
Bovidae were found also in Cervidae (Qureshi and
Blake 1995), indicating the conservativeness of satellite
DNA II which, however, does not concern the
sequence composition. Comparison of sat II sequences
using BLASTN searches revealed existing
sequence similarity only in species within the same
family. We found only one case of interfamily sequence
similarity (70 %) between sheep (GenBank:
XO3117) and Odocoileus virginianus (GenBank:
U49916) sat II DNA. These findings support the
hypothesis that evolution of the satellite II DNA
family in Bovidae and Cervidae occurred mainly
by base substitution from an ancestral 700-bp tandem
repeat (Tanaka et al. 1999).
3. Unclassified sequences
The unclassified centromeric clones were isolated
from M. kirkii (GenBank: KF787987) and
Strepsiceros strepsiceros (formerly Tragelaphus
strepsiceros; GenBank: KF787988). These were
similar in length (2,900 bp in M. kirkii cf 2,500 bp
in S. strepsiceros), their repetitive tandem arrangement,
and they also contained interspersed blocks of
LTR elements (350 bp in M. kirkii and 220 bp in
S. strepsiceros). These new repeats showed 86 %
sequence similarity within these two species, but no
similarity to any other satellite DNA in the
GenBank database was found. BLASTN searches
suggest high sequence orthology (82 % in M. kirkii,
84 % in S. strepsiceros) only to the “Muntiacus
muntjak vaginalis BAC clone SatM5 centromeric
sequence” (GenBank: EU433566). Our unclassified
centromeric clones lie in the region, which does not
represent muntjac satellite DNA described on
SatM5 clone by Cheng et al. (2009) and are not
present as repetitions in muntjac. Whereas these
sequences probably remained preserved in this
shape in most of the bovid species, they were amplified
during the evolution to high copy number
forming the main centromeric DNA in M. kirkii and
substantial satellite DNA in S. strepsiceros.
FISH analysis
The most common sat I clone of each specific tribal
representative was labelled with Orange-dUTP (Vysis,
O. Kopecna et al.
Paper 2.1.7 135
Richmond, UK) and used as a probe for FISH analysis
of species within and across various tribes. Table 2
summarizes the hybridization patterns obtained from
tribe-specific sat I probes to different chromosomal
groups (i.e. acrocentric and biarmed autosomes, the X
and Y chromosomes) in 36 bovid species. The sat I
probe gave large hybridization signals on the acrocentric
autosomes, while the signals were much weaker in the
Table 2 FISH patterns with various tribe-specific sat I DNA probes to the different chromosome groups (acrocentric and biarmed
autosomes, X and Y chromosomes) in the 36 Bovidae species analysed
Tribe Species analysed (2n) Acrocentric autosomes Biarmed autosomes X Y
Caprini Ammotragus lervia (58) +/− − − −
Ovis aries (54) + +/− − −
Ovibos moschatus (48) + − + −
Alcelaphini Damaliscus phillipsi (38) + + + −
Connochaetes taurinus (58) + − + −
Alcelaphus lichtensteinii (40) + + + ♀
Hippotragini Hippotragus niger (60) + + −
Hippotragus equinus (60) + + −
Adax nasomaculatus (58) + + + −
Oryx dammah (58) + − + −
Oryx gazella (56) + + + −
Oryx leucoryx (58) + − + −
Reduncini Redunca fulvorufula (56) + + + −
Kobus leche (48) + + + −
Kobus megaceros (52) + +/− + −
Kobus ellipsiprymnus (50) + +/− +/− −
Aepycerotini Aepyceros melampus (60) + + −
Cephalophini Cephalophus natalensis (60) + − ♀
Antilopini Nanger dama (38) +/− +/− + +
Gazella leptoceros (32♀, 33♂) +/− +/− + +
Antilope cervicapra (32♀, 33♂) + +/− − +
Antidorcas marsupialis (56) + − + −
Madoqua kirkii (46) + − + −
Neotragini Neotragus moschatus (56) + +/− +/− −
Oreotragini Oreotragus oreotragus (60) + + ♀
Tragelaphini Tragelaphus spekii (30) + − + −
Strepsiceros strepsiceros (32♀, 31♂) + − + −
Ammelaphus imberbis (38) + +/− + +
Tragelaphus euryceros (34♀, 33♂) + − + +
Nyala angasii (56♀, 55♂) + + + −
Taurotragus oryx (32♀, 31♂) + + + −
Bovini Bubalus bubalis (50), river type + +/− + −
Bos bonasus (60) + − −
Syncerus caffer (52) +/− +/− + −
Boselaphini Tetracerus quadricornis (38) + + ♀
Boselaphus tragocamelus (46) + +/− + −
(+) positive FISH signals on all chromosomes; (+/−) positive FISH signals only on several chromosomes or weak signal; (−) apparent
absence of FISH signals; (♀) female, no Y chromosome; note 2n=60, species have no biarmed chromosomes
Satellite DNA in Bovidae
136 Paper 2.1.7
biarmed autosomes demonstrating a reduction in the
size of the block of repeats following centric fusions in
derived karyotypes (Modi et al. 1996). A reduction of
the heterochromatin is considered to reflect the age of
the fusion event—the more recent, the greater the quantity
of heterochromatin still remaining (Iannuzzi et al.
1987; Chaves et al. 2000; Di Meo et al. 2006; Rubes
et al. 2008; Kopecna et al. 2012). All acrocentric X
chromosomes were FISH positive—the exception being
the Caprini. Biarmed X chromosomes showed no sat I
hybridization. Strikingly, however, in N. dama,
Tragelaphus spekii, Ammelaphus imberbis (formerly
Tragelaphus imberbis) and B. tragocamelus (all species
with X autosome translocations), sat I was localized to
the boundary between the two chromosomal compartments.
The Y chromosomes failed to show any hybridization
to the sat I probe in the majority of species,
although there were exceptions within Antilopini and
Tragelaphini. The hybridization motifs of tribe-specific
satellite DNAs on autosomes and sex chromosomes
were thoroughly described in Antilopini (Cernohorska
et al. 2011) and Tragelaphini (Rubes et al. 2008). In the
study of Cabelova et al. (2012), distribution of sat I
sequences on Y chromosomes was well documented
in representatives of nine bovid tribes.
The intensity of hybridization signals of the sat II
probe was relatively weak when compared to those
obtained with the sat I probe—the exception being the
biarmed autosomes where intensities of both satellite
probes were similar.
The unclassified centromeric clones isolated from
M. kirkii and S. strepsiceros genomic DNA hybridized
reciprocally resulting in identical patterns due to the
high sequence orthology. Within Bovidae, these clones
gave positive FISH results with only Tragelaphini and
the two Antilopini species (M. kirkii and Antidorcas
marsupialis). The hybridization patterns obtained with
the unclassified clones were similar to sat I probes
(except for M. kirkii), although signal intensity differed.
Results of FISH using tribe-specific sat I probes on
representatives of the 12 tribes of the family Bovidae are
shown in Table 3. In our experiments, the sat I probe
specific for a particular tribe always resulted in positive
in situ hybridization in all tested species of the tribe in
question. The finding that Bovini and Tragelaphini species
have similar sat I structure and homology is confirmed
by positive FISH results. Similarly, positive
cross-hybridization reflects significant sequence similarity
of sat I of Bovini, Tragelaphini and Boselaphini.
As anticipated, no positive FISH signals were detected
when species of the subfamily Bovinae (Bovini/
Tragelaphini/Boselaphini) were hybridized to those of
the subfamily Antilopinae. The Antilopinae sat I
(∼800 bp in Caprini, Alcelaphini, Hippotragini,
Reduncini, Aepycerotini, Cephalophini, Antilopini and
Neotragini or 1,300 bp in Oreotragini) had substantial
sequence similarity as demonstrated in the intensity
levels of cross-hybridization signals when compared
among Antilopinae specimens. This is consistent with
our phylogenetic analysis of sat sequences that grouped
the taxa into two distinct evolutionary clades, the
Bovinae and Antilopinae (see below). However, within
subfamilies, no correlation exists between the intensity
of FISH signal detected and the degree of sequence
similarity. An obvious explanation may be that this
reflects differences in copy number of repeat units
in the centromeric regions of the compared species.
Moreover, our satellite DNA shows that intraspecific
variability ranging from 70 to 96 %
exists among repetitive units. Clone sequences represent
one of many variable repeat units. A probe with
high homology to one subunit can show lower homology
to other subunits, lowering hybridization efficiency
(Cabelova et al. 2012).
Co-localization of the sat I, sat II and unclassified repeat
families
Double-colour FISH with tribe-specific sat I probes and
bovine sat II probe was conducted on the chromosomes
of several species of the subfamily Bovinae. Sat I and sat
II sequences were localized to corresponding sites in all
species. Sat I was found at centromeric regions, while
sat II mapped to the outer boundaries of the centromeric
regions, partly overlapping with sat I. In our FISH
experiments on various Bovinae species, we obtained
similar results as Tanaka et al. (1999, 2000) did on
species of the genus Bubalus.
The co-localization of the tribe sat I probes and ovine
sat II probes was similarly examined in species of the
subfamily Antilopinae. Both sat I and II usually hybridized
in the centromeric regions of chromosomes in
Alcelaphini, Hippotragini, Reduncini, Aepycerotini,
Cephalophini, Neotragini and the two Antilopini species
(A. cervicapra, G. leptoceros). A unique hybridization
pattern of sat I FISH signals at the centromeric regions
and sat II at the pericentromeric regions was noted in
Caprini and A. marsupialis, respectively. D’Aiuto et al.
O. Kopecna et al.
Paper 2.1.7 137
Table3ResultsofFISHwithtribe-specificsatIprobesonvariousspeciesrepresenting12bovidtribesandintertribesequencesimilarity(percentagevalues)obtainedbycomparisonofthe
mostfrequentclonesinrepresentativesofdistincttribes
Tribe,species
analysed
Origin(tribe,species)ofcentromericDNAprobe
Caprini
A.lervia
Alcelaphini
D.phillipsi
Hippotragini
H.niger
RedunciniR.
fulvorufula
Aepycerotini
A.melampus
Cephalophini
C.natalensis
Antilopini
G.dama
NeotraginiA.
marsupialis
Oreotragini
M.kirkii
Tragelaphini
N.moschatus
BoviniO
.oreotragus
Boselaphini
T.spekei
B.
bubalis
T.
quadricornis
Caprini+++++++++++++to++000+++000
Ammotragus
lervia
10080837675766968666972nohom.nohom.nohom.
Alcelaphini+++++++++to+++0to+++00to+0to+0+000
Damaliscus
phillipsi
801008073737467/6869657270nohom.nohom.nohom.
Hippotragini+++++++++++++++to+++++to+++++to++++0to+0to+0to+++to++++000
Hippotragus
niger
838010074757670/6870677374/71nohom.nohom.nohom.
Reduncini+++to
++++
++++to++++++++to+++00to+0to++to0+000
Redunca
fulvorufula
767374100757666/6869657171nohom.nohom.nohom.
Aepycerotini++to
+++
+++++++++to++++++++++0+++++++0to+000
Aepyceros
melampus
757375751007768/7172677071nohom.nohom.nohom.
Cephalophini++++++++++++++0to+000to++000
Cephalophus
natalensis
76747676771006972677173/72nohom.nohom.nohom.
Antilopini0to+000+0++++++++++to+++00000
Nangerdama6967/6870/6866/6868/71691007770/676766/65nohom.nohom.nohom.
+to++0to+0to+0to+++0to+++++++++++0to+0000
Antidorcas
marsupialis
6869706972727710069nohom.69/67nohom.nohom.nohom.
+00to+00to+0++++++++++0000
Madoquakirkii66656765676770/6769100nohom.66/65nohom.nohom.nohom.
Neotragini+++++to+++++0+000++++0000
Neotragus
moschatus
69727371707167nohom.nohom.100nohom.nohom.nohom.nohom.
Oreotragini++++++++to+++++to+++++to+++++++++++++++000
Oreotragus
oreotragus
72074/71717173/7266/6569/6766/65nohom.100nohom.nohom.nohom.
Tragelaphini00000000000+++++to++++
Tragelaphus
spekii
nohom.nohom.nohom.nohom.nohom.nohom.nohom.nohom.nohom.nohom.nohom.1006868
Bovini00000000to+000+to++++++++
Bubalusbubalisnohom.nohom.nohom.nohom.nohom.nohom.nohom.nohom.nohom.nohom.nohom.6810074
Boselaphini00000000000++++++
Satellite DNA in Bovidae
138 Paper 2.1.7
(1997) found this specific pattern in one Caprini species
(Ovis aries). The converse was observed in N. dama
where sat II localized to the centromeric regions and sat
I to the pericentromeric regions, respectively. In gazelles,
the distribution of sat I and sat II DNAs was
described in more details in the study by Cernohorska
et al. (2012).
Three satellite DNA probes (sat I, sat II and unclassified)
hybridized to discrete chromosomal domains in
M. kirkii (Fig. 1). Signals of the unclassified centromeric
clone from M. kirkii were bordered by sat I in the
pericentromeric areas of chromosomes, and sat II was
placed on the opposite proximal side. FISH of the three
satellite DNA probes on O. oreotragus chromosomes
revealed that the 1,300 bp sat I hybridized to the centromeric
regions of all chromosomes, whereas the
800 bp sat I clone gave strong signals on the distal
portions of the centromeric regions of half of the chromosomes
(Fig. 2). The patterns found for ovine sat II
were similar to those detected with the 800 bp clone, but
these were present on all chromosomes and with lower
hybridization intensity. In our FISH experiments, the
hybridization signals appeared as one compact dot in
centromeres, while at the pericentromeric regions lying
on the separate chromatids, the signals split into two
discrete dots. This phenomenon was observed regardless
of the type of satellite DNA used.
Phylogenetic analysis
Species-level phylogeny of Bovidae sat I DNA
sequences
The alignment of the sequences from 38 species isolated
in this study was 1,854 bp long, where 47.3 % of the
alignment consisted of gaps. The ML analysis converged
after 550 bootstrap replicates, and bootstrapping
was stopped. Convergence of the runs of the BI analysis
was demonstrated with the final values of the average
standard deviation of split frequencies <0.01, the potential
scale reduction factor approaching 1.000 for all
model parameters, and the 95 % highest posterior density
interval of the tree length that included its ML
estimate. These results show that the analyses found
the global optimum in the tree space, and the trees were
reliable and suitable for interpretation.
Both ML and BI trees were in accord in resolving the
relationships between investigated sat I DNA sequences
(Fig. 3). Here, the sequences from subfamilies
Table3(continued)
Tribe,species
analysed
Origin(tribe,species)ofcentromericDNAprobe
Caprini
A.lervia
Alcelaphini
D.phillipsi
Hippotragini
H.niger
RedunciniR.
fulvorufula
Aepycerotini
A.melampus
Cephalophini
C.natalensis
Antilopini
G.dama
NeotraginiA.
marsupialis
Oreotragini
M.kirkii
Tragelaphini
N.moschatus
BoviniO
.oreotragus
Boselaphini
T.spekei
B.
bubalis
T.
quadricornis
Tetracerus
quadricornis
nohom.nohom.nohom.nohom.nohom.nohom.nohom.nohom.nohom.nohom.nohom.6874100
(+)positiveFISHsignals,numberof+correspondstotheintensityofsignals;(0)apparentabsenceofFISHsignals;(nohom.)nohomologyfoundusingBLAST2programme;(67/68)two
blockswithdifferentsequencesimilarityvaluesfoundinsequencesusingBLAST2programme
O. Kopecna et al.
Paper 2.1.7 139
Antilopinae and Bovinae were clearly separated with
midpoint rooting of the phylogenies. This is in agreement
with the current phylogeny of Bovidae (Decker
et al. 2009; Hassanin et al. 2012; Ropiquet et al. 2009),
indicating, at this level, congruence between bovid phylogeny
based on sat I DNA sequences and other
markers. Within the subfamilies, sequences from all
tribes with multiple sampled members were monophyletic,
albeit their relationships were often unresolved.
Tribes in Bovinae diverged fast after differentiation of
the subfamily (Ropiquet et al. 2009; Hassanin et al.
2012; Bibi 2013), and the sat I DNA sequences provided
no resolution between them. Tribal level structure in
Antilopinae based on our data consistently showed basal
position of Oreotragini sequences in the group, followed
by a rapid divergence of sequences representing other
tribes as shown in the Bayesian phylogeny. This shows
better resolution compared to the previous knowledge of
the group where in phylogenies reconstructed from other
genomic sequences Oreotragini showed variable positions
at deeper divergences of Antilopinae (Ropiquet
et al. 2009; Bibi 2013). Aepycerotini were unstable with
no supported affinity in our analyses. The phylogenetic
position of the sole representative of the tribe, Aepyceros
melampus, is notoriously difficult to ascertain from
DNA sequence data. It is either in an unsupported
Fig. 1 a Co-hybridization of sat I clone (red) and the unclassified
centromeric clone (green), both isolated from Madoqua kirkii to
metaphase chromosomes of M kirkii. b FISH with sat II probe
derived from Antidorcas marsupialis to chromosomes of M. kirkii.
Inset shows co-hybridization of sat I (red), sat II (yellow) and the
unclassified centromeric clone (green) to a chromosome of
M. kirkii. Chromosomes were counterstained with DAPI (blue).
Scale bar represents 5 μm
Fig. 2 a Double-colour FISH with the 1,300 bp sat I clone (red)
and the 800 bp sat I clone (green), both isolated from Oreotragus
oreotragus to metaphase chromosomes of O. oreotragus. b FISH
with ovine sat II to chromosomes of O. oreotragus. Chromosomes
were counterstained with DAPI (blue). Scale bar represents 5 μm
Satellite DNA in Bovidae
140 Paper 2.1.7
position (Decker et al. 2009; Ropiquet et al. 2009;
Hassanin et al. 2012) or it forms a sister relationship
with Neotragini (Bibi 2013; Hassanin et al. 2012). In
recent analyses, including this study, it branches close to
the base of the Antilopinae, indicating that it is a relict
from an early divergence of the group. Sat I DNA
sequences from members of Antilopini, Cephalophini
and Neotragini formed a group, wherein the ML analysis
distinguished a sister relationship of Cephalophini
and Neotragini sequences that was absent in the BI
phylogeny. A close relationship between Cephalophini
and Neotragini seems to be unresolved in mitochondrial
phylogenies (Hassanin et al. 2012) or Neotragini form a
sister relationship with Aepycerotini (Bibi 2013). This
conflict in placement of Neotragini between mitochondrial
and sat DNA phylogenies indicates the need for
thorough investigation of nuclear markers to ascertain
the origin of the group. Similarly, sequences belonging
to the tribe Reduncini formed a basal group with respect
to those from the remaining tribes in the ML tree only.
0.1
Gazella_leptoceros_1
Antidorcas_marsupialis_1
Alcelaphus_lichtensteinii_1
Redunca_fulvorufula_1
Oryx_dammah_1
Bubalus_bubalis_1
Raphicerus_sharpei_1
Syncerus_caffer_1
Oreotragus_oreotragus_1
Madoqua_kirkii_1
Tragelaphus_spekii_1
Bos_bonasus_1
Damaliscus_phillipsi_1
Kobus_leche_1
Nanger_dama_1
Cephalophus_natalensis_1
Ammelaphus_imberbis_1
Oryx_gazella_1
Hippotragus_niger_1
Antilope_cervicapra_1
Taurotragus_oryx_1
Kobus_megaceros_1
Oryx_leucoryx_1
Connochaetes_taurinus_1
Tragelaphus_euryceros_1
Ammotragus_lervia_1
Aepyceros_melampus_1
Eudorcas_thomsoni_1
Kobus_ellipsiprymnus_1
Ovibos_moschatus_1
Neotragus_moschatus_1
Nyala_angasii_1
Strepsiceros_strepsiceros_1
Tetracerus_quadricornis_1
Boselaphus_tragocamelus_1
Adax_nasomaculatus_1
Ovis_aries
Hippotragus_equinus_1
1/95
1/100
-/82
0.95//98
1/100
1/100
1/100
-/-
1/100
0.98/100
0.97/76
-/89
1/100
1/100
0.98/83
1/89
1/100
1/100
1/100
1/100
1/100
-/-
1/100
-/-
1/84
-/89
1/100
-/81
-/73
1/100
-/-
1/100
Boselaphini
Bovini
Tragelaphini
Oreotragini
Cephalophini
Neotragini
Antilopini
Aepycerotini
Reduncini
Caprini
Alcelaphini
Hippotragini
Fig. 3 Bayesian phylogenetic tree based on species representative
sat I DNA sequences isolated in this study. Numbers above
branches represent Bayesian posterior probability/maximum
likelihood bootstrap support. Nodes indicated with a dash were
unsupported in the analysis (see “Materials and methods” for
details)
O. Kopecna et al.
Paper 2.1.7 141
Both analyses showed that Hippotragini and
Alcelaphini sat I DNA sequences were closely related
and sister to Caprini (Fig. 3).
Sat I DNA phylogeny with intraspecific variation
in Bovidae
The alignment that included within-species variation in
the sat I DNA sequences consisted of 41 species, 100
sequences, and it was 2,058 bp long where 52.9 % of
sites were gaps. Majority-rule-based bootstopping criterion
indicated 400 replicates as sufficient for the estimation
of the ML phylogeny. The BI analysis with this data
did not converge with the given MCMC settings, and
the number of generations was increased to 3 million
with the proportional increase in burn-in fraction.
The subfamily- and the tribal-level topology of the
sequences on the trees that included intraspecific variation
in the sat I DNA (see Supplementary Figs. 4 and 5)
was comparable to that described above. The main
difference was that the sequences from Oreotragini species
were no longer basal to those from the Antilopinae.
In trees with intraspecific variation, the Antilopinae sat I
DNA sequences exhibited a rapid burst of differentiation
that affected representatives from all the tribes
except Caprini, Alcelaphini and Hippotragini that
formed a supported group.
Interestingly, throughout the tree, the alternative sequences
obtained from the same individuals did not
form sister relationships within several tribes.
Antilopini, Bovini, Caprini, Hippotragini and
Tragelaphini conspecific sequences generally did not
group together. This could be explained by parallel
differentiation of sat DNA at variable loci. The sequences
from the tribe Reduncini displayed a different
topology that leads to an alternative interpretation. Here,
two lineages formed, and they exhibited long branches
similar to those distinguishing sequences from some
other tribes. One of these lineages is likely driven by a
long indel. Analyses, which included phylogenetic information
from indels, placed this lineage at the base of
the whole tree in a sister relationship with sat I DNA
from Tetracerus quadricornis (see Supplementary
Fig. 6). While this is a computational artefact caused
by apparent lack of shared evolutionary history between
the sequences that share large indels, the case of
Reduncini sat I DNA demonstrated the putative heterogeneity
of the dataset. The sequences were identified
here based on similarity, which is consistent with the
evolutionary divergence only in some cases. In
Reduncini, the two lineages represent disparate evolutionary
histories of multiple repeat units.
Elsewhere in the Antilopinae tree, the phylogenetic
information from indels stabilized the relationships between
tribal sat I DNA sequences. The sequences at the
subfamily level split into two groups, and each group
had a distinct ladder-like topology, meaning that tribal
sat DNA branched from the common ancestral sequence
successively. In the first group, Neotragini sat I DNA
sequences were basal, followed by a ladder-like differentiation
of sequences from Cephalophini, Aepycerotini
and Antilopini. Similarly, in the second group, sat I
DNA sequences from Reduncini were basal, and then
those from Caprini, Alcelaphini and Hippotragini diverged
(see Supplementary Fig. 6). These relationships
were indicated albeit often unresolved in an otherwise
highly resolved phylogeny based upon 40,843 genomewide
SNP constructed by Decker et al. (2009).
The phylogenetic relationships of sat DNA sequences
within family Bovidae showed relationships
roughly consistent with the current taxonomy (Groves
and Grubb 2011). Where our phylogeny diverged from
expectations, the phenomena could have resulted from
rapid divergence in early evolution of a group, incomplete
lineage sorting or analyses of paralogous repeat
units.
In conclusion, we isolated and examined so far unpublished
sat I sequences from non-domestic species
representative of all tribes of the family Bovidae. All
of them showed high sequence similarity to 1.714 or
1.715 satellite I DNA and, with few exceptions, were of
similar length. Affiliation of our sat I sequences to ovine
1.714 or bovine 1.715 sat DNA corresponds with their
assignment to the respective subfamily (Antilopinae vs.
Bovinae). Our sat I sequences showed sequence variability
sufficient for providing successful phylogenetic
analysis. However, high intraspecific variability enabled
the differentiation only at the subfamily and tribal levels.
On the other hand, sat II DNA sequences isolated in our
study are more conservative and lack the variability
essential for phylogenetic studies. Apart from sat I and
sat II sequences, we found new, as yet undescribed
centromeric clones in M. kirkii and S. strepsiceros that
showed no similarity to any other known satellite DNA.
To sum up, the present investigation extends what is
known of the structure, distribution and phylogenetic
spread of satellite repeats in Bovidae and particularly
non-domestic species.
Satellite DNA in Bovidae
142 Paper 2.1.7
Acknowledgments We thank Terence J. Robinson and Anne
Ropiquet for kindly providing several samples and for useful
comments on a draft of the manuscript. This work was supported
by the Ministry of Agriculture of the Czech Republic (project
MZE 0002716202), the Grant Agency of the Czech Republic
(grant P506/10/0421) and by the project “CEITEC—Central European
Institute of Technology” (CZ.1.05/1.1.00/02.0068) from
European Regional Development Funds.
References
Adega F, Chaves R, Guedes-Pinto H, Heslop-Harrison JS (2006)
Physical organization of the 1.709 satellite IV DNA family in
Bovini and Tragelaphini tribes of the Bovidae: sequence and
chromosomal evolution. Cytogenet Genome Res 114:140–
146
Bibi F (2013) A multi-calibrated mitochondrial phylogeny of
extant Bovidae (Artiodactyla, Ruminantia) and the importance
of the fossil record to systematics. BMC Evol Biol 13:
166
Buckland RA (1985) Sequence and evolution of related bovine
and caprine satellite DNAs. J Mol Biol 186:25–30
Buckland RA, Evans HJ (1978) Cytogenetic aspects of phylogeny
in the Bovidae. I. G-banding. Cytogenet Cell Genet 21:42–63
Cabelova K, Kubickova S, Cernohorska H, Rubes J (2012) Malespecific
repeats in wild Bovidae. J Appl Genet 53:423–433
Cernohorska H, Kubickova S, Vahala J, Robinson TJ, Rubes J
(2011) Cytotypes of Kirk’s dik-dik (Madoqua kirkii,
Bovidae) show multiple tandem fusions. Cytogenet
Genome Res 132:255–263
Cernohorska H, Kubickova S, Vahala J, Rubes J (2012) Molecular
insights into X; BTA5 chromosome rearrangements in the
tribe Antilopini (Bovidae). Cytogenet Genome Res 136:188–
198
Chaves R, Guedes-Pinto H, Heslop-Harrison J, Schwarzacher T
(2000) The species and chromosomal distribution of the
centromeric alpha-satellite I sequence from sheep in the tribe
Caprini and other Bovidae. Cytogenet Cell Genet 91:62–66
Chaves R, Guedes-Pinto H, Heslop-Harrison JS (2005)
Phylogenetic relationships and the primitive X chromosome
inferred from chromosomal and satellite DNA analysis in
Bovidae. Proc Biol Sci 272:2009–2016
Cheng YM, Li TS, Hsieh LJ, Hsu PC, Li YC, Lin CC (2009)
Complex genomic organization of Indian muntjac centromeric
DNA. Chromosome Res 17:1051–1062
D’Aiuto L, Barsanti P, Mauro S, Cserpan I, Lanave C, Ciccarese S
(1997) Physical relationship between satellite I and II DNA
in centromeric regions of sheep chromosomes. Chromosome
Res 5:375–381
Darriba D, Taboada GL, Doallo R, Posada D (2012) jModelTest 2:
more models, new heuristics and parallel computing. Nat
Methods 9:772
Decker JE, Pires JC, Conant GC et al (2009) Resolving the
evolution of extant and extinct ruminants with highthroughput
phylogenomics. Proc Natl Acad Sci U S A 106:
18644–18649
Di Meo GP, Perucatti A, Chaves R, Adega F, De Lorenzi L,
Molteni L, De Giovanni A, Incarnato D, Guedes-Pinto H,
Eggen A, Iannuzzi L (2006) Cattle rob(1;29) originating from
complex chromosome rearrangements as revealed by both
banding and FISH-mapping techniques. Chromosome Res
14:649–655
Gallagher DS Jr, Womack JE (1992) Chromosome conservation in
the Bovidae. J Hered 83:287–298
Goujon M, McWilliam H, Li W, Valentin F, Squizzato S, Paern J,
Lopez R (2010) A new bioinformatics analysis tools framework
at EMBL-EBI. Nucleic Acids Res 38:W695–W699
Groves C, Grubb P (2011) Ungulate taxonomy. The Johns
Hopkins University Press, Baltimore
Guindon S, Gascuel O (2003) A simple, fast and accurate method
to estimate large phylogenies by maximum-likelihood. Syst
Biol 52:696–704
Hassanin A, Douzery EJ (1999) Evolutionary affinities of the
enigmatic saola (Pseudoryx nghetinhensis) in the context of
the molecular phylogeny of Bovidae. Proc Biol Sci 7:893–
900
Hassanin A, Douzery EJ (2003) Molecular and morphological
phylogenies of Ruminantia and the alternative position of
the Moschidae. Syst Biol 52:206–228
Hassanin A, Delsuc F, Ropiquet A, Hammer C, Jansen van Vuuren
B, Matthee C, Ruiz-Garcia M, Catzeflis F, Areskoug V,
Nguyen TT, Couloux A (2012) Pattern and timing of diversification
of Cetartiodactyla (Mammalia, Laurasiatheria), as
revealed by a comprehensive analysis of mitochondrial genomes.
C R Biol 335:32–50
Iannuzzi L, Di Berardino D, Gustavsson I, Ferrara L, Di Meo GP
(1987) Centromeric loss in translocations of centric fusion
type in cattle and water buffalo. Hereditas 106:73–81
Jobse C, Buntjer JB, Haagsma N, Breukelman HJ, Beintema JJ,
Lenstra JA (1995) Evolution and recombination of bovine
DNA repeats. J Mol Evol 41:277–283
Kopecna O, Kubickova S, Cernohorska H, Cabelova K, Vahala J,
Rubes J (2012) Isolation and comparison of tribe-specific
centromeric repeats within Bovidae. J Appl Genet 53:193–
202
Kubickova S, Cernohorska H, Musilova P, Rubes J (2002) The use
of laser microdissection for the preparation of chromosomespecific
painting probes in farm animals. Chromosome Res
10:571–577
Lee C, Wevrick R, Fisher RB, Ferguson-Smith MA, Lin CC
(1997) Human centromeric DNA. Hum Genet 100:291–304
Louzada S, Paço A, Kubickova S, Adega F, Guedes-Pinto H,
Rubes J, Chaves R (2008) Different evolutionary trails in
the related genomes Cricetus cricetus and Peromyscus
eremicus (Rodentia, Cricetidae) uncovered by orthologous
satellite DNA repositioning. Micron 39:1149–1155
Macaya G, Cortadas J, Bernardi G (1978) An analysis of the
bovine genome by density-gradient centrifugation.
Preparation of the dG+dC-rich DNA components. Eur J
Biochem 84:179–188
Matthee CA, Davis SK (2001) Molecular insights into the evolution
of the family Bovidae: a nuclear DNA perspective. Mol
Biol Evol 18:1220–1230
Modi WS, Gallagher DS, Womack JE (1996) Evolutionary histories
of highly repeated DNA families among the Artiodactyla
(Mammalia). J Mol Evol 42:337–349
Modi WS, Ivanov S, Gallagher DS (2004) Concerted evolution
and higher-order repeat structure of the 1.709 (satellite IV)
family in bovids. J Mol Evol 58:460–465
O. Kopecna et al.
Paper 2.1.7 143
Nijman IJ, Lenstra JA (2001) Mutation and recombination in cattle
satellite DNA: a feedback model for the evolution of satellite
DNA repeats. J Mol Evol 52:361–371
Nowak RM (1999) Order Artiodactyla. In: Walker’s Mammals of
the World, vol 2, 6th edn. The Johns Hopkins University
Press, Baltimore, 1051–1238
Pattengale ND, Alipour M, Bininda-Emonds ORP, Moret BME,
Stamatakis A (2010) How many bootstrap replicates are
necessary? J Comput Biol 17:337–354
Pauciullo A, Kubickova S, Cernohorska H, Petrova K, Di
Berardino D, Ramunno L, Rubes J (2006) Isolation and
physical localization of new chromosome-specific centromeric
repeats in farm animals. Vet Med 51
Qureshi SA, Blake RD (1995) Sequence characteristics of a cervid
DNA repeat family. J Mol Evol 40:400–404
Rambaut A, Drummond AJ (2007) Tracer v1.4, Available from
http://beast.bio.ed.ac.uk/Tracer
Robinson TJ, Ropiquet A (2011) Examination of hemiplasy, homoplasy
and phylogenetic discordance in chromosomal evolution
of the Bovidae. Syst Biol 60:439–450
Ronquist F, Teslenko M, van der Mark P, Ayres DL, Darling A,
Höhna S, Larget B, Liu L, Suchard MA, Huelsenbeck JP
(2012) MrBayes 3.2: efficient Bayesian phylogenetic inference
and model choice across a large model space. Syst Biol
61:539–542
Ropiquet A, Li B, Hassanin A (2009) SuperTRI: a new approach
based on branch support analyses of multiple independent
data sets for assessing reliability of phylogenetic inferences.
C R Biol 332:832–847
Rubes J, Kubickova S, Pagacova E, Cernohorska H, Di Berardino
D, Antoninova M, Vahala J, Robinson TJ (2008)
Phylogenomic study of spiral-horned antelope by crossspecies
chromosome painting. Chromosome Res 16:935–
947
Saffery R, Earle E, Irvine DV, Kalitsis P, Choo KH (1999)
Conservation of centromere protein in vertebrates.
Chromosome Res 7:261–265
Sievers F, Wilm A, Dineen DG, Gibson TJ, Karplus K, Li W,
Lopez R, McWilliam H, Remmert M, Söding J, Thompson
JD, Higgins D (2011) Fast, scalable generation of highquality
protein multiple sequence alignments using Clustal
Omega. Mol Syst Biol 7:539
Skowronski J, Plucienniczak A, Bednarek A, Jaworski J, Bovine
1.709 satellite (1984) Recombination hotspots and dispersed
repeated sequences. J Mol Biol 177:399–416
Stamatakis A (2006) RAxML-VI-HPC: maximum likelihoodbased
phylogenetic analyses with thousands of taxa and
mixed models. Bioinformatics 22:2688–2690
Tanaka K, Matsuda Y, Masangkay JS, Solis CD, Anunciado RVP,
Namikawa T (1999) Characterization and chromosomal distribution
of satellite DNA sequences of the water buffalo
(Bubalus bubalis). J Hered 90:418–422
Tanaka K, Matsuda Y, Masangkay JS, Solis CD, Anunciado RV,
Kuro-o M, Namikawa T (2000) Cytogenetic analysis of the
tamaraw (Bubalus mindorensis): a comparison of R-banded
karyotype and chromosomal distribution of centromeric satellite
DNAs, telomeric sequence, and 18S-28S rRNA genes
with domestic water buffaloes. J Hered 91:117–121
Ugarković D, Plohl M (2002) Variation in satellite DNA profilescauses
and effects. EMBO J 21:5955–5959
Vaiman D, Billault A, Tabet-Aoul K, Schibler L, Vilette D, OustryVaiman
A, Soravito C, Cribiu EP (1999) Construction and
characterization of a sheep BAC library of three genome
equivalents. Mamm Genome 10:585–587
Wilson DE, Reeder DM (2005) Mammal species of the world.
Johns Hopkins University Press, Baltimore
Wurster DH, Benirschke K (1968) Chromosome studies in the
superfamily Bovoidea. Chromosoma 25:152–171
Satellite DNA in Bovidae
144 Paper 2.1.7
Paper 2.1.8
Wallace I. S., Shakesby A. J., Hwang J. H., Choi W. G., Martínková N., Douglas A. E.,
Roberts D. M. 2012. Acyrthosiphon pisum AQP2: A multifunctional insect aquaglyceroporin.
BBA Biomembranes 1818: 627-635.
– 145 –
Acyrthosiphon pisum AQP2: A multifunctional insect aquaglyceroporin
Ian S. Wallace a,1
, Ally J. Shakesby b
, Jin Ha Hwang a
, Won Gyu Choi a
, Natália Martínková b,c
,
Angela E. Douglas b,d
, Daniel M. Roberts a,
⁎
a
Department of Biochemistry & Cellular, and Molecular Biology, The University of Tennessee, Knoxville, Knoxville, TN, 37996–0840, USA
b
Department of Biology, University of York, York, YO10 5DD, UK
c
Institute of Vertebrate Biology, Academy of Sciences of the Czech Republic, v.v.i., Květná 8, 603 65 Brno, Czech Republic
d
Department of Entomology, Comstock Hall, Cornell University, Ithaca, NY 14850, USA
a b s t r a c ta r t i c l e i n f o
Article history:
Received 31 August 2011
Received in revised form 19 November 2011
Accepted 28 November 2011
Available online 8 December 2011
Keywords:
Aphid
Aquaporins
Buchnera aphidicola
Osmoregulation
Polyols
Symbiosis
Annotation of the recently sequenced genome of the pea aphid (Acyrthosiphon pisum) identified a gene
ApAQP2 (ACYPI009194, Gene ID: 100168499) with homology to the Major Intrinsic Protein/aquaporin superfamily
of membrane channel proteins. Phylogenetic analysis suggests that ApAQP2 is a member of an insectspecific
clade of this superfamily. Homology model structures of ApAQP2 showed a novel array of amino acids
comprising the substrate selectivity-determining “aromatic/arginine” region of the putative transport pore.
Subsequent characterization of the transport properties of ApAQP2 upon expression in Xenopus oocytes
supports an unusual substrate selectivity profile. Water permeability analyses show that the ApAQP2 protein
exhibits a robust mercury-insensitive aquaporin activity. However unlike the water-specific ApAQP1
protein, ApAQP2 forms a multifunctional transport channel that shows a wide permeability profile to a
range of linear polyols, including the potentially biologically relevant substrates glycerol, mannitol and sorbitol.
Gene expression analysis indicates that ApAQP2 is highly expressed in the insect bacteriocytes (cells
bearing the symbiotic bacteria Buchnera) and the fat body. Overall the results demonstrate that ApAQP2 is
a novel insect aquaglyceroporin which may be involved in water and polyol transport in support of the
Buchnera symbiosis and aphid osmoregulation.
© 2011 Elsevier B.V. All rights reserved.
1. Introduction
Major Intrinsic Proteins (MIPs) are an ancient class of integral membrane
protein channels that facilitate the selective bidirectional transport
of water and uncharged solutes across biological membranes [1].
The aquaporins represent the best characterized transporters in this
family, and these proteins play diverse roles in physiology by mediating
the bulk movement of water driven by osmotic and pressure gradients
[1–3]. Members of the MIP superfamily share a signature topology and
pore architecture, referred to as the “hourglass fold”. This topology
consists of six transmembrane α-helices related by a pseudo twofold
symmetry [4]. Additionally, two highly conserved loops between
transmembrane α-helices 2 and 3, and transmembrane α-helices 5
and 6 form two smaller α-helical segments (the “NPA” boxes) which
fold back into the protein and pack with the six transmembrane αhelices,
forming a 7th pseudo transmembrane helix [4]. The pore formed
by the packing of these helices resembles an hourglass with the transport
selectivity determined by a narrow constriction, the “aromatic/
arginine” (ar/R) selectivity filter [4,5]. The ar/R selectivity filter is
defined by four residues (one each from transmembrane helices 2
and 5, and two from the interhelical loop containing the second
NPA box) that mediate transport selectivity based on solute size
and hydrophobicity. Vertebrate MIPs generally fall into two broad
transport classes, water-selective aquaporins and multifunctional
aquaglyceroporins that differ in the amino acid composition of the
ar/R [3].
While the transport behavior and physiology of vertebrate aquaporins
have been the subject of intensive study [3], the structure,
function, and physiology of invertebrate MIPs are less well studied
[2]. In the case of insects (e.g. bed bugs, mosquitoes, aphids, plant
hoppers) that ingest large volumes of vertebrate blood or plant sap,
aquaporins have been implicated in the bulk water movement across
various cellular membranes in the gut, participating in volume and
osmotic homeostasis, and fluid excretion [2,6–9]. For example, a
water-selective aquaporin, ApAQP1, expressed in both the stomach
and closely juxtaposed distal intestine of the plant-sap feeding pea
aphid (Acyrthosiphon pisum) has been proposed to serve an osmoregulatory
role in water cycling [9].
Analysis of the recently sequenced genome of A. pisum [10] revealed
the presence of a second MIP gene encoding an aquaporin-like protein
(ApAQP2) that is represented among ESTs obtained from both whole
aphids [11] and isolated bacteriocytes [12]. Bacteriocytes are specialized
aphid cells that house and maintain intracellular symbiotic γBiochimica
et Biophysica Acta 1818 (2012) 627–635
⁎ Corresponding author. Tel.: +1 865 974 4070; fax: +1 865 974 6306.
E-mail address: drobert2@utk.edu (D.M. Roberts).
1
Present address: The Energy Biosciences Institute, 130 Calvin lab MC5230, Berkeley,
CA, 94720, USA.
0005-2736/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.bbamem.2011.11.032
Contents lists available at SciVerse ScienceDirect
Biochimica et Biophysica Acta
journal homepage: www.elsevier.com/locate/bbamem
146 Paper 2.1.8
proteobacteria Buchnera aphidicola (reviewed in [13]). The symbiosis
has a nutritional basis: Buchnera provide the insect with essential
amino acids that are in short supply in the aphid diet of plant phloem
sap [14,15]. In this study, molecular modeling and functional analysis
reveal that ApAQP2 is a multifunctional aquaglyceroporin channel
that shows unique ar/R pore structure and exhibits permeability to
a wide range of physiologically relevant polyols and is localized to
both bacteriocytes and the fat body (an insect organ that functions
in energy storage and immunity). The potential significance of this
second aquaporin channel for the physiology of the insect, including
its symbiosis with intracellular bacteria, is discussed.
2. Materials and methods
2.1. ApAQP2 cDNA cloning
Acyrthosiphon pisum clone LL01 was maintained on broad bean
Vicia faba cv. The Sutton, grown at 20 °C with a 16 h light/8 h dark
cycle. Aphids were homogenized in ice-cold TRIzol reagent (Invitrogen)
and RNA was extracted following manufacturer's instructions
(Invitrogen). To remove contaminating DNA, the RNA was incubated
with RNAse-free DNAseI (Roche) for 30 min at 37 °C then at 75 °C for
15 min before purification with the RNAeasy minikit (Qiagen) using
the RNA cleanup protocol.
For ApAQP2 cDNA cloning, first strand cDNA was synthesized from
RNA using Superscript II reverse transcriptase (Invitrogen) following
the manufacturer's instructions. ApAQP2 was amplified in 1× PCR
buffer, 1 mM MgSO4, 0.2 mM of each dNTP, 1 U KOD Hot Start Polymerase
(Toyobo), 2 μL template (ca. 0.5 μg) and 1 μM gene-specific
primers (Supplemental Table 1) with the following conditions: 31 cycles
of 94 °C for 1 min, 35 °C for 1 min, and 72 °C for 2 min 30 s. PCR
products were purified by electrophoresis in a 1% (w/v) agarose gel
and were cloned into the pCR®-Blunt II-TOPO® plasmid vector (Invitrogen)
following the manufacturer's instructions. Cloned PCR products
were identified as ApAQP2 by sequencing using an Applied
Biosystems 3130 Genetic Analyzer (University of York Technology
Facility, York, U.K.).
2.2. Real time Q-PCR analysis of ApAQP2 transcript abundance
Embryo, fat body, gut, bacteriocyte and head tissues were dissected
from 7-day-old final instar larvae using fine pins and scissors, and
the RNA was extracted from these tissues and parallel samples of
whole aphids, as described earlier. cDNAs were generated from
isolated RNA samples using Superscript II reverse transcriptase
(Invitrogen) and p(dN6) random hexamers (Roche). Control reactions
without reverse transcriptase (−RTase) were included in all
assays. The abundance of ApAQP2 transcripts was determined by
Q-PCR with an ABI Prism 7900 Sequence Detection System (Applied
Biosystems), using the comparative Ct method. The reaction
mixtures contained 1× Power SYBR Green MasterMix (Applied
Biosystems), 0.1 μM gene-specific primers (Supplemental Table 1),
and 2 μL cDNA template in a final reaction volume of 25 μL. Thermal
cycling conditions were 2 min at 50 °C, 10 min at 95 °C followed by
40 cycles of 15 s at 95 °C and 1 min at 60 °C. The assays included a
dissociation curve (95 °C for 30 s followed by a 60–95 C temperature
ramp in increments of 0.5 °C for 1 min each), which confirmed
that all detectable fluorescence was derived from specific products.
All experimental samples were assayed in triplicate, with template
free and -RTase controls. The relative expression of ApAQP2 was
assessed by determining the threshold cycle (Ct), and each transcript
was normalized to the expression of the ribosomal protein
L32 transcript (RPL32) standardized to the expression level of the
transcript in the whole aphid body.
2.3. Expression and functional analyses of ApAQP2 in Xenopus laevis
oocytes
The ApAQP2 open reading frame (ORF) was amplified by PCR from
the pCR Blunt-II TOPO vector (Invitrogen) with ExTaq polymerase
and gene-specific primers (Supplemental Table 1) containing BamHI
restriction sites, and was cloned into the BglII site of the pXβG-FLAG
vector by the approach described in [16]. The final expression construct
contained ApAQP2 open reading frame translationally fused to
an N-terminal FLAG epitope tag.
Capped cRNA was produced from XbaI-linearized Flag-ApAQP2/pXβG
and soybean nodulin 26/pXβG constructs by using the AmpliCap-Max T3
High Yield Message Maker Kit (Epicentre Technologies). Stage V and VI
X. laevis oocytes were microinjected and cultured as described previously
[9,16]. Experimental oocytes were injected with 46 nL of 1 ng/
nL of each test cRNA, and negative control oocytes were injected
with an equivalent volume of sterile RNase-free water. Expression
of ApAQP2 and nodulin 26 proteins in oocytes was quantified by
Western blot analysis of oocyte lysates using an anti-FLAG epitope
monoclonal antibody (Stratagene) for ApAQP2-injected oocytes, and
an antibody against soybean nodulin 26 for nodulin 26-injected oocytes
as previously described [16–18]. The osmotic water permeability (Pf) of
Xenopus oocytes was determined as previously described [16,18]. Oocytes
were placed in a 15 °C bath solution containing diluted (30%)
frog Ringers solution [16–18], and serial images of the oocytes were
collected as they began to swell in response to hypoosmotic challenge.
The rate of oocyte swelling (d[V/V0]/dt) determined by video microscopy
was used to calculate the osmotic permeability coefficient (Pf) using
the following equation:
Pf ¼
V0=S0 d V=V0½ Š=dtð Þ
Sreal
Ssphere
 
Vw osmin−osmoutð Þ
where V0 is the initial volume and So is the initial surface area of the oocyte,
osmin is the osmolarity on the inside of the oocyte, osmout is the
osmolarity of the bathing solution, Vw is the partial molar volume of
water (18 cm3
/mol), Sreal is the actual oocyte surface area, and Ssphere
is the theoretical oocyte surface area assuming a perfect sphere. A
value of Sreal/Ssphere of 9 was used in all calculations to correct for the
increase in oocyte plasma membrane area resulting from the presence
of folds and microvilli [18].
The effect of mercurials on transport was investigated by preincubating
ApAQP2-injected oocytes in frog Ringers solution supplemented
with 1 mM HgCl2 for 5 min prior to assay. The viability of
control and ApAQP2-expressing oocytes was verified by measurement
of the resting oolemma membrane potentials as described in
[19]. All oocytes possessed inwardly negative resting potentials between
−22 and −25 mV, consistent with the reported membrane
potential values of viable oocytes [20].
Xenopus oocytes expressing ApAQP2 were assayed for solute permeability
by two different methods. Nonradiolabeled solute uptake assays
were performed by measuring solute-induced swelling under isoosmotic
conditions as previously described [16]. The oocytes were placed into
a bath solution containing isoosmotic frog Ringers solution with the
NaCl component replaced with 200 mM test solute. In this assay, solute
uptake results in an inwardly-directed osmotic gradient that leads to
water uptake and oocyte swelling. Solute uptake is reported as an
oocyte swelling rate [d(V/V0)/dt], determined by video microscopy.
Radiolabeled glycerol and mannitol uptake assays were performed
by a method modified from [16]. Twelve oocytes were incubated
in 150 μL of 5 mM HEPES NaOH pH 7.6, 86 mM NaCl, 2 mM KCl,
5 mM MgCl2, 0.6 mM CaCl2 supplemented with 20 mM
3
H-glycerol
(14 μCi/mL) for glycerol uptake assays, or with 20 mM
14
C-mannitol
(1.4 μCi/mL) for mannitol uptake assays. All radioisotopic assays were
conducted for 10 min at 25 °C. The oocytes were rinsed four times
628 I.S. Wallace et al. / Biochimica et Biophysica Acta 1818 (2012) 627–635
Paper 2.1.8 147
with 10 mL of radioisotope-free, ice-cold assay buffer, and separated
into three groups of four oocytes in scintillation vials. The oocytes
were lysed in 300 μL of 10% (w/v) SDS and isotope uptake was quantified
by as in [16].
2.4. Molecular modeling
A homology model of ApAQP2 was constructed using the Molecular
Operating Environment software (MOE2009.10; Chemical Computing
Group, Montreal, Canada). The crystal structure of human AQP4 (pdb
3GD8 [21]) was chosen as the structural template because of its high
resolution (1.8 Å) and the high level of amino acid sequence identity
(33%) to the ApAQP2 amino acid sequence within transmembrane
helical regions that form the transport pore. Similar results were
obtained with other aquaporin structural templates (data not shown).
ApAQP2 was aligned with the AQP4 template by using the MOE structural
alignment tool. Homology models were constructed by using the
homology modeling facility in MOE and the CHARMM27 force field.
An ensemble of ten possible structures for ApAQP2 was generated and
ranked by packing score. The model with the most favorable packing
score (3.0373) was energy-minimized using the CHARMM27 force
field and distance-dependent dielectric down to an energy gradient
of 10− 5
kcal/mol/Å2
. The stereochemical quality of the final model
was assessed by using Ramachandran plot analysis and the Protein
Report structural analysis function in the MOE Protein Structure
Evaluation package as previously described [22] to determine disallowed
bond angles, bond lengths, and side-chain rotamers. The final
ApAQP2 model possessed two residue outliers in the Ramachandran
plot. Both residues were found in the unconserved and unstructured
region of the C loop which does not contribute to the formation of the
aquaporin fold and transport pore. Pore diameters of homology
models were calculated using the HOLE2.0 program [23] of the energy
minimized homology model structure using the simple.rad van
der Waals radius file.
2.5. Phylogenetic analysis
The protein sequences were aligned using ClustalX [24], checked
manually in BioEdit 5.0.9 [25] and truncated to remove the predicted
N and C terminal cytoplasmic tails, which could not be aligned with
confidence. Bayesian inference (BI) and maximum likelihood (ML)
analyses were conducted, following selection of the WAG+Γ+I model
[26], using the optimal instantaneous rate matrix estimated in
ProtTest 1.4 [27]. The gamma shape parameter α and proportion
of invariable sites were estimated as part of the analysis. Eight rate
categories were used for rate heterogeneity estimation. The BI analysis
was run in MrBayes 3.1.2 [28]. Data were processed in two partitions.
In the first partition, gaps in the amino-acid alignment were
treated as missing data, and phylogenetic information present in
the gaps was then encoded as a binary dataset of the same length
and analysed as the second partition. Each Metropolis-coupled
Markov chain Monte Carlo (MCMC) run was three million generations
long, sampled every 1000th step, and the first 30% of sampled
trees were discarded as burn-in. The runs were considered converged
when average standard deviation of split frequencies was
less than 0.01 and potential scale reduction factor approached 1.0.
The observed 95% confidence interval of the tree length did not include
ML tree length estimate and branch lengths were subsequently
calculated using ML approach on fixed BI tree topology in
RAxML 7.2.6 [29]. The ML analysis was performed in PhyML 3.0
[30] with BIONJ selected as the starting tree and NNI tree topology
search algorithm. Alignment gaps were treated as missing data and
bootstrap support was estimated from 100 parametric replicates.
Midpoint rooting was used in the analyses. Posterior probabilities≥0.95
in BI analyses and bootstrap support≥70% in ML analyses were designated
significant. Analyses were conducted on computational clusters
at the Institute of Vertebrate Biology AS CR, Brno, Czech Republic and
Bioportal, University of Oslo, Norway.
3. Results
3.1. Sequence and phylogenetic analysis of pea aphid aquaporin genes
Detailed examination of the pea aphid genome [Acyr 2.0 primary
assembly (NCBI)] resulted in the identification of three gene loci
encoding proteins with aquaporin/MIP homology. The first locus
corresponds to the aquaporin gene ApAQP1 (ACYPI006387, Gene ID:
100165436 on scaffold NW_003383975). ApAQP1 expression generates
two isoforms by alternative splicing of the 5′ exons encoding proteins of
272 and 250 amino acids, respectively. The dominant isoform expressed
in the gut is isoform-1 (272 amino acids, NP_001139376.1), which has
been demonstrated previously to function as a Hg-sensitive waterselective
aquaporin involved in water cycling and osmoregulation [9].
A second locus (Gene ID: 100573582, scaffold NW_003384268) encodes
an aquaporin-like protein in which the amino-terminal 212
amino acids is identical to ApAQP1 isoform-1. However this protein
diverges from the canonical aquaporin sequence with the 60 carboxyl
terminal residues of ApAQP1 replaced by 41 amino acids that lack
aquaporin/MIP pore forming determinants (the second NPA box and
the LE1 and LE2 residues of the ar/R selectivity filter), suggesting that
it is incapable of forming a functional aquaporin/MIP channel.
Another aquaporin-like gene (ACYPI009194, Gene ID: 100168499,
scaffold NW_003383567), which we refer to as ApAQP2, encodes two
transcriptional variants that encode separate protein isoforms: isoform-
1 (308 amino acids [Fig. 1]) and isoform-2 (275 amino acids), which
lacks the 33 N-terminal amino acids in isoform-1. Published EST data
indicate that both isoforms of this gene are expressed in aphids [11].
This study focused on isoform-1, building on evidence that it is expressed
in isolated bacteriocytes [12]. It contains all of the sequence and structural
characteristics of the MIP channel family, including six predicted transmembrane
domains and two canonical NPA boxes, all components of
the prototypical “hourglass fold” of the aquaporin/MIP superfamily
(Fig. 1).
The phylogenetic position of the ApAQP2 protein sequence was
investigated by comparison with 60 animal MIP sequences using
Bayesian (BI) and maximum likelihood (MI) methods. The tree topologies
obtained with the two methods were very similar, assigning
all but one of the sequences to one of four significantly supported
clades, termed A–D (Fig. 2 shows the BI tree). Clade A contained a
number of functionally characterized water-specific aquaporins from
both mammals and insects. This clade contains the three insect subfamilies
of aquaporins, BIBs, DRIPs, and PRIPs [2]. The previously characterized
A. pisum aquaporin ApAQP1 [9] clusters within the PRIP subfamily.
Clade C represents a group of functionally characterized mammalianlike
aquaglyceroporins that includes AQP3, AQP7, and AQP9 [31–33].
Clade D MIPs are similar to human AQP11 and AQP12, which have
been termed “superaquaporins”, and yet have poorly defined transport
properties [34].
ApAQP2 was assigned to clade B, which differs from the other
clades in that all members are of insect origin. Since ApAQP2 belongs
to an insect-specific clade, distinct from other MIP family members
with canonical aquaporin or aquaglyceroporin activities, structural
and functional analyses were undertaken to model the putative transport
pore and determine its transport selectivity.
3.2. Molecular modeling of ApAQP2
The structural properties of the putative transport pore of ApAQP2
were investigated by molecular modeling by using the Molecular Operating
Environment (MOE) software. A homology model of the
ApAQP2 protein was constructed utilizing the 1.8 Å X-ray structure
(PDB ID 3GD8 [21]) of human AQP4 as a modeling template (Fig. 3).
629I.S. Wallace et al. / Biochimica et Biophysica Acta 1818 (2012) 627–635
148 Paper 2.1.8
The structural alignment of the ApAQP2 homology model with the
modeling template was excellent with an average carbon backbone
root mean square deviation (rmsd) of 0.762 Å. All elements of the
aquaporin fold were apparent which allowed modeling of the predicted
pore determinant regions.
The selectivity-determining ar/R region consists of a tetrad of
residues from transmembrane α-helices 2 (H2) and 5 (H5), as well
as two residues from the 2nd NPA helical loop E (LE1 and LE2) [22]. In
the human AQP4 X-ray structure, these residues are Phe 77, His 201,
Ala 210, and Arg 216, respectively. This collection of ar/R residues,
particularly a conserved His at the H5 position, is characteristic of
water-selective aquaporins. In the AQP4 ar/R structure, water is
bound by four hydrogen bonds collectively contributed by the sides
chains of His 201, Arg 216, as well as by the backbone carbonyl of
Ala210 [21]. These residues provide the necessary structural features
to facilitate rapid water transport while the small ar/R diameter
excludes larger solutes, resulting in a high conductance waterspecific
channel.
The ApAQP2 homology model suggests that the Phe residue at H2
and the Arg residue at LE2 are conserved with respect to human AQP4
(Fig. 3B). Comparison of the predicted ar/R residues of Clade B MIPs
suggests that these two aquaporin-like ar/R residues are conserved
in other subfamily members as well (Table 1). However, ApAQP2 contains
a Ser substitution at H5, which is a highly conserved His in AQP4
and other water-specific aquaporins. The presence of a Ser, Cys or Ala
in the place of His is a characteristic feature of the Clade B insect MIPs
(Table 1). At the LE1 position, ApAQP2 possesses an unusual Asn
residue, whereas water-selective aquaporins and other Clade B MIPs
possess a smaller amino acid, typically Ala, Gly or Cys (Table 1).
Modeling of the predicted ApAQP2 pore diameter using the HOLE
program suggests that the ar/R region still forms a size constriction,
albeit with a larger pore diameter compared to AQP4 (Fig. 3C). Overall,
the novel ar/R region of ApAQP2 is atypical of water-specific aquaporins
and mammalian aquaglyceroporins, suggesting that this
protein may exhibit different functional properties and substrate selectivity
compared to these well-characterized MIP channels.
3.3. Functional analysis of ApAQP2 transport
To investigate the functional properties of ApAQP2, the protein
was expressed in X. laevis oocytes and subjected to water and solute
transport analyses. Assays of ApAQP2-injected oocytes indicated
that expression of this channel increases the osmotic water permeability
(Pf) of the oocyte plasma membrane 15-fold, a clear indication
of aquaporin activity (Fig. 4A). Western blot analysis confirmed that
this increase in Pf corresponded with the expression of the ApAQP2
channel (Fig. 4B). The effect of HgCl2, a classical aquaporin inhibitor,
was also investigated. Unlike many aquaporins, including the waterselective
ApAQP1 [9], the Pf of ApAQP2 oocytes treated with 1 mM
HgCl2 was not significantly different (p=0.265) from untreated controls
(Fig. 4C).
Due to the unusual composition of the ApAQP2 ar/R region, a
broader range of test solutes was assayed to investigate the channel
substrate selectivity. Similar to the well-characterized soybean
Fig. 1. Amino acid sequence of ApAQP2: The full length cDNA sequence of ApAQP2 showing the open reading frame and deduced amino acid sequence. The predicted transmembrane
domains are highlighted in dashed boxes, the NPC and NPA motifs in solid boxes, and the residues of the proposed ar/R selectivity filter in solid circles. The position of a
potential polyadenylation site in the 3′-untranslated region is underlined.
630 I.S. Wallace et al. / Biochimica et Biophysica Acta 1818 (2012) 627–635
Paper 2.1.8 149
nodulin-26 aquaglyceroporin (GmNod26 in Table 1) [18,35], ApAQP2expressing
oocytes showed an increased permeability to
3
H-glycerol
(Fig. 5A). However, ApAQP2-expressing oocytes also exhibited a 10fold
increase in permeability to the larger polyol mannitol, while the
nodulin 26-expressing oocytes effectively excluded this solute (Fig. 5B).
Oocyte solute-dependent swelling assays were also performed to
test ApAQP2 permeability to a wider range of polyols and model substrates.
In agreement with the data from radioisotopic uptake assays,
ApAQP2-expressing oocytes were permeable to mannitol as well as
the epimeric six carbon alditols galacitol and sorbitol (Fig. 6). Additionally,
ApAQP2-expressing oocytes were permeable to four carbon
(erythritol) and five carbon (arabinitol, ribitol, xyltiol) alditols. The
measured permeabilities for each solute were very similar, and the
ApAQP2-expressing oocytes exhibited very little transport preference
for alditol chain length or hydroxyl group stereochemistry. ApAQP2expressing
oocytes were also capable of transporting the uncharged
test compounds formamide and urea, which are also common substrates
for many aquaglyceroporins. However, ApAQP2-expressing
oocytes were not permeable to the cyclic six-carbon polyol inositol.
Overall, these results indicate a unusual substrate profile for
ApAQP2 which forms a high-conductance, Hg-insensitive water channel
that is also permeated by a broad but defined range of linear
polyols.
3.4. Expression analysis of ApAQP2 in A. pisum
Previous studies have indicated that ApAQP2 is expressed in whole
aphid samples [11] and bacteriocytes [12]. To quantitate more precisely
the expression pattern of ApAQP2, transcript levels in dissected
pea aphid organs was investigated using real-time Q-PCR expression
0.5
human AQP2 (4502179)
Acyrthosiphon pisum ApAQP2 (ACYPI009194)
Nasonia vitripennis (156537707)
Tribolium castaneum (91092636)
Macaca mulatta (109097304)
Aedes aegypti (157104245)
Acyrthosiphon pisum ApAQP1 (ACYPI006387)
Dermacentor variabilis (114153752)
Pediculus humanus corporis (242019817)
Nasonia vitripennis (156540966)
Rhodnius prolixus (10862886)
Apis mellifera (110768288)
Bombyx mori (110171871)
Cicadella viridis (1279358)
Aedes aegypti (157107365)
Dipodomys merriami AQP4 (14578585)
Tribolium castaneum (91078198)
human AQP1 (37694062)
Pediculus humanus corporis (242019914)
Apis mellifera (66514115)
Ornithorhynchus anatinus (149411283)
Gallus gallus AQP4 (52345956)
Mus musculus AQP4 (148669655)
Anguilla anguilla AQP3 (28207812)
Nasonia vitripennis (156540968)
human AQP10 (55663099)
Drosophila melanogaster (45445703)
Pediculus humanus corporis (242007078)
Aedes aegypti (157136835)
human AQP6 (86792455)
Haematobia irritans (32469580)
human AQP0 (6912506)
Tribolium castaneum (91082857)
Lutzomyia longipalpis (157674501)
human AQP3 (4826645)
Drosophila melanogaster (24652751)
Aedes aegypti (157136837)
Osmerus mordax (108949240)
Strongylocentrotus purpuratus (115665420)
Drosophila melanogaster (20130305)
human AQP5 (4502183)
Aedes aegypti (157108306)
Drosophila melanogaster (19922828)
Bufo marinus (3821907)
Aedes aegypti (32469581)
human AQP7 (4502187)
human AQP9 (157266307)
human AQP11 (27370565)
Pediculus humanus corporis (242009228)
Nasonia vitripennis (156538323)
Mus musculus AQP12 (29243944)
human AQP8 (45446752)
Aedes aegypti (157108310)
Drosophila melanogaster (17136672)
human AQP4 (4502181)
Drosophila melanogaster (28573652)
Anopheles gambiae (158299454)
Pediculus humanus corporis (242018018)
Drosophila melanogaster (21627267)
Apis mellifera (66514152)
Pediculus humanus corporis (242015045)
clade D
clade C
clade B
clade A
1.0
1.0
1.0
1.0
0.99
1.0
1.01.0
1.0
1.0
1.0
1.0
1.0
1.0
0.95
1.0
0.99
1.0
1.0
1.0
1.0
1.0
1.0
1.0
0.97
0.99
0.97
0.99
0.98
1.0
1.0
0.99
0.99
1.0
Fig. 2. Phylogenetic analysis of aphid aquaporins: A phylogenetic tree of animal aquaporin sequences was generated by Bayesian inference based on the Blosum62 rate matrix with
gamma distribution of rate variation across sites. Numbers above edges denote statistically significant posterior probability (≥0.95). Genus and species names are indicated in
italics with NCBI protein sequence identifiers indicated in parentheses. The sequences comprising the four phylogenetically supported clades (A–D) are labeled by vertical bars
in the right margin. The sequence identifiers for ApAQP2, as well as the water-specific ApAQP1 [9] are shown in bold.
631I.S. Wallace et al. / Biochimica et Biophysica Acta 1818 (2012) 627–635
150 Paper 2.1.8
analysis (Fig. 7). ApAQP2 expression levels were significantly overrepresented
in the fat body (5.1-fold) and bacteriocytes (19.3-fold),
and under-represented in the aphid gut.
4. Discussion
This study has identified an aphid aquaglyceroporin gene, ApAQP2,
that is preferentially expressed in the bacteriocytes and fat body of
the pea aphid Acyrthosiphon pisum. ApAQP2 is a member of an insectspecific
clade (clade B MIPs) that is phylogenetically distinct from
other known animal aquaporin and aquaglyceroporin sequences. In
accordance with this observation, structural analysis of an ApAQP2
protein homology model indicates it contains novel substitutions
within the proposed selectivity-determining ar/R region. Similar to
water-selective aquaporins, Clade B MIPs contain a conserved Phe
at the H2 position and the invariant Arg residue at the LE2 position
of the ar/R region. However, the Clade B ar/R possesses small neutral
hydrophilic amino acids (Ser or Ala) at the H5 position that results in
an overall wider and more hydrophilic selectivity filter compared to
other insect and mammalian aquaporins (Table 1). ApAQP2 is unique
among other clade B MIPs and most aquaporins since it contains an
unusual Asn residue at the LE1 position. This positions an additional
side chain within the ar/R selectivity filter that could potentially create
new hydrogen bonding contacts for transported solutes.
Functional analysis of ApAQP2 in Xenopus oocytes indicates that the
protein possesses a multifunctional aquaglyceroporin activity capable
of transporting both water and neutral polyol substrates. ApAQP2 is
highly permeable to water, inducing a 15-fold increase in the Pf of the
oolemma upon expression in Xenopus oocytes. However, unlike most
aquaporins, including the water-selective aphid aquaporin ApAQP1
[9], ApAQP2 water permeability was not sensitive to the common aquaporin
channel blocker HgCl2. Although an unusual property, aquaporin
channels that are insensitive to mercury ions have been documented
[36,37]. The inability of Hg2+
to block these transporters could be the
result of the absence a Cys residue near the transport pore.
While ApAQP2 transports glycerol in a manner similar to established
animal (e.g., AQP3 [31]), higher plant (e.g., nodulin 26 [35]),
protist (e.g., PfAQP [38]) and bacterial (e.g., GlpF [39]) aquaglyceroporins,
it differs significantly from these proteins with regard to its
polyol transport behavior. For example, the well characterized
E. coli GlpF shows limited ability to transport longer polyol substrates.
E. coli GlpF transports the five carbon polyol ribitol at a rate
similar to the established biological substrate glycerol, but exhibits
strong size and stereoselectivity preferences for other polyols, with
epimers of ribitol (e.g., xylitol and arabinitol) showing low permeability,
and the six carbon compounds mannitol and sorbitol showing
complete impermeability [40]. Additionally, a multifunctional
aquaglyceroporin isolated from Plasmodium falciparum (PfAQP) exhibited
Fig. 3. Homology modeling analysis of ApAQP2 pore-forming residues: A homology model of ApAQP2 generated with the Molecular Operating Environment (MOE) software using
the experimental AQP4 structure (pdb 3GD8) as a structural template. A. Superimposition of the structural model of ApAQP2 (fushia) and the AQP4 structure (yellow). The water
molecules in the AQP4 pore are indicated by light blue spheres to show the position of the transport pore. The relative positions of the extracellular space (ex) and the cytosol (cyt)
are indicated. B. The residues comprising the ar/R region of ApAQP2 (fushia) are shown viewed perpendicular to the transport pore axis from the extracellular face of the protein
superimposed upon the corresponding residues in the AQP4 structure (yellow). Each ApAQP2 ar/R amino acid is labeled with the single letter amino acid designation as well as the
residue index. The ar/R positions proceed counterclockwise as follows: H2, H5, LE1, LE2. The position of the transport substrate (water) bound to the ar/R region is indicated by the
light blue sphere. C. Comparison of the ar/R regions of the AQP4 and ApAQP2 structures with HOLE. The calculated pore radius along the z-coordinate of the pore across the ar/R
region is shown.
Table 1
Conservation of pore-forming residues of insect clade B MIP sequences.
Proteina
Ar/R residuesb
NPA motifsc
H2 H5 LE1 LE2 NPA1 NPA2
Selective aquaporins
HsAQP4 F H A R NPA NPA
ApAQP1 F H A R NPA NPA
Aquaglyceroporins
GmNod26 W V A R NPA NPA
HsAQP3 F G Y R NPA NPA
PfAQP W G F R NLA NPS
Glyceroporins
EcGlpF W G F R NPA NPA
Clade B insect MIPs
ApAQP2 F S N R NPC NPA
AQP-Bom2 F S A R NPS NPA
AaMIP1 F A A R NPA NPA
AaMIP2 F S A R NPS NPA
PhcMIP1 F A C R NPA NPA
PhcMIP2 F A G R NPS NPA
NvMIP1 F A C R NPA NPV
NvMIP2 F C C R NPA NPA
TcMIP F S A R NPA NTA
AmMIP F A C R NPA NPA
DmCG4019 F A G R NPA NPA
DmCG17664 F S A R NPA NPV
RpMIP F S A R NPV NPV
a
The protein sequence name for each MIP isoform is indicated. Genus and species
designations are as follows: Hs, Homo sapiens; Gm, Glycine max; Pf, Plasmodium
falciparum; Ec, Escherichia coli; Ap, Acrythosiphon pisum; Rp, Rhodnius prolixus;
Aa, Aedes aegypti;Phc, Pediculus humanus corporis; Nv, Nassonia vitripennis; Tc, Tribolium
castaneum; Am, Apis melifera; Dm, Drosophila melanogaster. Accession number
for these proteins are as follows: HsAQP4 (NP_001641), ApAQP1 (NP_001139376),
GmNod26 (CAA28471), HsAQP3 (NP_004916), PfAQP (AAN35922); EcGlpF, (BAE77383);
ApAQP2 (NP_001139377), AQP-Bom2 (BAE97427), AaMIP1 (XP_001650170), AaMIP2
(XP_001650168), PhcMIP1 (XP_002428189), PhcMIP2 (XP_002430355), NvMIP1
(XP_001601253), NvMIP2 (XP_001601231), TcMIP (XP_970728), AmMIP (XP_624194),
DmCG4019 (NP_611813), DmCG17664 (NP_788433), and RpMIP (CAC13959).
b
The residues comprising the helix 2 (H2), helix 5 (H5), loop E position 1 (LE1) and loop
E position 2 (LE2) residues of the ar/R region are indicated for each sequence. A representative
sequence alignment showing the position of each ar/R residue is shown in Supplementary
Fig. 1.
c
The residues comprising the N-terminal (NPA1) and C-terminal (NPA2) NPA motifs are
shown for each sequence.
632 I.S. Wallace et al. / Biochimica et Biophysica Acta 1818 (2012) 627–635
Paper 2.1.8 151
robust permeability to five carbon polyols with a preference for arabitol
and xylitol. However, PfAQP was impermeable to the five carbon stereoisomer
ribitol [38] and was also impermeable to the larger six carbon
polyol mannitol [41]. In contrast, ApAQP2 shows strong permeability to
a wide variety of linear polyols (C3 to C6) with little apparent preference
for hydroxyl group stereochemistry, indicating a fundamental difference
between in the transport pores of these classes of aquaglyceroporins.
Vertebrate aquaglyceroporins are a structurally and functionally
conserved class of MIPs that cluster into a distinct phylogenetic
group (Clade C) with a defined ar/R selectivity filter (Table 1). Interestingly,
clade C MIPs are poorly represented in invertebrates whereas
a diverse array of Clade B MIPs is represented in all insect species
examined. This suggests that insects have acquired structurally and
functionally distinct aquaglyceroporins compared to their vertebrate
animal counterparts. It is unknown whether the broad array of
polyol transport properties exhibited by ApAQP2 is shared by other
clade B MIPs. Transport behavior of another clade B MIP, encoded
by AQP-Bom2 from the silkworm Bombyx mori, shows that it is able
to transport water, glycerol and urea [42], but its selectivity and ability
to transport larger polyols (e.g., mannitol or sorbitol) have not yet
been assessed.
The variety of substrates transported by ApAQP2 also raises the
question of the physiological function of water and polyol transport
through this multifunctional channel. Based on Q-PCR expression
analysis, ApAQP2 is highly expressed in aphid bacteriocytes as well
as in fat body cells. Both of these cell types are bathed in the hemolymph
and actively engage in exchange of metabolites and solutes
with this circulatory fluid. The osmotic pressure of the aphid hemolymph
is tightly regulated within narrow limits, thereby protecting
the fat body, bacteriocytes and other internal organs from the high
and variable osmotic pressure of plant phloem sap ingested by the insect
into the gut lumen [43,44]. One potential role for the robust
aquaporin activity of ApAQP2 could be the maintenance of osmotic
equilibrium between the hemolymph and bacteriocytes or fat body
cells through rapid adjustments in water content.
With respect to aphid physiology, the polyol transport activity of
ApAQP2 could also participate in carbohydrate metabolism and
transport related to stress biology as well as the symbiosis with
the Buchera bacteria. The accumulation of polyols including mannitol,
sorbitol and erythritol in insects [45–47] has been linked to the
ability of these compounds to protect proteins and membrane structures
from denaturation and disruption in response to osmotic and
temperature (cold or heat) stresses. In the case of aphids, both mannitol
(Aphis gossypii [48]) and sorbitol (Acyrthosiphon pisum [49]) accumulate
to high concentrations, particularly in response to heat
stress, and may play a role in thermotolerance [49,50]. Polyols have
been proposed to be synthesized from fructose precurors in the hemolymph
via a ketone reductase activity isolated from whole aphids
[50], although the discovery of a putative aldose reductase gene
(ACYPI005685) expressed in the bacteriocytes and fat body ([51],
Fig. 4. Water permeability analysis of Xenopus oocytes expressing ApAQP2: A. Xenopus oocytes were injected 46 ng of ApAQP2 cRNA (white bar) or soybean nodulin 26 cRNA
(hatched bar) and osmotic water permeability (Pf) was determined by the oocyte swelling assay. Oocytes injected with sterile RNase free-water were used as negative controls
(black bar). Error bars represent SEM (n=10 oocytes). B. Western blot analyses of oocyte lysates (20 μg total protein per lane). Top panel, Coomassie blue stained SDS-PAGE
gel, middle panel, Western blot with anti-FLAG antibody, bottom panel, Western blot with anti-nodulin 26 antibody [55]. The positions of the molecular weight markers are indicated
to the left of each panel. C. Negative control and ApAQP2-expressing oocytes were pre-treated with 1 mM HgCl2 (solid black bars) prior to the standard water permeability
assay described in Materials and methods. Untreated oocytes (hatched bars) were used as controls. Error bars represent SEM (n=6–9 oocytes).
Fig. 5. Analysis of glycerol and mannitol permeabilities of ApAQP2 by radioisotopic substrate uptake: Xenopus oocytes injected with ApAQP2 (white bars) or nodulin-26 (hatched
bars) cRNAs were assayed for uptake of A.
3
H-glycerol or B.
14
C-mannitol. Oocytes injected with sterile RNase-free water (black bars) were used as negative controls. Error bars represent
SEM (n=4). Asterisks over individual bars signify that transport rates are significantly higher than control oocytes (*pb0.05; **pb0.01; ***pb0.001).
633I.S. Wallace et al. / Biochimica et Biophysica Acta 1818 (2012) 627–635
152 Paper 2.1.8
Douglas, unpublished data) suggests additional potential sites of
polyol biosynthesis. Since polyols are largely impermeable to lipid
bilayers, the presence of a mannitol or a sorbitol facilitator such as
ApAQP2 may be essential for the transport and partitioning of
these critical protective polyols within bacteriocytes and fat body
cells.
With respect to the bacteriocytes, ApAQP2 may have a symbiotic
function. Most of the bacteriocyte cytoplasm is occupied by Buchnera
cells, each of which is enclosed individually in a membrane of aphid
origin (“the symbiosome membrane”). All metabolites required by
the Buchnera and products of Buchnera metabolism are, of necessity,
transported across the symbiosome membrane. The Buchnera genome
is much reduced (0.64 Mb) and of small gene content (620
genes) [52], but it has retained the genes for a GlpF-like glycerol
faciliator and a mannitol phosphotransferase system MtlAD, suggesting
conservation of a polyol transport system. The potential significance
of mannitol in the symbiosis is further underscored in silico
analysis of the Buchnera metabolic network which suggests that it
could represent an important carbon source for the endosymbiont
[53,54]. In addition, the ability to take up and partition polyols between
the aphid bacteriocyte cell cytosol, the internal space of the
symbiosome, and Buchnera cells, may aid in osmoregulation. Based
on the permeability profile of ApAQP2, and the observation that its
expression is enriched in bacteriocytes, a potential transport function
of these physiologically relevant polyols can be postulated. The
determination of the ApAQP2 subcellular localization in bacteriocytes,
including the determination of whether it is a symbiosome
membrane protein, will provide valuable insight into these potential
functions of this unusual aquaglyceroporin.
In summary, this study shows that the pea aphid aquaporin ApAQP2
is a multifunctional MIP that exhibits unprecedented substrate selectivity
to both water and a wide range of potentially relevant linear polyols.
Transport analyses suggest potential physiological functions of the protein
in osmoregulation, as well as carbon nutrition of the insect and its
symbiotic bacteria, and also provide a basis to investigate the transport
properties of related MIPs in other insects. As a final note, it is also important
to recognize that ApAQP2 is apparently expressed as two transcript
variants that encode different protein isoforms that have the
same core aquaporin-like pore structure but differ in the length of
their cytosolic amino terminal regions. Further investigation is needed
to determine the biological significance of these alternative splice variants
in Acyrthosiphon pisum physiology.
Supplementary materials related to this article can be found online
at doi:10.1016/j.bbamem.2011.11.032.
Acknowledgements
The authors acknowledge the assistance of Tian Li in the molecular
modeling analyses. This study is supported in part by the National
Science Foundation grant MCB-0618075, and a BBSRC Research Fellowship
(BB/C520898) and the Sarkaria Institute of Insect Physiology
and Toxicology.
References
[1] C. Hachez, F. Chaumont, Aquaporins: a family of highly regulated multifunctional
channels, Adv. Exp. Med. Biol. 679 (2010) 1–17.
[2] E.M. Campbell, A. Ball, S. Hoppler, A.S. Bowman, Invertebrate aquaporins: a review,
J. Comp. Physiol. B 178 (2008) 935–955.
[3] L.S. King, D. Kozono, P. Agre, From structure to disease: the evolving tale of aquaporin
biology, Nat. Rev. Mol. Cell Biol. 5 (2004) 687–698.
[4] T. Walz, Y. Fujiyoshi, A. Engel, The AQP structure and functional implications,
Handb. Exp. Pharmacol. 190 (2009) 31–56.
[5] B.L. de Groot, H. Grubmuller, Water permeation across biological membranes:
mechanism and dynamics of aquaporin-1 and GlpF, Science 294 (2001) 2353–2357.
[6] L. Duchesne, J.F. Hubert, J.M. Verbavatz, D. Thomas, P.V. Pietrantonio, Mosquito
(Aedes aegypti ) aquaporin, present in tracheolar cells, transports water, not glycerol,
and forms orthogonal arrays in Xenopus oocyte membranes, Eur. J. Biochem.
270 (2003) 422–429.
[7] M. Echevarria, R. Ramirez-Lorca, C.S. Hernandez, A. Gutierrez, S. Mendez-Ferrer, E.
Gonzalez, J.J. Toledo-Aral, A.A. Ilundain, G. Whittembury, Identification of a new
water channel (Rp-MIP) in the Malpighian tubules of the insect Rhodnius prolixus,
Pflugers Arch. 442 (2001) 27–34.
[8] F. Le Caherec, S. Deschamps, C. Delamarche, I. Pellerin, G. Bonnec, M.T. Guillam, D.
Thomas, J. Gouranton, J.F. Hubert, Molecular cloning and characterization of an
insect aquaporin functional comparison with aquaporin 1, Eur. J. Biochem. 241
(1996) 707–715.
[9] A.J. Shakesby, I.S. Wallace, H.V. Isaacs, J. Pritchard, D.M. Roberts, A.E. Douglas, A
water-specific aquaporin involved in aphid osmoregulation, Insect Biochem.
Mol. 39 (2009) 1–10.
[10] I.A.G. Consortium, Genome sequence of the pea aphid Acyrthosiphon pisum, PLoS
Biol. 8 (2010) e1000313.
Fig. 6. ApAQP2 permeability to a series of polyol transport substrates: Xenopus oocytes
were injected with ApAQP2 cRNA, and were assayed for permeability to various solutes
by video microscopy using an oocyte swelling assay (white bars). Oocytes injected
with sterile RNase-free water (black bars) served as a negative control. Assays were
conducted by immersion of oocytes in an isoosmotic Ringer's solution containing the
test solutes shown above each bar. The swelling rates were measured as a change in
oocyte volume [(dV/V0)/dt] due to the uptake of each test solute followed by the
osmotically-driven uptake of water. Error bars represent SEM (n=10–24 oocytes).
Fig. 7. Quantitative PCR analysis of ApAQP2 gene expression in aphid tissues: Expression
of the ApAQP2 gene was normalized to the RPL32 reference gene and standardized
to the ApAQP2 transcript expression level over the entire aphid body. Significant differences
in expression (*) were determined by multiple t-tests with Bonferroni correction
for five tests (critical probability α=0.01).
634 I.S. Wallace et al. / Biochimica et Biophysica Acta 1818 (2012) 627–635
Paper 2.1.8 153
[11] B. Sabater-Munoz, F. Legeai, C. Rispe, J. Bonhomme, P. Dearden, C. Dossat, A. Duclert,
J.P. Gauthier, D.G. Ducray, W. Hunter, P. Dang, S. Kambhampati, D. Martinez-Torres,
T. Cortes, A. Moya, A. Nakabachi, C. Philippe, N. Prunier-Leterme, Y. Rahbe, J.C. Simon,
D.L. Stern, P. Wincker, D. Tagu, Large-scale gene discovery in the pea aphid Acyrthosiphon
pisum (Hemiptera), Genome Biol. 7 (2006) R21.
[12] A. Nakabachi, S. Shigenobu, N. Sakazume, T. Shiraki, Y. Hayashizaki, P. Carninci,
H. Ishikawa, T. Kudo, T. Fukatsu, Transcriptome analysis of the aphid bacteriocyte,
the symbiotic host cell that harbors an endocellular mutualistic bacterium,
Buchnera, Proc. Natl. Acad. Sci. U. S. A. 102 (2005) 5477–5482.
[13] A.E. Douglas, Phloem-sap feeding by animals: problems and solutions, J. Exp. Bot.
57 (2006) 747–754.
[14] E.A. Gunduz, A.E. Douglas, Symbiotic bacteria enable insect to use a nutritionally
inadequate diet, P R Soc B 276 (2009) 987–991.
[15] N.A. Moran, J.P. McCutcheon, A. Nakabachi, Genomics and evolution of heritable
bacterial symbionts, Annu. Rev. Genet. 42 (2008) 165–190.
[16] I.S. Wallace, D.M. Roberts, Distinct transport selectivity of two structural subclasses
of the nodulin-like intrinsic protein family of plant aquaglyceroporin
channels, Biochemistry 44 (2005) 16826–16834.
[17] J.F. Guenther, N. Chanmanivone, M.P. Galetovic, I.S. Wallace, J.A. Cobb, D.M. Roberts,
Phosphorylation of soybean nodulin 26 on serine 262 enhances water permeability
and is regulated developmentally and by osmotic signals, Plant Cell 15 (2003) 981–991.
[18] R.L. Rivers, R.M. Dean, G. Chandy, J.E. Hall, D.M. Roberts, M.L. Zeidel, Functional
analysis of nodulin 26, an aquaporin in soybean root nodule symbiosomes, J.
Biol. Chem. 272 (1997) 16256–16261.
[19] E.D. Vincill, K. Szczyglowski, D.M. Roberts, GmN70 and LjN70. Anion transporters
of the symbiosome membrane of nodules with a transport preference for nitrate,
Plant Physiol. 137 (2005) 1435–1444.
[20] N. Dascal, The use of Xenopus oocytes for the study of ion channels, CRC Crit. Rev.
Biochem. 22 (1987) 317–387.
[21] J.D. Ho, R. Yeh, A. Sandstrom, I. Chorny, W.E. Harries, R.A. Robbins, L.J. Miercke,
R.M. Stroud, Crystal structure of human aquaporin 4 at 1.8 A and its mechanism
of conductance, Proc. Natl. Acad. Sci. U. S. A. 106 (2009) 7437–7442.
[22] I.S. Wallace, D.M. Roberts, Homology modeling of representative subfamilies
of Arabidopsis major intrinsic proteins. Classification based on the aromatic/arginine
selectivity filter, Plant Physiol. 135 (2004) 1059–1068.
[23] O.S. Smart, J.M. Goodfellow, B.A. Wallace, The pore dimensions of gramicidin A,
Biophys. J. 65 (1993) 2455–2460.
[24] J.D. Thompson, T.J. Gibson, F. Plewniak, F. Jeanmougin, D.G. Higgins, The
CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment
aided by quality analysis tools, Nucleic Acids Res. 25 (1997) 4876–4882.
[25] T.A. Hall, BioEdit: a user-friendly biological sequence alignment editor and analysis
program for Windows 95/98/Nt, Nucleic Acids Symp. Ser. 41 (1999) 95–98.
[26] S. Whelan, N. Goldman, A general empirical model of protein evolution derived
from multiple protein families using a maximum-likelihood approach, Mol. Biol.
Evol. 18 (2001) 691–699.
[27] F. Abascal, R. Zardoya, D. Posada, ProtTest: selection of best-fit models of protein
evolution, Bioinformatics 21 (2005) 2104–2105.
[28] F. Ronquist, J.P. Huelsenbeck, MrBayes 3: Bayesian phylogenetic inference under
mixed models, Bioinformatics 19 (2003) 1572–1574.
[29] A. Stamatakis, RAxML-VI-HPC: maximum likelihood-based phylogenetic analyses
with thousands of taxa and mixed models, Bioinformatics 22 (2006) 2688–2690.
[30] S. Guindon, O. Gascuel, A simple, fast, and accurate algorithm to estimate large
phylogenies by maximum likelihood, Syst. Biol. 52 (2003) 696–704.
[31] M. Echevarria, E.E. Windhager, G. Frindt, Selectivity of the renal collecting duct
water channel aquaporin-3, J. Biol. Chem. 271 (1996) 25079–25082.
[32] K. Ishibashi, M. Kuwahara, Y. Gu, Y. Tanaka, F. Marumo, S. Sasaki, Cloning and
functional expression of a new aquaporin (AQP9) abundantly expressed in the
peripheral leukocytes permeable to water and urea, but not to glycerol, Biochem.
Biophys. Res. Commun. 244 (1998) 268–274.
[33] K. Ishibashi, M. Kuwahara, Y. Kageyama, A. Tohsaka, F. Marumo, S. Sasaki, Cloning
and functional expression of a second new aquaporin abundantly expressed in
testis, Biochem. Biophys. Res. Commun. 237 (1997) 714–718.
[34] K. Ishibashi, S. Hara, S. Kondo, Aquaporin water channels in mammals, Clin. Exp.
Nephrol. 13 (2009) 107–117.
[35] R.M. Dean, R.L. Rivers, M.L. Zeidel, D.M. Roberts, Purification and functional reconstitution
of soybean nodulin 26. An aquaporin with water and glycerol transport
properties, Biochemistry 38 (1999) 347–353.
[36] M.J. Daniels, T.E. Mirkov, M.J. Chrispeels, The plasma membrane of Arabidopsis
thaliana contains a mercury-insensitive aquaporin that is a homolog of the tonoplast
water channel protein TIP, Plant Physiol. 106 (1994) 1325–1333.
[37] H. Hasegawa, T. Ma, W. Skach, M.A. Matthay, A.S. Verkman, Molecular cloning of a
mercurial-insensitive water channel expressed in selected water-transporting
tissues, J. Biol. Chem. 269 (1994) 5497–5500.
[38] E. Beitz, S. Pavlovic-Djuranovic, M. Yasui, P. Agre, J.E. Schultz, Molecular dissection
of water and glycerol permeability of the aquaglyceroporin from Plasmodium
falciparum by mutational analysis, Proc. Natl. Acad. Sci. U. S. A. 101 (2004)
1153–1158.
[39] C. Maurel, J. Reizer, J.I. Schroeder, M.J. Chrispeels, M.H. Saier Jr., Functional characterization
of the Escherichia coli glycerol facilitator, GlpF, in Xenopus oocytes, J.
Biol. Chem. 269 (1994) 11869–11872.
[40] D. Fu, A. Libson, L.J. Miercke, C. Weitzman, P. Nollert, J. Krucinski, R.M. Stroud,
Structure of a glycerol-conducting channel and the basis for its selectivity,
Science 290 (2000) 481–486.
[41] T. Zeuthen, B. Wu, S. Pavlovic-Djuranovic, L.M. Holm, N.L. Uzcategui, M. Duszenko,
J.F. Kun, J.E. Schultz, E. Beitz, Ammonia permeability of the aquaglyceroporins from
Plasmodium falciparum, Toxoplasma gondii and Trypansoma brucei, Mol. Microbiol.
61 (2006) 1598–1608.
[42] N. Kataoka, S. Miyake, M. Azuma, Aquaporin and aquaglyceroporin in silkworms,
differently expressed in the hindgut and midgut of Bombyx mori, Insect Mol. Biol.
18 (2009) 303–314.
[43] A.E. Douglas, The nutritional physiology of aphids, Adv. Insect. Physiol. 31 (2003)
73–140.
[44] T.L. Wilkinson, D.A. Ashford, J. Pritchard, A.E. Douglas, Honeydew sugars and osmoregulation
in the pea aphid Acyrthosiphon pisum, J. Exp. Biol. 200 (1997)
2137–2143.
[45] L. Lalouette, V. Kostal, H. Colinet, D. Gagneul, D. Renault, Cold exposure and associated
metabolic changes in adult tropical beetles exposed to fluctuating thermal
regimes, FEBS J. 274 (2007) 1759–1767.
[46] M.R. Michaud, J.B. Benoit, G. Lopez-Martinez, M.A. Elnitsky, R.E. Lee, D.L. Denlinger,
Metabolomics reveals unique and shared metabolic changes in response to heat
shock, freezing and desiccation in the Antarctic midge, Belgica antarctica, J. Insect
Physiol. 54 (2008) 645–655.
[47] D. Doucet, V.K. Walker, W. Qin, The bugs that came in from the cold: molecular adaptations
to low temperatures in insects, Cell. Mol. Life Sci. 66 (2009) 1404–1418.
[48] D.L. Hendrix, M.E. Salvucci, Polyol metabolism in homopterans at high temperatures:
accumulation of mannitol in aphids (Aphididae: Homoptera) and sorbitol
in whiteflies (Aleyrodidae: Homoptera), Comp. Biochem. Phys. A 120 (1998)
487–494.
[49] G. Burke, O. Fiehn, N. Moran, Effects of facultative symbionts and heat stress on
the metabolome of pea aphids, ISME J. 4 (2010) 242–252.
[50] M.E. Salvucci, G.R. Wolfe, D.L. Hendrix, Purification and properties of an unusual
NADPH-dependent ketose reductase from the silverleaf whitefly, Insect Biochem.
Mol. 28 (1998) 357–363.
[51] A. Poliakov, C.W. Russell, L. Ponnala, H.J. Hoops, Q. Sun, A.E. Douglas, K.J. van Wijk,
Large-scale label-free quantitative proteomics of the pea aphid-Buchnera symbiosis,
Mol. Cell. Proteomics 10 (2011) M110.007039.
[52] S. Shigenobu, H. Watanabe, M. Hattori, Y. Sakaki, H. Ishikawa, Genome sequence
of the endocellular bacterial symbiont of aphids Buchnera sp APS, Nature 407
(2000) 81–86.
[53] G.H. Thomas, J. Zucker, S.J. Macdonald, A. Sorokin, I. Goryanin, A.E. Douglas, A
fragile metabolic network adapted for cooperation in the symbiotic bacterium
Buchnera aphidicola, BMC Syst. Biol. 3 (2009).
[54] S.J. Macdonald, G.H. Thomas, A.E. Douglas, Genetic and metabolic determinants of
nutritional phenotype in an insect-bacterial symbiosis, Mol. Ecol. 20 (2011)
2073–2084.
[55] Y. Zhang, D.M. Roberts, Expression of soybean nodulin 26 in transgenic tobacco.
targeting to the vacuolar membrane and effects on floral and seed development,
Mol. Biol. Cell 6 (1995) 109–117.
635I.S. Wallace et al. / Biochimica et Biophysica Acta 1818 (2012) 627–635
154 Paper 2.1.8
Paper 2.2.1
Martínková N., McDonald R. A., Searle J. B. 2007. Stoats (Mustela erminea) provide
evidence of natural overland colonisation of Ireland. Proceedings of the Royal Society
B-Biological Series 274: 1387-1393.
– 155 –
Stoats (Mustela erminea) provide evidence
of natural overland colonization of Ireland
Nata´lia Martı´nkova´1,2
, Robbie A. McDonald3
and Jeremy B. Searle1,*
1
Department of Biology (area 2), University of York, PO Box 373, York YO10 5YW, UK
2
Institute of Vertebrate Biology, Academy of Science of the Czech Republic, Studenec 122,
675 02 Konesˇı´n, Czech Republic
3
Quercus, School of Biological Sciences, Queen’s University Belfast, 97 Lisburn Road, Belfast BT9 7BL, UK
The current Irish biota has controversial origins. Ireland was largely covered by ice at the Last Glacial
Maximum (LGM) and may not have had land connections to continental Europe and Britain thereafter.
Given the potential difficulty for terrestrial species to colonize Ireland except by human introduction, we
investigated the stoat (Mustela erminea) as a possible cold-tolerant model species for natural colonization of
Ireland at the LGM itself. The stoat currently lives in Ireland and Britain and across much of the Holarctic
region including the high Arctic. We studied mitochondrial DNA variation (1771 bp) over the whole
geographical range of the stoat (186 individuals and 142 localities), but with particular emphasis on the
British Isles and continental Europe. Irish stoats showed considerably greater nucleotide and haplotype
diversity than those in Britain. Bayesian dating is consistent with an LGM colonization of Ireland and
suggests that Britain was colonized later. This later colonization probably reflects a replacement event,
which can explain why Irish and British stoats belong to different mitochondrial lineages as well as different
morphologically defined subspecies. The molecular data strongly indicate that stoats colonized Ireland
naturally and that their genetic variability reflects accumulation of mutations during a population
expansion on the island.
Keywords: cytochrome b; D-loop; last glacial maximum; mitochondrial DNA; phylogeography
1. INTRODUCTION
Although Britain and Ireland are neighbouring islands, the
difference in their current biota is astonishing. Britain has
a fauna and flora that is somewhat restricted but broadly
similar to the nearby areas of continental Europe. By
contrast, Ireland is very species poor, but has some
unexpected forms among those present. Most noteworthy
are those that otherwise occur in southwest Europe,
collectively known as the ‘Lusitanian element’, including
species such as the Kerry slug (Geomalacus maculosus) and
Mackay’s Heath (Erica mackaiana; Corbet 1961). Moore
(1987) articulated these unusual features of the biogeography
of Ireland as ‘the Irish question’, recognizing that
there is a need to understand how the Irish fauna and flora
could have developed in this way.
The unusual nature of the Irish biota, and particularly
its distinctiveness from the British, is particularly seen
clearly in mammals (Yalden 1999). For instance, in
mainland Britain, non-commensal small mammals are
represented by three species of vole, three species of shrew,
three species of mouse and two small carnivores, all of
which apparently colonized naturally at around the end of
the last glaciation. Only four of these eleven species are
found in Ireland, three with a long-standing presence (the
pygmy shrew (Sorex minutus), the wood mouse (Apodemus
sylvaticus) and the stoat (Mustela erminea)) and the fourth
(the bank vole; Clethrionomys glareolus) is a twentieth
century introduction. Furthermore, phylogeographic
studies have shown that the pygmy shrews in Ireland
have greater molecular similarity with populations in
northern Iberia than those in Britain (Mascheretti et al.
2003). A comparable result has been obtained with the
pine marten (Martes martes; Davison et al. 2001),
providing further examples of the Lusitanian element, as
applied to genetic forms within species of mammal.
Mountain hares (Lepus timidus) in Ireland also have
more molecular similarity to populations in continental
Europe than to those in Britain (Hamill et al. 2006).
Few of the species currently present in Britain and
Ireland would have been present at the Last Glacial
Maximum (LGM; 19–23 kyr ago), when both landmasses
were predominantly covered by ice (Andersen & Borns
1997; Mix et al. 2001), so colonization has been largely
since that time. This colonization to Britain was over a
land bridge to continental Europe which persisted until
7500 years ago at the very latest (Yalden 1999). The
restricted fauna of Ireland clearly suggests that there was
not such a long-lasting land connection between Ireland
and either Britain and/or continental Europe. Indeed,
while there would have been a land connection between
Ireland, Britain and continental Europe at the LGM
(owing to lowered sea levels; Andersen & Borns 1997),
there are doubts whether Ireland was connected to either
landmass at all thereafter (Lambeck & Purcell 2001). If
there was no land bridge, that would explain the restricted
fauna and flora in Ireland (Stuart & Van WijngaardenBakker
1985). It would also suggest that the organisms
which do live there have, to a large extent, been brought by
Proc. R. Soc. B (2007) 274, 1387–1393
doi:10.1098/rspb.2007.0334
Published online 3 April 2007
Electronic supplementary material is available at http://dx.doi.org/10.
1098/rspb.2007.0334 or via http://www.journals.royalsoc.ac.uk.
* Author for correspondence (jbs3@york.ac.uk).
Received 22 December 2006
Accepted 9 March 2007
1387 This journal is q 2007 The Royal Society
156 Paper 2.2.1
humans who first appeared in Ireland in the Mesolithic
(Woodman et al. 1997). The Lusitanian element would
therefore suggest a particular importance of cultural links
in early human history between Ireland and southwest
Europe in introducing species to Ireland (Corbet 1961).
The Irish question can be fully answered only by the
consideration of the colonization history of a wide range of
species of animals and plants. However, given the
uncertainty surrounding natural overland colonization, it
is worthwhile to study the best candidates for this and
establish that it occurred at all. From the palaeontological
record, two such candidates for natural colonization are
the mountain hare and the stoat. Single fossil specimens of
the mountain hare and the stoat have been dated to the
Late Glacial period, i.e. after the LGM but before the
known occurrence of Mesolithic people in Ireland
(Woodman et al. 1997; McCormick 1999). Clearly, if
these two species occurred in Ireland before the records of
people, it suggests natural colonization. Both these cases
underline the uncertainty of the presence of a land bridge
to Ireland after the LGM. Given their cold tolerance, the
stoat and mountain hare are species that may have been
able to survive LGM conditions in Ireland (Stuart & Van
Wijngaarden-Bakker 1985). If these species did manage to
colonize Ireland during the LGM, then the speculation
about whether or not there was a land bridge subsequent
to the LGM is irrelevant to their colonization, i.e. they
would already have been there. Indeed, when we suggest
‘colonization’ at the LGM, it is also possible that the
species were already present in Ireland at this time and the
colonization was an even earlier event.
While single radiocarbon-dated records of the two
species are encouraging, further evidence of natural
colonization is desirable. Here, we adopt a complementary
phylogeographic approach, i.e. we examine whether the
molecular variation in modern specimens is consistent
with natural colonization, as opposed to human introduction.
Our study is based on the stoat because we believe it
to be a particularly good model, and unlike the mountain
hare (Hamill et al. 2006), it has not been subjected to any
previous phylogeographic analysis relating to Ireland
(Kurose et al. 2005).
Stoats are found over a very wide range of temperature
conditions from warm temperate to arctic (King 1991).
They currently occur in the high Arctic of Greenland and
Canada, feeding on lemmings (Gilg et al. 2003), and at the
LGM, it is known from fossils that lemmings survived in
Ireland (Woodman et al. 1997), providing a potential food
supply. Although fossils of stoats have yet to be found from
the LGM in Ireland, the fossil record in continental
Europe indicates their probable presence because the
species is one of the glacial faunal elements, occurring in
assemblages together with mammoth, woolly rhinoceros,
spotted hyena and reindeer (Sommer & Benecke 2004). A
stoat vertebra has also been found in a deposit believed to
be ca 15 kyr old on the island of Andøya off northern
Norway (Fjellberg 1978), at a time when Fennoscandia
still had substantial ice cover (Andersen & Borns 1997).
One of the great advantages of the stoat for phylogeographic
analysis is its wide and continuous contemporary
distribution in Ireland, Britain and continental Europe.
Not only it is thermally adaptable but the species is also
generalized and common; in Britain, for instance, the stoat
is the most abundant wild carnivore (Harris et al. 1995).
Although there have been some recent introductions of the
stoat to Shetland and New Zealand for biological control
(King 1991), there is no reason to believe there have been
accidental or deliberate introductions of the stoat to Britain
or Ireland, or movements around continental Europe.
Therefore, there is every expectation that the stoats present
during the Late Glacial (based on the radiocarbon-dated
specimens) survived in Ireland through to the present-day
without anyextinctions or introductions. Here, weexamine
whether the molecular data are consistent with these
expectations of natural colonization based on a comparison
of Irish and British stoats within the context of the
European and Holarctic ranges of the species.
2. MATERIAL AND METHODS
(a) Sample collection, processing and sequencing
A total of 197 tissue and skin samples collected from stoats
from 153 localities in Eurasia and Greenland successfully
yielded sequences. Details of these samples, which largely
came from museum skin collections, are given in the
electronic supplementary material. DNA isolation, PCR
amplification and sequencing were described in detail
previously (Martı´nkova´ & Searle 2006). The full set
of primers used for amplification and sequencing of
mitochondrial genes for cytochrome b (cyt b), Thr-tRNA,
Pro-tRNA and partial D-loop are listed in the table 1. We
Table 1. Primers and PCR conditions. (The new primers designed for this study include the indication of the position of their 30
end within the dog complete mitochondrial sequence (Kim et al. 1998). Ta, annealing temperature; Cb, cytochrome b; D, partial
D-loop (and flanking tRNA sequences); F, forward; R, reverse.)
primer name sequence (50
–30
) Ta reference
Cb-M1-F CTC ACA TGG AAT CTA ACC ATG AC 56 Kurose et al. (2000)
Cb-L14474-F CTT TAT CTG CCT ATT CCT ACA CG 56 this study
Cb-H14549-R CTA TGA ATG CAG TTG CTA TGA C 55 this study
Cb-Mustela07-F TTC ATC ATT TCA GCA CTA GCA GCA GTC 55 Fleming & Cook (2002)
Cb-Mustela06-R GTG GAA TGG GAT TTT GTC AGA GTC GGA 55 Fleming & Cook (2002)
Cb-L14897-F GCC CTA TTC CTT ATT CTA ACA C 55 this study
Cb-H14939-R GGC TGG GAT ATA GTT GTC TGG 56 this study
D-Cbz-F ATG AAT TGG AGG ACA ACC AGT 55 Kurose et al. (1999)
Cb-MR1-R TCT TCC TTG AGT CTT AGG GAG 56 Kurose et al. (2000)
D-L15454-F CTG ACA TTC TAA CTA AAC TAT TCC 55 this study
D-H15777-R TGA AGT AAG AAC CAG ATG CCA G 55 this study
D-MER-R CGC GGG TGG TGT ATA AAT AT 55 Kurose et al. (1999)
1388 N. Martı´nkova´ et al. Colonization of Ireland by stoats
Proc. R. Soc. B (2007)
Paper 2.2.1 157
took particular care to ensure authenticity of sequences from
the museum samples using methods derived from laboratory
protocols for handling ancient DNA (Martı´nkova´ &
Searle 2006).
(b) Data analysis
Chromatogram contigs were assembled in SEQUENCHER v. 4.2
(Gene Codes) and sequences were aligned manually with
BIOEDIT v. 7.0.5.2 (Hall 1999). Haplotypes of mitochondrial
(mt) DNA sequences of stoats were identified by DAMBE v.
4.2.13 (Xia & Xie 2001), and haplotype and nucleotide
diversity and total and net divergence were calculated with
DNASP v. 4.00.5 (Rozas et al. 2003). For network construction,
the 49 cyt b haplotypes that were obtained from our 197
specimens were combined with published haplotypes from
Russia, Japan and North America (14 GenBank sequences,
12 new haplotypes and 13 new localities). From the 197
stoats, we were able to obtain concatenated mtDNA
sequences from 186 individuals, yielding 76 haplotypes.
Median-joining networks were constructed in NETWORK v.
4.1.1.0 (Bandelt et al. 1999) using either the complete cyt b
sequences or the concatenated sequence of the cyt b, ThrtRNA,
Pro-tRNA and the partial D-loop excluding an
ambiguous part of the TnCn stretch. The sequences reported
in this paper have been deposited in the GenBank database
(Accession numbers: EF088939–EF089135).
(c) Analysis of population size changes
Mismatch distributions were constructed from the stoat
sequence data with ARLEQUIN v. 3.00 (Excoffier et al. 2005)
and compared with those expected under the sudden
population expansion model (Rogers & Harpending 1992)
using goodness-of-fit statistics based on the sum of square
deviations from 10 000 bootstrap replicates (Schneider &
Excoffier 1999). The population expansion hypothesis was
further tested with ARLEQUIN by Fu’s (1997) Fs index of
selective neutrality that is sensitive also to demographic
population expansion.
(d) Estimation of mutation rate
The mutation rate of the whole mtDNA concatenated region
of the stoat was estimated under a molecular clock assumption
from the Irish haplotype dataset by Bayesian coalescent
analysis in BEAST v. 1.3 (Drummond & Rambaut 2006). The
lower constraint of the age of the Irish population was set to
13 kyr ago, which is the age of the Late Glacial fossil found in
Ireland (Woodman et al. 1997). This relates to a stoat from
Killavullen Cave, Co. Cork that Woodman et al. (1997) dated
as 10 680G110 14
C years old. In terms of calendar years, this
is ca 12 750–12 900 years old based on the IntCal04 radiocarbon
calibration curve (Reimer et al. 2004). The upper age
constraint was selected as the time when Ireland was fully
covered by an ice-sheet 40 kyr ago (Bowen et al. 2002), an
event also supported by a gap in the Irish fossil record
(Woodman et al. 1997). The mutation rate was then calculated
from a range of plausible alternative trees in the Markov chain
Monte Carlo (MCMC) algorithm search in which root height
fits the constraint range. The MCMC was run for 50 million
steps and the first 10% were discarded as burn-in.
(e) Bayesian molecular dating
As the mismatch distribution and Fu’s (1997) Fs indicated that
the Irish, British and continental European populations of
stoat exhibited independent demographic histories, their ages
were estimated using an exponential growth model. They were
calculated by Bayesian coalescent analysis using MCMC in
BEAST with the previously estimated mutation rate and the age
constraints on the Irish population. BEAST records the age of a
particular node in various trees after the MCMC convergence
in which the node appears. Hence, the age estimate is not
constrained to a precisely defined evolutionary history (tree
topology and branch lengths), but rather explores the available
tree space making the method particularly suitable for dating
recent divergence events. Five independent MCMC runs were
made with 10 million steps each, sampled every 1000th step.
The results of all runs were then combined to obtain the ages of
the most recent common ancestors of Irish, British and
European stoat populations, respectively, with the first 10%
discarded as burn-in.
3. RESULTS
We analysed complete cyt b sequences from a total of 211
stoats (166 localities), which yielded 61 haplotypes
(figure 1a and electronic supplementary material). All
British cyt b haplotypes (which came from stoats collected
throughout the mainland and in Islay and Shetland)
formed a monophyletic lineage (figure 1a), 85% of which
shared the same haplotype (no. 39). Similarly, Irish stoats
(also collected throughout the country and in the Isle of
Man) formed a monophyletic but distinct lineage with a
single dominant haplotype (no. 17). There is little
indication of genetic structure in continental Eurasia.
Some of the haplotypes identified in Alaska (nos 56–60)
were similar to those from Siberia and Japan (no. 44,
nos 49–55, no. 61). Fleming & Cook (2002) identified this
‘Beringian’ clade, and we found that it forms part of an
extensive continental Eurasian lineage. Other North American
haplotypes remain distinct and consistent with Fleming
and Cook’s ‘Continental (American)’ and ‘British Columbia
islands’ clades (figure 1a).
We identified 76 haplotypes in 186 stoats (142 localities)
from a concatenated sequence of mtDNA, comprising
1771 bp of cyt b, Thr-tRNA, Pro-tRNA and the partial
D-loop, excluding an ambiguous part of the TnCn stretch
(figure 1b and electronic supplementary material). British
and Irish populations retained their distinctiveness and
each remained monophyletic (figure 1b).
Considering the concatenated sequence, we found
markedly lower nucleotide and haplotype diversity in the
British population than in Irish or continental European
populations. Nucleotide (pGs.d.) and haplotype (hGs.d.)
diversity were as follows: Irish population: pZ0.0026G
0.0002, hZ0.96G0.01, nZ52; British population:
pZ0.0007G0.0001, hZ0.58G0.08, nZ53; and continental
European population: pZ0.0031G0.0002,
hZ0.97G0.01, nZ74. The total (Dxy) and net (Da)
divergences between Irish and continental European
populations of 0.43 and 0.15%, respectively, were similar
in magnitude to the divergences between Britain and
continental Europe (DxyZ0.45% and DaZ0.26%).
However, the British and Irish populations were more
divergent from each other (DxyZ0.61% and DaZ0.45%)
than either was from the continental European population.
There is no reason to believe that the greater
genetic diversity of Irish stoats is the result of multiple
colonization events; mismatch distributions of all populations
significantly fitted the sudden population
Colonization of Ireland by stoats N. Martı´nkova´ et al. 1389
Proc. R. Soc. B (2007)
158 Paper 2.2.1
(a)(b)
Figure1.Therelatednessanddiversityofgeneticlineagesofthestoat,depictedbymedian-joiningnetworksofsequencesobtainedfromindividualsthroughoutthespeciesrange.(a)cytb
sequences(1140bp),thearrowindicatestheconnectiontoBeringianandNorthAmericancytbhaplotypes(inset),haplotypeno.44isseparatedfromhaplotypeno.1by14substitutionsand
fromhaplotypeno.2by37substitutions,respectively.(b)Concatenatedsequencesofcytb,Thr-tRNA,Pro-tRNAandpartialD-loop,excludingtheambiguouspartoftheTnCnregion
(1771bp),thearrowindicatestheconnectiontotheGreenlandhaplotypeno.16(inset)separatedfromhaplotypeno.68by50substitutions.Colourcodingisusedtodistinguishindividual
Europeancountries,AsiaandNorthAmerica.Furtherdetailsofthenumberedhaplotypesareavailableintheelectronicsupplementarymaterial.
1390 N. Martı´nkova´ et al. Colonization of Ireland by stoats
Proc. R. Soc. B (2007)
Paper 2.2.1 159
expansion model, which describes growth of a single
founder population (Ireland: sum of square deviations
SSDZ0.0021, pZ0.46; Britain: SSDZ0.0131, pZ0.56;
continental Europe: SSDZ0.00067, pZ0.87). Also
consistent with population expansion, Fu’s (1997) Fs
values were negative and significantly different from zero
for the Irish (FsZK10.188, pZ0) and continental
European lineages (FsZK28.548, pZ0) but non-significant
for the British lineage (FsZK2.207, pZ0.13).
The mutation rate of the concatenated mtDNA region
consistent with the age constraints of the Irish population
sampled over a range of plausible alternative trees was
7.244!10K8
substitutions siteK1
yrK1
(14.5% per Myr).
This high rate is expected when measuring within-species
divergences younger than 1 Myr (Ho et al. 2005). Analysis
with this mutation rate showed that the age of the most
recent common ancestor of the Irish lineage is greater than
that for the British lineage, though non-significantly so
(figure 2). With 95% confidence, the most recent common
ancestor of the Irish lineage is 14 270–34 050 years old
(mean estimate, 23 510), whereas that of the British
lineage is 4858–20 710 years old (mean estimate, 12 240).
4. DISCUSSION
The molecular diversity data that we obtained are extraordinary.
Even though the landmass of Britain is 2.5 times
larger than Ireland and is located closer to continental
Europe, we found stoats to be far less genetically variable in
Britain than Ireland. Indeed, for the concatenated cyt b and
D-loop dataset, the haplotype and nucleotide diversities for
Ireland are similar to those for continental Europe, while
those in Britain are considerably lower.
These data argue strongly against the likelihood of
introduction of stoats to Ireland. Generally, populations
arising from introductions show lower genetic diversity
than those that have been colonized naturally. The
numbers of the individuals colonizing tend to be smaller
and introductions tend to be more recent than natural
colonizations. We have, for instance, found low levels of
genetic diversity among stoats introduced to New Zealand
(N. Martı´nkova´ et al. 2007, unpublished data) and bank
voles introduced to Ireland (P. Stuart et al. 2007,
unpublished data). On the basis of the molecular data, it
would seem positively perverse to suggest that stoats in
Ireland are an introduction while those in Britain are a
natural colonization (and, to our knowledge, there
has never been any suggestion that stoats were introduced
to Britain).
Therefore, the molecular dataare supportiveofthe fossil
evidence for a natural colonization of Ireland by stoats. We
applied a Bayesian method to date the origins of the Irish
and British populations of stoats from the molecular data,
using the palaeontological record for Ireland to set age
constraints. The Irish and British populations fit into
separate monophyletic lineages. The constraints
determined the 95% CI that we obtained for the age of
the Irish lineage: 14 270–34 050 years, which is, of course,
compatible with an LGM origin for the lineage. These
dates are also compatible with colonization over a Late
Glacial land bridge from continental Europe to southwest
England and Ireland, which is the most probable time for
such a land bridge if it existed (Lambeck & Purcell 2001).
The estimate of the age of the British lineage is
4858–20 710 years old, suggesting a more recent colonization
than the Irish one. Again, given that the British
colonization is natural, it is difficult to argue that an earlier
colonization of Irish stoats should be an introduction.
Clearly, the dating of the British lineage is consistent with
the natural colonization of Britain by stoats before the
flooding of the English Channel.
Some of our most interesting data relate to stoats from
the Isle of Man, an island in the Irish Sea, approximately
halfway between Ireland and Britain. The Isle of Man has
a similar range of non-commensal small mammals as
Ireland, with the same three species (wood mouse, pygmy
shrew and stoat) out of the eleven found in Britain (Yalden
1999). However, the Isle of Man was connected longer to
Britain than to Ireland (Innes et al. 2004). Hence, the Manx
stoats might be expected to be more similar to the British
than the Irish. In fact, in terms of mtDNA, the Manx stoats
fall within the Irish lineage. Morphologically, the Irish and
Manx stoats are also more similar. They are classified
together as a distinct subspecies, Mustela erminea hibernica,
on the basis of pelage characteristics (Miller 1912).
A scenario can be developed which best explains our
molecular data for stoats in the British Isles. Following
Stuart & Van Wijngaarden-Bakker (1985), we believe that
stoats are sufficiently cold tolerant to have survived close
to the British–Irish ice sheet at the LGM, and therefore
were present on the exposed landmass of Ireland, having
colonized at that time or earlier. Thus, we propose that the
progenitors of the Irish mtDNA lineage and morphotype
were cold tolerant. Consistent with that, mtDNA
haplotypes from high latitudes (Scandinavia) and altitudes
(Swiss and Italian Alps) are among those most closely
related to modern Irish haplotypes (figure 1). We believe
that these cold-tolerant stoats followed the retreating ice at
the end of the LGM and spread throughout Britain,
Ireland and the Isle of Man. The stoats in Ireland would
have become isolated with rising sea level after the LGM
and would have increased in population size from a small
number of founders (hence the signal for population
expansion). At a later time, the Isle of Man population of
stoats would also have become separated from the British.
0
20
40
60
80
100
120
0.02 0.04 0.06 0.08 0.10 0.12
age (Myr)
posteriorprobabilitydensity
Britain
Ireland
continental Europe
Figure 2. Estimates of the age of the Irish, British and
continental European stoat populations. These are shown as
posterior probability density functions estimated by the
Bayesian coalescent analysis using a mutation rate of
0.07244 bpK1
MyrK1
and age constraints of the Irish
population: 13–40 kyr. Five independent runs of MCMC
were performed for 10 million steps each, with the first 1
million steps discarded as burn-in.
Colonization of Ireland by stoats N. Martı´nkova´ et al. 1391
Proc. R. Soc. B (2007)
160 Paper 2.2.1
However, Ireland and the Isle of Man would have become
islands long before Britain was itself separated from
continental Europe. They perhaps became islands early
enough that cold-sensitive species were totally unable to
colonize them overland, hence the paucity of species in
both Ireland and the Isle of Man.
There is a need to explain why British stoats are
currently a different mtDNA lineage and morphotype
from those found in Ireland and the Isle of Man. We
believe that the continued land bridge with continental
Europe is the key. This land bridge would have been
available to stoats during the warm periods of the Late
Glacial and Postglacial. For instance, during the Postglacial
period between the end of the Younger Dryas
(11 400 years ago; Walker et al. 2003) and the flooding of
the English Channel (7500 years ago at the very latest;
Yalden 1999), the climate in Britain would have been
similar to today (Andersen & Borns 1997). Therefore, we
suggest that an mtDNA lineage and morphotype of stoats
adapted to such warm conditions would have been able to
enter Britain and replace the cold-tolerant lineage that was
originally present and continues to survive in the isolated
populations of Ireland and the Isle of Man. In this way, we
can not only explain the morphological and mitochondrial
differences between British and Irish stoats, but also clarify
why the British mtDNA lineage is younger and less
variable than the mtDNA lineage in Ireland.
There is evidence from the literature for replacement
processes of the type we suggest. A phylogeographic study
by Piertney et al. (2005) indicated partial replacement of
one mtDNA lineage of water voles (Arvicola terrestris) by
another in Britain, and ancient DNA studies have
demonstrated sequential replacement of mtDNA lineages
in brown bears (Ursus arctos) occupying Beringia (Barnes
et al. 2002). There is also evidence that different mtDNA
haplotypes may adapt individuals to different temperature
conditions (Fontanillas et al. 2005). Different external
morphologies may also of course be related to different
temperature adaptations. Within the crow species (Corvus
corone), the all-black carrion crow (Corvus corone corone) is
expanding in Britain at the expense of the grey-and-black
hooded crow (C. c. cornix) apparently in response to
climate change (Cook 1975). The hooded crow is now
found only in the north of Scotland. However, in a clear
parallel with what we suggest for the stoat, it is the hooded
crow (i.e. the form that is being displaced) that is found in
Ireland, even though the climate in Ireland is similar to
southern Britain.
In terms of the Irish question, we do not claim to have
‘answered’ it. There is a need to study a wide variety of
species before there will be a comprehensive understanding
of the peculiarities of Ireland’s biota. Human introductions
have undoubtedly been very important in generating the
Irish fauna and flora, and it is actually very difficult to
demonstrate convincingly that any particular terrestrial
species did manage to colonize naturally. We believe that
the weight of evidence strongly favours such a natural
colonization by the stoat and that this colonization was very
early, probably around the LGM. Our data suggest that
there is no need to propose a Late Glacial land bridge to
explain the natural occurrence of stoats in Ireland.
Our studies of the stoat provide another very interesting
example of a terrestrial mammal with strong genetic
differences between Irish and British populations to add to
pine martens (Davison et al. 2001), pygmy shrews
(Mascheretti et al. 2003) and mountain hares (Hamill
et al. 2006). However, in the case of the stoat, we suggest a
new explanation for this discrepancy between Britain and
Ireland: that there was originally the same genetic type in
Britain and Ireland, and that the much long-lasting
connection with continental Europe allowed a replacement
event in Britain. Replacement may be important in
other examples of within-species genetic differences
between Ireland and Britain. It could also explain some
of the differences in the species lists for the two countries.
We are grateful to all people who provided samples for this
study as listed in the electronic supplementary material. We
thank I. Barnes, D. Fo¨rster, I˙. Gu¨ndu¨z, J. Provan,
A. Rambaut and B. Shapiro for their technical advice,
S. Martı´nek for help with the figures and S. Bearhop,
A. Douglas, J. Herman, C. Maggs and J. Provan for their
comments on earlier versions of the manuscript. This work
was supported by the Natural Environment Research
Council. R.M. is supported by the Quercus partnership
between Queen’s University Belfast and the Environment &
Heritage Service, Northern Ireland.
REFERENCES
Andersen, B. G. & Borns Jr, H. W. 1997 The Ice Age world.
Oslo, Norway: Scandinavian University Press.
Bandelt, H.-J., Forster, P. & Ro¨hl, A. 1999 Median-joining
networks for inferring intraspecific phylogenies. Mol. Biol.
Evol. 16, 37–48.
Barnes, I., Matheus, P., Shapiro, B., Jensen, D. & Cooper, A.
2002 Dynamics of Pleistocene population extinctions in
Beringian brown bears. Science 295, 2267–2270. (doi:10.
1126/science.1067814)
Bowen, D. Q., Philips, F. M., McCabe, A. M., Knutz, P. C. &
Sykes, G. A. 2002 New data for the Last Glacial
Maximum in Great Britain and Ireland. Quaternary Sci.
Rev. 21, 89–101. (doi:10.1016/S0277-3791(01)00102-0)
Cook, A. 1975 Changes in carrion-hooded crow hybrid zone
and possible importance of climate. Bird Study 22,
165–168.
Corbet, G. B. 1961 Origin of the British insular races of small
mammals and of the ‘Lusitanian’ fauna. Nature 191,
1037–1040. (doi:10.1038/1911037a0)
Davison, A., Birks, J. D. S., Brookes, R. C., Messenger, J. E.
& Griffiths, H. I. 2001 Mitochondrial phylogeography and
population history of pine martens Martes martes
compared with polecats Mustela putorius. Mol. Ecol. 10,
2479–2488. (doi:10.1046/j.1365-294X.2001.01381.x)
Drummond, A. J. & Rambaut, A. 2006 BEAST v1.0, http://
evolve.zoo.ox.ac.uk/beast/.
Excoffier, L., Laval, G. & Schneider, S. 2005 ARLEQUIN ver.
3.0: an integrated software package for population
genetics data analysis. Evol. Bioinform. Online 1, 47–50.
Fjellberg, A. 1978 Fragments of Middle Weichselian fauna on
Andøya, north Norway. Boreas 7, 39.
Fleming, M. A. & Cook, J. A. 2002 Phylogeography of
endemic ermine (Mustela erminea) in southeast Alaska.
Mol. Ecol. 11, 795–807. (doi:10.1046/j.1365-294X.2002.
01472.x)
Fontanillas, P., De´praz, A., Giorgi, M. S. & Perrin, N. 2005
Nonshivering thermogenesis capacity associated to mitochondrial
DNA haplotypes and gender in the greater
white-toothed shrew, Crocidura russula. Mol. Ecol. 14,
661–670. (doi:10.1111/j.1365-294X.2004.02414.x)
Fu, Y.-X. 1997 Statistical tests of neutrality of mutations
against population growth, hitchhiking and background
selection. Genetics 147, 915–925.
1392 N. Martı´nkova´ et al. Colonization of Ireland by stoats
Proc. R. Soc. B (2007)
Paper 2.2.1 161
Gilg, O., Hanski, I. & Sittler, B. 2003 Cyclic dynamics in a
simple vertebrate predator–prey community. Science 302,
866–868. (doi:10.1126/science.1087509)
Hall, T. A. 1999 BIOEDIT: a user-friendly biological sequence
alignment editor and analysis program for Windows
95/98/NT. Nucleic Acids Symp. Ser. 41, 95–98.
Hamill, R. M., Doyle, D. & Duke, E. J. 2006 Spatial patterns
of genetic diversity across European subspecies of the
mountain hare, Lepus timidus L. Heredity 97, 355–365.
(doi:10.1038/sj.hdy.6800880)
Harris, S., Morris, P., Wray, S. & Yalden, D. 1995 A review of
British mammals: population estimates and conservation status
of British mammals other than cetaceans. Peterborough, UK:
Joint Nature Conservation Committee.
Ho, S. Y. W., Phillips, M. J., Cooper, A. & Drummond, A. J.
2005 Time dependency of molecular rate estimates and
systematic overestimation of recent divergence times. Mol.
Biol. Evol. 22, 1561–1568. (doi:10.1093/molbev/msi145)
Innes, J. B., Chiverrell, R. C., Blackford, J. J., Davey, P. J.,
Gonzalez, S., Rutherford, M. M. & Tomlinson, P. R. 2004
Earliest Holocene vegetation history and island biogeography
of the Isle of Man, British Isles. J. Biogeogr. 31,
761–772. (doi:10.1111/j.1365-2699.2003.01048.x)
Kim, K. S., Lee, S. E., Jeong, H. W. & Ha, J. H. 1998 The
complete nucleotide sequence of the domestic dog (Canis
familiaris) mitochondrial genome. Mol. Phylogenet. Evol.
10, 210–220. (doi:10.1006/mpev.1998.0513)
King, C. M. 1991 Stoat. In The handbook of British mammals
(eds G. B. Corbet & S. Harris), pp. 377–387, 3rd edn.
Oxford, UK: Blackwell.
Kurose, N., Masuda, R. & Yoshida, M. C. 1999 Phylogeographic
variation in two mustelines, the least weasel
Mustela nivalis and the ermine M. erminea of Japan,
based on mitochondrial DNA control region sequences.
Zool. Sci. 16, 971–977. (doi:10.2108/zsj.16.971)
Kurose, N., Abramov, A. V. & Masuda, R. 2000 Intrageneric
diversityofthecytochromebgeneandphylogenyofEurasian
species of the genus Mustela (Mustelidae, Carnivora). Zool.
Sci. 17, 673–679. (doi:10.2108/zsj.17.673)
Kurose, N., Abramov, A. V. & Masuda, R. 2005 Comparative
phylogeography between the ermine Mustela erminea and
the least weasel M. nivalis of Palaearctic and Nearctic
regions, based on analysis of mitochondrial DNA control
region sequences. Zool. Sci. 22, 1069–1078. (doi:10.2108/
zsj.22.1069)
Lambeck, K. & Purcell, A. P. 2001 Sea-level change in the
Irish Sea since the Last Glacial Maximum: constraints
from isostatic modelling. J. Quaternary Sci. 16, 497–506.
(doi:10.1002/jqs.638)
Martı´nkova´, N. & Searle, J. B. 2006 Amplification success
rate of DNA from museum skin collections: a case study of
stoats from 18 museums. Mol. Ecol. Notes 6, 1014–1017.
(doi:10.1111/j.1471-8286.2006.01482.x)
Mascheretti, S., Rogatcheva, M. B., Gu¨ndu¨z, I., Fredga, K. &
Searle, J. B. 2003 Phylogeography of the Eurasian pygmy
shrew Sorex minutus: clues on its mode of colonisation of
Ireland. Proc. R. Soc. B 270, 1593–1599. (doi:10.1098/
rspb.2003.2406)
McCormick, F. 1999 Early evidence for wild animals in
Ireland. In The Holocene history of the European vertebrate
fauna. Modern aspects of research (ed. N. Benecke),
pp. 355–371. Rahden, Germany: Verlag Marie Leidorf
GmbH.
Miller, G. S. 1912 Catalogue of the mammals of western Europe.
London, UK: British Museum.
Mix, A. C., Bard, E. & Schneider, R. 2001 Environmental
processes of the ice age: land, oceans, glaciers (EPILOG).
Quaternary Sci. Rev. 20, 627–657. (doi:10.1016/S0277-
3791(00)00145-1)
Moore, P. D. 1987 Snails and the Irish question. Nature 328,
381–382. (doi:10.1038/328381a0)
Piertney, S. B., Stewart, W. A., Lambin, X., Telfer, S., Aars, J.
& Dallas, J. F. 2005 Phylogeographic structure and
postglacial evolutionary history of water voles (Arvicola
terrestris) in the United Kingdom. Mol. Ecol. 14,
1435–1444. (doi:10.1111/j.1365-294X.2005.02496.x)
Reimer, P. J. et al. 2004 IntCal04 terrestrial radiocarbon age
calibration, 0–26 cal ky BP. Radiocarbon 46, 1029–1058.
Rogers, A. R. & Harpending, H. 1992 Population-growth
makes waves in the distribution of pairwise genetic
differences. Mol. Biol. Evol. 9, 552–569.
Rozas, J., Sa´nchez-DelBarrio, J. C., Messeguer, X. & Rozas,
R. 2003 DNASP, DNA polymorphism analysis by coalescent
and other methods. Bioinformatics 19, 2496–2497.
(doi:10.1093/bioinformatics/btg359)
Schneider, S. & Excoffier, L. 1999 Estimation of demographic
parameters from the distribution of pairwise
differences when the mutation rates vary among sites:
application to human mitochondrial DNA. Genetics 152,
1079–1089.
Sommer, R. & Benecke, N. 2004 Late- and post-glacial
history of the Mustelidae in Europe. Mammal Rev. 34,
249–284. (doi:10.1111/j.1365-2907.2004.00043.x)
Stuart, A. J. & Van Wijngaarden-Bakker, L. 1985 Quaternary
vertebrates. In The Quaternary history of Ireland (eds K. J.
Edwards & W. P. Warren), pp. 221–250. London, UK:
Academic Press.
Walker, M. J. C., Coope, G. R., Sheldrick, C., Turney,
C. S. M., Lowe, J. J., Blockley, S. P. E. & Harkness, D. D.
2003 Devensian lateglacial environmental changes in
Britain: a multi-proxy environmental record from Llanilid,
South Wales, UK. Quaternary Sci. Rev. 22, 475–520.
(doi:10.1016/S0277-3791(02)00247-0)
Woodman, P., McCarthy, M. & Monaghan, N. 1997 The
Irish quaternary fauna project. Quaternary Sci. Rev. 16,
129–159. (doi:10.1016/S0277-3791(96)00037-6)
Xia, X. & Xie, Z. 2001 DAMBE: data analysis in molecular
biology and evolution. J. Hered. 92, 371–373. (doi:10.
1093/jhered/92.4.371)
Yalden, D. 1999 The history of British mammals. London, UK:
Poyser.
Colonization of Ireland by stoats N. Martı´nkova´ et al. 1393
Proc. R. Soc. B (2007)
162 Paper 2.2.1
Paper 2.2.2
Seifertová M., Bryja J., Vyskočilová M., Martínková N., Šimková A. 2012. Multiple
Pleistocene refugia and post-glacial colonization in the European chub (Squalius cephalus)
revealed by combined use of nuclear and mitochondrial markers. Journal of Biogeography
39: 1024-1040.
– 163 –
ORIGINAL
ARTICLE
Multiple Pleistocene refugia and postglacial
colonization in the European chub
(Squalius cephalus) revealed by combined
use of nuclear and mitochondrial markers
Ma´ria Seifertova´1
, Josef Bryja1,2
, Martina Vyskocˇilova´1
, Nata´lia Martı´nkova´2,3
and Andrea Sˇimkova´1
*
1
Department of Botany and Zoology, Faculty of
Science, Masaryk University, Kotla´rˇska´ 2,
61137 Brno, Czech Republic, 2
Institute of
Vertebrate Biology, Academy of Sciences of the
Czech Republic, Kveˇtna´ 8, 60365 Brno, Czech
Republic, 3
Institute of Biostatistics and
Analyses, Masaryk University, Kamenice 3,
62500 Brno, Czech Republic
*Correspondence: Andrea Sˇimkova´, Department
of Botany and Zoology, Faculty of Science,
Masaryk University, Kotla´rˇska´ 2, 61137 Brno,
Czech Republic.
E-mail: simkova@sci.muni.cz
ABSTRACT
Aim To analyse patterns of nuclear and mitochondrial genetic variation in the
European chub, Squalius cephalus (Linnaeus, 1758), in order to understand the
evolutionary history of this species and to test biogeographical hypotheses for the
existence of co-distributed European freshwater fish species.
Location Rivers in Europe (Finland, Poland, Czech Republic, France, Bulgaria,
Spain, Italy).
Methods We genotyped 12 polymorphic microsatellite markers derived from
310 individuals collected from across the distribution of S. cephalus in Europe
(including a total of 15 populations) and sequenced mitochondrial DNA
(mtDNA) from a subset of 75 individuals. Sequences of mtDNA cytochrome b
were analysed using both phylogenetic (median-joining networks) and
population genetic methods (tests for demographic history, mismatch
distributions, Bayesian coalescent analysis). Geographical structure in
microsatellite loci was examined using a distance method (FST), factorial
correspondence analysis (FCA) and a Bayesian clustering method (structure).
Results The mtDNA network showed a clear split into four different haplogroup
lineages: Western (separated into Atlantic and Danubian sublineages), Eastern,
Aegean (occurring in two distinct sublineages in the Balkans and in Spain) and
Adriatic. Our results indicate recent population expansion in the Eastern and
Western Atlantic lineages and the admixture of two previously separate
sublineages (Atlantic and Danubian) in the Western lineage. Bayesian structure
analysis as well as FCA results roughly corresponded to the mtDNA-based
structure, separating the sampled individuals into almost non-overlapping
groups.
Main conclusions Our results support hypotheses suggesting origins of extant
lineages of freshwater fishes in multiple refugia and the subsequent post-glacial
colonization of Europe via different routes. We confirmed the previously
proposed two-step expansion scenario from the Danube refuge, the existence of a
secondary (Atlantic) refuge during the last glaciation (probably in the Rhone
River) and population expansion of this lineage. Conspicuous divergences among
Mediterranean populations reflect their different origin, as well as their low
contribution to the recent genetic pool of chub in central Europe.
Keywords
Cytochrome b, Europe, freshwater fishes, glacial refugia, microsatellites,
phylogeography, population structure.
Journal of Biogeography (J. Biogeogr.) (2012) 39, 1024–1040
1024 http://wileyonlinelibrary.com/journal/jbi ª 2011 Blackwell Publishing Ltd
doi:10.1111/j.1365-2699.2011.02661.x
164 Paper 2.2.2
INTRODUCTION
The most important processes influencing the evolution of
biota in Europe throughout the past 2 million years (Myr)
were the climatic oscillations that have driven periodic
movements of ice sheets (Hewitt, 1999). Repeated range
contractions were followed by range expansions from climatically
more favourable regions (so-called glacial refugia), these
events leaving a signature within the genome of many species
(Hewitt, 1996). In freshwater fishes, evolutionary history is
closely related to the biotic and geological evolution of a
region, because their dispersal is dependent on direct connections
between hydrographic basins, and because the history of
basin interconnections reflects the underlying geological
development of landscapes (Lundberg, 1993). Recent phylogeographical
studies have brought many new insights into the
post-glacial history of European freshwater fish species. Studies
have often demonstrated the existence of evolutionarily
distinct groups within each species, which suggests multiple
refugia during glacial cycles (e.g. Nesbo et al., 1999; Kotlik &
Berrebi, 2001; Gum et al., 2005; Ha¨nfling et al., 2009).
Generally, the number of refugia defines the number of recent
major genetic lineages, and the geographical distribution of
these lineages provides clues to the routes of post-glacial
colonization (Ha¨nfling et al., 2009).
The European chub, Squalius cephalus (Linnaeus, 1758),
belonging to the Cyprinidae family and the Leuciscinae
subfamily, has become a convenient model organism for the
testing of biogeographical hypotheses concerning European
freshwater ecosystems, especially due to its occurrence
throughout most of Europe and a postulated Mesopotamian
origin during the Pliocene (Durand et al., 1999a, 2000;
Zardoya & Doadrio, 1999). Moreover, chub distribution in
Europe is not likely to be influenced by intentional introduction
because of its low importance in commercial aquaculture,
and, similar to other primary freshwater fishes, this species has
a low capacity for trans-watershed dispersal (Zardoya &
Doadrio, 1999). Thus, the distribution of this fish species can
be postulated to closely reflect both the geological and climatic
historical development of hydrographic basins (Ketmaier et al.,
2004), including episodes of isolation and interconnection
processes (Durand et al., 1999a; Sanjur et al., 2003).
Durand et al. (1999a, 2000) identified four highly divergent
mitochondrial DNA (mtDNA) phylogeographical lineages of
chub in Europe – Western, Eastern and two Mediterranean
lineages (Adriatic and Aegean) – that are probably descendants
resulting from the rapid radiation of a widespread ancestor.
This marked population diversification provides strong evidence
of the eradication of S. cephalus from most of Europe
during Pleistocene glacial maxima and its survival in four
refugia, i.e. the Adriatic side of the Balkans, eastern Greece
(Aegean rivers), the southern tributaries of the Danube, and
the peripheries of the Black and Caspian seas. Two sources of
post-glacial recolonization of central and northern Europe
have been proposed on the basis of mtDNA analyses (Durand
et al., 1999a). First, the Western (Danube) lineage reached
central and western Europe in two steps. This lineage entered
western Europe during the Riss–Wu¨rm interglacial period
(c. 130–110 ka) and survived the next glacial period in western
European rivers (the Rhine and the Rhone); the next
expansion started from these refugia at the end of the Wu¨rm
period (10 ka). Second, expansion of the Eastern lineage
(Ponto–Caspian) into northern Europe was in a single step
from the Caspian refuge (Durand et al., 1999a).
Nevertheless, the reconstruction of species history often
depends on the markers studied. The majority of the
postulated hypotheses elucidating the phylogeography and
post-glacial colonization routes of freshwater fishes in the
Palaearctic are inferred only from the geographical distribution
of mtDNA variation (Ha¨nfling & Brandl, 1998; Durand et al.,
1999a, 2000; Laroche et al., 1999; Nesbo et al., 1999; Bernatchez,
2001; Kotlik & Berrebi, 2001; Salzburger et al., 2003). To
date, only a few studies have addressed historical processes
using nuclear markers (Triantafyllidis et al., 2002; Makinen
et al., 2006) or used a comprehensive approach that combined
both highly polymorphic microsatellite and mitochondrial
markers (Gum et al., 2005; Barluenga et al., 2006; Bryja et al.,
2010a). Although mtDNA or allozymes were previously
considered to be almost ideal for genealogical and evolutionary
studies of animal populations (Avise et al., 1987; Avise, 1991),
microsatellites have, in the last two decades, become a more
convenient molecular tool for inferring finer levels of population
structure and dynamics for a species across its native
range (Estoup et al., 1998). First, the principal advantage of
microsatellites lies in their high level of polymorphism, high
mutation rate, and ability to distinguish fine-scale population
genetic structure. Second, microsatellites (as nuclear markers)
are biparentally inherited, whereas mtDNA is maternally
inherited without recombination and reflects only the matrilineal
history (Zhang & Hewitt, 2003). Third, studies concentrating
only on mtDNA have often been criticized because the
mitochondrial genome acts as a single genetic locus, providing
only a single ‘gene tree’ which might not accurately reflect the
‘organismal tree’ (Degnan, 1993). The accuracy of mtDNA
gene trees is compromised especially through hybridization
that may lead to the introgression of mitochondrial genomes,
which is not a rare process in nature (e.g. Bryja et al., 2010b;
Rodriguez et al., 2010). However, using only microsatellites
might not be optimal for phylogeographical inference mainly
due to fast mutation rates, complicated evolutionary relationships
among alleles, variable mutation rates among organisms
and even between varieties, and the questionable neutrality of
some microsatellite sequences (Zhang & Hewitt, 2003).
The aim of the present study was to clarify the evolutionary
history of S. cephalus in Europe by integrating phylogeographical
patterns of nuclear and mtDNA. We sampled 15
populations and assayed geographical variation across both
microsatellites (12 loci; 310 samples) and mitochondrial
sequence data [cytochrome b (cyt b); 75 samples]. We
expected that microsatellite markers would reveal a pattern of
variation on a more recent temporal scale, while analysis of
mtDNA sequences would provide a historical perspective
Squalius cephalus phylogeography in Europe
Journal of Biogeography 39, 1024–1040 1025
ª 2011 Blackwell Publishing Ltd
Paper 2.2.2 165
helpful for the interpretation of microsatellite results. In
particular, this study addressed the following three objectives:
(1) to re-analyse the phylogeographical structure of chub
populations in Europe, previously analysed using only
mtDNA sequences; (2) to compare the patterns of population
structure inferred from microsatellites and mtDNA to detect
possible past hybridization and admixture zones; and (3) to
investigate whether the phylogeographical distribution of
chub populations analysed using microsatellites is consistent
with the hypothesis of multiple-refugia origin, followed by
the differential post-glacial colonization of central and
northern Europe, as previously predicted for European
freshwater fishes. In addition, we included the populations
from Mediterranean areas (Adriatic, Iberian and Aegean
populations), where the taxonomy of S. cephalus and closely
related taxa is controversial (Kottelat & Freyhof, 2007). This
may eventually contribute to a taxonomic revision of the
Mediterranean chub.
MATERIALS AND METHODS
Sample collection and DNA extraction
A geographically and genetically representative total of 310
specimens of S. cephalus were sampled (Fig. 1) at 15 sites
across Europe by electro-fishing during the years 2004–06
(Table 1; for more details see Seifertova´ et al., 2008). In the
Mediterranean sampling areas, Kottelat & Freyhof (2007)
recently described several new species of the genus Squalius on
the basis of morphological differences. In this study, however,
we consider all collected individuals as S. cephalus in accordance
with the studies of Durand et al. (1999a) and Zardoya &
Doadrio (1999), and hereafter we use the term ‘S. cephalus
complex’ to strengthen its unclear taxonomical situation. Fin
clips were taken from each individual and kept in absolute
ethanol until the extraction of total genomic DNA using a
DNeasyÒ
Tissue Kit (Qiagen, Hilden, Germany).
Phylogeographical analysis of mtDNA
Amplification of an 840-bp fragment of cyt b was performed
using the primers LACB (Schmidt & Gold, 1993) and HBCB
(Dowling et al., 2002) in five individuals of each population.
Polymerase chain reaction (PCR) protocols and sequencing
procedures are described in Seifertova´ et al. (2008). Nucleotide
sequences obtainedfrom oursamples were aligned with available
cyt b sequences of European chub retrieved from GenBank (see
Appendix S1 in Supporting Information) in BioEdit 7.0.5.3.
(Hall, 1999) using ClustalW multiple alignment.
Phylogenetic analysis of mtDNA sequences was not performed
in the present study because such analyses have already
been described elsewhere (i.e. Durand et al., 1999a, 2000;
Seifertova´ et al., 2008). Haplotype relationships were obtained
with the median-joining (MJ) algorithm in the software
package network 4.5.1.6 (Bandelt et al., 1999) using an equal
transition/transversion ratio. The cyt b haplotypes and their
frequencies were identified using DnaSP 5 (Librado & Rozas,
2009). The frequencies of the published haplotypes were
retrieved from the original studies (Durand et al., 1999a,b).
Genetic diversity and historical demography of major
mtDNA lineages
DnaSP 5 software was used to estimate the genetic diversity of
the main phylogenetic lineages observed in this study in
S. cephalus, i.e. number of sequences (n), number of polymorphic
sites (Np), number of haplotypes (Nh), haplotype
diversity (Hd), nucleotide diversity (p, expressed as percentages,
i.e. 0.001 = 0.1%) and average number of nucleotide
differences (k).
Figure 1 Sampled localities and geographical
distribution of nuclear genetic variation in
European chub (Squalius cephalus). Pie diagrams
on the map correspond to the proportion
of chub population in each of seven
genetic clusters defined with the model-based
clustering method (for K = 7) implemented
in structure (Pritchard et al., 2000). The
size of circles represents the mean allelic
richness of 12 microsatellite loci corrected for
sample size.
M. Seifertova´ et al.
1026 Journal of Biogeography 39, 1024–1040
ª 2011 Blackwell Publishing Ltd
166 Paper 2.2.2
Furthermore, we analysed the population history of clades
by using an array of statistics that were originally introduced as
tests of neutrality. We used the D* and F* tests (Fu & Li, 1993)
in DnaSP 5 to analyse the population history of individual
lineages. The critical values obtained by simulations in Fu and
Li’s tests were used to determine statistical significance
(significant values at P < 0.05). Fu’s FS (Fu, 1997), one of
the most powerful neutrality tests for detecting expansion in
non-recombining genomic regions, and Tajima’s D (Tajima,
1989) were estimated in Arlequin 3.5 (Excoffier et al., 2005)
and their significance was tested using 1000 simulations. A
significant departure from the null hypothesis can be interpreted
as the result of demographic history (population
decrease or expansion), population structure and/or selection.
Negative values of Fu’s FS and Tajima’s D indicate population
size expansion, while significantly positive values of FS indicate
a deficit of rare haplotypes and significant positive values of
Tajima’s D often suggest an admixture (Rand, 1996).
The hypothesis of recent population growth was evaluated
in Arlequin 3.5 by calculating the mismatch distribution of
pairwise differences (Rogers & Harpending, 1992) between all
haplotypes within the main genetic lineages. A unimodal
distribution is assumed to be the signature of a recent
population size and, in a geographical context, range expansion,
in contrast to multimodal or ragged distribution, which
is typical in populations with stable demography (Rogers &
Harpending, 1992). A bimodal distribution often results from
the admixture of two previously separate lineages. We tested
whether the data fitted the sudden demographic expansion
model using Harpending’s raggedness index (rg; Harpending,
1994) and the sum of squared deviations (SSD) between the
observed and expected mismatch from 2000 parametric
bootstrap replicates.
Molecular dating
Molecular dating using Bayesian coalescent analysis requires
reliable time constraints within the ingroup for dating events
that are expected to be younger than 1–2 Ma, because of the
possible decay of the evolutionary rate during the time period
(Ho et al., 2005). In this study, we used three time constraints,
a fossil record, and two biogeographical assumptions based on
previous knowledge. The constraint based on the fossil record
was applied to the root of the tree. The lower limit was set to
0.7 Ma based on a S. cephalus fossil found in Germany, and the
upper limit was 6.56 Ma as dated for a S. aff. cephalus fossil
from Greece (Bo¨hme & Ilg, 2003). Two star-like patterns were
observed in the MJ network. One was found in the Dniester
drainage area and one in the Rhone drainage. Both regions
were previously identified as glacial refugia for chub (Durand
et al., 1999a). We assume that the star-like pattern observed in
our data represents a sudden population expansion following
climate amelioration. To provide for the uncertainty of the
evolutionary history, we included the Riss–Wu¨rm interglacial
and Wu¨rm glacial periods (0.01–0.13 Ma) as the time
constraint for sequences of the central haplotypes of the
respective stars and sequences derived from them by 1 bp.
The Bayesian coalescent analysis was run in beast 1.6.1
(Drummond&Rambaut,2007),usingallsequences generatedin
this study and related published sequences of the S. cephalus
species complex, which amounted to 300 sequences in total. All
time constraints were applied as priors with the uniform
distribution. The Markov chain Monte Carlo (MCMC) was
run for 10–20 million generations, sampling the chain every
1000th step. MCMC convergence was estimated from the
likelihood trace in Tracer 1.4 (Rambaut & Drummond, 2007)
and effective sample size for all parameters was > 100. We used
Table 1 Characteristics of Squalius cephalus samples in this study, including mitochondrial DNA (mtDNA), phylogenetic lineages, population
sampled, geographical location and hydrographic system, sample size (n) for microsatellite analyses, and symbols and numbers (in
parentheses) of mtDNA haplotypes (five individuals from each locality were sequenced).
mtDNA
lineage
Population
ID Location River Basin n mtDNA haplotypes
Eastern Mus Finland Mustijoki Baltic 21 H2 (5)
Ker Finland Keravanjoki Baltic 21 H2 (5)
Vis Poland Mroga/Vistula Baltic 21 H2 (5)
Western
Danube Odr Czech Republic Oder Baltic 20 H18 (5)
Elb Czech Republic Elbe North Sea 18 H16 (2), H21 (2), H23 (1)
Dan1 Czech Republic Jihlava/Danube Black Sea 19 H16 (2), H23 (1), H21 (1), H24 (1)
Dan2 Czech Republic Svitava/Danube Black Sea 19 H16 (2), H21 (2), H22 (1)
Dan3 Bulgaria Vidbol/Danube Black Sea 21 H16 (3), H17 (1), H18 (1)
Atlantic Rho1 France Le Buech/Rhone Mediterranean Sea 20 H12 (5)
Rho2 France L’Arc/Rhone Mediterranean Sea 19 H12 (5)
Aegean
Balkan Str Bulgaria Struma Aegean 23 H19 (4), H20 (1)
Spain Flu Spain Ser/Fluvia Mediterranean Sea 23 H25 (5)
Tor Spain Santa Coloma/Tordera Mediterranean Sea 21 H25 (5)
Adriatic Po Italy Ticino/Po Adriatic 22 H7 (1), H29 (3), H30 (1)
Omb Italy Merse/Ombrone Tyrrhenian Sea 22 H29 (2), H31 (2), H32 (1)
Squalius cephalus phylogeography in Europe
Journal of Biogeography 39, 1024–1040 1027
ª 2011 Blackwell Publishing Ltd
Paper 2.2.2 167
both a strict molecular clock and a relaxed clock model, with
lognormal distribution forconstant population andYuleprocess
tree priors. We compared the four alternative models with Bayes
factors, estimating marginal likelihoods for the models.
Microsatellite genotyping and intra-population
variability
A total of 310 samples were genotyped at 12 polymorphic
microsatellite loci (N7F8, N7G5, N7K4, LC03, LC04, E01,
LC93, LC32, LC293, LC27, LC166 and LC288). Details about
PCR conditions, primer sequences, fluorescent dyes and
fragment analysis are included in Vyskocˇilova´ et al. (2007).
The frequency of null alleles (NF) for each locus and population
was computed using Freena (Chapuis & Estoup, 2007).
Observed (HO) and non-biased expected heterozygosities
(HE; Nei, 1978) were calculated for each locus and for each
population using genetix 4.05.2 (Belkhir et al., 2004). genepop
3.4 (Raymond & Rousset, 1995) was used to test for
deviation from the Hardy–Weinberg equilibrium (HWE) for
each locus and population, and to test linkage disequilibrium
between all pairs of loci among populations and within each
population using the Fisher exact test. Probability values (P)
for each pairwise comparison were estimated by a Markov
chain method (1000 de-memorization, 100 batches and 1000
iterations). Significance levels of all multiple statistical tests
were corrected using the false discovery rate (FDR) approach
(Benjamini & Hochberg, 1995) implemented in the Qvalue
package of the software R (Storey et al., 2004). The allelic
richness (AR) corrected by the rarefaction method was
calculated using the fstat 2.9.3.2 program (Goudet, 1995)
in order to investigate differences in the number of alleles
among populations independent of sample size.
Population structure and differentiation
To quantify genetic structure, we calculated pairwise FST
ENA
values in the Freena software, using the so-called ENA
method, because null alleles are known to overestimate the
genetic differentiation between pairs of populations. This
method efficiently corrects FST estimates for the positive bias
introduced by the presence of null alleles (Chapuis & Estoup,
2007). Confidence intervals (95% CI) for mean F-statistics
were generated by bootstrap resampling across loci.
Considering that chub have a large geographical distribution
and long evolutionary history, the relevance of molecular
information of microsatellites was tested. We compared FST
and RST (Slatkin, 1995) values with the aim of analysing the
main causes of population differentiation in microsatellites
(i.e. whether the differentiation is caused by drift or by
mutations). The calculation of these values shares equal
assumptions when differentiation is caused solely by drift,
whereas RST is expected to be larger than FST under the
contribution of stepwise-like mutations (Hardy et al., 2003).
We used SPAGeDi 1.3 (Hardy & Vekemans, 2002) to test the
null hypothesis of there being no contribution of stepwise
mutations to genetic differentiation. We permuted allele sizes,
i.e. we produced the variable pRST, and compared it with the
real RST values also calculated in SPAGeDi. If the real RST is
significantly higher than pRST, stepwise mutations have played
a significant role in genetic differentiation. One thousand
random permutations of allele sizes provided a distribution of
pRST values, 95% CIs covering the 25th to the 975th ordered
values, and P-values testing if RST > pRST (Hardy et al., 2003).
The pattern of gene frequency differentiation between
populations was represented by a factorial correspondence
analysis (FCA) implemented in genetix 4.05.2 (Belkhir et al.,
2004). Additionally, the program structure 2.2 (Pritchard
et al., 2000) was used to perform an individual-based Bayesian
cluster analysis, which allowed us to infer and delineate the most
probable number of genetically homogeneous groups of
sampled individuals (K) with no a priori assumptions of
population structure, and to assign individuals to the inferred
groups. We performed a series of 10 independent runs for each
K ranging from 1 to 10 using 500,000 MCMC replications with a
burn-in period of 50,000 chains. Admixture models with
correlated allele frequencies were used. To infer the most
appropriate number of genetic groups in our dataset, we used
the ad hoc statistic DK method proposed by Evanno et al.
(2005). Additionally, we forced the assignment of individuals to
clusters at values of K beyond the number considered to
maximize the posterior probability of the data P(X|K). This
approach can be used to reconstruct the hierarchical relationships
among populations, as well as to distinguish between
processes that are likely to shape this structure (e.g. Flanders
et al., 2009; Bryja et al., 2010a). We utilized the program
clumpp 1.1.1 (Jakobsson & Rosenberg, 2007) in order to assess
the average pairwise similarity (H) of runs with the Greedy
algorithm and 10,000 random input orders of the 10 independent
structure runs. Results from clumpp were imported into
distruct 1.1 (Rosenberg, 2004) for graphical representation.
RESULTS
Phylogeographical structure and historical
demography based on mtDNA
Using 75 individuals from 15 populations sequenced for
mtDNA, 18 different cyt b haplotypes of 840 bp were identified
(GenBank accession numbers EU791864–EU791885). Detailed
information on the cyt b sequences is included in Seifertova´
et al. (2008). To perform phylogeographical and historical
demography analyses, we completed our dataset of cyt b with
24 sequences retrieved from GenBank (AJ006887–AJ006902,
AJ002319, AJ002342, AJ002343, AJ002344, AJ002346 and
AJ002347) and corresponding frequency data from Durand
et al. (1999a,b). A total of 275 individuals (Appendix S2) were
used for network calculation and the final dataset comprised
33 haplotypes of 600-bp cyt b sequences with 91 polymorphic
sites (Appendix S1). No indels or stop codons were found,
suggesting that no nuclear copies of cyt b were included.
The MJ network showed a clear split into four different
M. Seifertova´ et al.
1028 Journal of Biogeography 39, 1024–1040
ª 2011 Blackwell Publishing Ltd
168 Paper 2.2.2
haplogroups (Fig. 2), which correspond to four previously
described phylogenetic lineages, i.e. Western, Eastern, Aegean
and Adriatic (sensu Durand et al., 1999a). Two of these
lineages can be further subdivided to sublineages that are
geographically separated, increasing the number of mtDNA
clusters to six. First, haplotype 25 (yellow in Fig. 2; hereafter
called Aegean–Spain) of the Aegean lineage was observed in
the Ebro, Fluvia and Tordera rivers in Spain, while remaining
haplotypes of this lineage (violet in Fig. 2; Aegean–Balkan)
occur in the Evros and Struma rivers in the Balkan Peninsula
(Fig. 2). Second, the Western lineage can be divided into
Western–Atlantic (grey in Fig. 2) and Western–Danubian
sublineages (blue in Fig. 2). The distribution of these two
sublineages overlaps in the Elbe and Rhine rivers (Fig. 2). Two
common haplotypes (H2 from the Eastern lineage and H12
from the Western–Atlantic lineage) were found in 142 out of
275 (51.64%) chub individuals. As a consequence, the
Western–Atlantic lineage and partially also the Eastern lineage
displayed a star-like pattern of haplotype distribution, indicating
the recent descent of the satellite haplotypes from the
ancestral haplotypes H12 and H2 (Fig. 2).
Measures of mitochondrial DNA diversity, estimates of
demographic parameters and neutrality tests calculated for each
lineage are shown in Table 2. Because the Aegean lineage
comprised the haplotypes from the Aegean area (Aegean–Balkan
sublineage) and also from the Iberian area (Aegean–Spain
sublineage), the analyses were performed separately for both
sublineages. The highest values of diversity were observed for the
Adriatic lineage. The Western–Atlantic sublineage showed
significant negative D* and F* values as well as a large and
significant negative value of FS (Table 2). The mismatch
distributions for Eastern (Fig. 3a), Western–Danubian (Fig. 3b)
and Western–Atlantic lineages (Fig. 3c) were unimodal, which
is a typical sign of populations having undergone a recent
expansion. A bimodal pattern for the Aegean–Balkan lineage
(Fig. 3d) was observed, which may suggest the admixture of two
previously separate lineages in the Aegean region (Valqui et al.,
2010). A ragged pattern of mismatch distribution was observed
for the Adriatic lineage (Fig. 3e), indicating that populations are
at demographic equilibrium (Rogers & Harpending, 1992).
The model with a relaxed uncorrelated molecular clock with
lognormal distribution and Yule process tree prior had the
highest likelihood amongst the tested models. Comparison of
marginal likelihoods with Bayes factors did not strongly favour
the model over the one assuming the same clock and constant
population size tree prior (Bayes factor = 5.5). We further
used the latter model to avoid overparameterization. The
median mutation rate with this model was 0.0631 mutations
per lineage per site per Myr, which represents a sequence
divergence of c. 13% Myr)1
. The most recent common
ancestor of the paraphyletic Western–Danubian lineage was
0.16 Ma [95% highest posterior density (HPD) 0.03–0.36].
The Western–Atlantic lineage was included within the group
and was 0.04 Ma (HPD: 0.01–0.09). The Eastern lineage was
0.09 (HPD: 0.02–0.19) Ma, the Aegean–Spain 0.03 Ma (HPD:
0.003–0.09), the Aegean–Balkan 0.14 Ma (HPD: 0.02–0.33)
and the Adriatic lineage 0.18 Ma (HPD: 0.03–0.39) (Fig. 4).
The time of population expansion, estimated from the s
parameter of the mismatch distribution analysis (t = s/
2llocus), was 0.04 Ma (95% CI: 0.005–0.05) for the Eastern
lineage, 0.02 Ma (95% CI: 0.01–0.03) for the Western–
Danubian lineage, and 0.04 Ma (95% CI: 0.006–0.04) for the
Figure 2 Network constructed using mitochondrial DNA cytochrome b gene sequences of European chub (Squalius cephalus). The
haplotypes are represented by circles, with diameters proportional to their frequencies (for detailed information see Table 1). The haplotypes
and rivers included in the sampling within this study are shown in bold italic. Branch lengths correspond to the number of mutation steps
between haplotypes. In the case of long branches, the number of mutation steps is represented by a number above the branches. Sequences
obtained from GenBank are included in the network with their respective frequencies (as given in the original publication; see the text for
more details). Colour legend of mtDNA lineages: orange, Eastern; blue, Western–Danubian; grey, Western–Atlantic; violet, Aegean–Balkan;
yellow, Aegean–Spain; green, Adriatic.
Squalius cephalus phylogeography in Europe
Journal of Biogeography 39, 1024–1040 1029
ª 2011 Blackwell Publishing Ltd
Paper 2.2.2 169
Western–Atlantic lineage. The remaining lineages did not
exhibit the signal of a single recent population expansion in the
mismatch analyses.
Genetic diversity and structure assessed
by microsatellites
A total of 257 different alleles were detected at the 12
microsatellite loci in the 310 chub individuals (2–52 alleles per
locus; Appendix S2). The non-random association of genotypes
at two loci (N7F8, LC93) in the Flu locality at the 0.05
level after FDR correction was found. The mean frequencies of
null alleles per population were lower than 5% in all but one
locus (the only exception was locus E01 with an average of
6.61% of null alleles per population; Appendix S2). A high
frequency of null alleles (> 15%) was detected only in the Elb
(15.8%), Rho1 (22.9%) and Rho2 (33.6%) populations at the
locus E01 and in the Omb population (17.1%) at the locus
Table 2 Measures of mitochondrial DNA diversity, estimates of demographic parameters and neutrality tests calculated for four lineages of
Squalius cephalus identified in European rivers in this study. The results of the observed mismatch distribution against a sudden expansion
distribution included the raggedness rg statistic and the sum of squared deviations (SSD).
Eastern Western–Danubian Western–Atlantic Aegean–Balkan Aegean–Spain Adriatic
n 79 51 83 19 12 31
Np 6 6 5 13 0 20
Nh 4 7 6 5 1 10
Hd (SD) 0.330 (0.063) 0.740 (0.035) 0.118 (0.048) 0.637 (0.105) – 0.862 (0.034)
p (SD) 0.170 (0.038) 0.193 (0.020) 0.020 (0.008) 0.715 (0.152) – 0.999 (0.074)
k 1.020 1.161 0.120 4.292 – 5.983
D* 1.139 )1.414 )4.137** (P = 0.002) 0.727 – 0.370
F* 0.757 )1.256 )4.027** (P = 0.002) 0.788 – 0.551
FS 1.329 )1.445 )8.450*** (P < 0.001) 3.126 – 1.232
D )0.375 )0.329 )1.934* (P = 0.03) 0.566 – 0.675
SSD 0.057 0.006 0.0002 0.105 – 0.021
rg 0.411 0.087 0.602 0.189 – 0.037
n, number of sequences; Np, number of polymorphic sites; Nh, number of haplotypes; Hd, haplotype diversity; p, nucleotide diversity (expressed as
percentages, i.e. 0.001 = 0.1%); k, average number of nucleotide differences; SD, standard deviation; FS, Fu’s statistic; D, Tajima’s D-test. The
significant tests are shown with asterisks as follows: *P < 0.05, **P < 0.01, ***P < 0.001.
Figure 3 Mismatch distribution among mitochondrial DNA (mtDNA) haplotypes from five major Squalius cephalus mtDNA lineages
observed in Europe: (a) Eastern lineage, (b) Western–Danubian lineage, (c) Western–Atlantic lineage, (d) Aegean–Balkan lineage, (e)
Adriatic lineage. The expected frequency is based on the sudden expansion model. Grey bars indicate the observed values and black lines
show the expected distribution under the sudden expansion model.
M. Seifertova´ et al.
1030 Journal of Biogeography 39, 1024–1040
ª 2011 Blackwell Publishing Ltd
170 Paper 2.2.2
LC166. Deviation from HWE expectation was found for the
three populations Dan3, Rho1 and Rho2, which could be the
result of the presence of null alleles at locus E01 in the Rho1
and Rho2 (see above) populations and at locus LC32 in the
Dan3 population (14.4%). When these two loci were removed
from the dataset, no deviation from HWE for these populations
was observed. The mean observed and expected heterozygosities
and allelic richness estimated by the rarefaction
method for the smallest sample size (17 diploid individuals)
for each population are shown in Table 3. Mean AR per locus
was highly correlated with HE (Spearman correlation,
r = 0.964, P < 0.001). The highest values of genetic diversity
were found in the populations of Adriatic (Po and Omb) and
Western–Danubian lineages (Elb, Odr, Dan1, Dan2 and Dan3)
(Fig. 1).
Microsatellites revealed a considerable level of genetic
differentiation over all populations (global FST
ENA
= 0.287;
P = 0.0001, 95% CI = 0.224–0.373). Pairwise FST
ENA
values
ranged from 0.017 (Dan2 and Elb; Dan1 and Dan2) to 0.560
(Mus and Tor) and most of them (94.3%) were significantly
higher than zero, suggesting a distinct genetic structure of
European chub populations (Table 4). The multilocus RST
value was significantly higher than mean pRST. This difference
was also significant at 5 out of 12 loci when analysed separately
per locus (Appendix S2), suggesting the important role of
stepwise mutations in population differentiation. When we
analysed differences between pairwise RST and pairwise pRST,
we found important effects of mutations only in pairs of
population samples originating from different mtDNA lineages
(six groups were considered; Eastern, Western–Danubian,
Western–Atlantic, Adriatic, Aegean–Balkan, Aegean–Spain).
This difference was significant in 49 out of 89 comparisons
(55%). However, none of the 16 pairwise comparisons within
mtDNA lineages showed significant differences, suggesting that
genetic differences within lineages were caused predominately
by drift (Table 4).
Using a FCA of microsatellite genotypes, the sampled
individuals were clearly separated into the four different
groups with almost no overlap (Fig. 5a). Eastern populations
(i.e. Mus, Ker and Vis) were separated from all remaining
samples on the first axis (explaining 21.06% of total variance).
Two Iberian populations (i.e. Flu and Tor) and the Bulgarian
population from the Aegean basin (i.e. Str) were separated
from the remaining populations on the second axis (explaining
21.06% of variance). The Czech (Odr, Elb, Dan1, Dan2),
Italian (Po, Omb), French (Rho1, Rho2) and Bulgarian
Danubian (Dan3) populations clustered together in the
analysis using the whole dataset. The third axis (14.84% of
variance) only deepened the differences among the abovedescribed
groups and did not yield the separation of further
populations (not shown). When we performed more detailed
analysis using only the samples from the last-mentioned large
group, three clusters became visible, i.e. Italian, French and
Czech+Bulgarian (Fig. 5b).
Bayesian analyses in structure based on the highest DK
value, i.e. ln P(X/K), revealed seven population clusters
(Appendix S3). The seven clusters mostly corresponded to
geographical areas, with the exception of two closely situated
Bulgarian localities (Dan3 and Str), which formed separate
clusters (Figs 1 & 6). When analysing the individual assignments
beyond the best K, we found that the first split (K = 2)
Figure 4 Chronogram of the European chub (Squalius cephalus)
cytochrome b sequences inferred from the Bayesian coalescent
analysis assuming an uncorrelated relaxed clock with lognormal
distribution and constant population size. Node bars represent the
95% highest posterior density intervals of node ages; collapsed
lineages are colour-coded according to Fig. 2. Sequences not
included in collapsed lineages were downloaded from GenBank.
The scale bar is calibrated to million years ago (Ma).
Table 3 The measures of microsatellite genetic diversity in 15
populations of Squalius cephalus in European rivers.
Population HO HE AR
Mus 0.413 0.402 3.441
Ker 0.500 0.509 3.776
Vis 0.524 0.478 3.041
Odr 0.612 0.635 7.293
Elb 0.694 0.729 8.530
Dan1 0.679 0.707 8.585
Dan2 0.617 0.662 7.670
Dan3* 0.647 0.688 8.453
Str 0.601 0.589 6.509
Po 0.708 0.716 8.888
Omb 0.687 0.702 8.902
Rho1* 0.491 0.533 3.857
Rho2* 0.535 0.574 4.722
Flu 0.522 0.519 3.803
Tor 0.258 0.283 2.864
Mean 0.566 0.582 6.022
HO, observed heterozygosity; HE, expected heterozygosity, which represents
a non-biased estimate according to Nei (1978); AR, mean allelic
richness per locus.
*Significant departure (P < 0.05) from the Hardy–Weinberg equilibrium
(after false discovery rate correction for multiple comparisons).
Squalius cephalus phylogeography in Europe
Journal of Biogeography 39, 1024–1040 1031
ª 2011 Blackwell Publishing Ltd
Paper 2.2.2 171
separated eastern populations (Mus, Ker and Vis) from all
others; with increasing K (from 2 to 10) the populations from
Mediterranean areas were successively assigned to separate
groups (Fig. 6). The Iberian populations (Flu and Tor) already
started to separate at K = 2. The southern Bulgarian population
from the Struma Basin (Str) formed a cluster with the
Italian populations (Po and Omb) at K = 4 and 5, whereas this
Bulgarian population was separated from Italian populations
and formed a separate group at K = 6. The French populations
(Rho1 and Rho2) were separated from the Czech populations
(Odr, Elb, Dan1 and Dan2) and Bulgarian population (Dan3)
at K = 5. The Bulgarian population from the lower Danube
(Dan3) formed a single cluster at K = 7. Further increasing K
up to 10 did not introduce new information, except partial
separation of the population from the Oder River (Odr) at
K = 8.
Neither FCA nor structure allowed us to correct efficiently
for null alleles at microsatellite loci, but both analyses also
provided very similar results when run using datasets in which
the loci with null alleles (E01 and LC166) were excluded.
DISCUSSION
In the present study, the range-wide genetic diversity and
population structure of S. cephalus were, for the first time,
investigated using a combination of mtDNA and nuclear
markers. Comparison of the phylogeographical history of chub
(estimated by cytochrome b sequences) with its recent
population structure (assessed by nuclear microsatellites)
allowed us to obtain a better estimate of the chub’s
evolutionary history. A notable result of our study is evidence
of at least seven genetically separated subpopulations of
S. cephalus in Europe. Their genetic differentiation may be
Table 4 Pairwise FST
ENA
values (below the diagonal) and pairwise RST values (above the diagonal, ANOVA approach) calculated for all
microsatellite loci across European populations of Squalius cephalus.
Mus Ker Vis Odr Elb Dan1 Dan2 Dan3 Str Po Omb Rho1 Rho2 Flu Tor
Mus 0.138 0.124 0.323 0.303 0.364 0.387 0.318 0.500 0.602 0.541 0.622 0.675 0.724 0.885
Ker 0.133 0.261 0.417 0.415 0.464 0.482 0.392 0.590 0.680 0.626 0.707 0.749 0.781 0.909
Vis 0.181 0.148 0.308 0.256 0.306 0.363 0.323 0.309 0.528 0.500 0.516 0.562 0.609 0.764
Odr 0.400 0.358 0.344 0.006 0.049 0.006 )0.003 0.377 0.274 0.254 0.246 0.244 0.424 0.666
Elb 0.340 0.296 0.292 0.034n.s.
0.008 0.016 0.023 0.347 0.237 0.225 0.220 0.235 0.384 0.651
Dan1 0.377 0.332 0.330 0.042 0.018n.s.
0.006 0.063 0.338 0.100 0.126 0.098 0.102 0.228 0.489
Dan2 0.403 0.358 0.357 0.029 0.017n.s.
0.017n.s.
0.014 0.401 0.167 0.162 0.178 0.171 0.360 0.626
Dan3 0.394 0.349 0.363 0.073 0.035n.s.
0.039 0.045 0.363 0.296 0.283 0.309 0.310 0.470 0.690
Str 0.463 0.407 0.428 0.302 0.252 0.262 0.281 0.269 0.561 0.586 0.591 0.631 0.645 0.791
Po 0.384 0.336 0.357 0.135 0.094 0.116 0.121 0.094 0.236 0.034 0.105 0.104 0.217 0.488
Omb 0.371 0.330 0.345 0.127 0.087 0.107 0.111 0.087 0.249 0.028 0.103 0.107 0.255 0.521
Rho1 0.468 0.413 0.418 0.167 0.143 0.150 0.159 0.175 0.359 0.203 0.203 )0.007 0.180 0.563
Rho2 0.447 0.391 0.393 0.132 0.119 0.130 0.129 0.157 0.336 0.181 0.177 0.022n.s.
0.239 0.660
Flu 0.441 0.392 0.410 0.267 0.216 0.229 0.247 0.261 0.413 0.273 0.275 0.338 0.325 0.262
Tor 0.560 0.504 0.529 0.451 0.397 0.404 0.432 0.436 0.545 0.429 0.433 0.529 0.510 0.149
For FST
ENA
values, a value significantly different from zero was tested by bootstrapping over loci. n.s., not significant (P > 0.05) after false discovery rate
correction. Significant differences between pairwise RST and pairwise pRST (i.e. important effect of mutations on population differentiation) are shown in
bold. FST
ENA
, a measure of population differentiation corrected for the presence of null alleles (Chapuis & Estoup, 2007); RST, allele size-based measure of
population differentiation assuming stepwise mutation process (Slatkin, 1995); pRST, RST computed after allele size permutation (Hardy et al., 2003).
Figure 5 Two-dimensional factorial correspondence analysis
(FCA) of nuclear microsatellite variation: (a) analysis using the
whole European chub (Squalius cephalus) dataset (the first axis
explains 21.06% of total variance and the second axis explains
17.66% of total variance); (b) detailed analysis using only the subset
of individuals clustered in Group 1 in (a) (the first axis explains
29.74% of total variance and the second axis explains 24.78% of total
variance). Different symbols represent main genetic groups.
M. Seifertova´ et al.
1032 Journal of Biogeography 39, 1024–1040
ª 2011 Blackwell Publishing Ltd
172 Paper 2.2.2
caused by a combination of different factors, i.e. post-glacial
colonization from different refugia or recent evolutionary
processes such as random drift or isolation by distance.
Population-genetic structure of chub in Europe –
comparison of mtDNA and nuclear markers
Hitherto, there have been few studies of European freshwater
fish species using both nuclear and mitochondrial markers to
investigate the population structure from the phylogeographical
perspective (e.g. Gum et al., 2005; Barluenga et al., 2006;
Bryja et al., 2010a). Using only mitochondrial markers can lead
to biased estimates of species phylogeny (because of sampling
only one locus), while using both microsatellites and mtDNA
sequences allows us to obtain a more precise view of species
history. Thus, this latter approach is recommended in order to
increase the performance of phylogenetic studies (e.g. Rodriguez
et al., 2010).
Our sampling covered most of the geographical area of
chub distribution in Europe. In accordance with previous
studies (Durand et al., 1999a, 2000), our mtDNA analysis,
including both published cyt b sequences and those obtained
by this study, confirmed the existence of four main
mitochondrial lineages of S. cephalus, i.e. Western, Eastern,
Adriatic and Aegean lineages. Network analysis further
divided two of them into geographically distinct sublineages:
the Western lineage into Atlantic (occurring mainly in France
and Germany) and Danubian (occurring mainly in the
Danube Basin, the Elbe and the Oder) sublineages; and the
Aegean lineage into Balkan (the Evros and Struma rivers) and
Spanish (rivers in northern Spain) sublineages, increasing the
number of mtDNA groups to six. Likewise, microsatellite
analysis revealed the strong genetic differentiation of chub
populations even on a relatively small geographical scale,
which is, in most cases, in congruence with mitochondrial
data (e.g. the strict separation of Danube and Struma
populations in Bulgaria). In addition, microsatellites, due to
their high variability and fast mutation rate, showed the
substantial divergence between hydrologically connected
populations. This seems to be a common feature of European
freshwater species, and it suggests the important role of
genetic drift or local selection in shaping the genetic structure
of non-migratory fishes (Triantafyllidis et al., 2002; Gum
et al., 2005; Bryja et al., 2010a). Almost all pairwise comparisons
of populations showed highly significant values
(P < 0.001) of FST. The only exceptions were several pairs
Figure 6 Bayesian population assignments of 310 European chub (Squalius cephalus) individuals using 12 microsatellite loci considering a
given K (from 2 to 10) in structure. Black lines separate individuals sampled from different geographical locations. The phylogenetic tree
indicates a schematic topology of mitochondrial DNA sequences based on Bayesian inference in beast.
Squalius cephalus phylogeography in Europe
Journal of Biogeography 39, 1024–1040 1033
ª 2011 Blackwell Publishing Ltd
Paper 2.2.2 173
Table5OverviewoftheresultsofselectedphylogeographicalstudiesoffreshwaterEuropeanfishspecieswithsimilarenvironmentalpreferencesanddistributioninEurope.
SpeciesReferencesMarkerPhylogeographicalstructureSuggestedrefugia
Europeanchub(Squalius
cephalus),Cyprinidae
Durandetal.(1999a,
2000),thisstudy
mtDNA,
microsatellites
1.Westernlineage(two-stepexpansionscenario)SoutherntributariesoftheDanube(secondaryAtlantic
refugiuminRhine/Rhonedrainages)
2.EasternlineagePeripheryoftheBlackandCaspianseas
3.AegeanlineageEasternGreece(Aegeanrivers)
4.AdriaticlineageAdriaticsideoftheBalkans
Europeanbitterling
(Rhodeusamarus),
Cyprinidae
Bohlenetal.(2006),
Bryjaetal.(2010a)
mtDNA,
microsatellites
1.Danubiancluster(Westernlineage)RefugiuminthelowerDanube
2.Dnieper–Dniester–Vistulacluster(Easternlineage)NorthoftheBlackSea(colonizationofNEEurope)
3.AegeanclusterSeveralrefugiaaroundtheAegeanSea
Varione(Telestessouffia),
Cyprinidae
Salzburger
etal.(2003)
mtDNA1.ItalianlineageRefugiuminItaly
2.Alpinelineage(Rhone/VarandDanube/Rhine
sublineages)
Danubianrefugium
Barbel(Barbusbarbus),
Cyprinidae
Kotlik&Berrebi
(2001)
mtDNA1.Lineage:EBulgariaandNAnatoliaN/WBlackSearefugium
2.Lineage:WandcentralEurope(two-step
expansionscenario)
Danubianrefugium(secondaryAtlanticrefugiumin
southernFrance–Rhine/Rhone)
Europeangrayling
(Thymallusthymallus),
Salmonidae
Gumetal.(2005),
Gum(2009)
mtDNA,
microsatellites
1.LineageN/NEEuropeNorthoftheBlackorCaspianSeabasin
2a.LineageSW/W–Rhine/Maindrainage
andEngland
PersistencewithintheRhine/Mainsystemduring
thelastmajorPleistoceneglaciation
2b.CentralandEEuropefromElbedrainagetothe
Vistula,inSScandinaviaandLithuania
Ponto-Caspianrefugium
3.DanubedrainagesDanubianrefugium
4.AdriaticlineageAdriaticrefugium
Europeancatfish(Silurus
glanis),Siluridae
Kriegetal.(2000),
Triantafyllidisetal.(2002)
mtDNA,
microsatellites
NopatternofgeographicalstructuringOneglacialrefugiumaroundthePonto-Caspianregion
(colonizationofCentralandSEurope)
Spinedloach(Cobitis
taenia),Cobitidae
Cullingetal.(2006)mtDNA1.Lineage:N/WEuropeThreerefugesinthePonto-Caspianarea:
1.colonizationofEuropewestwardtoEngland
2.Lineage:N,EandN/WEurope2.ColonizationofEuropespreadingnorthwardinto
Russiaandmovingwest
3.Lineage:lowersouthernBugRiver3.NearbyBlackSearefugiumdidnotcontributeto
thecolonizationofEurope
Europeanperch(Perca
fluviatilis),Percidae
Nesboetal.(1999)mtDNA1.Lineage:SEuropeDanubianrefugium
2.Lineage:EEuropeRefugiuminSEEurope–Black/CaspianSeadrainages
3.Lineage:WEuropeandPolandRefugiuminWEuropeanRivers–positionofthis
refugiumisunclear
4.Lineage:NorwayRefugiuminNEEurope–ice-dammedlakeseastof
theicesheet
Bullhead(Cottusgobio),
Cottidae
Englbrechtetal.(2000),
Ha¨nflingetal.(2002)
mtDNA,
microsatellites
1.Lineage:Danube,Elbe,Rhine,MainDanubianrefugium
2.Lineage:Oder,VistulaPonto/Caspianrefugium
3–7.Lineage:LowerRhine,Seine,Adour,EnglandPersistencewithintheBritishIslesandwithinthe
drainagesoftheRiversElbeandMain
mtDNA,mitochondrialDNA.
M. Seifertova´ et al.
1034 Journal of Biogeography 39, 1024–1040
ª 2011 Blackwell Publishing Ltd
174 Paper 2.2.2
of central European populations, which could be explained
by short divergence time and/or by short geographical
distances allowing gene flow among them. Based on the
allele permutation test (Hardy et al., 2003), we indicated that,
in most cases, RST provides a better description of population
structure than FST. This result suggests that genetic differences
among basins may be caused also by mutations at
microsatellite loci, while the differences within basins are
caused predominately by drift.
The analysis of population genetic structure based on nuclear
markers revealed seven major groups of chub populations in
Europe, which is in line with the mitochondrial evidence
supporting the multiple-refugia assumption (Durand et al.,
1999a, 2000). However, when using Bayesian analysis on
microsatellite data in structure, the pattern observed for
K = 4 (Fig. 5) was not completely congruent with the separation
of chub populations into four main mtDNA phylogenetic
lineages, i.e. the Aegean population from the Struma River was
associated with geographically close populations of the Adriatic
region, while Iberian populations belonging to the Aegean
mtDNA lineage formed a separate group. At K = 6, however,
the structure model showed exactly the same differentiation
as the network analysis of mtDNA. The best model of genetic
structure using microsatellites (K = 7) additionally showed the
subdivision of Danube Basin populations into two groups (i.e.
central European and Balkan) (Fig. 1).
Roughly congruent phylogeographical patterns in mtDNA
and nuclear markers were also found in other freshwater fishes,
e.g. European grayling (Gum, 2009), bullhead (Englbrecht
et al., 2000; Ha¨nfling et al., 2002) and burbot (Barluenga et al.,
2006). However, contrasting results between both marker types
have also occasionally been found (e.g. Bryja et al., 2010a).
This discordance may indicate past evolutionary processes
undetectable with mtDNA, such as hybridization among
highly divergent lineages in contact zones, perhaps resulting
in mtDNA introgression or replacement.
Inferences on the evolutionary history of Squalius
cephalus in Europe and the phylogeography of chub
populations in the Mediterranean region
Primary freshwater fishes, due to their dependence upon water
routes, most reliably reflect the historical geographical attributes
of a region, including connections among hydrographic basins,
and the process of isolation and interconnection among rivers
and lakes (Levy et al., 2009). On the basis of the comprehensive
phylogenetic and biogeographical analyses of the Leuciscinae
subfamily performed by Perea et al. (2010), the chub (S. cephalus)
belongs to the Euroasiatic (Central European and Anatolian)
group of the genus Squalius. The diversification of this
group is dated in the Upper Miocene (9.16 Ma) and the
existence of the four contemporary phylogenetic lineages
Figure 7 Geographical location of sampling sites and schematic overview of the refugia suggested for the European chub (Squalius
cephalus). Arrows indicate possible colonization pathways. Symbols refer to the affiliation of haplotypes of a population to one of the
mitochondrial DNA lineages. PC, Ponto-Caspian refugium; D, Danubian refugium; Ad, Adriatic refugium; Ae, Aegean refugium; At,
secondary Atlantic refugium. The presence of a possible Iberian refugium is indicated by a question mark.
Squalius cephalus phylogeography in Europe
Journal of Biogeography 39, 1024–1040 1035
ª 2011 Blackwell Publishing Ltd
Paper 2.2.2 175
of chub in Europe is explained as the result of an ancient
allopatric fragmentation that occurred during the late Pliocene
(3–2.5 Ma) (Durand et al., 2000). Therefore, the current
distribution and phylogenetic structure of chub in Europe
may be connected with several historical events, such as the
Messian Lago Mare stage, the rise of humidity at the beginning
of the Pliocene in the Mediterranean region, and the glaciations
and uplift of the Alps during the Pleistocene (Perea et al., 2010).
Nevertheless, it is generally accepted that the phylogeographical
structure of widely distributed European freshwater fish species
is a result of Pleistocene glacial periods. During the extremely
cold last glaciation (115–10 ka), most central and western
European basins were covered by ice. Consequently, most
freshwater fish species disappeared from these areas and
survived in several different refugia, including the lower
Danube, around the Black Sea, and in the Aegean region (see
Table 5 summarizing phylogeographical studies of freshwater
fish species). The subsequent colonization of central and western
Europe from these refugia explains the recent origin and
expansion of lineages in central and western Europe. However,
these lineages could not have reached the Mediterranean
peninsulas – probably as a result of the uplift of the Alps during
the Pleistocene – explaining the divergence of Mediterranean
populations. It is highly probably that Alpine mountains
prevented Mediterranean populations from moving northwards;
therefore, the populations from the Mediterranean area
did not contribute to the post-Pleistocene colonization of nonMediterranean
Europe (Durand et al., 1999a; Perea et al., 2010).
The evolutionary history of the European chub is consistent
with this scenario. Being unable to survive in periglacial areas
of Europe, the species colonized the northern part of its
distribution recently. Our network analysis indicated at least
two foci that formed star-like phylogenies, typical for sudden
population expansion, such as those that would accompany
post-glacial colonization. Additionally, both nuclear and
mitochondrial markers showed the high divergence of Mediterranean
(Adriatic, Aegean–Balkan, and Aegean–Spain) lineages
of chub from non-Mediterranean lineages. This indicates
their long independent evolutionary history linked to the
different geological history of freshwater bodies in the
Mediterranean region. Likewise, we observed high nucleotide
diversity for both lineages, which is typical for freshwater
species living in non-glaciated areas (Durand et al., 1999a) and
long-term demographic stability (i.e. multimodal mismatch
distribution).
Durand et al. (1999a) proposed the existence of two
southern refugia for chub in the Mediterranean area, i.e. the
Adriatic side of the Balkans and Aegean rivers (Fig. 7). As was
previously mentioned, microsatellite markers showed slightly
inconsistent results regarding the relationships among Mediterranean
populations contrary to mtDNA. This could contribute
to a better understanding of the evolutionary history of
chub in this region. First, the link between the Struma
population belonging to the Aegean lineage and the populations
of Adriatic lineage revealed by nuclear markers in this
study (see structure analysis for K = 4; Fig. 6) may indicate
the location of the refugium for the Italian populations on the
Adriatic side of the Balkans (sensu Durand et al., 1999a) or
may be a result of trans-Adriatic exchanges between the
western Balkans and eastern Italy during the Wu¨rmian
regression (Fig. 7) (Levy et al., 2009). Generally, the ichthyofauna
of northern Italy and the northern rivers of the Balkan
Peninsula show high affinities, which is explained by more
recent events, such as expansion of the Po River during the
Last Glacial Maximum (c. 20 ka) as a result of a drop in sea
level (Perea et al., 2010). Second, the close phylogenetic affinity
of Iberian populations of chub to the Aegean lineage was
revealed by analyses of mitochondrial cyt b (Zardoya &
Doadrio, 1999; Durand et al., 2000; Sanjur et al., 2003;
Seifertova´ et al., 2008), but strong differentiation of Iberian
populations at nuclear microsatellites may indicate their
independent histories as well as the possible existence of a
Pleistocene glacial refugium in the Iberian Peninsula. The fact
that the Aegean lineage has the oldest most recent common
ancestor further supports its complicated history, which may
include, for example, strong founder events followed by lineage
sorting. It could also be the result of ancient introgression or
hybridization with several highly endemic Squalius species,
mostly with parapatric distributions in this region, and of very
limited connections between adjacent areas in France where
the Western–Atlantic mtDNA lineage is distributed (Fig. 1).
This supports the study of Durand et al. (2000), which
suggested that hybridization between the chub and endemic
Squalius species restricted to Mediterranean areas may be
considered a widespread evolutionary phenomenon. Therefore,
additional analyses of different DNA markers are
necessary in the future to elucidate the phylogenetic relationships
between the Iberian and Aegean populations.
Biogeography of Central/Western/Eastern European
populations and post-glacial colonization of
non-Mediterranean Europe by chub
The general scenario of the colonization of non-Mediterranean
Europe by freshwater fishes comprises two main routes, i.e. the
‘eastern colonization route’ including the northward spread of
lineages from the Ponto-Caspian refugia, and the ‘western
colonization route’ often occurring in two waves (Durand
et al., 1999a; Kotlik & Berrebi, 2001; Bryja et al., 2010a; review
in Table 5). Partitioning of the Eastern populations of
European chub in both analyses (mtDNA and microsatellites)
clearly confirmed the survival of the Eastern lineage during
Pleistocene glaciations in a separate refugium. Durand et al.
(1999a) postulated the location of this refugium to be on the
periphery of the Black and Caspian seas (Fig. 7), whence this
Eastern lineage probably entered the Baltic areas as far as the
Oder and northern Elbe during the Holocene. Low mtDNA
and microsatellite variability, partial star-like phylogeny with a
young most recent common ancestor, and unimodal mismatch
distribution observed for the Eastern lineage reflects the recent
origin of the colonization of the Baltic drainages. The
Ponto-Caspian region was also identified as a separate
M. Seifertova´ et al.
1036 Journal of Biogeography 39, 1024–1040
ª 2011 Blackwell Publishing Ltd
176 Paper 2.2.2
refugium for other freshwater fishes with distribution and
environmental preferences similar to chub, e.g. spined loach
(Cobitis taenia), grayling (Thymallus thymallus), bitterling
(Rhodeus amarus), perch (Perca fluviatilis) and catfish (Silurus
glanis) (Table 5).
Likewise, confirmation of the existence of the Western–
Danubian lineage by both markers is in accordance with the
previously reported importance of the Danubian refugium
(lower Danube) for the survival of European freshwater fish
fauna during glacial periods, e.g. barbell (Barbus barbus),
varione (Telestes souffia), bitterling, perch and grayling
(Table 5, Fig. 7). However, not all freshwater fish had refugia
in the lower Danube area. For example, recent dispersion from
a single glacial refugium around the Ponto-Caspian region is
proposed for the European catfish, and several refugia in the
Ponto-Caspian area are proposed as sources for the colonization
of central Europe by the spined loach, as the Danube is
not part of its current distribution (Table 5).
The observed isolation of the Rhone populations from the
ancestral Western lineage, using mtDNA and microsatellites in
the present study, supports the two-step expansion scenario
from the Danubian refuge and the existence of a secondary
glacial (Atlantic) refugium probably located in the Rhone
Basin during the Last Glacial Maximum. The present results
show that the Western–Atlantic lineage recently underwent
rapid population expansion as indicated by demographic
analyses and the star-like shape of the haplotype network. This
is also supported by the greater reduction in mtDNA genetic
diversity in this lineage relative to the Western–Danubian
lineage. The predominance of a single haplotype (i.e. H12)
found over a wide geographical area is more compatible with
non-equilibrium conditions such as previous habitat reduction
or with population bottlenecks due to glacial episodes. A
similar pattern of low genetic diversity within the Western
lineage identified using mtDNA was observed in other fish
species, e.g. burbot (Lota lota) (Barluenga et al., 2006).
Therefore, it seems that reduction of genetic diversity could
represent a general pattern in western European rivers.
The next typical feature of post-glacial dispersion common
for European freshwater fishes is that the main mtDNA
lineages came into secondary contact in the same central
European areas (Bernatchez, 2001; Kotlik & Berrebi, 2001;
Gum et al., 2005; Bryja et al., 2010a). Our mtDNA and
microsatellite data do not provide evidence for secondary
contact between Western and Eastern lineages in central
Europe, which is in contrast with Durand et al. (1999a), who
reported the Western–Atlantic, Western–Danubian and Eastern
mtDNA lineages of chub co-occurring in the Elbe Basin.
The absence of signs of admixture could be caused by the low
number of sites sampled in the present study in the areas of
potential contact zones (i.e. only one locality is sampled from
the Elbe River). Following previous studies (Durand et al.,
1999a, 2000), the populations of the Baltic Basin belong to the
Eastern lineage. This assumption was not confirmed in our
study, as both markers revealed a Western origin for the Oder
population belonging to the Baltic Basin. Human factors (e.g.
accidental transport related to aquaculture) or geographical
proximity to Western lineage populations are the most
plausible explanations for this finding. The same pattern, i.e.
the Danubian origin of fish in the upper Oder River, was
recently also found in the European bitterling (Rhodeus
amarus) (J. Bryja & M. Reichard, unpublished data).
Similar to the work of Durand et al. (1999a), analyses of
nuclear as well as mtDNA data revealed no introgression
between two Bulgarian populations (the River Struma and the
lower Danube). This regional clustering could reflect different
colonization events or may be the result of limited connection
between these areas due to geographical barriers to dispersal.
In the bitterling (Bryja et al., 2010a), population structure
analysis using microsatellites has shown separation of the
south-eastern Danubian population in Bulgaria from other
Danubian populations in central Europe, which is in accordance
with our results (see Fig. 6, K = 7). However, the
Danubian population of R. amarus in Bulgaria formed a
cluster with the geographically proximate Bulgarian population
from the River Struma, which contrasts with our results.
Taxonomic implication
Our comprehensive sampling revealed six major lineages
within the European chub. Comparing these data with
previously published patterns for congeneric species, we found
that some of these taxa corresponded to lineages as defined in
this paper, whereas others remained separated (Fig. 4).
Namely, Squalius orientalis was included in the Eastern lineage,
Squalius laietanus in the Aegean–Spain, Squalius cf. orpheus in
the Aegean–Balkan, Squalius prespensis and Squalius albus in
the Adriatic, and Squalius vardarensis in the Western–Danubian
(Zardoya & Doadrio, 1999; Durand et al., 2000; Freyhof
et al., 2005; Perea et al., 2010). The Western–Atlantic lineage
did not contain any previously published sequences attributed
to a different species. Additional taxa formed branches separate
from lineages recognized here (Zardoya & Doadrio, 1999;
Durand et al., 2000; Perea et al., 2010).
Our Bayesian coalescent analysis showed that the model
assuming a phylogenetic dataset that used the Yule process to
model species diversification and extinction was better, but not
strongly favoured, compared with the model that assumes an
intra-specific dataset with constant population size. This
supports our treatment of all included populations as one
species. Given the complexity of information available today
and the lack of consensus in the literature, we believe that a
thorough systematic revision of the S. cephalus complex is
warranted to establish possible taxonomic ranks or synonymy
of the multiple taxa.
ACKNOWLEDGEMENTS
We thank all anonymous referees for comments that greatly
improved the manuscript, and Pavel Jurajda, Marke´ta
Ondracˇkova´, Teodora Trichkova, Milen Vassilev, Alexis Ribas
Salvador, Antoni Arrizabalaga, Andre´ Gilles, Rene´ Chappaz,
Squalius cephalus phylogeography in Europe
Journal of Biogeography 39, 1024–1040 1037
ª 2011 Blackwell Publishing Ltd
Paper 2.2.2 177
Paolo Galli, Jussi Pennanen, Miroslaw Przybylski and Grziegorz
Zieba for their help with fish sampling in the different
sampling localities and the opportunity to use their laboratories.
This study was funded by the Czech Science Foundation,
project no. 524/07/0188. Travel costs for the different field
studies and material collections were funded by the Research
Project of the Masaryk University, Brno, project no. MSM
0021622 416 and also supported by the Ichthyoparasitology
Centre of Excellence, project no. LC 522, funded by the
Ministry of Education, Youth and Sports of the Czech
Republic. J.B. was supported by the Biodiversity Research
Centre (LC06073). The bioinformatic analyses (Bayesian
coalescent analysis) were conducted at the computational
cluster of the Institute of Vertebrate Biology, AS CR, Brno. We
are very grateful to Matthew Nicholls for the English revision
of the draft.
REFERENCES
Avise, J.C. (1991) Ten unorthodox perspectives on evolution
prompted by comparative population genetic findings on
mitochondrial DNA. Annual Review of Genetics, 25, 45–69.
Avise, J.C., Arnold, J., Ball, R.M., Bermingham, E., Lamb, T.,
Neigel, J.E., Reeb, C.A. & Saunders, N.C. (1987) Intraspecific
phylogeography – the mitochondrial DNA bridge
between population genetics and systematics. Annual Review
of Ecology and Systematics, 18, 489–522.
Bandelt, H.J., Forster, P. & Rohl, A. (1999) Median-joining
networks for inferring intraspecific phylogenies. Molecular
Biology and Evolution, 16, 37–48.
Barluenga, M., Sanetra, M. & Meyer, A. (2006) Genetic admixture
of burbot (Teleostei: Lota lota) in Lake Constance from two
European glacial refugia. Molecular Ecology, 15, 3583–3600.
Belkhir, K., Borsa, P., Goudet, J., Chikhi, L. & Bonhomme, F.
(2004) Genetix version. 4.0, logiciel sous Windows TM pour la
genetique des populations. Laboratoire Genome et Populations,
CNRS UPR 9060, Universite´ de Montpellier II,
Montpellier, France.
Benjamini, Y. & Hochberg, Y. (1995) Controlling the false
discovery rate – a practical and powerful approach to
multiple testing. Journal of the Royal Statistical Society B:
Methodological, 57, 289–300.
Bernatchez, L. (2001) The evolutionary history of brown trout
(Salmo trutta L.) inferred from phylogeographic, nested
clade, and mismatch analyses of mitochondrial DNA variation.
Evolution, 55, 351–379.
Bohlen, J., Sˇlechtova´, V., Bogutskaya, N. & Freyhof, J. (2006)
Across Siberia and over Europe: phylogenetic relationships
of the freshwater fish genus Rhodeus in Europe and the
phylogenetic position of R. sericeus from the River Amur.
Molecular Phylogenetics and Evolution, 40, 856–865.
Bo¨hme, M. & Ilg, A. (2003) fosFARbase. Available at: http://
www.wahre-staerke.com/ (accessed 25 May 2011).
Bryja, J., Smith, C., Konecˇny´, A. & Reichard, M. (2010a)
Range-wide population genetic structure of the European
bitterling (Rhodeus amarus) based on microsatellite and
mitochondrial DNA analysis. Molecular Ecology, 19, 4708–
4722.
Bryja, J., Granjon, L., Dobigny, G., Patzenhauerova´, H.,
Konecˇny´, A., Duplantier, J.M., Gauthier, P., Colyn, M.,
Durnez, L., Lalis, A. & Nicolas, V. (2010b) Plio-Pleistocene
history of West African Sudanian savanna and the phylogeography
of the Praomys daltoni complex (Rodentia):
the environmental/geography/genetic interplay. Molecular
Ecology, 19, 4783–4799.
Chapuis, M.P. & Estoup, A. (2007) Microsatellite null alleles
and estimation of population differentiation. Molecular
Biology and Evolution, 24, 621–631.
Culling, M.A., Janko, K., Boron, A., Vasil’ev, V.P., Cote, I.M. &
Hewitt, G.M. (2006) European colonization by the spined
loach (Cobitis taenia) from Ponto-Caspian refugia based on
mitochondrial DNA variation. Molecular Ecology, 15, 173–
190.
Degnan, S.M. (1993) The perils of single-gene trees – mitochondrial
versus single-copy nuclear-DNA variation in
white-eyes (Aves, Zosteropidae). Molecular Ecology, 2, 219–
225.
Dowling, T.E., Tibbets, C.A., Minckley, W.L. & Smith, G.R.
(2002) Evolutionary relationships of the Plagopterins
(Teleostei: Cyprinidae) from cytochrome b sequences.
Copeia, 3, 665–678.
Drummond, A.J. & Rambaut, A. (2007) BEAST: Bayesian
evolutionary analysis by sampling trees. BMC Evolutionary
Biology, 7, 214.
Durand, J.D., Persat, H. & Bouvet, Y. (1999a) Phylogeography
and postglacial dispersion of the chub (Leuciscus cephalus) in
Europe. Molecular Ecology, 8, 989–997.
Durand, J.D., Templeton, A.R., Guinand, B., Imsiridou, A. &
Bouvet, Y. (1999b) Nested clade and phylogeographic
analyses of the chub, Leuciscus cephalus (Teleostei, Cyprinidae),
in Greece: implications for Balkan peninsula biogeography.
Molecular Phylogenetics and Evolution, 13, 566–580.
Durand, J.D., Unlu, E., Doadrio, I., Pipoyan, S. & Templeton,
A.R. (2000) Origin, radiation, dispersion and allopatric
hybridization in the chub (Leuciscus cephalus). Proceedings of
the Royal Society B: Biological Sciences, 267, 1687–1697.
Englbrecht, C.C., Freyhof, J., Nolte, A., Rassmann, K., Schliewen,
U. & Tautz, D. (2000) Phylogeography of the bullhead
Cottus gobio (Pisces: Teleostei: Cottidae) suggests a prePleistocene
origin of the major central European populations.
Molecular Ecology, 9, 709–722.
Estoup, A., Rousset, F., Michalakis, Y., Cornuet, J.M., Adriamanga,
M. & Guyomard, R. (1998) Comparative analysis of
microsatellite and allozyme markers: a case study investigating
microgeographic differentiation in brown trout
(Salmo trutta). Molecular Ecology, 7, 339–353.
Evanno, G., Regnaut, S. & Goudet, J. (2005) Detecting the
number of clusters of individuals using the software
structure: a simulation study. Molecular Ecology, 14,
2611–2620.
Excoffier, L., Laval, G. & Schneider, S. (2005) Arlequin
(version 3.0): an integrated software package for population
M. Seifertova´ et al.
1038 Journal of Biogeography 39, 1024–1040
ª 2011 Blackwell Publishing Ltd
178 Paper 2.2.2
genetics data analysis. Evolutionary Bioinformatics, 1, 47–
50.
Flanders, J., Jones, G., Benda, P., Dietz, Ch., Zhang, S., Li, G.,
Sharifi, M. & Rossiter, S.J. (2009) Phylogeography of the
greater horseshoe bat, Rhinolophus ferrumequinum: contrasting
results from mitochondrial and microsatellite data.
Molecular Ecology, 18, 306–318.
Freyhof, J., Lieckfeldt, D., Pitra, C. & Ludwig, A. (2005)
Molecules and morphology: evidence for introgression of
mitochondrial DNA in Dalmatian cyprinids. Molecular
Phylogenetics and Evolution, 37, 347–354.
Fu, Y.X. (1997) Statistical tests of neutrality of mutations
against population growth, hitchhiking and background
selection. Genetics, 147, 915–925.
Fu, Y.X. & Li, W.H. (1993) Statistical tests of neutrality of
mutations. Genetics, 133, 693–709.
Goudet, J. (1995) FSTAT (version 1.2): a computer program to
calculate F-statistics. Journal of Heredity, 86, 485–486.
Gum, B. (2009) Conservation genetics and management
implications for European grayling, Thymallus thymallus:
synthesis of phylogeography and population genetics. Fisheries
Management and Ecology, 16, 37–51.
Gum, B., Gross, R. & Kuehn, R. (2005) Mitochondrial and
nuclear DNA phylogeography of European grayling (Thymallus
thymallus): evidence for secondary contact zones in
Central Europe. Molecular Ecology, 14, 1707–1725.
Hall, T.A. (1999) BioEdit: a user-friendly biological sequence
alignment editor and analysis program for Windows 95/98/
NT. Nucleic Acids Symposium Series, 41, 95–98.
Ha¨nfling, B. & Brandl, R. (1998) Genetic and morphological
variation in a common European cyprinid, Leuciscus cephalus
within and across central European drainages. Journal of
Fish Biology, 52, 706–715.
Ha¨nfling, B., Hellemans, B., Volckaert, F.A.M. & Carvalho,
G.R. (2002) Late glacial history of the cold-adapted freshwater
fish Cottus gobio, revealed by microsatellites. Molecular
Ecology, 11, 1717–1729.
Ha¨nfling, B., Dumpelmann, C., Bogutskaya, N.G., Brandl, R. &
Brandle, M. (2009) Shallow phylogeographic structuring of
Vimba vimba across Europe suggests two distinct refugia
during the last glaciation. Journal of Fish Biology, 75, 2269–
2286.
Hardy, O.J. & Vekemans, X. (2002) SPAGeDi: a versatile
computer program to analyse spatial genetic structure at the
individual or population levels. Molecular Ecology Notes, 2,
618–620.
Hardy, O.J., Charbonnel, N., Fre´ville, H. & Heuertz, M. (2003)
Microsatellite allele sizes: a simple test to assess their significance
on genetic differentiation. Genetics, 163, 1467–
1482.
Harpending, H.C. (1994) Signature of ancient population
growth in a low-resolution mitochondrial DNA mismatch
distribution. Human Biology, 66, 591–600.
Hewitt, G.M. (1996) Some genetic consequences of ice ages,
and their role in divergence and speciation. Biological Journal
of the Linnean Society, 58, 247–276.
Hewitt, G.M. (1999) Post-glacial re-colonization of European
biota. Biological Journal of the Linnean Society, 68, 87–112.
Ho, S.Y.W., Phillips, M.J., Cooper, A. & Drummond, A.J.
(2005) Time dependency of molecular rate estimates and
systematic overestimation of recent divergence times.
Molecular Biology and Evolution, 22, 1561–1568.
Jakobsson, M. & Rosenberg, N.A. (2007) CLUMPP: a cluster
matching and permutation program for dealing with label
switching and multimodality in analysis of population
structure. Bioinformatics, 23, 1801–1806.
Ketmaier, V., Bianco, P.G., Cobollia, M., Krivokapic, M.,
Caniglia, R. & De Matthaeis, E. (2004) Molecular phylogeny
of two lineages of Leuciscinae cyprinids (Telestes and Scardinius)
from the peri-Mediterranean area based on cytochrome
b data. Molecular Phylogenetics and Evolution, 32,
1061–1071.
Kotlik, P. & Berrebi, P. (2001) Phylogeography of the barbel
(Barbus barbus) assessed by mitochondrial DNA variation.
Molecular Ecology, 10, 2177–2185.
Kottelat, M. & Freyhof, J. (2007) Handbook of European
freshwater fishes. Kottelat, Cornol, Switzerland.
Krieg, F., Triantafyllidis, A. & Guymard, R. (2000) Mitochondrial
DNA variation in European populations of Silurus
glanis. Journal of Fish Biology, 56, 713–724.
Laroche, J., Durand, J.D., Bouvet, Y., Guinand, B. & Brohon,
B. (1999) Genetic structure and differentiation among
populations of two cyprinids, Leuciscus cephalus and Rutilus
rutilus, in a large European river. Canadian Journal of
Fisheries and Aquatic Sciences, 56, 1659–1667.
Levy, A., Doadrio, I. & Almada, V.C. (2009) Historical biogeography
of European leuciscins (Cyprinidae): evaluating
the Lago Mare dispersal hypothesis. Journal of Biogeography,
36, 55–65.
Librado, P. & Rozas, J. (2009) DnaSP v5: a software for
comprehensive analysis of DNA polymorphism data. Bioinformatics,
25, 1451–1452.
Lundberg, J.G. (1993) African–South American freshwater fish
clades and continental drift, problems with a paradigm.
Biotic relationships between Africa and South America (ed. by
P. Goldblatt), pp. 156–198. Yale University Press, New
Haven, CT.
Makinen, H.S., Cano, J.M. & Merila, J. (2006) Genetic relationships
among marine and freshwater populations of the
European three-spined stickleback (Gasterosteus aculeatus)
revealed by microsatellites. Molecular Ecology, 15, 1519–
1534.
Nei, M. (1978) Estimation of average heterozygosity and genetic
distance from a small number of individuals. Genetics,
89, 583–590.
Nesbo, C.L., Fossheim, T., Vollestad, L.A. & Jakobsen, K.S.
(1999) Genetic divergence and phylogeographic relationships
among European perch (Perca fluviatilis) populations
reflect glacial refugia and postglacial colonization. Molecular
Ecology, 8, 1387–1404.
Perea, S., Bohme, M., Zupancic, P., Freyhof, J., Sanda, R.,
Ozulug, M., Abdoli, A. & Doadrio, I. (2010) Phylogenetic
Squalius cephalus phylogeography in Europe
Journal of Biogeography 39, 1024–1040 1039
ª 2011 Blackwell Publishing Ltd
Paper 2.2.2 179
relationships and biogeographical patterns in Circum-Mediterranean
subfamily Leuciscinae (Teleostei, Cyprinidae)
inferred from both mitochondrial and nuclear data. BMC
Evolutionary Biology, 10, 265.
Pritchard, J.K., Stephens, M. & Donnelly, P. (2000) Inference
of population structure using multilocus genotype data.
Genetics, 155, 945–959.
Rambaut, A. & Drummond, A.J. (2007) Tracer v1.4. Available
at: http://beast.bio.ed.ac.uk/Tracer (accessed 20 May 2011).
Rand, D.M. (1996) Neutrality tests of molecular markers and
the connection between DNA polymorphism, demography,
and conservation biology. Conservation Biology, 10, 665–671.
Raymond, M. & Rousset, F. (1995) GENEPOP (version 1.2):
population genetics software for exact tests and ecumenicism.
Journal of Heredity, 86, 248–249.
Rodriguez, F., Pe´rez, T., Hammer, S.E., Albornoz, J. &
Dominguez, A. (2010) Integrating phylogeographic patterns
of microsatellite and mtDNA divergence to infer the evolutionary
history of chamois (genus Rupicapra). BMC Evolutionary
Biology, 10, 222.
Rogers, A.R. & Harpending, H. (1992) Population growth
makes waves in the distribution of pairwise genetic differences.
Molecular Biology and Evolution, 9, 552–569.
Rosenberg, N.A. (2004) distruct: a program for the graphical
display of population structure. Molecular Ecology Notes, 4,
137–138.
Salzburger, W., Brandstatter, A., Gilles, A., Parson, W., Hempel,
M., Sturmbauer, C. & Meyer, A. (2003) Phylogeography
of the vairone (Leuciscus souffia, Risso 1826) in Central
Europe. Molecular Ecology, 12, 2371–2386.
Sanjur, O.I., Carmona, J.A. & Doadrio, I. (2003) Evolutionary
and biogeographical patterns within Iberian populations of
the genus Squalius inferred from molecular data. Molecular
Phylogenetics and Evolution, 29, 20–30.
Schmidt, T.R. & Gold, J.R. (1993) Complete sequence of the
mitochondrial cytochrome b gene in the cherryfin shiner,
Lythrurus roseipinnis (Teleostei, Cyprinidae). Copeia, 3, 880–883.
Seifertova´, M., Vyskocˇilova´, M., Morand, S. & Sˇimkova´, A.
(2008) Metazoan parasites of freshwater cyprinid fish
(Leuciscus cephalus): testing biogeographical hypotheses of
species diversity. Parasitology, 135, 1417–1435.
Slatkin, M. (1995) A measure of population subdivision based
on microsatellite allele frequencies. Genetics, 139, 457–462.
Storey, J.D., Taylor, J.E. & Siegmund, D. (2004) Strong control,
conservative point estimation and simultaneous conservative
consistency of false discovery rates: a unified
approach. Journal of the Royal Statistical Society B: Methodological,
66, 187–205.
Tajima, F. (1989) Statistical method for testing the neutral
mutation hypothesis by DNA polymorphism. Genetics, 123,
585–595.
Triantafyllidis, A., Krieg, F., Cottin, C., Abatzopoulos, T.J.,
Triantaphyllidis, C. & Guyomard, R. (2002) Genetic
structure and phylogeography of European catfish (Silurus
glanis) populations. Molecular Ecology, 11, 1039–1055.
Valqui, J., Hartl, G.B. & Zachos, F.E. (2010) Non-invasive
genetic analysis reveals high levels of mtDNA variability
in the endangered South-American marine otter (Lontra
felina). Conservation Genetics, 11, 2067–2072.
Vyskocˇilova´, M., Sˇimkova´, A. & Martin, J.F. (2007) Isolation
and characterization of microsatellites in Leuciscus cephalus
(Cypriniformes, Cyprinidae) and cross-species amplification
within the family Cyprinidae. Molecular Ecology Notes, 7,
1150–1154.
Zardoya, R. & Doadrio, I. (1999) Molecular evidence on the
evolutionary and biogeographical patterns of European
cyprinids. Journal of Molecular Evolution, 49, 227–237.
Zhang, D.X. & Hewitt, M. (2003) Nuclear DNA analyses in
genetic studies of populations: practice, problems and
prospects. Molecular Ecology, 12, 536–584.
SUPPORTING INFORMATION
Additional Supporting Information may be found in the
online version of this article:
Appendix S1 Haplotypes of cytochrome b (600 bp) in
Squalius cephalus analysed in this study.
Appendix S2 Geographical distribution and accession numbers
of cytochrome b haplotypes (Table S2.1), and genetic
variability across microsatellite loci genotyped (Table S2.2) in
Squalius cephalus.
Appendix S3 Results of the structure analysis for K = 1–
10 for Squalius cephalus.
As a service to our authors and readers, this journal provides
supporting information supplied by the authors. Such materials
are peer-reviewed and may be re-organized for online
delivery, but are not copy-edited or typeset. Technical support
issues arising from supporting information (other than
missing files) should be addressed to the authors.
BIOSKETCH
This work is a part of the PhD study of Ma´ria Seifertova´ on
host–parasite interactions in a geographical context using
fishes as model hosts. The research team included specialists
working on molecular ecology, phylogeny and phylogeography,
and modelling of DNA sequence evolution.
Author contributions: M.S., J.B. and A.Sˇ. conceived the ideas;
M.S. and A.Sˇ. collected the data; M.S., J.B., N.M. and M.V.
analysed the data; M.S., J.B., N.M. and A.Sˇ. wrote the
manuscript.
Editor: Brett Riddle
M. Seifertova´ et al.
1040 Journal of Biogeography 39, 1024–1040
ª 2011 Blackwell Publishing Ltd
180 Paper 2.2.2
Paper 2.2.3
Martínková N., Barnett R., Cucchi T., Struchen R., Pascal M., Pascal M., Fischer M. C.,
Higham T., Brace S., Ho S. Y. W., Quéré J.-P., O’Higgins P., Excoffier L., Heckel G.,
Hoelzel A. R., Dobney K. M., Searle, J. B. 2013. Divergent evolutionary processes associated
with colonization of offshore islands. Molecular Ecology 22: 5205-5220.
– 181 –
Divergent evolutionary processes associated with
colonization of offshore islands
NATÁLIA MARTÍNKOVÁ,*†1
ROSS BARNETT,*‡§1
THOMAS CUCCHI,§¶**1
RAHEL
STRUCHEN,†† MARINE PASCAL,‡‡ MICHEL PASCAL,‡‡ MARTIN C. FISCHER,††§§
THOMAS HIGHAM,¶¶ SELINA BRACE,*** SIMON Y. W. HO,††† JEAN-PIERRE QU ER E,‡‡‡ PAUL
O’ HIGGINS,§§§ LAURENT EXCOFFIER,††§§ GERALD HECKEL,††§§ A. RUS HOELZEL,‡
KEITH M. DOBNEY§¶2
and JEREMY B. SEARLE*¶¶¶2
*Department of Biology, University of York, Wentworth Way, York YO10 5DD, UK, †Institute of Vertebrate Biology, Academy
of Sciences of the Czech Republic, Květná 8, Brno 603 65, Czech Republic, ‡School of Biological and Biomedical Sciences,
Durham University, South Road, Durham, DH1 3LE, UK, §Department of Archaeology, Durham University, South Road,
Durham, DH1 3LE, UK, ¶Department of Archaeology, University of Aberdeen, St Mary’s, Elphinstone Road, Aberdeen AB24
3UF, UK, **Museum national d’histoire naturelle, case postale 56 (b^atiment d’anatomie comparee), 55 rue Buffon, F-75231 Paris
Cedex 05, Paris, France, ††Computational and Molecular Population Genetics (CMPG), Institute of Ecology and Evolution,
University of Bern, Baltzerstrasse 6, Bern CH-3012, Switzerland, ‡‡Equipe Ecologie des Invasions Biologiques, INRA, Campus
de Beaulieu 35 000, Rennes Cedex, France, §§Swiss Institute of Bioinformatics, Genopode, Lausanne CH-1015, Switzerland,
¶¶Oxford Radiocarbon Accelerator Unit, Research Laboratory for Archaeology and the History of Art (RLAHA), Dyson Perrins
Building, South Parks Road, Oxford OX1 3QY, UK, ***School of Biological Sciences, Royal Holloway, University of London,
Egham TW20 0EX, UK, †††School of Biological Sciences, University of Sydney, Sydney, NSW 2006, Australia, ‡‡‡UMR
CBGP (INRA/IRD/Cirad/Montpellier SupAgro), INRA, Campus international de Baillarguet, CS 30016, F-34988 Montferriersur-Lez
cedex, France, §§§Centre for Anatomical and Human Sciences, Hull York Medical School (HYMS), University of York,
John Hughlings Jackson Building, York YO10 5DD, UK, ¶¶¶Department of Ecology and Evolutionary Biology, Corson Hall,
Cornell University, Ithaca, NY 14853-2701, USA
Abstract
Oceanic islands have been a test ground for evolutionary theory, but here, we focus on
the possibilities for evolutionary study created by offshore islands. These can be colonized
through various means and by a wide range of species, including those with low
dispersal capabilities. We use morphology, modern and ancient sequences of cytochrome
b (cytb) and microsatellite genotypes to examine colonization history and evolutionary
change associated with occupation of the Orkney archipelago by the common vole
(Microtus arvalis), a species found in continental Europe but not in Britain. Among possible
colonization scenarios, our results are most consistent with human introduction at
least 5100 BP (confirmed by radiocarbon dating). We used approximate Bayesian computation
of population history to infer the coast of Belgium as the possible source and estimated
the evolutionary timescale using a Bayesian coalescent approach. We showed
substantial morphological divergence of the island populations, including a size increase
presumably driven by selection and reduced microsatellite variation likely reflecting
founder events and genetic drift. More surprisingly, our results suggest that a recent and
widespread cytb replacement event in the continental source area purged cytb variation
there, whereas the ancestral diversity is largely retained in the colonized islands as a
genetic ‘ark’. The replacement event in the continental M. arvalis was probably triggered
Correspondence: Jeremy B. Searle, Fax: +1-607-255-8088; E-mail: jeremy.searle@cornell.edu
1
These authors contributed equally.
2
These authors contributed equally.
© 2013 The Authors. Molecular Ecology Published by John Wiley & Sons Ltd.
This is an open access article under the terms of the Creative Commons Attribution License,
which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
Molecular Ecology (2013) 22, 5205–5220 doi: 10.1111/mec.12462
182 Paper 2.2.3
by anthropogenic causes (land-use change). Our studies illustrate that small offshore
islands can act as field laboratories for studying various evolutionary processes over relatively
short timescales, informing about the mainland source area as well as the island.
Keywords: demographic analysis, genetic replacement, island colonization, Microtus arvalis,
phylogeography
Received 5 May 2013; revision received 2 July 2013; accepted 9 July 2013
Introduction
Marine islands were pivotal settings for the development
of evolutionary theory (Wallace 1880) and continue to
inspire evolutionary biologists today (Grant 2008). They
tend to be ecologically simple and distinctive compared
with mainland settings, with fewer competitors and predators,
a different food supply and environmental conditions
and a smaller living space, all factors that promote
rapid and perhaps substantial evolutionary change relative
to mainland source populations. Studies on islands
have therefore provided insights into evolutionary processes
such as species radiations (Emerson 2002) and the
generation of evolutionary novelty (Lachaise et al. 2000).
Marine islands are also interesting for the varied
ways in which they can be colonized. First, for islands
separating from a mainland, some species may already
be present at island formation. Second, islands may be
colonized by natural ‘sweepstake’ dispersal (Simpson
1940). Finally, species may be introduced into islands
by humans.
Many of the classic systems for the study of island
evolution, particularly species radiations, have involved
oceanic islands or archipelagos colonized by sweepstake
dispersal, for example Darwin’s finches on the Galapagos
and Hawaiian drosophila. However, here we consider
offshore islands, which can be colonized by all three
mechanisms, providing rich opportunities for evolutionary
studies, notably of poor dispersers such as small
mammals (Berry 1996).
Offshore islands around the Atlantic coast of Europe
were formed by end-glacial sea level rise, and thus,
small mammals currently found on them may have
already been present prior to island separation, and
perhaps even before the last glacial maximum (LGM; c.
20 000 BP: Ehlers & Gibbard 2004), if local conditions
were compatible with survival (Beirne 1952; Martınkova
et al. 2007). Alternatively, the small mammals could
have been introduced by humans, for example, within
bedding/fodder for livestock (Corbet 1961). Sweepstake
dispersal, by rafting on floating masses of vegetation
(Morrison et al. 1992) or walking over ice (White &
Searle 2008), is also a possibility.
In this study, we aim to distinguish between these different
modes of colonization of a European offshore
archipelago (Orkney) by a small mammal (the common
vole Microtus arvalis) and to document evolutionary
change in that species postcolonization. Microtus arvalis is
a ground-living, grass-eating small rodent that is
common in agricultural meadows (Niethammer & Krapp
1982). The species is widespread in western continental
Europe, although it is absent from Britain and Fennoscandia
(Fig. 1a). Therefore, the whole British landmass
separates the M. arvalis populations in continental Europe
and on Orkney. The only species of Microtus in mainland
Britain and other islands off mainland Britain is Microtus
agrestis, which, like M. arvalis, has a wide continental
European range (Shenbrot & Krasnov 2005). The two
species belong to unrelated lineages within Microtus
(Fink et al. 2010; Martınkova & Moravec 2012).
The colonization history of a population can be
inferred through comparisons with potential source populations.
If the presence of the Orkney vole is explained
by glacial survival or sweepstake dispersal, then surviving
representatives of the source populations are no
longer available, because there are no extant populations
suitably close to Orkney. However, human introduction
can come from a distance, and the most likely source area
is represented by the closest coastal part of the continental
European range of the same mitochondrial lineage as
found on Orkney, that is, northern France and Belgium
(Haynes et al. 2003; Tougard et al. 2008). For this reason,
we focus on this potential source area in our analyses
and consider whether such a human introduction is feasible
or whether one of the other modes of colonization is
more likely.
To explore the colonization history of the Orkney vole
and the evolutionary change since colonization, we compare
populations from Orkney and the potential source
area using mitochondrial DNA sequences and microsatellites.
We also use archaeological deposits of the species
from Orkney for ancient DNA (aDNA) sequencing and
radiocarbon dating (two Orkney vole bones have previously
been radiocarbon-dated [Hedges et al. 1987] using
technology now outdated), to obtain a more detailed
temporal dimension to the colonization history and to
© 2013 The Authors. Molecular Ecology Published by John Wiley & Sons Ltd.
5206 N. MARTINKOVÁ ET AL.
Paper 2.2.3 183
support a demographic analysis. Finally, we use geometric
morphometrics of a molar tooth as a marker for morphological
evolution. This work provides a more
sophisticated assessment of colonization and evolution of
the Orkney vole than has previously been possible (Berry
& Rose 1975; Corbet 1986; Haynes et al. 2003; Thaw et al.
2004; Gorman & Reynolds 2008). More generally, this
case study provides new insights into not only short-term
0.60 0.65 0.70 0.75 0.80
LogCS
Orkney
Continental
(a)
(b)
(e)
(c)
Guernsey
(d)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
0.09
0.06
0.03
0.00
–0.03
–0.06
–0.09
0.090.060.030.00–0.03–0.06–0.09
PC1 (30.8%)
PC2(13.9%)
200 km
10 km
Fig. 1 The morphological variation of the
first lower molar (M1) among modern
Microtus arvalis populations. (a) Sampling
localities within its western and central
European range (Shenbrot & Krasnov
2005) (in grey). Note that the three Spanish
localities and the three Italian localities
are treated as single entries in Fig. 1e
and Table S1 (Supporting information).
(b) Position of the 18 landmarks and 12
semi-landmarks on the M1 occlusal view.
(c) Scatter plot of the first two principal
component scores depicting overall M1
shape variation. (d) Diagram showing
the M1 shape change associated with the
PC1 scores by 0.1 units in the positive
direction (black) against the mean shape
(grey). (e) Box plot showing log-transformed
centroid size variation of the M1.
© 2013 The Authors. Molecular Ecology Published by John Wiley & Sons Ltd.
COLONIZATION OF OFFSHORE ISLANDS 5207
184 Paper 2.2.3
evolutionary changes in offshore islands, but also the
mainland areas with which they are being compared.
Materials and methods
Specimens
Details of all Microtus arvalis specimens used in this
study are listed in Tables 1 and 2 and Tables S1–S3
(Supporting information) and mapped in Figs 1 and 2
and Figs S1 and S2 (Supporting information). Geometric
morphometrics were applied to 553 modern specimens
from 27 localities, while 651 M. arvalis specimens from
70 localities were used for the analysis of modern DNA
(125 specimens were used in both cytb and microsatellite
studies). Ancient DNA analysis was conducted on
37 archaeological specimens from eight localities. Nineteen
of these were among 23 archaeological specimens
used for radiocarbon dating, calibrated using Calib Rev.
5.0.1 (Stuiver & Reimer 1993) and IntCal04 (Reimer et al.
2004), with dates before present (BP) standardized to
1950 AD.
Geometric morphometrics
Modern M. arvalis from populations in Orkney and continental
Europe (Fig. 1a) were compared by geometric
morphometrics of dental phenotype. Digital photographs
of the first lower molar (M1) were used to
record two-dimensional Cartesian coordinates of 18
anatomical landmarks at the bases and tips of the
lingual and labial cusps as well as 12 equidistant semilandmarks
on a manually drawn curve along the
anterior loop (Fig. 1b).
Digitization of landmarks and semi-landmarks was
performed using TPSdig 2 (Rohlf 2010a). Position, orientation
and scaling information from the raw coordinates
were standardized by a generalized Procrustes
analysis (GPA) using TPSrelw 1.49 (Rohlf 2010b). To
combine landmarks and semi-landmarks in the GPA,
Table 1 Microsatellite comparison of Microtus arvalis in Orkney and continental Europe, including diversity indices for all
populations sampled and approximate Bayesian computation selection of most likely continental population for the Orkney colonization.
For map of localities, see Fig. S1 (Supporting information)
Locality
Country/Orkney
Island Latitude Longitude
Number
of specimens mtDNA lineage AR H ΔC
Stalhille Belgium 51.21 3.07 26 Western-North 5.69 (1.90) 0.72 (0.15) 2.03
Pihen les Gu^ınes France 50.87 1.79 22 Western-North 7.02 (2.08) 0.78 (0.17) 2.24
Daubeuf France 49.78 0.52 20 Western-North 6.65 (2.51) 0.73 (0.25) 2.24
Alflen Germany 50.18 7.04 22 Western-North 6.62 (1.66) 0.80 (0.09) 2.27
Clermont-Ferrand France 45.78 3.08 23 Western-North 6.06 (2.21) 0.70 (0.23) 2.62
Avallon France 47.85 3.90 22 Western-North 6.74 (1.84) 0.78 (0.13) 2.63
Aiffres France 46.29 À0.41 22 Western-South 7.59 (3.34) 0.73 (0.27) 2.66
Fressenneville France 50.07 1.58 24 Western-North 6.84 (2.23) 0.73 (0.23) 2.71
Thaon France 49.26 À0.46 24 Western-North 7.19 (2.76) 0.74 (0.24) 2.72
Ste Marie du Mont France 49.38 À1.23 20 Western-North 6.72 (2.88) 0.69 (0.28) 2.91
Schiltach Germany 48.30 8.34 20 Western-North 5.30 (1.58) 0.74 (0.11) 2.94
Veurne Belgium 51.07 2.66 26 Western-North 4.09 (0.95) 0.65 (0.11) 3.02
Baie d’Aiguillon France 46.30 1.17 15 Western-South 7.76 (3.38) 0.74 (0.26) —
St Jean le Thomas France 48.73 À1.51 10 Western-North 4.07 (2.13) 0.59 (0.24) —
Dinteloord Netherlands 51.64 4.37 12 Central 4.68 (1.59) 0.66 (0.24) —
Heerenveen Netherlands 52.96 5.93 13 Central 6.04 (2.08) 0.75 (0.23) —
Harray Stenness Mainland, Orkney 59.02 À3.20 23 Western-North 5.01 (2.31) 0.62 (0.26) —
Settiscarth Mainland, Orkney 59.05 À3.10 26 Western-North 4.35 (1.88) 0.60 (0.27) —
St Ola Mainland, Orkney 58.94 À2.95 19 Western-North 3.79 (1.94) 0.50 (0.31) —
Whitemill Bay Sanday, Orkney 59.30 À2.55 19 Western-North 2.11 (1.60) 0.23 (0.26) —
Wind Wick S Ronaldsay, Orkney 58.76 À2.94 20 Western-North 1.83 (1.14) 0.21 (0.26) —
Grimness S Ronaldsay, Orkney 58.82 À2.91 20 Western-North 2.17 (1.21) 0.24 (0.26) —
Loch of Swartmill Westray, Orkney 59.29 À2.92 21 Western-North 2.16 (1.50) 0.26 (0.31) —
Ness Westray, Orkney 59.24 À2.87 23 Western-North 1.73 (1.19) 0.19 (0.29) —
AR: mean allelic richness over loci. H: mean heterozygosity over loci; values in parentheses are standard deviations. DC: average
distance between the observed and 1000 simulated summary statistics computed over the three Mainland Orkney populations for
each continental population (c) to find the most likely source population for the colonization of Orkney by M. arvalis (smallest value).
Only samples of 19 or more individuals were used for this analysis. The populations on Mainland Orkney were considered to best
represent the colonized area, because of retention of high diversity (see also Fig. 3).
© 2013 The Authors. Molecular Ecology Published by John Wiley & Sons Ltd.
5208 N. MARTINKOVÁ ET AL.
Paper 2.2.3 185
the semi-landmarks along the anterior loop curve were
constrained to slide along an estimated tangent at
each sliding point using the bending energy method
(Bookstein 1997).
The overall size parameter for the M1 is centroid size
(square root of the sum of squared distances of landmarks
and semi-landmarks from the centroid) and
comparisons between populations use a box plot of logtransformed
values produced with ‘R’ v. 2.13.1 (R
Development Core Team 2011). The shape variables are
the Procrustes coordinates obtained after the GPA,
and variation among populations is displayed by principal
component analysis (PCA) using MorphoJ 1.05c
(Klingenberg 2011).
Cytochrome b sequencing (modern DNA)
A total of 283 specimens of M. arvalis were sequenced
from 18 localities in Orkney and 50 localities in
continental Europe (particularly from around the potential
area of human introduction: northern France and
Belgium and inland from there). Total genomic DNA
was isolated using the DNeasy Tissue Kit (Qiagen,
Halden, Germany) and PCR-amplified using previously
described primers (Table S4, Supporting information;
Jaarola et al. 2004). PCR products were purified with
QIAquick PCR purification kit (Qiagen) and commercially
sequenced using BigDye Terminator Sequencing
chemistry (Applied Biosystems, Foster City, CA, USA)
with newly designed primers (Table S4, Supporting
information) and run on ABI PRISM 3730xl sequencers
(Applied Biosystems). Complete cytb sequences
(1143 bp) were generated (GU190383–GU190665).
Cytochrome b sequencing (ancient DNA)
All aDNA extractions were performed in a laboratory
in Durham where no modern molecular biology or
post-PCR work is undertaken and where Microtus were
analysed for the first time. Before use, all materials and
work surfaces were wiped with 10% bleach, and the
workspace was UV-irradiated overnight. Samples of Microtus
bone were excised using a scalpel blade and
crushed within aluminium foil. The bone powder was
incubated overnight on a rotator at 55 °C in 500 lL of
extraction buffer (0.5M pH 8.0 EDTA, 0.1M pH 8.0 Tris,
0.05% w/v SDS) with 8 lL of proteinase K (0.3 mg/
mL). Digested samples were then extracted using the
Qiagen QIAquick PCR purification method, as
described in Nichols et al. (2007). Final elutions of
aDNA were collected in 50 lL of TE buffer following
the QIAquick protocol and stored at À20 °C. Negative
extraction controls (lacking bone powder) were also
performed in parallel in a ratio of approximately 1:10.
Ancient DNA was successfully obtained from 37
specimens (of 190 attempted) from Neolithic to Viking
Table 2 14
C dates and their calibrated age ranges for 23 Microtus arvalis mandibles collected from Orkney
Laboratory number Sample reference Site name Island 14
C age BP 95.4% (2s) cal age ranges
OxA 18324 R44 Point of Cott Westray 4555 Æ 40 cal BP 5050–5437
OxA 18782 R37 Point of Cott Westray 4459 Æ 33 cal BP 4967–5288
OxA 18325 R45 Point of Cott Westray 4451 Æ 38 cal BP 4884–5287
OxA-18668 R177 Quanterness Mainland 4414 Æ 27 cal BP 4869–5257
OxA-18784 R179 Quanterness Mainland 4400 Æ 33 cal BP 4861–5213
OxA-18786 R191 Skara Brae Mainland 4199 Æ 33 cal BP 4622–4844
OxA-20309 R84 Skara Brae Mainland 4145 Æ 29 cal BP 4574–4823
OxA-18664 R11 Skara Brae Mainland 4124 Æ 28 cal BP 4529–4815
OxA-18663 R3 Skara Brae Mainland 3946 Æ 27 cal BP 4294–4515
OxA-18787 R194 Skara Brae Mainland 3939 Æ 32 cal BP 4256–4514
OxA-18669 R193 Skara Brae Mainland 3906 Æ 27 cal BP 4249–4419
OxA-18785 R189 Skara Brae Mainland 3884 Æ 31 cal BP 4185–4418
OxA-18666 R23 Holm of Papa Westray Westray 4089 Æ 29 cal BP 4448–4808
OxA-18665 R20 Holm of Papa Westray Westray 4054 Æ 28 cal BP 4435–4784
OxA 18328 R126 Pierowall Quarry Westray 4000 Æ 45 cal BP 4298–4781
OxA 18783 R124 Pierowall Quarry Westray 3824 Æ 34 cal BP 4094–4406
OxA 18327 R99 Pierowall Quarry Westray 3822 Æ 38 cal BP 4092–4406
OxA-18350 R58 Howe Mainland 1860 Æ 28 cal BP 1721–1869
OxA-18351 R59 Howe Mainland 1849 Æ 27 cal BP 1714–1865
OxA-18667 R62 Howe Mainland 1469 Æ 24 cal BP 1308–1396
OxA-20310 RH3 Green Hill, South Walls Hoy 1100 Æ 24 cal BP 958–1060
OxA-20481 RH2 Green Hill, South Walls Hoy 993 Æ 27 cal BP 798–962
OxA 18326 R29 Earl’s Bu Mainland 966 Æ 29 cal BP 795–932
© 2013 The Authors. Molecular Ecology Published by John Wiley & Sons Ltd.
COLONIZATION OF OFFSHORE ISLANDS 5209
186 Paper 2.2.3
Age Orkney and Late Medieval Belgium (Table S3, Fig.
S2, Supporting information). Partial cytb sequences
(1130 bp) were generated through 8–9 overlapping
PCRs (Table S4, Supporting information), with the missing
cytb sequence at the 5′-end. Each PCR used 2 lL of
aDNA extract in a 25 lL volume with Hi-Fidelity Platinum
Taq (Invitrogen, Carlsbad, CA, USA). Purification
and sequencing of PCR products followed the protocols
used for modern DNA.
As further checks on the aDNA authenticity additional
to the negative extraction and PCR controls, two
samples were internally replicated from bone powder
(R59 and R60), producing identical sequences. All
sequences with unique mutations not found in modern
Orkney voles were amplified at least twice for confirmation.
The regions of overlap between the PCR products
were checked and found to be consistent. The
assembled cytb coding regions for all ancient voles produced
a readable protein sequence with no frameshift
or nonsense mutations. M. arvalis is known to possess a
nuclear copy of cytb (DeWoody et al. 1999) that is easily
identifiable due to unique fixed mutations and insertions
and was not amplified from any of the aDNA
extracts.
Six samples (R44, R58, R125, R131, R189 and R191)
were sent to Egham where the entire procedure, including
extraction and amplification, was replicated. Two
samples (R125 and R131) were sequenced in their
entirety and produced identical sequences to those in
Durham. Radiocarbon dating and high collagen levels
found in 19 samples from Orkney that had been successful
for aDNA extraction give corroborative evidence
for good biochemical preservation of the samples.
Microsatellites
A total of 492 modern specimens from eight localities in
Orkney and 16 localities in continental Europe from
around the most probable area of introduction (northern
France and Belgium and inland from there; Fig. S1,
Supporting information) were typed at 14 nuclear microsatellite
loci (Heckel et al. 2005; Jaarola et al. 2007;
Walser & Heckel 2008): CRB5, INb, MAG6, MAG25,
MAR003, MAR016, MAR063, MAR076, MAR080,
MAR102, MAR113, MM1, MM8 and Moe02. PCR amplifications
were performed with the Qiagen Multiplex kit
using Gene Amp PCR System 9700 (Applied Biosystems),
PCR fragments were separated on an ABI 3100
0.01
1.0
1.0
1.0
1.0
1.0
1.0
.99
.87
.80
1.0
Western-North (Orkney)
Western-North (Continental)
Western-South
Italian
Central
Eastern
M. arvalis obscurus
200 km
Fig. 2 Bayesian phylogenetic tree depicting all complete cytochrome b haplotypes of modern Microtus arvalis and localities where
each phylogroup is found. Lineages labelled following previous authors (Haynes et al. 2003; Tougard et al. 2008): Orkney (red) within
Western-North (yellow), Western-South (green), Central (dark blue), Eastern (light blue), Italian (purple), M. a. obscurus (grey). Posterior
probability is presented for major clades only. Outgroup sequence in tree: Microtus levis, AY513819 (Jaarola et al. 2004). Grey
shading: distribution of M. arvalis (Shenbrot & Krasnov 2005) where three obscurus localities are off the map to the east. Black arrow
shows the most likely source locality of the Orkney voles, based on microsatellite data (see Table 1).
© 2013 The Authors. Molecular Ecology Published by John Wiley & Sons Ltd.
5210 N. MARTINKOVÁ ET AL.
Paper 2.2.3 187
sequencer (Applied Biosystems), and their lengths
determined using GENEMAPPER 3.7 (Applied Biosystems).
For each locus and each population, tests for departure
from Hardy–Weinberg equilibrium were performed
with ARLEQUIN 3.12 (Excoffier et al. 2005), and
significance levels were corrected for multiple testing
using the sequential Bonferroni–Holm procedure (Rice
1989). There were no significant departures from equilibrium.
Mean allelic richness over loci (AR) was calculated
with FSTAT version 2.9.3.2 (Goudet 1995) and the
mean heterozygosity (H) over loci and population
differentiation (FST) with Arlequin.
Phylogenetic analysis
For cytb, sequence chromatograms were assembled in
Sequencher 4.5 (GeneCodes, Ann Arbor, MI, USA) and
aligned manually in BioEdit (Hall 1999). Newly obtained
modern DNA sequences were analysed together with
previously published modern sequences (Martin et al.
2000; Haynes et al. 2003; Borkowska & Ratkiewicz 2008;
Tougard et al. 2008). Using the 1143-bp alignment, this
yielded a total of 395 sequences distributed among 173
haplotypes. These were derived from 124 localities (106
from continental Europe, 18 from Orkney). Microtus levis
was used as an outgroup (Martınkova & Moravec 2012).
A Bayesian phylogenetic tree of all modern sequences
with the complete (1143 bp) cytb was estimated in
MrBayes 3.1.2 (Ronquist & Huelsenbeck 2003). To avoid
long-tree artefact (Marshall 2010), the Γ-distribution
shape parameter a was fixed to 1.0736. The value was
obtained from the GTR+Γ substitution model selected
by Akaike information criterion in MrModeltest 2.3
(Posada & Crandall 1998; Nylander 2004). The two separate
Markov chain Monte Carlo (MCMC) analyses each
comprised one cold and eleven heated chains. Samples
were drawn from the posterior every 1000 steps over
10 million steps. The chain temperature was 0.06, and
two chain swaps were attempted every step to optimize
mixing. The first 3000 sampled trees were discarded
as burn-in, with the resulting average standard deviation
of split frequencies equal to 0.007. The 95%
credibility interval of the tree length included a maximum-likelihood
estimate of the tree length obtained
from RAXML 7.2 (Stamatakis 2006).
Median-joining networks were constructed in Network
4.2 (Bandelt et al. 1999) with an equal transition/transversion
ratio. The analysis was carried out using the 1130-bp
alignment because aDNA sequences were included.
Demographic analysis
For cytb, nucleotide and haplotype diversities (Nei
1987) and Ramos-Onsins & Rozas’s (2002) R2 were
determined using DNASP 5.1 (Rozas et al. 2003). The
neutrality test statistics Tajima’s (1989) D and Fu’s
(1997) FS were estimated in Arlequin. This software
was also used for mismatch distribution analysis (Rogers
& Harpending 1992) to detect and date population
expansions.
Radiocarbon-dated and modern M. arvalis cytb
sequences were analysed using the program BEAST 1.5.1
(Drummond & Rambaut 2007), utilizing the 1130-bp
alignment because of inclusion of aDNA sequences. For
the Orkney data set, the GTR+Γ model of nucleotide
substitution was selected using the Akaike information
criterion. All BEAST analyses used this model, along with
a strict molecular clock. A comparison using Bayes factors
selected the constant-size coalescent prior as the
most appropriate demographic model. With ancient
sequences in the data set, the potential for undetected
sequence errors to influence the analyses was also
modelled in BEAST (Rambaut et al. 2009). All other
parameters were co-estimated with the phylogeny, with
samples drawn from the posterior every 1000 steps over
a total of 10 million steps. The first 1 000 000 steps were
discarded as burn-in. Acceptable mixing and convergence
to stationarity were checked using the program
TRACER 1.4.1 (Rambaut & Drummond 2007).
For modern DNA sequences from continental European
populations, we used a demographic model simulating
population expansion and used Bayes factors to
compare a strict molecular clock, uncorrelated lognormal
relaxed clock and uncorrelated exponential relaxed
clock (Drummond et al. 2006). A model of exponential
growth was identified as the best-fitting demographic
model using Bayes factors. Samples were drawn from
the Markov chain every 10 000 steps over a total of
100 million steps, with the first 30% of sampled trees
discarded as burn-in. Estimated effective sample sizes
were >200. Mean mutation rate was fixed to that
obtained from the tip-dated sequence analysis described
above, and each lineage that was expected to exhibit
unique demographic history in the target time frame
was analysed separately.
For the specific question of time of colonization of
Orkney, we used the program IMa (Hey & Nielsen
2007) to estimate the splitting time between the modern
Orkney sample (N = 57 for mtDNA, N = 114 for microsatellite
loci) and European mainland samples from the
potential area of human introduction: northern France,
Belgium and nearby areas of Germany (N = 46 for
mtDNA, N = 92 for microsatellite loci). Initial runs were
performed to estimate parameters, followed by three
replicate runs with different random number seeds to
check for consistency of results (results from the final
run reported). Input data were 1143-bp cytb sequences
(assuming the HKY mutation model and the mean
© 2013 The Authors. Molecular Ecology Published by John Wiley & Sons Ltd.
COLONIZATION OF OFFSHORE ISLANDS 5211
188 Paper 2.2.3
mutation rate estimated in BEAST; priors on the range of
mutation rate scalars were set one order of magnitude
above and below) and 14 microsatellite loci. The inheritance
scalar was set at 0.25. Metropolis coupling was
implemented using 20 chains and a geometric heating
model (term 1: 0.99; term 2: 0.95). The burn-in period
was 7 h. Upper bounds were set for prior distributions
for theta, migration rate and the splitting time. The final
run saved 499 469 trees per locus and ran for 705 h
(50 million steps following burn-in). Convergence was
tested by ensuring that effective sample sizes exceeded
50 and parameter trend lines were flat.
We analysed the microsatellite data using approximate
Bayesian computation (ABC) to estimate demographic
parameters of the colonization history of
Orkney by M. arvalis. Comparisons were made between
the continental European and Mainland Orkney
populations listed in Table 1 (and mapped in Fig. S1,
Supporting information), except that the four continental
populations with small sample sizes (15 individuals
or fewer) were excluded. Again, we focused on populations
from the region of continental Europe that most
likely represented a source for the introduction. We
specified an ABC model with 11 demographic parameters
in which the Orkney population diverged from a
continental population after a bottleneck. Moreover,
both the Orkney and the continental populations were
allowed to pass through independent bottlenecks. The
bottleneck of the continental population could occur
either before or after the colonization of Orkney. For
simplicity, the model was simulated with instantaneous
growth after all three bottleneck events. Further details
are given in Data S1, Table S5 and Fig. S3 (Supporting
information).
Results
Characteristics of the Orkney voles
Among extensive samples of Microtus arvalis collected
from across their European range, it is clear that the
island populations on Orkney (seven samples) and
Guernsey (one sample) are highly distinctive compared
with their continental counterparts in first lower molar
morphology (Fig. 1). This relates to both size (the island
voles generally had larger teeth: Fig. 1e) and shape (the
Orkney voles were divergent from continental and insular
European M. arvalis in the principal components
scatter plot, with the M1 displaying a relatively broader
anterior loop: Fig. 1c, d).
At the 14 microsatellite loci screened, the mean number
of alleles per locus and heterozygosities were systematically
lower in the Orkney populations than in
northern France, Belgium and nearby areas of Germany
(Table 1), the areas of continental Europe from where
introduced voles most likely originated. There was no
overlap in heterozygosity values and the small overlap
in mean number of alleles per locus related to Mainland
Orkney populations, which had distinctly higher diversity
values than other Orkney Island populations.
Overall FST between population samples was very
high (0.382, P < 0.0001) in agreement with previous
studies on the species (e.g. Heckel et al. 2005). Pairwise
comparisons between Orkney populations ranged from
FST = 0.094 to 0.682 (mean: 0.412) and between 0.007
and 0.346 for the continental populations (mean: 0.177;
Table S6, Supporting information).
Special features of the mitochondrial DNA variation
Together with previously published results, our new data
confirm that M. arvalis of the same (Western-North) cytb
lineage found today in France, Belgium, the Netherlands,
Germany and Switzerland also occurs on Orkney (Fig. 2
and Table S2, Supporting information). Fig. 2 shows very
clearly that the coastline of France and Belgium provides
possible maritime access of the Western-North cytb lineage
to Orkney, consistent with our contention that this
is the most likely source region for an introduction of
M. arvalis there.
In the Bayesian phylogenetic tree (Fig. 2) and the
median-joining network (Fig. 3), the Orkney cytb
sequences form an unsupported monophyletic clade
within the Western-North lineage. However, within
that context, the sequences are very distinct from those
from coastal Belgium/France and other regions in continental
Europe, separated by at least four mutations.
Thus, despite detailed sampling of the Western-North
lineage in general and along the coast of France and
Belgium in particular (Fig. 2), we found no cytb
sequences there that are clearly ancestral to those of the
Orkney voles.
Instead, over this coastal region, an area of more than
10 000 km2
(Table 3), a single mtDNA haplotype (and
sequences separated by one mutation from it) is dominant.
None of these haplotypes relate closely to the
Orkney M. arvalis haplotypes (Fig. 3). This ‘starburst’ of
the coastal French and Belgian sequences suggests their
recent expansion and dispersal. A significant deviation
from neutral expectation for Tajima’s D and Fu’s Fs, a
small Rozas’s R2 and a significant signal in pairwise
DNA sequence comparisons in the mismatch distribution
are consistent with this (Table 3).
Considering now the Orkney sequences, the same
cytb lineage and two of the same haplotypes as those in
modern specimens were also found in M. arvalis dating
back to the Neolithic period (Fig. 3; Table S3, Supporting
information). Interestingly, the central Orkney
© 2013 The Authors. Molecular Ecology Published by John Wiley & Sons Ltd.
5212 N. MARTINKOVÁ ET AL.
Paper 2.2.3 189
haplotype found in archaeological specimens from
Mainland Orkney and Westray was detected in modern
specimens from Westray and Burray not Mainland
Orkney. A second haplotype was found in both archaeological
and modern sequences from Mainland Orkney.
Both archaeological and modern sequences from Mainland
Orkney are represented by many haplotypes in the
network (9 and 12, respectively), and there are five
archaeological and two modern haplotypes from Westray
(Fig. 3).
Inland
Coastal
Ancient
Guernsey
Continental
Orkney
South Ronaldsay
Modern
Burray
Orkney mainland
Rousay
Sanday
Westray
Sample age
1
25
56
3 substitutions
Fig. 3 Median-joining network showing all modern and ancient DNA haplotypes in the Western-North phylogroup of Microtus
arvalis. The phylogroup is defined in Fig. 2. Node size is proportional to haplotype frequency and edge length to number of substitutions
separating the haplotypes. Guernsey is an offshore island also analysed in the geometric morphometric analysis (see Fig. 1).
Table 3 Modern mtDNA comparison of Microtus arvalis in Orkney and continental Europe
Coastal France and Belgium sampled* Orkney Mainland All Orkney Islands sampled
Total area (km2
) 10 439 52 73
Number of individuals sequenced 68 42 97
Haplotype diversity (h Æ SD) 0.785 Æ 0.039 0.864 Æ 0.029 0.911 Æ 0.108
Nucleotide diversity (p Æ SD) 0.00127 Æ 0.00017 0.00356 Æ 0.00201 0.00416 Æ 0.00227
Tajima’s D À1.986 À1.102 À1.167
Fu’s FS À10.035 À1.455 À3.014
Rozas’s R2 0.1616 0.0739 0.0587
Mismatch distribution (SSD) 0.0042 0.0112 0.0037
Characteristics of the modern mtDNA sequences of M. arvalis from the Orkney archipelago and coastal regions of France and
Belgium (where the Orkney M. arvalis most likely originated, see text) relating to the population expansion in the two areas.
For neutrality test statistics and the mismatch distribution, significant values (P < 0.01) consistent with population expansion are
given in bold.
*This included all localities within 100 km of the coast. The area sampled was estimated conservatively as a rectangle with one side
given by the distance between the end localities along the coast and the other by the mean distance to the coast of all localities.
© 2013 The Authors. Molecular Ecology Published by John Wiley & Sons Ltd.
COLONIZATION OF OFFSHORE ISLANDS 5213
190 Paper 2.2.3
A possible source area for the introduction of
M. arvalis to Orkney
We used ABC on the microsatellite data to compare 12
current populations (both coastal and inland) from
northern France, Belgium and nearby areas of Germany
as possible source populations for Orkney voles (Table 1
and Fig. S1, Supporting information). Our results confirmed
that, if M. arvalis were introduced to Orkney
from this source region, they most likely originated from
coastal Belgium/France (three of the four best-supported
populations: Table 1) rather than inland continental populations.
The best-supported modern source population
is from Stalhille on the Belgian coast (Fig. 2 and Table 1)
and geographically the closest point to Orkney within
the distribution of the Western-North lineage.
Time of colonization of Orkney
We obtained new 14
C dates for 23 specimens from
archaeological sites dating from the Late Neolithic
period to the Viking Age (Table 2). The estimated dates
are consistent with those of the archaeological contexts
from which their bones were recovered. They show
definitively that M. arvalis was present on Orkney at
least 5100 BP. Two molecular analyses suggest that they
did not arrive much earlier than this. An ABC analysis
of microsatellites shows modal point estimates of time
of colonization of Orkney that vary between 3000 and
5700 BP (between Stalhille, Belgian coast and three localities
on Mainland Orkney). An IMa analysis (Hey &
Nielsen 2007) of ancient and modern mtDNA sequences
and microsatellites yields an estimated colonization
time of 4515 years (90% CI: 2568–9019). Both analyses
assume a single generation per year. Multiple generations
(for which there is no evidence: Gorman &
Reynolds 2008) would imply a more recent postNeolithic
colonization, which does not fit well with our
new 14
C dates (voles on Orkney at latest 5100 BP) and
aDNA sequencing of archaeological specimens (same
genetic lineage as the modern specimens).
While the radiocarbon and molecular dating indicate
that voles arrived on Orkney about 5000 BP, the time to
the most recent common ancestor (tMRCA) for the
Orkney cytb sequences is much earlier, c. 15 400 years
(95% CI of 8100–25 400 years), based on Bayesian coalescent
analysis incorporating modern and ancient DNA
data. Using the same mutation rate (l = 3.27 9 10À7
mutations/site/year) as for Orkney sequences, the
tMRCA for the sequences from coastal France/Belgium
(the putative source area for the Orkney voles) is only
2356 years (95% CI of 723–4941 years). Mismatch analysis
(Rogers & Harpending 1992) provides further
evidence of the recentness of the derivation of the cytb
sequences in this potential source area for the Orkney
voles, showing an expansion beginning 1860 BP (95%
CI: 1201–2665).
Discussion
Cytb mutation rate
Estimates of mutation rates for phylogeographical studies
have traditionally been based on fossil calibrations,
often founded on very incomplete information. For
Microtus, Triant and DeWoody (2006) used a fossil calibration
point of 1.5 Ma to provide a mutation rate
for cytb of approximately 8 9 10À8
mutations/site/year.
Although this is probably the best-substantiated fossilbased
estimate of the cytb mutation rate for Microtus in
the literature, we considered it inadequate for our
needs. This is because the fossil calibration point likely
has errors (it is not based on precisely dated geological
events or fossil transitions) and is geologically very
much earlier than the time frame for our study. Instead,
we were able to calibrate our rate estimate using the
aDNA sequences from radiocarbon-dated specimens as
part of a Bayesian coalescent analysis. This method has
the advantage of producing an estimate of the cytb
mutation rate relevant to precisely the population that
we were studying and the temporal scale of interest.
The mutation rate we obtained and used here
(3.27 9 10À7
mutations/site/year) was about four times
higher than that of Triant and DeWoody (2006), but
comparable to other aDNA-based estimates of mutation
rate for a variety of species (Ho et al. 2011). Navascues
& Emerson (2009) caution against possible upward bias
of such aDNA-based estimates under extreme demographic
scenarios, but there is a growing consensus in
the literature that aDNA-based estimates are better than
traditional fossil-based mutation rates, which are too
low for dating divergences and demographic events
that have occurred over the last few thousand years
(Ho & Larson 2006; de Bruyn et al. 2011). With regard
to our aDNA-based cytb mutation rate for M. arvalis, it
is extremely close to that obtained for another Microtus,
which used a very recent geological event as a calibration
point (M. agrestis; 3.89 9 10À7
mutations/site/year;
Herman & Searle 2011). This provides independent support
for our aDNA-based estimate, although, as with all
molecular dating, interpretations have to be viewed
with caution.
Colonization history of Orkney voles
It is striking that there is substantial cytb variation in
M. arvalis in Mainland Orkney and over the whole
archipelago (Fig. 3, Table 3). Island populations often
© 2013 The Authors. Molecular Ecology Published by John Wiley & Sons Ltd.
5214 N. MARTINKOVÁ ET AL.
Paper 2.2.3 191
show low genetic diversity (Frankham 1997). This can
relate to small population sizes and/or population
bottlenecks associated with colonization of islands,
particularly by sweepstake dispersal or human introduction.
The high cytb diversity could indicate that the
Orkney population of M. arvalis represents an island
relict of a previously continuous mainland population,
perhaps dating back to before the LGM (Beirne 1952).
This would fit with the long tMRCA for the molecular
variation on Orkney, potentially dating back to
25 400 BP (within the 95% CI).
However, there are strong arguments against glacial
survival of the Orkney vole population. First, the IMa
analysis based on microsatellites and cytb and the ABC
analysis based on microsatellites provide a date of arrival
around 5000 BP, considerably more recent than the
LGM. Second, all the other species of small mammals
on Orkney are most reasonably viewed as human introductions
(Yalden 1982), necessitating a special case for
M. arvalis as a glacial survivor. Third, it is very difficult
to make this special case given that M. arvalis is not a
species currently associated with arctic or even moderately
high latitude conditions. Its range extends eastward
beyond Lake Baikal and yet barely traverses
north of the 60th parallel (Fig. 2; Shenbrot & Krasnov
2005). Orkney was under or near a glacial ice sheet at
the LGM (Bowen et al. 2002) and M. arvalis is not part
of the fossil fauna known from Britain from the last glacial
period (Yalden 1999; Currant & Jacobi 2001).
Fourth, M. arvalis is not currently found in Britain
(Fig. 2). It is therefore contrary to think that M. arvalis
should be a glacial relict on Orkney rather than
M. agrestis, when the latter occurs further north in Eurasia
(beyond the 70th parallel) and is distributed
throughout Britain, including on many offshore islands,
while M. arvalis only occurs on Orkney. There have
been no land connections between mainland Britain
and Orkney after conditions ameliorated following the
LGM (Yalden 1982), and hence why M. agrestis (and
other wide-ranging small mammals in Britain, such as
common shrews Sorex araneus and bank voles Myodes
glareolus) failed to colonize Orkney. If M. arvalis did not
survive on Orkney itself during the last glacial period,
the absence of the species in Britain means there are no
grounds to suggest sweepstake colonization from there.
It is conceivable that there could have been sweepstake
colonization of Orkney from Doggerland, the landmass
connecting Britain, the Low Countries and Denmark
until about 8000 BP (Weninger et al. 2008) – but this
would require the survival of small mammals on floating
mats of vegetation over a substantial marine gap
between Doggerland and Orkney.
Thus, human introduction is by far the most likely
explanation for the occurrence of M. arvalis on Orkney.
From the IMa and ABC dates, this introduction at about
5000 BP fits well with the earliest radiocarbon dates for
archaeological M. arvalis from Neolithic contexts
(5100 years old: Table 2) and the beginnings of the Neolithic
culture on Orkney (5600 BP: Ritchie 2001; Schulting
et al. 2010). Voles could have been brought to Orkney
by Mesolithic hunter-gatherers, as early as c. 9000 BP,
but no vole remains have been found in the one excavated
Mesolithic site on Orkney (Lee & Woodward
2009), in contrast to their abundance at Neolithic and
later sites (Yalden 1999; Thaw et al. 2004).
If, as appears most likely, the voles were introduced
by Neolithic settlers about 5000 BP, various other implications
flow from our molecular data, which are of considerable
archaeological interest.
First, the introduction implies long-distance maritime
travel by Neolithic people between continental Europe
and Orkney, extending on findings from elsewhere (e.g.
Broodbank 2006). Our study highlights the Belgian
coastline as the most reasonable source of the Orkney
voles on the basis of available genetic data. This
suggests Neolithic cultural linkages between Belgium
and Orkney, of worthwhile focus for future archaeological
investigation. Microtus arvalis were not introduced
successfully into mainland Britain, which is consistent
with relatively direct transport to Orkney from the continental
source area.
Second, if the introduction occurred about 5000 BP,
then, because the tMRCA for the Orkney voles is so
long (15 400 years), substantial numbers of female
voles must have been introduced to explain the cytb
variation observed in modern and archaeological Orkney
voles. High genetic diversity is already evident in
the 16 aDNA sequences dating to 4200 BP or earlier,
separated by up to 10 mutations (Fig. 3 and Table S3,
Supporting information) and which produce an estimate
for the tMRCA (14 780 years; 95% CI of 4681–
36 379 years) similar to that of the full ancient and
modern data set.
To explain a substantial number of voles arriving
accidentally on Orkney implies transport of plentiful
grass livestock bedding/fodder in which the vole stowaways
could have survived. This in turn may suggest
the direct movement of livestock as part of the proposed
Neolithic linkage between Belgium and Orkney.
Alternatively, deliberate transport of voles onto
Orkney could explain the large numbers introduced
(Thaw et al. 2004). It is conceivable that voles were
taken as food items, pets or for cultural/religious
purposes – M. arvalis is docile in captivity (Berry 2000),
so could theoretically have been ‘tamed’. This suggestion
of deliberate transportation of small rodents has a
precedent: it has been argued that Pacific rats (Rattus
exulans), now present on islands throughout Oceania,
© 2013 The Authors. Molecular Ecology Published by John Wiley & Sons Ltd.
COLONIZATION OF OFFSHORE ISLANDS 5215
192 Paper 2.2.3
were intentionally conveyed by Polynesians as a food
source (Matisoo-Smith & Robins 2004).
Evolutionary processes affecting voles on Orkney and
the continental source area
Compared with continental European M. arvalis, those
on Orkney and another offshore island (Guernsey) are
divergent in terms of tooth morphology, including
increased tooth size. For Orkney, this divergence may
have occurred over c. 5000 years, if introduced during
the Neolithic. In addition to having larger teeth, the
Orkney and Guernsey M. arvalis have a larger body size
than continental voles (Gorman & Reynolds 2008).
Quick-evolving rodent gigantism has been described
previously on islands of the northeast Atlantic (Corbet
1961; Angerbjorn 1986), but not within such a precisely
defined time frame. A range of selective factors have
been proposed to explain this gigantism, including an
absence of small mammalian predators (Lomolino
1985), and a genetic basis for gigantism has been identified
in island house mice (Chan et al. 2012). Elsewhere,
we further explore the dynamics of morphological evolution
for the Orkney M. arvalis using archaeological
specimens (T. Cucchi, R. Barnett, N. Martınkova, et al.
submitted) extending substantially on previous studies
(Berry & Rose 1975; Corbet 1986).
Despite their similarity in large tooth and body size,
Guernsey and Orkney voles exhibit distinctive mtDNA
haplotypes (Fig. 3). It is therefore most reasonable to consider
that the Guernsey and Orkney voles attained their
large body size independently. It is not clear whether
Guernsey was colonized naturally before it became an
island, as part of the continental European late glacial/
postglacial species expansion (Haynes et al. 2003; Heckel
et al. 2005; Tougard et al. 2008), or whether the voles were
introduced by people after it became an island (Gorman
& Reynolds 2008). However, given that Guernsey voles
are likely to come from the same general (northern
France/Belgium) source area as the Orkney voles and
that they are also different in mtDNA from current northern
France/Belgium populations, there would be much
interest in further detailed comparison of Orkney, Guernsey
and northern France/Belgium voles.
In addition to the operation of selection in the evolution
of Orkney voles suggested by morphology, stochastic
processes appear to have been important based on
microsatellites. The population on the largest island,
Mainland Orkney, has retained much of the microsatellite
variation found in continental Europe, while all the
other Orkney Islands (which are considerably smaller:
Fig. 1) show very low levels of microsatellite variation,
consistent with founder events and genetic drift. Similar
stochastic processes can also explain microsatellite variation
among Scottish Island populations of common
shrew (White & Searle 2007a).
Our findings with regard to morphology and microsatellites
in M. arvalis are unsurprising in comparison
with previous studies on island small mammals, but
the results from our mtDNA analyses are more unexpected.
Although the cytb sequences from Orkney and
the proposed source area for the Orkney colonization
both belong to the Western-North lineage of M. arvalis,
the sequences are remarkably divergent given the time
frame for colonization. Also, it might have been
expected that (as for the microsatellites) variability
would have been lower on Orkney than in continental
Europe. In fact, the opposite is the case. Taking either
the principal island (Mainland Orkney) or the whole
archipelago, mtDNA diversity is higher in Orkney than
in coastal France/Belgium (Table 3). Our dating analysis
also shows that the mtDNA sequences in coastal
France/Belgium have a much more recent derivation
than the Orkney sequences.
So, here we are seeing another facet of evolution in
association with the colonization of offshore islands, in
this case occurring in the mainland population. The
presence of derived sequences in coastal France/Belgium
suggests a replacement event in M. arvalis, with
one mtDNA type (the current type) replacing another
(the Orkney type), similar to aDNA findings in other
species (Barnes et al. 2002; Pergams et al. 2003; Hofreiter
et al. 2007). The fact that there is an affiliation between
coastal Belgium and Orkney on the basis of microsatellite
genotypes argues against a complete population
replacement (e.g. by extinction–recolonization) as an
explanation for the mtDNA result. Instead, within-population
processes of selective sweeps or genetic drift are
implicated, and more likely expressed in the mtDNA
data, as a single locus with small effective population
size than in the microsatellite data. We cannot be sure
what environmental factors promoted the replacement.
There could, for instance, have been a local, unrecorded
disease outbreak. However, it is notable that the
replacement occurred over a period when M. arvalis
populations would have changed dramatically due to
human land-use change, and this appears the most
likely driver of the replacement. Over several thousand
years, sustained forest clearance in continental Europe
(Rackham 1998; Cyprien et al. 2004) created new agricultural
habitats and associated selection pressures that
essentially expanded the opportunities for M. arvalis as
a species that particularly exploits managed grassland
(Niethammer & Krapp 1982). In such a habitat, M. arvalis
populations can undergo massive population expansions
and crashes (Delattre et al. 1992) that reduce longterm
effective population size, promoting genetic
change through drift. On Orkney (which saw the rapid
© 2013 The Authors. Molecular Ecology Published by John Wiley & Sons Ltd.
5216 N. MARTINKOVÁ ET AL.
Paper 2.2.3 193
decline of low shrubs and tree species with the arrival
of Neolithic farmers: Bunting 1996), M. arvalis utilizes a
range of open habitats and does not show the same dramatic
population fluctuations as seen in parts of continental
Europe (Gorman & Reynolds 2008).
Offshore islands as field laboratories
There has been a tendency to view offshore island populations
of small mammals (and other organisms with
low density and low dispersal) as genetic deviants from
the ‘norm’. This is because studies of various species
have shown results similar to ours for morphology
(substantial change) and microsatellites (loss of variation)
(Lomolino 1985; Frankham 1997; Boessenkool et al.
2007; Millien 2011). These have included detailed studies
on small mammals such as wood mouse Apodemus
sylvaticus (Angerbjorn 1986; Michaux et al. 1996),
masked shrew Sorex cinereus (Stewart & Baker 1992)
and common shrew (White & Searle 2007a,b, 2008).
However, as we have demonstrated with our M. arvalis
mtDNA studies, island populations can also represent
genetic ‘arks’, retaining the ancestral genetic variation,
while evolutionary and other processes on the mainland
may lead to a loss of that ancestral variation. Islands may
have importance therefore in conservation of genetic variation.
A further example involving human introduction
of a small mammal onto an offshore island is provided
by the Eurasian red squirrel Sciurus vulgaris. Thus, Irish
red squirrels have genetic variants that apparently derive
by introduction from Britain, but these are now absent in
that source population (Finnegan et al. 2008; Searle 2008).
For low density and low dispersal organisms such as
small mammals, we suggest that genetic surveys of mainland
areas should, where available, include populations
from neighbouring offshore islands. It is very likely that
those island populations will provide a new perspective
on the temporal and spatial dynamics of the genetic
variation in that region.
The ‘ark’ concept that we discuss here is of course
more general. Populations colonizing new areas will take
the genetic and nongenetic characteristics of the source
population, and some of those characteristics may subsequently
be lost in the source population but retained in
the population in the new area. In this way, for instance,
the United States is a ‘linguistic ark’ for various English
words that would have been common in the British Isles
at the time of settlement of North America by the British,
but which have subsequently fallen into disuse in the
homeland (e.g. ‘fall’ meaning ‘autumn’).
Returning to genetic characteristics of offshore islands,
in addition to their potential as genetic ‘arks’, they also
hold potential as field laboratories to study genetic
change in the islands themselves. Compared with the
classic evolutionary studies on oceanic islands, those
based on offshore islands will tend to view events over
shorter timescales and thus provide a different perspective
on evolutionary processes. Offshore islands are particularly
valuable for studying initial stages of
diversification, with the opportunity (as in the current
study together with T. Cucchi, R. Barnett, N. Martınkova,
et al. submitted) to follow island populations from their
foundation to the present day using advanced genetic
and morphometric tools as applied to modern and
ancient populations of different ages. Extremely accurate
dating of ancient populations may be possible (e.g. in
archaeological settings). With this short time duration
and close proximity to the mainland, there is also a
greater chance to find the precise source area for the
island colonization, which allows interesting comparison
of evolutionary processes on the mainland and island.
This brings us back to the value of offshore island populations
in interpreting mainland processes. Offshore
islands are an underutilized resource for evolutionary
analysis, with great potential. In some ways, they represent
study systems intermediate between those in a continental
setting and those on oceanic islands; they have the
simplicity of the oceanic island system yet are clearly relevant
to continental situations.
Acknowledgements
We dedicate this article to Michel Pascal, our outstanding coauthor,
who died on 5 January 2013, and to Anne Brundle,
who gave us access to much archaeological material and who
died in 2012. We acknowledge receipt of a Marie Curie Intra
European Fellowship (to N.M.), support from the Swiss
National Science Foundation (projects 31003A-127377, 3100A0-
112072 and 3100-126074) to L.E. and G.H., funding from SYNTHESYS2
made available by the European Community
Research Infrastructure under FP7 (‘Synthesis of Systematic
Resources’, 226506-CP-CSA-Infra) to S.B., a Wellcome Trust
University award to K.M.D. (GR071037) and overarching funding
from the Arts and Humanities Research Council (project
grant 119396). We thank V. Bretille, A. Frantz, M. Fuster, N.
Gould, J. Herman, E. Jones, S. Martınek, R. Marwick, J. Michaux,
S. Montuire, J. Pauperio, C. Scott, C. Tougard, B. Walther
and N. Wheale for field specimens, T. White for assistance
with IMa runs, and A. Ritchie, L. Shepherd and A. Sheridan
for archaeological advice. We are grateful to the following for
museum and archaeological samples: J. Barrett (MacDonald
Institute, University of Cambridge), A. Brundle (Orkney
Museum), C. David (Guernsey Museum), A. Ervynck (Flemish
Heritage Institute), L. Gordon (Smithsonian Institute), J. Herman
(National Museums of Scotland), D. Lee (Orkney College),
R. Sabin (British Museum - Natural History, London) and
G. Veron (Museum national d’histoire naturelle, Paris).
References
Angerbjorn A (1986) Gigantism in island populations of wood
mice (Apodemus) in Europe. Oikos, 47, 47–56.
© 2013 The Authors. Molecular Ecology Published by John Wiley & Sons Ltd.
COLONIZATION OF OFFSHORE ISLANDS 5217
194 Paper 2.2.3
Bandelt HJ, Forster P, R€ohl A (1999) Median-joining networks
for inferring intraspecific phylogenies. Molecular Biology and
Evolution, 16, 37–48.
Barnes I, Matheus P, Shapiro B, Jensen D, Cooper A (2002)
Dynamics of Pleistocene population extinctions in Beringian
brown bears. Science, 295, 2267–2270.
Beirne BP (1952) The Origin and History of the British Fauna.
Methuen, London.
Berry RJ (1996) Small mammal differentiation on islands. Philosophical
Transactions of the Royal Society of London. B, Biological
Sciences, 351, 753–764.
Berry RJ (2000) Orkney Nature. Poyser, London.
Berry RJ, Rose FEN (1975) Islands and the evolution of Microtus
arvalis (Microtinae). Journal of Zoology, 177, 395–409.
Boessenkool S, Taylor SS, Tepolt CK, Komdeur J, Jamieson IG
(2007) Large mainland populations of South Island robins
retain greater genetic diversity than offshore island refuges.
Conservation Genetics, 8, 705–714.
Bookstein FL (1997) Landmark methods for forms without
landmarks: morphometrics of group differences in outline
shape. Medical Image Analysis, 1, 225–243.
Borkowska A, Ratkiewicz M (2008) Sex-related spatial structure
and effective population size in the common vole, Microtus
arvalis, as revealed by mtDNA analysis. Annales Zoologici
Fennici, 45, 255–262.
Bowen DQ, Phillips FM, McCabe AM, Knutz PC, Sykes GA
(2002) New data for the last glacial maximum in Great Britain
and Ireland. Quaternary Science Reviews, 21, 89–101.
Broodbank C (2006) The origins and early development of
Mediterranean maritime activity. Journal of Mediterranean
Archaeology, 19, 199–230.
de Bruyn M, Hoelzel AR, Carvalho GR, Hofreiter M (2011)
Faunal histories from Holocene ancient DNA. Trends in Ecology
and Evolution, 26, 405–413.
Bunting MJ (1996) The development of heathland in Orkney,
Scotland: pollen records from Loch of Knitchen (Rousay)
and Loch of Torness (Hoy). The Holocene, 6, 193–212.
Chan YF, Jones FC, McConnell E et al. (2012) Parallel selection
mapping using artificially selected mice reveals body weight
control loci. Current Biology, 22, 794–800.
Corbet GB (1961) Origin of British insular races of small mammals
and of Lusitanian fauna. Nature, 191, 1037–1040.
Corbet GB (1986) Temporal and spatial variation of dental pattern
in the voles, Microtus arvalis, of the Orkney Islands. Journal
of Zoology, 208, 395–402.
Currant A, Jacobi R (2001) A formal mammalian biostratigraphy
for the Late Pleistocene of Britain. Quaternary Science
Reviews, 20, 1707–1716.
Cyprien AL, Visset L, Carcaud N (2004) Evolution of vegetation
landscapes during the Holocene in the central and
downstream Loire basin (Western France). Vegetation History
and Archaeobotany, 13, 181–196.
Delattre P, Giraudoux P, Baudry J et al. (1992) Land use
patterns and types of common vole (Microtus arvalis) population
kinetics. Agriculture, Ecosystems & Environment, 39,
153–168.
DeWoody JA, Chesser RK, Baker RJ (1999) A translocated
mitochondrial cytochrome b pseudogene in voles (Rodentia:
Microtus). Journal of Molecular Evolution, 48, 380–382.
Drummond AJ, Rambaut A (2007) BEAST: Bayesian evolutionary
analysis by sampling trees. BMC Evolutionary Biology, 7, 214.
Drummond AJ, Ho SYW, Phillips MJ, Rambaut A (2006)
Relaxed phylogenetics and dating with confidence. PLoS
Biology, 4, e88.
Ehlers J, Gibbard PL (2004) Quaternary Glaciations – Extent and
Chronology. Part I: Europe. Elsevier, Amsterdam.
Emerson BC (2002) Evolution on oceanic islands: molecular
phylogenetic approaches to understanding pattern and process.
Molecular Ecology, 11, 951–966.
Excoffier L, Laval G, Schneider S (2005) Arlequin (version 3.0):
an integrated software package for population genetics data
analysis. Evolutionary Bioinformatics Online, 1, 47–50.
Fink S, Fischer MC, Excoffier L, Heckel G (2010) Genomic
scans support repetitive continental colonization events during
the rapid radiation of voles (Rodentia: Microtus): the utility
of AFLPs versus mitochondrial and nuclear sequence
markers. Systematic Biology, 59, 548–572.
Finnegan LA, Edwards CJ, Rochford JM (2008) Origin of, and
conservation units in, the Irish red squirrel (Sciurus vulgaris)
population. Conservation Genetics, 9, 1099–1109.
Frankham R (1997) Do island populations have less genetic
variation than mainland populations? Heredity, 78, 311–327.
Fu YX (1997) Statistical tests of neutrality of mutations against
population growth, hitchhiking and background selection.
Genetics, 147, 915–925.
Gorman ML, Reynolds P (2008) Orkney and Guernsey vole.
In: Mammals of the British Isles: Handbook, 4th edn (eds
Harris S, Yalden DW), pp. 107–110. The Mammal Society,
Southampton.
Goudet J (1995) FSTAT (Version 1.2): a computer program to
calculate F-statistics. Journal of Heredity, 86, 485–486.
Grant PR (ed.) (2008) Evolution on Islands. Oxford University
Press, Oxford.
Hall TA (1999) BioEdit: a user-friendly biological sequence
alignment editor and analysis program for Windows 95/98/
NT. Nucleic Acid Symposium Series, 41, 95–98.
Haynes S, Jaarola M, Searle JB (2003) Phylogeography of the
common vole (Microtus arvalis) with particular emphasis on
the colonization of the Orkney archipelago. Molecular Ecology,
12, 951–956.
Heckel G, Burri R, Fink S, Desmet JF, Excoffier L (2005) Genetic
structure and colonization processes in European populations
of the common vole, Microtus arvalis. Evolution, 59, 2231–2242.
Hedges REM, Housley RA, Law IA, Perry C, Gowlett JAJ
(1987) Radiocarbon dates from the Oxford AMS system:
archaeometry datelist 6. Archaeometry, 29, 289–306.
Herman JS, Searle JB (2011) Post-glacial partitioning of mitochondrial
genetic variation in the field vole. Proceedings of the
Royal Society. B, Biological Sciences, 278, 3601–3607.
Hey J, Nielsen R (2007) Integration within the Felsenstein equation
for improved Markov chain Monte Carlo methods in
population genetics. Proceedings of the National Academy of
Sciences USA, 104, 2785–2790.
Ho SYW, Larson G (2006) Molecular clocks: when times are
a-changin’. Trends in Genetics, 22, 79–83.
Ho SYW, Lanfear R, Phillips MJ et al. (2011) Bayesian
estimation of substitution rates from ancient DNA
sequences with low information content. Systematic Biology,
60, 366–375.
Hofreiter M, Munzel S, Conard NJ et al. (2007) Sudden replacement
of cave bear mitochondrial DNA in the late Pleistocene.
Current Biology, 17, R122–R123.
© 2013 The Authors. Molecular Ecology Published by John Wiley & Sons Ltd.
5218 N. MARTINKOVÁ ET AL.
Paper 2.2.3 195
Jaarola M, Martınkova N, G€und€uz _I et al. (2004) Molecular
phylogeny of the speciose vole genus Microtus (Arvicolinae,
Rodentia) inferred from mitochondrial DNA sequences.
Molecular Phylogenetics and Evolution, 33, 647–663.
Jaarola M, Ashford RT, Ratkiewicz M, Brunhoff C, Borkowska
A (2007) Isolation and characterization of polymorphic
microsatellite loci in the field vole, Microtus agrestis, and their
cross-utility in the common vole, Microtus arvalis. Molecular
Ecology Notes, 7, 1029–1031.
Klingenberg CP (2011) MorphoJ: an integrated software package
for geometric morphometrics. Molecular Ecology
Resources, 11, 353–357.
Lachaise D, Harry M, Solignac M et al. (2000) Evolutionary
novelties in islands: Drosophila santomea, a new melanogaster
sister species from Sao Tome. Proceedings of the Royal Society
of London. B, Biological Sciences, 267, 1487–1495.
Lee DHJ, Woodward NL (2009) Links House, Stronsay, Orkney
(Phase III). Data Structure Report, ORCA.
Lomolino MV (1985) Body size of mammals on islands - the
island rule reexamined. American Naturalist, 125, 310–316.
Marshall DC (2010) Cryptic failure of partitioned Bayesian
phylogenetic analyses: lost in the land of long trees. Systematic
Biology, 59, 108–117.
Martin Y, Gerlach G, Schl€otterer C, Meyer A (2000) Molecular
phylogeny of European muroid rodents based on complete
cytochrome b sequences. Molecular Phylogenetics and Evolution,
16, 37–47.
Martınkova N, Moravec J (2012) Multilocus phylogeny of arvicoline
voles (Arvicolini, Rodentia) shows small tree terrace
size. Folia Zoologica, 61, 254–267.
Martınkova N, McDonald RA, Searle JB (2007) Stoats (Mustela
erminea) provide evidence of natural overland colonization of
Ireland. Proceedings of the Royal Society. B, Biological Sciences,
274, 1387–1393.
Matisoo-Smith E, Robins JH (2004) Origins and dispersals of
Pacific peoples: evidence from mtDNA phylogenies of the
Pacific rat. Proceedings of the National Academy of Sciences
USA, 101, 9167–9172.
Michaux JR, Libois R, Fons R (1996) Differenciation genetique
et morphologique du mulot, Apodemus sylvaticus, dans le bassin
mediterraneen occidental. Vie Milieu, 46, 193–203.
Millien V (2011) Mammals evolve faster on smaller islands.
Evolution, 65, 1935–1944.
Morrison ML, Block WM, Jehl JR Jr, Hall LS (1992) Terrestrial
vertebrates of the Mono Lake Islands, California. Great Basin
Naturalist, 52, 328–334.
Navascues M, Emerson BC (2009) Elevated substitution rate
estimates from ancient DNA: model violation and bias of
Bayesian methods. Molecular Ecology, 18, 4390–4397.
Nei M (1987) Molecular Evolutionary Genetics. Colombia University
Press, New York.
Nichols C, Herman J, Gaggiotti OE et al. (2007) Genetic isolation
of a now extinct population of bottlenose dolphins
(Tursiops truncatus). Proceedings of the Royal Society. B, Biological
Sciences, 274, 1611–1616.
Niethammer J, Krapp F (1982) Microtus arvalis (Pallas, 1779) Feldmaus.
In: Handbuch der S€augetiere Europas 2 (1). Rodentia:
II (eds Niethammer J, Krapp F), pp. 284–318. Akademische
Verlagsgesellschaft, Wiesbaden.
Nylander JAA (2004) MrModeltest v2. Evolutionary Biology
Centre, Uppsala University, Uppsala.
Pergams ORW, Barnes WM, Nyberg D (2003) Rapid change in
mouse mitochondrial DNA. Nature, 423, 397.
Posada D, Crandall KA (1998) MODELTEST: testing the model
of DNA substitution. Bioinformatics, 14, 817–818.
R Development Core Team (2011) R: A Language and Environment
for Statistical Computing. Retrieved from http://www.
r-project.org
Rackham O (1998) Implications of historical ecology for conservation.
In: Conservation Science and Action (ed. Sutherland W),
pp. 152–175. Blackwell, Oxford.
Rambaut A, Drummond AJ (2007) Tracer v1.4. University of
Oxford, Oxford.
Rambaut A, Ho SYW, Drummond AJ, Shapiro B (2009) Accommodating
the effect of ancient DNA damage on inferences of
demographic histories. Molecular Biology and Evolution, 26,
245–248.
Ramos-Onsins SE, Rozas J (2002) Statistical properties of new
neutrality tests against population growth. Molecular Biology
and Evolution, 19, 2092–2100.
Reimer PJ, Baillie MGL, Bard E et al. (2004) IntCal04 terrestrial
radiocarbon age calibration, 0-26 cal kyr BP. Radiocarbon, 46,
1029–1058.
Rice WR (1989) Analyzing tables of statistical tests. Evolution,
43, 223–225.
Ritchie A (2001) Knap of Howar, Papa Westray. Discovery and
Excavation in Scotland, 2000, 124–125.
Rogers AR, Harpending H (1992) Population growth makes
waves in the distribution of pairwise genetic differences.
Molecular Biology and Evolution, 9, 552–569.
Rohlf FJ (2010a) TpsDig 2-Thin Plate Spline Digitizer. Ecology &
Evolution, State University at Stony Brook, New York.
Rohlf FJ (2010b) TpsRelw 1.49-Thin Plate Spline Relative Warp.
Ecology & Evolution, State University at Stony Brook, New
York.
Ronquist F, Huelsenbeck JP (2003) MrBayes 3: Bayesian phylogenetic
inference under mixed models. Bioinformatics, 19,
1572–1574.
Rozas J, Sanchez-DelBarrio JC, Messeguer X, Rozas R (2003)
DnaSP, DNA polymorphism analyses by the coalescent and
other methods. Bioinformatics, 19, 2496–2497.
Schulting RA, Sheridan A, Crozier R, Murphy E (2010) Revisiting
Quanterness: new AMS dates and stable isotope data
from an Orcadian chamber tomb. Proceedings of the Society of
Antiquaries of Scotland, 140, 1–50.
Searle JB (2008) The genetics of mammalian invasions: a
review. Wildlife Research, 35, 185–192.
Shenbrot GI, Krasnov BR (2005) An Atlas of the Geographic Distribution
of the Arvicoline Rodents of the World (Rodentia, Muridae:
Arvicolinae). Pensoft Publishers, Sofia, Moscow.
Simpson GG (1940) Mammals and land bridges. Journal of the
Washington Academy of Sciences, 30, 137–163.
Stamatakis A (2006) RAxML-VI-HPC: Maximum likelihoodbased
phylogenetic analyses with thousands of taxa and
mixed models. Bioinformatics, 22, 2688–2690.
Stewart DT, Baker AJ (1992) Genetic differentiation and biogeography
of the masked shrew in Atlantic Canada. Canadian
Journal of Zoology, 70, 106–114.
Stuiver M, Reimer PJ (1993) Extended 14
C database and revised
CALIB radiocarbon calibration program. Radiocarbon, 35,
215–230.
© 2013 The Authors. Molecular Ecology Published by John Wiley & Sons Ltd.
COLONIZATION OF OFFSHORE ISLANDS 5219
196 Paper 2.2.3
Tajima F (1989) Statistical method for testing the neutral mutation
hypothesis by DNA polymorphism. Genetics, 123,
585–595.
Thaw S, Jaarola M, Searle JB, Dobney KM (2004) The origin of
the Orkney vole Microtus arvalis orcadensis: a proxy for reconstructing
human movements. In: Atlantic Connections and
Adaptations (eds Housley RA, Coles G), pp. 114–119. Oxbow,
Oxford.
Tougard C, Renvoise E, Petitjean A, Quere J-P (2008) New
insight into the colonization processes of common voles:
inferences from molecular and fossil evidence. PLoS ONE, 3,
e3532.
Triant DA, DeWoody JA (2006) Accelerated molecular evolution
in Microtus (Rodentia) as assessed via complete mitochondrial
genome sequences. Genetica, 128, 95–108.
Wallace AR (1880) Island Life. Macmillan, London.
Walser B, Heckel G (2008) Microsatellite markers for the common
vole (Microtus arvalis) and their cross-species utility.
Conservation Genetics, 9, 479–481.
Weninger B, Schulting R, Bradtm€oller M et al. (2008) The catastrophic
final flooding of Doggerland by the Storegga Slide
tsunami. Documenta Praehistorica, 35, 1–24.
White TA, Searle JB (2007a) Genetic diversity and population
size: island populations of the common shrew, Sorex araneus.
Molecular Ecology, 16, 2005–2016.
White TA, Searle JB (2007b) Factors explaining increased body
size in common shrews (Sorex araneus) on Scottish islands.
Journal of Biogeography, 34, 356–363.
White TA, Searle JB (2008) The colonization of Scottish islands
by the common shrew, Sorex araneus (Eulipotyphla: Soricidae).
Biological Journal of the Linnean Society, 94, 797–808.
Yalden DW (1982) When did the mammal fauna of the British
Isles arrive? Mammal Review, 12, 1–57.
Yalden DW (1999) The History of British Mammals. Poyser,
London.
K.M.D. and J.B.S. conceived, initiated and coordinated
the project. T.C., M.F., G.H., N.M., M.P., Ma.P., J.-P.Q.
and J.B.S. organized and collected field specimens, R.B.,
T.C. and K.M.D. organized and obtained museum and
archaeological samples. R.B., T.C. and N.M. conducted
laboratory work supported by K.M.D., A.R.H., G.H.
and J.B.S. and with major contributions from S.B. and
T.H. R.B., T.C., N.M. and R.S. analysed the data supported
by K.M.D., L.E., P.O’H., A.R.H., G.H., S.Y.W.H.
and J.B.S. J.B.S. led the writing of the text to which all
authors contributed.
Data accessibility
The DNA sequences have been deposited in GenBank
(GU190383-GU190665). The TreeBASE entry for the
phylogenetic tree of cytb sequences shown in Fig. 2, the
microsatellite genotypes, the input file for the IMa analysis,
the modern and ancient cytb sequences, and the
morphological coordinates collected from 2D images of
skulls are all available through DRYAD:
doi:10.5061/dryad.9rf5m.
Supporting information
Additional supporting information may be found in the online
version of this article.
Data S1 Material and Methods.
Table S1 List of modern M. arvalis samples used for morphometrics,
including country and site of origin.
Table S2 List of all modern M. arvalis used for cytb analysis
and collected for this study, including country and site of origin;
arranged according to cytb haplotype.
Table S3 List of all ancient specimens of M. arvalis that successfully
provided a cytb sequence with details of location collected,
calibrated age range (where obtained) and GenBank
Accession Number for the sequence.
Table S4 List of primers used for the amplification of cytochrome
b from M. arvalis.
Table S5 Prior distributions of the ABC model parameters (as
illustrated in Fig. S3).
Table S6 Pairwise FST values between population samples
analysed with microsatellites (see Table 1 and Fig. S1).
Fig. S1 Map showing distribution of population samples of
modern M. arvalis used for microsatellite typing, labelled for
mtDNA lineage. Population names as listed in Table 1: 1 – Heerenveen,
2 – Dinteloord, 3 – Stalhille, 4 – Veurne, 5 – Pihen
les Gu^ınes, 6 – Fressenneville, 7 – Daubeuf, 8 – Thaon, 9 – Ste
Marie du Mont, 10 – St Jean du Thomas, 11 – Baie d’Aiguillon,
12 – Aiffres, 13 – Avallon, 14 – Clermont-Ferrand, 15 – Alflen,
16 – Schiltach, 17 – Loch of Swartmill, 18 – Ness, 19 – Whitemill
Bay, 20 – Settiscarth, 21 – Harray Stenness, 22 – St Ola, 23
– Grimness, 24 – Wind Wick.
Fig. S2 Map showing Orkney localities where ancient specimens
of M. arvalis successfully provided either radiocarbon
dates and/or cytb sequences. Localities as listed in Tables 2
and S3: 1 - Holm of Papa Westray, Westray, 2 - Point of Cott,
Westray, 3 - Pierowall Quarry, Westray, 4 - Quanterness,
Mainland, 5 - Earl’s Bu, Mainland, 6 - Howe, Mainland, 7 Skara
Brae, Mainland, 8 - Green Hill, South Walls, Hoy.
Fig. S3 Diagram illustrating the ABC model parameters.
© 2013 The Authors. Molecular Ecology Published by John Wiley & Sons Ltd.
5220 N. MARTINKOVÁ ET AL.
Paper 2.2.3 197
Paper 2.3.1
Irwin N. R., Bayerlová M., Missa O., Martínková N. 2012. Complex patterns of host
switching in New World arenaviruses. Molecular Ecology 21: 4137-4150.
– 199 –
Complex patterns of host switching in New World
arenaviruses
NANCY R. IRWIN,* MICHAELA BAYERLOVA´ ,† OLIVIER MISSA,* and NATA´ LIA
MARTI´NKOVA´ ,†‡
*Department of Biology, University of York, POBOX 373, York, YO10 5DD, UK, †Institute of Biostatistics and Analyses,
Masaryk University, Brno, Czech Republic, ‡Institute of Vertebrate Biology, v.v.i., Academy of Sciences of the Czech Republic,
Brno, Czech Republic
Abstract
We empirically tested the long-standing hypothesis of codivergence of New World
arenaviruses (NWA) with their hosts. We constructed phylogenies for NWA and all
known hosts and used them in reconciliation analyses. We also constructed a
phylogenetic tree of all Sigmodontinae and Neotominae rodents and tested whether
viral–host associations were phylogenetically clustered. We determined host geographical
overlap to determine to what extent opportunity to switch hosts was limited by host
relatedness or physical proximity. With the exception of viruses from North America, no
phylogenetically codivergent pattern between NWA and their hosts was found. We
found that different virus clades were clustered differently and that Clade B with
members pathogenic to humans was randomly distributed across the rodent phylogeny.
Furthermore, viral relatedness within Clade B was significantly explained by the
geographic overlap of their hosts’ ranges rather than host relatedness, indicating that
they are capable of host switching opportunistically. This has important bearings on
their potential to become panzootic. Together, these analyses suggest that NWA have not
codiverged with their hosts and instead have evolved predominantly via host switching.
Keywords: AxParafit, codivergence, cophylogeny, Phylocom, reconciliation, viral haemorrhagic
fever
Received 23 November 2011; revision received 1 May 2012; accepted 2 May 2012
Introduction
Arenaviruses (Arenaviridae) are zoonotic viruses, some
of which cause severe diseases in humans, primarily
characterized by haemorrhagic fevers and ⁄ or meningitis
with high mortality rates (Peters et al. 1996; Jay et al.
2005). Arenaviruses are lipid-enveloped bi-segmented
ambisense single-stranded RNA viruses classified into
two main groups, separated both serologically and geographically,
into the Old World (Lymphocytic choriomeningitis
virus) and New World (Tacaribe) complexes
(Bowen et al. 1997). The Old World arenaviruses
(OWA) contain 10 species, two of which cause diseases
in humans with high mortality rates (Ogbu et al. 2007;
Briese et al. 2009; de Bellocq et al. 2010). The New
World arenaviruses (NWA) contain 23 viral species that
are subdivided into four well-established lineages; lineages
A, B and C occur in Latin America (Bowen et al.
1996) and lineage A-Recombinant (A-Rec) in North
America (Charrel et al. 2001; Archer & Rico-Hesse
2002). Five viruses of lineage B, Junı´n, Machupo, Guanarı´to,
Sabia´ and Chapare, cause severe haemorrhagic
fever with a 10–33% case mortality (Gonzalez et al.
1996; Delgado et al. 2008). Viruses of lineages A, A-Rec
and C are not known to be pathogenic to humans, with
the exception of Flexal, which caused nonfatal infections
in laboratory workers (Peters et al. 1996) and Whitewater
Arroyo, which was implicated in three fatalities
(Enserink 2000).
The natural hosts of arenaviruses are rodents (Kunz
2009), with the exception of the Tacaribe virus discovered
Correspondence: Nancy R. Irwin, Fax: 44 (0)1904 328505;
E-mail: nancy.irwin@york.ac.uk
Ó 2012 Blackwell Publishing Ltd
Molecular Ecology (2012) 21, 4137–4150 doi: 10.1111/j.1365-294X.2012.05663.x
200 Paper 2.3.1
in two Artibeus bat species (Downs et al. 1963). Infected
rodents excrete the viruses through their urine and saliva
(Childs & Peters 1993b). Humans get infected either
by cuts or by bites or via inhalation or ingestion of
excreta particles (Dylla et al. 2008). People with regular
direct or indirect contact with wildlife, agricultural
workers (Childs et al. 1995), laboratory workers (Rousseau
et al. 1997) and in domestic residences (Kilgore
et al. 1995) are particularly susceptible. There are also
rare incidences of human-to-human infection by either
close contact (McCormick et al. 1987), nosocomial infection
(Carey et al. 1972; Briese et al. 2009) or organ transplantation
(Palacios et al. 2008). Feral rodents have
infected hamsters and guinea pigs for the pet trade
(Skinner & Knight 1979) and laboratory mice (Skinner
et al. 1977; Smith et al. 1984), which, in some cases,
have gone on to infect humans (Bowen et al. 1975;
Smith et al. 1984; Rousseau et al. 1997). Such virulent
properties mean that the pathogenic arenaviruses have
been classified in the highest category for infectious
agents (Category A) and even developed with the
intent to be used as a biological weapon (Borio et al.
2002).
The arenavirus genome consists of two singlestranded
RNA segments designated as the small segment
(S) and the large segment (L) (Bowen et al. 1997).
The S-segment (3500 bp) encodes the nucleocapsid
protein (NP) and glycoprotein precursor (GP). The L
segment (7100 bp) encodes the zinc-binding protein
(Z protein) and the RNA-dependent RNA polymerase
(L protein). Proteins of each segment are separated by
intergenic noncoding regions. The GP undergoes posttranslational
cleavage to generate envelope proteins G1
and G2 and a signal peptide (Charrel et al. 2002). The
GP protein is involved in cell entry to the hosts and is
therefore critical to the transmission of the virus (Radoshitzky
et al. 2008). To date, two cell receptors have
been identified as targets for cell entry. The first, the
alpha-dystroglycan (a-DG) receptor is used by OWA
(Cao 1998) as well as Latino and Oliveros viruses from
the NWA (Spiropoulou et al. 2002). The second is the
alpha transferrin receptor 1 protein (TfR1), an iron
transport receptor, used by Clade B of the NWA (Radoshitzky
et al. 2007; Flanagan et al. 2008; Abraham et al.
2009). These viruses also exploit an independent cell
entry pathway, currently not identified (Flanagan et al.
2008). It is currently not known which receptors Clade
A or A-Rec of NWA use to enter the cell (Spiropoulou
et al. 2002; Reignier et al. 2008).
Rodent species that are considered the primary hosts
are found to suffer from a persistent infection of a single
virus species and are capable of both vertical and
horizontal transmission (Childs & Peters 1993b). The
ability to survive with arenavirus infection is interpreted
as evidence of a long evolutionary adaptation
between the virus and its host (Childs & Peters 1993b;
Briese et al. 2009; Kunz 2009). Each virus is thought to
be associated with a particular host, and therefore, the
distribution of the host is likely to determine the distribution
of the virus (Charrel & de Lamballerie 2010).
The only globally distributed arenavirus is Lymphocytic
chorio-meningitis virus mediated by the distribution of
its host, the house mouse Mus musculus (Salazar-Bravo
et al. 2002). Rodents from the subfamily Murinae host
OWA, whereas in the New World, arenaviruses infect
members from the endemic subfamilies Sigmodontinae
and Neotominae (Bowen et al. 1997). Given that both
New and Old World arenaviruses and their respective
hosts form distinct evolutionary units, co-evolution
dating back to when the rodent groups diverged
(20–30 Mya) is often hypothesized (Childs & Peters
1993a; Bowen et al. 1997; Mills et al. 1997; Salazar-Bravo
et al. 2002; Gonzalez et al. 2007). Yet, there is evidence
that some viruses have switched host, in that some
viruses are found in more than one host species, for
example Bear Canyon virus infects both Neotoma
macrotis and Peromyscus californicus (Fulhorst et al. 2002;
Cajimat et al. 2007), and some host species are known
to harbour more than one virus from different clades
present in the same region, for example Zygodontomys
brevicauda in Venezuela has both Pirital virus (Clade A)
and pathogenic Guanarı´to (Clade B) (Weaver et al.
2000).
Bowen et al. (1997) hypothesized that the evolution of
arenaviruses would be best explained by co-evolution
(i.e. codivergence) with the occasional host switching,
but they lacked a phylogeny of New World rodents to
empirically test their hypothesis. Phylogenetic reconciliation
analysis has previously been limited to OWA
(Hugot et al. 2001; Hugot 2003; Coulibaly-N’Golo et al.
2011) and a study that included only 12 of the known
23 NWA (Jackson & Charleston 2004). Despite the
repeated assertion of a long history of codivergence (Jay
et al. 2005; Gonzalez et al. 2007; Radoshitzky et al.
2008), the studies showed no evidence of congruence
between the viral and host phylogenies. While Jackson
& Charleston (2004) concluded that no cophylogenetic
association existed, Hugot et al. (2001) concluded that
diffuse co-evolution (i.e. host switching on related
hosts) occurred on top of an ancient association
between arenaviruses and their hosts. In the most
recent assessment of the OWA, the authors concluded
that multiple host switching and extinction events may
have obscured any co-evolutionary signal (CoulibalyN’Golo
et al. 2011). With increasing availability of
molecular data from both the NWA and rodents,
improved host knowledge and advances in methodology,
AxParafit (Stamatakis et al. 2007), phylogenetic
4138 N. R. IRWIN ET AL.
Ó 2012 Blackwell Publishing Ltd
Paper 2.3.1 201
clustering (Webb et al. 2008), phylogenetically constrained
Mantel tests (Harmon & Glor 2010), it is now
possible to critically reassess the codivergence hypothe-
sis.
In this article, we tested the codivergence hypothesis
for New World arenaviruses. We found no evidence
that NWA have codiverged with their hosts, but instead
that they have evolved via host switching. We found
that each NWA clade is differently associated with their
hosts and demonstrate either host relatedness or geographic
proximity of their hosts has been the drivers of
host switching.
Materials and methods
We downloaded all available nucleotide sequences of
arenaviruses from GenBank. We used complete coding
region sequences of GP, NP, L and Z proteins for phylogenetic
analyses. We analysed all available mitochondrial
sequences of the complete cytochrome b gene of
all known hosts as well as all Sigmodontine and Neotomine
rodents of the Americas from GenBank. Nucleotide
sequences were aligned according to amino acid
alignment using global translation alignment in Geneious
Pro v4.7 (Biomatters Ltd.) (Drummond et al. 2009).
For each protein, we identified identical sequences and
removed them from the alignment in DAMBE 4.5 (Xia
& Xie 2001).
Host–pathogen associations
We assembled all known wild natural hosts of arenaviruses
from a literature review (host taxa = 24, associations
= 33; Table S1, Supporting information). The hosts
for the viruses Chapare and Sabia´ are unknown and are
therefore not included in any analysis. Likewise, missing
host sequence data for Oecomys paricola and Artibeus
lituratus prevented their viral associations being
included. The poor taxonomic knowledge of South
American rodents may lead to misleading associations
and patterns if virus collectors in the field did not collect
a sample of the host at the same time. Here, taxonomic
revision has changed the taxonomic names of
five of the hosts of NWA (see Supporting information).
All other associations are firmly established in the literature
(Table S1, Supporting information).
Phylogenetic reconstruction
We reconstructed phylogenetic trees for each protein
individually using the GTR+C+I substitution model
selected by all criteria in Modeltest 3.7 (Posada & Crandall
1998). No sequence data were available for the Pinhal
virus; therefore, 22 taxa were used in the GP
(1626 bp, n = 62) and NP (1770 bp, n = 64; Table S2,
Supporting information) phylogenetic analysis. Tonto
Creek, Catarina, Big Brushy Tank and Skinner Tank
viruses are also missing data for the L and Z proteins,
and these were therefore reconstructed with the remaining
18 NWA taxa (L protein, 6978 bp, n = 37; Z protein,
333 bp, n = 51; Table S2, Supporting information).
Bayesian inference (BI) analysis was run with Markov
chain Monte Carlo (MCMC) for two million generations
in MrBayes 3.1 (Ronquist & Huelsenbeck 2003), discarding
10% of sampled trees as burn-in. Maximum likelihood
(ML) analysis was executed using GTRCAT
model in RAxML 7.1.0 (Stamatakis et al. 2004). Node
support for ML trees was estimated from 10,000 rapid
bootstrap replicates. In the NWA analysis, an OWA
Mopeia virus was used as the outgroup. Trees were
visualized in FigTree v1.2 (Drummond & Rambaut
2007). To estimate the host phylogeny, we constructed a
ML phylogenetic tree from 523 of complete mitochondrial
cytochrome b (1,140 bp) sequences of up to three
sequences per taxon for known hosts and all rodents
from subfamilies Sigmodontinae and Neotominae
(Table S3, Supporting information) using the same
methodology as used for viruses with 3,000 rapid bootstrap
replicates. This phylogenetic tree includes two
hosts from different orders, a mustelid and a bat.
Reconciliation analysis
We tested whether the viruses were distributed nonrandomly
with respect to their hosts using two complementary
approaches: coassociation analysis and
phylogenetic clustering analysis.
Coassociation analysis. New World arenaviruses contain
eight viruses that have multiple hosts. Coassociation on
mapping the pathogen association onto the host phylogeny
(e.g. Treemap) therefore cannot be used, as it is
impossible to infer host order (Jackson & Charleston
2004). We therefore tested the coassociation of arenaviruses
with their known hosts using ParaFit (Legendre
et al. 2002) as implemented in the optimized algorithm
of AxParafit (Stamatakis et al. 2007) used with the
CopyCat GUI (Meier-Kolthoff et al. 2007). Parafit uses
patristic distances from both the host and virus phylogenies
to test whether the null hypothesis of independence
is significantly rejected, both globally (across the
whole tree) and individually (for each viral association).
Unlike other cophylogeny methods, it does not infer an
evolutionary scenario. The advantages of this method
are that it can handle large data sets, viruses with multiple
hosts and has been shown to have reliable type-I
and type-II errors (Legendre et al. 2002). We used a
host tree of 21 taxa and tested associations for each
HOST SWITCHING IN NEW WORLD ARENAVIRUSES 4139
Ó 2012 Blackwell Publishing Ltd
202 Paper 2.3.1
pruned viral protein tree individually (GP viruses = 22,
associations = 31; NP viruses = 22, associations = 31; L
viruses = 18, associations = 27 and Z viruses = 18, associations
= 27). Significance of individual host–virus
associations was calculated from 9999 permutations per
row of the association matrix. We corrected for multiple
comparisons by using a false discovery approach (Benjamini
& Hochberg 1995).
We tested the sensitivity of our results to missing
data by removing one host species from the analysis
and re-running the AxParafit analysis to evaluate
whether the patterns of significance changed. We
repeated this for all 21 host taxa. The number of associations
varied from 28 to 30 depending on which host
was removed.
We tested if the global significance of NWA with
their hosts was driven by a clade of viruses by removing
each viral clade from the coassociation analysis
(removing Clade A-Rec; associations = 19, viruses = 13,
hosts = 14; removing Clade A, associations = 25,
viruses = 15, hosts = 18; removing Clade B; associations
= 19, viruses = 14, hosts = 15 or removing Clade
C; associations = 29, viruses = 18, hosts = 21).
Clustering analysis. New World arenaviruses phylogenetic
host specificity can also be examined using indices
of phylogenetic diversity (Poulin et al. 2011). Patterns of
association at different taxonomic levels can be tested,
as the host tree is not restricted to only known host
taxa. Here, we constructed a host phylogeny of all
available American rodent species in the subfamilies
Sigmodontinae and Neotominae and pruned it to one
individual per species (n = 269). We also retained the
non-rodent hosts, a bat and a mustelid, in the phylogeny
(n = 271). We tested whether hosts of arenaviruses
were randomly spread, over-dispersed or clustered
across the host phylogeny using Phylocom (Webb et al.
2008). We did this by calculating the mean phylogenetic
distance (MPD) and the nearest phylogenetic taxon distance
(MNTD) of the virus hosts and compared these
values to those obtained from randomizing host species
labels across the host phylogeny. Abundance of viral
species found on each host was incorporated into the
test, to account for host species that can be infected by
more than one virus (either from a single or multiple
clades). Two-tailed P-values were calculated from 9,999
permutations. We included Pinhal in the analysis as
although there is no sequence data for the virus, the
host Calomys tener is known. We therefore tested the
significance of 32 NWA viral associations found in 22
hosts across all rodents of the subfamilies Sigmodontinae
and Neotominae. We then tested each NWA clade
separately by pruning the host phylogeny to the
rodents that occur in the countries where each arenavirus
clade is known. Distributions of each rodent species
were determined from the Global Mammal Assessment
(IUCN 2010) and the Global Biodiversity Information
Facility data portal (GBIF 2012) (Table S3, Supporting
information).
Geographical overlap between hosts
To evaluate whether host specificity was primarily constrained
by host relatedness or whether there was a
spatial component, we examined host range overlap of
all virus pairs. The distribution of each rodent host species
(n = 19) was downloaded from the Global Mammal
Assessment (IUCN 2010). Each file was then processed
using R (R Development Core Team, 2.12 2010) to create
geo-referenced maps of host distributions for each
arenavirus lineage. These maps were then used to calculate
the amount of geographical overlap between all
hosts (in km2
, then log transformed after adding a constant
of 1 for all subsequent analyses). A genetic distance
matrix was calculated for both viruses (using GP
sequences, n = 19) and hosts (n = 19) using Geneious
Pro v4.7 (Drummond et al. 2009). A number of Mantel
tests (simple and partial) were then performed using 29
associations. The simple Mantel tests assessed whether
the genetic distance between two viruses was significantly
associated with the genetic distance of their
respective hosts and ⁄ or the geographic overlap existing
between their respective hosts. However, because we
can expect closely related hosts to overlap more than
distantly related hosts, partial Mantel tests were also
performed to ascertain whether the above associations
were direct or indirect ones. To account for viruses that
have more than one host, we calculated the average
range overlap between their respective hosts using an
arithmetic mean (on the log transformed overlap). Similarly,
the average genetic distance between two sets of
hosts was calculated using a geometric mean (after adding
a constant of 0.01, to account for some hosts being
shared by two viruses).
To test for global associations across the whole viral
tree, their host and geographic ranges, a Mantel test
informed by phylogeny (Lapointe & Garland 2001) was
used to reduce the risk of type-I error associated with
the classical Mantel test (Harmon & Glor 2010). This permutation
procedure swaps closely related species more
often than distantly related species across the phylogenetic
tree (Lapointe & Garland 2001). However when
assessing relationships within each clade of viruses, the
classical Mantel test was applied instead as the small
sample size (number of viruses in a single clade) would
not guarantee a sufficient level of permutation (i.e. the
identity of the viruses in the permutated trees would
remain too often the original identities) with a Mantel
4140 N. R. IRWIN ET AL.
Ó 2012 Blackwell Publishing Ltd
Paper 2.3.1 203
test informed by phylogeny. To minimize type-I error,
the partial Mantel tests (whether phylogeny informed or
not) were performed using the permutation procedure
on raw data (virus genetic distance) (Legendre 2000).
Analyses were completed in R using modified algorithms
from Harmon & Glor (2010). To assess the level
of significance, 9999 permutations were used.
Results
Phylogeny of New World arenaviruses
All phylogenetic analyses supported four previously
recognized evolutionary lineages A, B, C and A-Rec
(Fig. S1A–D, Supporting information). The GP phylogeny
places Clade A-Rec as sister to Clade B, whereas
the NP, Z and L protein place it as sister to A. This
well-known incongruence has been shown to be the
result of a recombinant progenitor (Charrel et al. 2001;
Archer & Rico-Hesse 2002). Our larger phylogenetic
trees support this conclusion. The rodent phylogenetic
tree supported reciprocal monophyly of subfamilies Sigmodontinae
and Neotominae, but relationships between
genera of Sigmodontinae were poorly resolved (Fig. 1)
as has been demonstrated in larger studies using multiple
loci (Engel et al. 1998; Jansa & Weksler 2004).
Host–viral associations
Coassociation analysis. Global co-evolutionary association
of NWA with their known hosts was statistically significant
for GP (ParafitGlobal Test Statistic = 51.105,
P = 0.0004), NP (ParafitGlobal = 27.644, P = 0.0003) and
L protein (ParafitGlobal = 34.656, P = 0.0239), but not
significant for Z protein (ParafitGlobal = 22.978,
P = 0.1452). However, the significance of individual
hosts associations varied. In the host–virus associations
based on GP sequences, 22 of 31 host–virus associations
were not significantly explained by topology (Fig. 2).
Three of these associations were found to be on the border
of significance (P = 0.05) but were not significant
after correction for multiple testing. The only significant
associations found in NWA were those from Clade ARec
hosted by woodrats of the genus Neotoma. Further
analysis of the other proteins showed that the same
nine A-Rec viruses were significantly associated with
their hosts for the NP, but no cophylogenetic signal was
found for any virus with any host for the L and Z proteins
(Fig. S2, Supporting information). The results
were very similar for analyses based on the GP and NP
trees. In subsequent analyses, we chose to use the GP
data set as it is known to be associated with host cell
entry recognition (Radoshitzky et al. 2008) and was
0.09
90-100
80-89
70-79
60-69
50-59
40-49
30-39
20-29
10-19
0-9
A
B
C
A-Rec
Figure 1 Phylogenetic tree of all Sigmodontinae
and Neotominae rodents
with hosts of each New World arenavirus
indicated. A ML phylogenetic tree
of mitochondrial cytochrome b of all
available species (n = 523). Bootstrap
support is indicated by grey-scale coding
of respective branches. Host species
(Table S1, Supporting information)
shown for each New World arenaviruses
(NWA) clade.
HOST SWITCHING IN NEW WORLD ARENAVIRUSES 4141
Ó 2012 Blackwell Publishing Ltd
204 Paper 2.3.1
therefore more likely to recover host–virus cophylogeny
than the NP.
Testing the sensitivity of the coassociation analysis. Removing
16 of the hosts did not change the pattern of significance
in the coassociation matrix. The same nine A-Rec
viruses remained significantly associated with their
host. However, we found that removing either nonrodent
host (the mustelid or the bat) resulted in the Bear
Canyon virus becoming significantly coassociated with
Peromyscus californicus. We found that removing one of
the Neotoma species that had multiple viral associations
(N. micropus, N. mexicana or N. albigula) resulted in all
associations previously seen in A-Rec viruses no longer
remaining significant. The global statistic for coassociation
remained significant after removing viral Clade A,
B or C (P = 0.0001, P = 0.0002, P = 0.0001). However,
removing Clade A-Rec caused the global association
statistic to become nonsignificant (P = 0.41414).
Clustering analysis. New World arenaviruses were randomly
distributed across the rodent tree (n = 22, MPD
P = 0.137; MNTD P = 0.442). Clade A-Rec (n = 7)
viruses were significantly clustered with their hosts for
both analyses (MPD P > 0.001; MNTD P = 0.039). Clade
C, limited to only 3 viruses, were not significant in
either analyses (MPD P > 0.270; MNTD P = 0.162).
Clade A (n = 6) were significantly clustered when compared
to all available hosts (MPD P = 0.026), but not
clustered into particularly closely related taxa (MNTD
P = 0.475). Clade B (n = 11), however, were significantly
over-dispersed across the phylogenetic tree, for both
indices (MPD P = 0.021, MNTD P = 0.047) when the
two hosts of different orders were included in the analysis.
Removing these two taxa, Clade B was shown to
be randomly distributed amongst the rodents (MPD
P = 0.164 and MNTD P = 0.964). These results are summarized
in Table 1.
Distribution trends for each virus clade
The potential extent of each NWA clade has been
mapped by plotting the distribution of their known
hosts (Fig. 3). Hosts of Clade A arenaviruses are all
distributed in South America (not further south than
Uruguay) and part of Central America (not further
west than Costa Rica) (Fig. 3). Hosts of Clade B arenaviruses
are located in South and Central America (south
of Mexico). However, two subclades of Clade B
appeared separated by a major barrier to dispersal, the
Amazon river. The hosts of subclade B1 (Tacaribe, Junı´n,
Machupo) (Charrel et al. 2002) are all located south
of the Amazon River except for Tacaribe whose hosts
are widespread tropical bat species (Artibeus lituratus
and Artibeus jamaicensis). The hosts of subclade B2
(Amapari, Cupixi and Guanarı´to) (Charrel et al. 2002)
are all located north of the Amazon River, except for
one host, Hylaeamys megacephalus, occurring from Venezuela
to southern Brazil. This species, however, is a
complex of three lineages; the true host is therefore
likely to be an unnamed species with a distribution
north of the Amazon river (Miranda et al. 2007). Hosts
P(ParaFitLink1) < 0.05
P(ParaFitLink1) > 0.05
0.3
Paraná AF485261 NC_010756
Junín NC_005081
Guanaríto NC_005077
Pirital NC_005894
Flexal NC_010757
Amapari NC_010247
Pichinde NC_006447
Machupo NC_005078
Cupixi NC_010254
Oliveros OVU34248 NC_010248
Allpahuayo NC_010253
Tacaribe NC_004293
Whitewater Arroyo NC_010700
Big Brushy Tank EF619035
Tamiami AF485263 NC_010701
Skinner Tank EU123328
Tonto Creek EF619033
Latino AF512830 NC_010758
Catarina DQ865244
Bear Canyon NC_010256
0.3
AY273914 Bolomys temchuki
AF435110 Sigmodon hispidus
EU579521 Zygodontomys brevicauda
AF108699 Oecomys bicolor
DQ227457 Oligoryzomys fulvescens
AF298848 Neotoma mexicana
AF293397 Sigmodon alstoni
DQ781304 Neotoma macrotis
AF186800 Neotoma cinerea
AF155393 Peromyscus californicus
AF186814 Neotoma albigula
EU579512 Sooretamys angouya
AF376474 Neotoma micropus
DQ869432 Artibeus planirostris trinitatis
EU579504 Neacomys spinosus
AY033183 Calomys callosus
EU579505 Nephlomys albigularis
AY033190 Calomys laucha
AF385602 Calomys musculinus
EF987754 Galictis cuja
EU579499 Hylaeamys megacephalus
B
A-Rec
C
A
Figure 2 Host–virus associations between New World arenaviruses and their known hosts. 31 associations are drawn on trees
pruned to one individual per species from a Bayesian glycoprotein arenavirus tree (n = 22) (Fig. S1A, Supporting information) and
known hosts (n = 21) (Fig. 1). When there were no data for a host, we chose the most closely related species to represent the association
as indicated by grey type (Table S1, Supporting information). The four New World arenaviruses (NWA) clades are indicated at
the base of each clade; the Neotominae rodents (dark grey) and the Sigmodotinae rodents (pale grey). Significance of host–virus associations
were calculated from 9,999 permutations, using ParaFit, (Legendre et al. 2002). Full lines represent significant cophylogenetic
association between specific host-arenavirus pairs, dashed lines represent nonsignificant relationships.
4142 N. R. IRWIN ET AL.
Ó 2012 Blackwell Publishing Ltd
Paper 2.3.1 205
of Clade C all occur south of the Amazon River, but,
interestingly, the host of Oliveros (Necromys benefactus)
is restricted to central Argentina and does not overlap
with the other Clade C hosts (Calomys callosus and
C. tener). Other hosts for Clade C may therefore exist in
North Argentina and Paraguay. Hosts of Clade A-Rec
all occur in North America and Mexico, except for Neotoma
mexicana, which also occurs in Central America as
far east as Guatemala (Fig. 3). These maps extend the
areas where NWA are predicted to occur, in particular
Clade A-Rec’s hosts are overlapping with Clade A’s
hosts in Central America. NWA are likely to be present
in Central America, which has been sparsely sampled
to date.
Table 1. Clustering analysis of New World arenaviruses (NWA)
Virus
No. of
viruses
No. of
host taxa
No. of
nonhosts MPD
MPD
random
MPD
P value MNTD
MNTD
random
MNTD
P value
NWA 32 22 249 0.5717 0.5268 0.1372 0.3129 0.2889 0.442
NWA (restricted to rodents) 30 20 249 0.5094 0.5241 0.6302 0.2962 0.0348 0.558
Clade A-Rec 11 7 103 0.2781 0.4596 0.000 0.2496 0.3849 0.039
Clade A 6 6 140 0.3642 0.4635 0.026 0.3540 0.4004 0.475
Clade B 12 11 246 0.6166 0.5033 0.021 0.4378 0.3409 0.047
Clade B (restricted to rodents) 11 9 246 0.4409 0.4909 0.164 0.3561 0.3584 0.964
Clade C 3 3 125 0.3095 0.3713 0.270 0.3432 0.4888 0.162
The phylogenetic tree of all rodents of the Sigmodontinae and Neotominae subfamilies (Fig. 1) was pruned to one sequence per
species (n = 269) and included two nonrodent hosts (n = 271). Clustering analysis tested the significance of phylogenetic distance
(MPD, Mean Phylogenetic Distance) and the distance of the nearest neighbour (MNTD, Mean Nearest phylogenetic Taxon Distance)
between known hosts by comparison with the values generated by 9,999 permutations according to a two-tailed test. Clustering
analysis of all NWA across all rodents of these subfamilies was tested. Each virus clade was also tested independently against the
phylogeny restricted to their sympatric rodents (Table S3, Supporting information). Two hosts of Clade B virus are found in
different orders of mammals; the analysis was therefore also run with and without these taxa for NWA as a whole and for Clade B.
A-Rec
A
B C
Figure 3 Distribution map of hosts of New World arenavirus Clades. Distributions of hosts (Table S1, Supporting information) were
downloaded from the IUCN Global Mammal Assessment Team (IUCN 2010) and processed using R (R Development Core Team,
2.12 2010). Each clade is indicated as follows; figure to the left, Clade A-Rec –stripes (7 host species) Clade B—(9 host species); figure
to the right, Clade A—dots (6 host species), Clade C— (3 host species).
HOST SWITCHING IN NEW WORLD ARENAVIRUSES 4143
Ó 2012 Blackwell Publishing Ltd
206 Paper 2.3.1
Is viral phylogeny correlated to host phylogeny
or to host geographical overlap?
Across all NWA, as the genetic distance between two
arenaviruses increased, so did the genetic distance
between their respective hosts (partial r = 0.291, phylogenetically
constrained Mantel P = 0.015) (Fig. 4). In
turn, as the genetic distance between the hosts of two
viruses increased, the spatial overlap existing between
them tended to decrease (partial r = )0.578, phylog.
constr. Mantel P < 0.001). The genetic distance between
two viruses, however, was not significantly correlated
to the amount of spatial overlap existing between the
hosts of these viruses (partial r = )0.147, phylog. constr.
Mantel P = 0.069). At the clade level, however, the patterns
varied widely and differed markedly from the
overall pattern. Clade C had too few viruses to enable
analysis to be completed at clade level. Only the viral
genetic distance was significantly correlated to the spatial
overlap of their respective hosts for Clades A and B
(contrary to the overall pattern) (Table 2). For Clade ARec,
only the correlation between the viruses’ genetic
distance and the genetic distance of their hosts was sig-
nificant.
Discussion
We empirically tested whether NWA had codiverged
with their hosts using three complementary approaches:
reconciliation analysis, phylogenetic clustering and
Mantel tests. We found no evidence to support codivergence
of NWA with their hosts. Instead, the evidence
suggests that host switching has occurred frequently,
though to a different extent in the different clades.
Phylogenetic analyses of host-virus associations
In cophylogenetic analysis of NWA confined to known
hosts, we found a globally significant association
between the viral and host phylogenies with all viral
protein trees (GP, NP, L and Z). However, it was only
in the larger data sets with more structured trees (NP
and GP analysis) that nine individual Clade A-Rec viral
associations were significantly associated with their
hosts. Testing the sensitivity of the AxParafit analysis
showed the results to be generally robust. The ParaFit
global statistic only became nonsignificant after removing
Clade A-Rec, demonstrating that this clade was
driving the global significant association. Furthermore,
removing any host that had multiple viral associations
with A-Rec viruses caused all other associations to
become nonsignificant.
The power to discriminate a significant association
(and thus avoid type-II error) in Parafit is associated
with completeness of knowledge of all the hosts and
viruses in the system (Legendre et al. 2002). Nine NWA
have been discovered in the past decade (Charrel & de
Lamballerie 2010), making it unlikely that all hosts and
viruses are currently known. Unsupported phylogenies,
Figure 4 Mantel tests of virus genetic distance to host genetic
distance and host area overlap. Simple and partial Mantel tests
of each relationship are presented with their corresponding r
value and significance level. See Table 2 for clade-specific
results.
Table 2. Assessment of geographic overlap and host phylogeny on viral phylogeny
Lineage
Virus genetic distance
vs. Host genetic distance
Virus genetic distance
vs. Host area overlap
Host genetic distance
vs. Host area overlap
A r = 0.029 r = )0.586* r = 0.019
part. r = 0.050 part. r = )0.587* part. r = 0.045
A-Rec r = 0.581* r = )0.372 r = )0.232
part. r = 0.548* part. r = )0.300 part. r = )0.021
B r = 0.680 r = )0.757** r = )0.546**
part. r = 0.487 part. r = )0.627** part. r = )0.066
Coefficients of correlation (simple and partial) between the genetic distance separating two arenaviruses, the genetic distance
separating their respective hosts and the amount of spatial overlap existing between these hosts. Significant results are indicated by a
star, *P < 0.05, **P > 0.001.
4144 N. R. IRWIN ET AL.
Ó 2012 Blackwell Publishing Ltd
Paper 2.3.1 207
for example rapid radiations, also add to uncertainty in
permutation testing procedures. Here, despite the viral
phylogeny being well supported, the Sigmodontinae
higher taxonomy was difficult to confidently resolve
(Jansa & Weksler 2004) and may have contributed to
the lack of statistical significance. While the improved
algorithms for permutation tests of reconciliation analysis
enable multiple hosts and large number of taxa associations
to be tested, if one or both of the phylogenies
have uncertainty, other statistical approaches are
needed to assess the nature of host–virus relationships.
Clustering analysis
To explore viral–host associations further, we examined
whether patterns of phylogenetic clustering were significantly
different from chance expectations across all the
potential Sigmodontinae and Neotominae rodent hosts
using different indices of relatedness. We found that
NWAs collectively have invaded a wide range of rodent
hosts, which were randomly distributed across the
rodent phylogeny. However, different patterns were
recovered for each Clade when analysis was restricted
to rodents within their geographic range. Clade A-Rec
was the most tightly constrained clade of NWA viruses,
being significantly clustered with their hosts. Clade A
was less constrained, being clustered in subsets of the
rodent tree but beyond that, not in particularly closely
related taxa. Worryingly, Clade B with pathogenic
members was the least constrained being randomly distributed
across the rodent tree and even able to infect
hosts of different mammalian orders.
Geographic proximity analysis
We examined further the relationship between viral
phylogeny and host geographical overlap with simple
and partial Mantel tests to see whether we could tease
out any explanatory factor that differed between clades.
Viral phylogeny was explained by host phylogeny only
for Clade A-Rec, which was congruent with all other
analyses. Host relatedness is therefore a factor that has
constrained host switching in Clade A-Rec, with few
exceptions. Geographical overlap of hosts was more
important than host relatedness in shaping the evolution
of both Clades A and B, indicating that physical
contact with a new host to infect was more limiting
than the need to find a sufficiently related host. Host
range is clearly a rather crude proxy for virus distribution,
as true viral distribution will be dependant on
many factors including host demography and pathogen
prevalence (e.g. Mills 1994). This makes our results
even more striking, as it was expected a priori that phylogenetic
relatedness of hosts should constrain host
switching because of host defence and cell entry mecha-
nisms.
Refuting the codivergence hypothesis for NWA
A long-standing codiverged relationship has been
asserted for NWA because Clade B viruses infected on
the whole Sigmodontinae, while Clade A-Rec viruses
primarily infected Neotoma species in the subfamily
Neotominae (Webb et al. 1970; Arata & Gratz 1975;
Childs & Peters 1993b; Bowen et al. 1997; Mills et al.
1997; Mills 1999; Salazar-Bravo et al. 2002; Charrel et al.
2003; Cajimat et al. 2007; Gonzalez et al. 2007). The odd
viruses such as Bear Canyon or Tamiami that did not
follow this pattern were therefore thought to represent
rare host switches to a different group of rodents (Cajimat
et al. 2007). Here, we found host switching to be
the predominant pattern of association.
Classically, codivergence implies a pattern of high
host specificity where one host is infected by one parasite
(Fahrenholz’s Rule) (Paterson & Banks 2001). However,
a comprehensive literature review showed that
one quarter of the NWA infected multiple hosts and
one-third of the host species were infected by more
than one NWA virus.
Doubts about the validity of the codivergence hypothesis
have also been raised from the observation that
several subfamilies of rodents sister to the host subfamilies
are devoid of arenaviruses (Salazar-Bravo et al.
2002). While LCMV (in the Old World) has been
recorded in Arvicolinae (Laakkonen et al. 2006) and
Cricetinae (Skinner & Knight 1979), wild infections in
Gerbillinae, Tylomyinae and Deomyinae have never
been recorded. This either can be interpreted as lack of
field sampling in these groups, extinction of viral lineages
(Paterson & Banks 2001), or as evidence that arenaviruses
did not coevolve with their hosts. While
sampling deficiency is always possible, extinction across
three major lineages seems unlikely.
Likewise, topological incongruence between hosts
and viruses could be explained by extinction events or
lack of sampling (Charleston & Perkins 2006). Although
extinctions will have contributed to the NWA system,
we believe it has not dominated the pattern of coassociation
given that (i) Clade A and B are so widely distributed
across the Sigmodontinae and Neotominae tree it
would imply widespread extinctions across a large
range of taxa; and (ii) physical proximity can already
significantly explain the host species that Clades A and
B have managed to invade.
Temporal incongruence between the viral and host
trees would also support the rejection of the codivergence
hypothesis. However, limited heterochronous
data for NWA mean that molecular dating is currently
HOST SWITCHING IN NEW WORLD ARENAVIRUSES 4145
Ó 2012 Blackwell Publishing Ltd
208 Paper 2.3.1
not available. However, substitution rates for OWA
(Coulibaly-N’Golo et al. 2011) have been estimated at
the typical rate for RNA viruses of 10)3
to 10)4
substitutions
per site per year (Domingo & Holland 1997)
resulting in dates for the time to the most recent common
ancestor (TMRCA) of between 2500 and 7000 ybp
(Coulibaly-N’Golo et al. 2011). These dates are orders of
magnitude younger than the 22 million years postulated
by the codivergence hypothesis (Bowen et al. 1997). A
growing body of RNA virus studies with heterochronous
data has also produced speciation events much
younger than expected (Duffy et al. 2008; Holmes 2008;
Ramsden et al. 2009; Paga´n et al. 2010), which has led
some to suggest a methodological problem with inferring
dates from short time series, especially for the deeper
nodes (Worobey et al. 2010; Ho et al. 2011). Dating
inferred from host speciation events based on encapsulated
RNA is also orders of magnitude older than those
based on molecular clock estimates (Horie et al. 2010;
Katzourakis 2010; Worobey et al. 2010). This paradox
can be explained by high lineage turnover in rapidly
evolving populations (Holmes 2003, 2008). The molecular
clock approach calculates substitution rates based
on the genetic variation of current circulating viruses.
It does not provide the absolute age of when these
viruses split from their sister taxon, but the TMRCA
of the viruses sampled (Holmes 2003, 2008). The shallow
dates for RNA viruses reflect this. The rapid turnover
is therefore likely to remove any signal of ancient
history or association between RNA viruses and their
hosts.
Each NWA clade has a unique pattern of relationships
with their hosts, indicating that each clade has
evolved differently. Clade A-Rec is comprised of specialists
that infect only a single host or a small number
of closely related hosts. The general pattern of cophylogeny
for Clade A-Rec viruses with their hosts does
not imply codivergence but can be equally explained by
host switching (Holmes & Price 1980). In a system dominated
by ‘preferential’ host switching, viruses are more
likely to switch successfully on closely related hosts as
they are preadapted genetically, physiologically and
ecologically to the original host (Charleston & Robertson
2002). These viruses then become adapted to the
new host, speciating and obscuring the host-switching
signal (Brook & Mclennan 1991). Even the phylogenetically
distant host switches seen in Bear Canyon and
Tamiami are encompassed, as the concept does not
imply that there will be no distant host switching only
that it will occur less frequently (Charleston & Robertson
2002). Conversely, the proximity analysis suggested
that ecological opportunity to infect new hosts rather
than relatedness was the driver for host switching for
the viruses of Clades A and B.
The different pattern of association shown for each
lineage of NWA implies that each lineage is constrained
differently to enter new hosts. This hypothesis is supported
by experimental and structural studies, which
have demonstrated that cell entry mechanisms for
NWA are key to understanding the potential for host
spread and disease outbreak (Flanagan et al. 2008; Choe
et al. 2011). For example, currently nonpathogenic
viruses of Clade B with only minor changes in the apical
domain of the GP used as the cell entry receptor target
could emerge as new diseases (Abraham et al. 2009,
2010). Pathogenic viruses with the ability to infect a
wide range of hosts are of serious concern, but experimental
infection has shown that Clade B viruses were
unable to infect some nonhost species such as the house
mouse and black rat, and that each virus species had a
different ability to infect native hosts and humans
(Radoshitzky et al. 2008; Abraham et al. 2010). Switching
to a distantly related host may therefore remain a
rare phenomenon followed by rapid adaptation to the
new host and ultimately resulting in viral speciation.
Owing to the difficulties of screening highly infectious
viruses in the wild that are both spatially and
temporally patchy in their distribution (Jay et al. 2005),
sequencing the five interaction motifs of the TfR1 receptor
(Abraham et al. 2010) of potential hosts may be a
targeted approach to predict new hosts. Evidence from
the phylogenetic clustering and reconciliation analysis
here, as well as laboratory infections (Lukashevich et al.
2002; Abraham et al. 2010) and examination of the cell
receptors (Abraham et al. 2009), would suggest that
bats, monkeys and other Sigmodontinae and Neotominae
rodents, currently not known to be hosts of Clade
B, should be screened to understand better the potential
for host switching and disease spread. The ability of
both OWA and NWA to switch to unrelated hosts
including humans makes the development of a general
vaccine and ⁄ or therapeutic approach (Larson et al.
2008; Kotturi et al. 2009; Botten et al. 2010) all the more
important especially as not all viruses may yet be
known.
Summary
We found no evidence that NWA have codiverged
(associated by descent) with their hosts. Similarly, the
pattern of association of Old World arenaviruses
(OWA) with their known hosts could not be
distinguished from random (Coulibaly-N’Golo et al.
2011). However, this led the authors to conclude that
the frequent host switching may have obscured an
ancient codivergent pattern of association (CoulibalyN’Golo
et al. 2011). Here, we would have drawn a similar
conclusion for NWA (with the possible exception of
4146 N. R. IRWIN ET AL.
Ó 2012 Blackwell Publishing Ltd
Paper 2.3.1 209
Clade A-Rec), but the addition of phylogenetic clustering
analysis demonstrated that each clade of NWA was
associated differently with the hosts available to them
and that only Clade B was truly randomly distributed.
We found that the phylogenetic clustering approach on
all the potential hosts was more robust than reconciliation
analysis limited to known hosts. Likewise, the
Mantel tests suggested that the potential for host
switching differed between clades. The diversification
of two clades was driven by geographic proximity of
their hosts rather than their relatedness, indicating host
switching was opportunistic, whereas the diversification
of Clade A-Rec was driven by host relatedness. We
expect these approaches to be informative for other
pathogen–host systems as more molecular data become
available for construction of large host phylogenies.
Acknowledgements
NRI was supported by a Daphne Jackson Fellowship sponsored
by the Natural Environmental Research Council. Biomatters
sponsored NRI with a license to use Geneious Pro. NM
was supported by Academy of Sciences of the Czech Republic
(grant AV0Z60930519). The data were analysed on the computational
cluster of the Institute of Vertebrate Biology and at
Bioportal of the University of Oslo. We thank the subeditor Dr.
Roman Biek and the three anonymous referees for their comments,
which helped improve the paper.
References
Abraham J, Kwong J, Albarin˜o C et al. (2009) Host-species
transferrin receptor 1 orthologs are cellular receptors for
nonpathogenic New World clade B arenaviruses. PLoS
Pathogens, 5, e1000358.
Abraham J, Corbett KD, Farzan M, Choe H, Harrison SC (2010)
Structural basis for receptor recognition by New World
hemorrhagic fever arenaviruses. Nature Structural &
Molecular Biology, 17, 438–444.
Arata AA, Gratz NG (1975) The structure of rodent faunas
associated with arenaviral infections. Bulletin of the World
Health Organization, 52, 621–627.
Archer A, Rico-Hesse R (2002) High genetic divergence and
recombination in Arenaviruses from the americas. Virology,
304, 274–281.
de Bellocq JG, Borremans B, Katakweba A et al. (2010)
Sympatric occurrence of 3 arenaviruses, Tanzania. Emerging
infectious diseases, 16, 692–695.
Benjamini Y, Hochberg Y (1995) Controlling the false discovery
rate: a practical and powerful approach to multiple testing.
Journal of Royal Statistical Society, 57, 289–300.
Borio L, Inglesby T, Peters CJ et al. (2002) Hemorrhagic fever
viruses as biological weapons: medical and public health
management. JAMA, 287, 2391–2405.
Botten J, Whitton JL, Barrowman P et al. (2010) A multivalent
vaccination strategy for the prevention of Old World
arenavirus infection in humans. Journal of Virology, 84, 9947–
9956.
Bowen GS, Calisher CH, Winkler WG et al. (1975) Laboratory
studies of a lymphocytic choriomeningitis virus outbreak in
man and laboratory animals. American Journal of
Epidemiology, 102, 233–240.
Bowen M, Peters C, Nichol S (1996) The phylogeny of New
World (Tacaribe complex) arenaviruses. Virology, 219, 285–
290.
Bowen M, Peters C, Nichol S (1997) Phylogenetic analysis of
the Arenaviridae: patterns of virus evolution and evidence
for cospeciation between arenaviruses and their rodent
hosts. Molecular Phylogenetics and Evolution, 8, 301–316.
Briese T, Paweska J, Mcmullan L et al. (2009) Genetic detection
and characterization of Lujo virus, a new hemorrhagic fever–
associated arenavirus from Southern Africa. PLoS Pathogens,
4, e1000455.
Brook DR, Mclennan DA (1991) Phylogeny, Ecology and Behavior;
A Research Program in Comparative Biology. The University of
Chicago Press, Chicago, IL, USA.
Cajimat MNB, Milazzo ML, Hess BD, Rood MP, Fulhorst
CF (2007) Principal host relationships and evolutionary
history of the North American arenaviruses. Virology, 367,
235–243.
Cao W (1998) Identification of -dystroglycan as a receptor for
lymphocytic choriomeningitis virus and lassa fever virus.
Science, 282, 2079–2081.
Carey D, Kemp G, White H et al. (1972) Lassa fever.
Epidemiological aspects of the 1970 epidemic, Jos, Nigeria.
Transactions of the Royal Society of Tropical Medicine and
Hygiene, 66, 402.
Charleston MA, Perkins SL (2006) Traversing the tangle:
algorithms and applications for cophylogenetic studies.
Journal of Biomedical Informatics, 39, 62–71.
Charleston MA, Robertson DL (2002) Preferential host
switching by primate lentiviruses can account for
phylogenetic similarity with the primate phylogeny.
Systematic Biology, 51, 528–535.
Charrel RN, de Lamballerie X (2010) Zoonotic aspects of
arenavirus infections. Veterinary Microbiology, 140, 213–220.
Charrel R, de Lamballerie X, Fulhorst C (2001) The whitewater
arroyo virus: natural evidence for genetic recombination
among Tacaribe serocomplex viruses (family Arenaviridae).
Virology, 283, 161–166.
Charrel R, Feldmann H, Fulhorst C et al. (2002) Phylogeny of
New World arenaviruses based on the complete coding
sequences of the small genomic segment identified an
evolutionary lineage produced by intrasegmental
recombination. Biochemical and Biophysical Research
Communications, 296, 1118–1124.
Charrel R, Lemasson J, Garbutt M et al. (2003) New insights
into the evolutionary relationships between arenaviruses
provided by comparative analysis of small and large
segment sequences. Virology, 317, 191–196.
Childs J, Peters CJ (1993) Ecology and epidemiology of
arenaviruses and their hosts. In: The Arenaviridae (ed Salvato
MS), pp. 331–384. Plenum press, New York.
Childs JE, Peters CJ (1993b) Ecology and epidemiology of
arenaviruses and their hosts. In: The Arenaviridae (ed Salvato
MS), pp. 331–384. Plenum press, New York.
Childs JE, Mills J, Glass GE (1995) Rodent-borne hemorrhagic
fever viruses: a special risk for mammalogists? Journal of
Mammalogy, 76, 664–680.
HOST SWITCHING IN NEW WORLD ARENAVIRUSES 4147
Ó 2012 Blackwell Publishing Ltd
210 Paper 2.3.1
Choe H, Jemielity S, Abraham J, Radoshitzky SR, Farzan M
(2011) Transferrin receptor 1 in the zoonosis and
pathogenesis of New World hemorrhagic fever arenaviruses.
Current Opinion in Microbiology, 14, 476–482.
Coulibaly-N’Golo D, Allali B, Kouassi SK et al. (2011) Novel
arenavirus sequences in Hylomyscus sp. and Mus
(Nannomys) setulosus from Coˆte d’Ivoire: implications for
evolution of arenaviruses in Africa. PLoS One, 6, e20893.
Delgado S, Erickson B, Agudo R et al. (2008) Chapare virus, a
newly discovered arenavirus isolated from a fatal
hemorrhagic fever case in Bolivia. PLoS Pathogens, 4,
e1000047.
Domingo E, Holland XX (1997) RNA virus mutations and
fitness for survival. Annual Reviews in Microbiology, 51, 151–
178.
Downs WG, Anderson CR, Spence L, Aitken THG, Greenhall
AH (1963) Tacaribe virus, a new agent isolated from
Artibeus bats and mosquitoes in Trinidad, West Indies. The
American Journal of Tropical Medicine and Hygiene, 12, 640.
Drummond AJ, Rambaut A (2007) BEAST: bayesian
evolutionary analysis by sampling trees. BMC Evolutionary
Biology, 7, 214.
Drummond AJ, Ashton B, Buxton S, Cheung M, Cooper A,
Duran C, Field M, Heled J, Kearse M, Markowitz S, Moir R,
Stones-Havas S, Sturrock S, Thierer T, Wilson A (2009)
Geneious v 4.7. Available at http://www.geneious.com/.
Duffy S, Shackelton LA, Holmes EC (2008) Rates of
evolutionary change in viruses: patterns and determinants.
Nature Reviews Genetics, 9, 267–276.
Dylla DE, Michele DE, Campbell KP, McCray Jr PB (2008)
Basolateral entry and release of New and Old World
arenaviruses from human airway Epithelia. Journal of
Virology, 82, 6034–6038.
Engel SR, Hogan KM, Taylor JF, Davis SK (1998) Molecular
systematics and paleobiogeography of the South American
sigmodontine rodents. Molecular Biology and Evolution, 15,
35–49.
Enserink M (2000) Emerging Diseases: new arenavirus blamed
for recent deaths in California. Science, 289, 842–843.
Flanagan ML, Oldenburg J, Reignier T et al. (2008) New world
clade B arenaviruses can use transferrin receptor 1 (TfR1)dependent
and -independent entry pathways, and
glycoproteins from human pathogenic strains are associated
with the use of TfR1. Journal of Virology, 82, 938–948.
Fulhorst C, Bennett S, Milazzo M et al. (2002) Bear Canyon
virus: an arenavirus naturally associated with the California
mouse (Peromyscus californicus). Emerging Infectious Diseases,
8, 717–721.
GBIF (2012) Biodiversity occurrence data published by The
Museum of Vertebrate Zoology, Oklahoma Museum of
Natural History, Museum of Southwestern Biology
Albuquerque, Michigan State University Museum, Mammal
Collection of the University of Kansas Biodiversity Research
Centre, Senckenberg Collection Mammalia SMF, Coleccio´n
de Mamı´feros de la Sierra Volca´nica Transversal de Me´xico
(UAM-I), The Field Museum of Natural History, Yale
Peabody Museum <data.gbif.org>.
Gonzalez J, Bowen M, Nichol S, Rico-Hesse R (1996) Genetic
characterization and phylogeny of Sabia virus, an emergent
pathogen in Brazil. Virology, 221, 318–324.
Gonzalez J, Emonet S, de Lamballerie X, Charrel R (2007)
Arenaviruses. Current Topics in Microbiology and Immunology,
315, 253–288.
Harmon LJ, Glor RE (2010) Poor statistical performance of the
mantel test in phylogenetic comparative analyses. Evolution,
64, 2173–2178.
Ho SYW, Lanfear R, Bromham L et al. (2011) Time-dependent
rates of molecular evolution. Molecular Ecology, 20, 3087–
3101.
Holmes E (2003) Molecular clocks and the puzzle of RNA
virus origins. Journal of Virology, 77, 3893–3897.
Holmes EC (2008) Evolutionary history and phylogeography
of human viruses. Annual Reviews in Microbiology, 62, 307–
328.
Holmes JC, Price PW (1980) Parasite communities: the roles of
phylogeny and ecology. Systematic Zoology, 29, 203–213.
Horie M, Honda T, Suzuki Y et al. (2010) Endogenous nonretroviral
RNA virus elements in mammalian genomes.
Nature, 463, 84–87.
Hugot JP (2003) L’e´volution des Arenaviridae et de leurs hoˆtes
muride´s re´sulte-t-elle d’e´ve`nements de capture ou de
processus de coe´volution? Evolution of the Old World
Arenaviridae and their rodent hosts: generalized hosttransfer
or association by descent? Bulletin de l’Acade´mie
Ve´te´rinaire de France 156, 37–44.
Hugot JP, Gonzalez JP, Denys C (2001) Evolution of the Old
World Arenaviridae and their rodent hosts: generalized hosttransfer
or association by descent? Infection Genetics
Evolution, 1, 13–20.
IUCN (2010) IUCN Red List of Threatened Species. Version
2010. http://www.iucnredlist.org.
Jackson AP, Charleston MA (2004) A cophylogenetic
perspective of RNA-virus evolution. Molecular Biology and
Evolution, 21, 45–57.
Jansa SA, Weksler M (2004) Phylogeny of muroid rodents:
relationships within and among major lineages as
determined by IRBP gene sequences. Molecular Phylogenetics
and Evolution, 31, 256–276.
Jay M, Glaser C, Fulhorst C (2005) The arenaviruses. Journal of
the American Veterinary Medical Association, 227, 904–915.
Katzourakis A (2010) PLoS genetics: endogenous viral elements
in animal genomes. PLoS Genetics, 6, e1001191.
Kilgore P, Peters C, Mills J et al. (1995) Prospects for the
control of Bolivian hemorrhagic fever. Emerging infectious
diseases, 1, 97–99.
Kotturi MF, Botten J, Sidney J et al. (2009) A multivalent and
cross-protective vaccine strategy against arenaviruses
associated with human disease. PLoS Pathogens, 5, e1000695.
Kunz S (2009) Receptor binding and cell entry of Old World
arenaviruses reveal novel aspects of virus–host interaction.
Virology, 387, 245–249.
Laakkonen J, Kallio-Kokko H, Oktem MA et al. (2006)
Serological survey for viral pathogens in Turkish rodents.
Journal of wildlife diseases, 42, 672–676.
Lapointe F-J, Garland T (2001) A generalized permutation
model for the analysis of cross-species data. Journal of
Classification, 18, 109–127.
Larson R, Dai D, Hosack V, Tan Y, Bolken T (2008)
Identification of a broad-spectrum arenavirus entry inhibitor.
Journal of Virology, 82, 10768–10775.
4148 N. R. IRWIN ET AL.
Ó 2012 Blackwell Publishing Ltd
Paper 2.3.1 211
Legendre P (2000) Comparison of permutation methods for the
partial correlation and partial Mantel tests. Journal of
Statistical Computation and Simulation, 67, 37–73.
Legendre P, Desdevises Y, Bazin E (2002) A statistical test
for host-parasite coevolution. Systematic Biologists, 51, 217–
234.
Lukashevich IS, Djavani M, Rodas JD et al. (2002) Hemorrhagic
fever occurs after intravenous, but not after intragastric,
inoculation of rhesus macaques with lymphocytic
choriomeningitis virus. Journal of Medical Virology, 67, 171–186.
McCormick JB, Webb P, Krebs J, Johnson K, Smith ES (1987) A
prospective study of the epidemiology and ecology of Lassa
fever. The Journal of Infectious Diseases, 155, 437–444.
Meier-Kolthoff JP, Auch A, Huson DH, Goeker M (2007)
CopyCat: cophylogenetic analysis tool. Bioinformatics, 23,
898–900.
Mills JN, Ellis BA, Childs JE et al. (1994) HPrevalence of
infection with Junin virus in rodent populations in the
epidemic area of Argentine hemorrhagic fever. The American
Journal of Tropical Medicine and Hygiene, 51, 554–562.
Mills JN (1999) The role of rodents in emerging human
disease: examples from the hantaviruses and arenaviruses.
Ecologically-based Rodent Management, 134–160.
Mills J, Bowen M, Nichol S (1997) African arenavirusescoevolution
between virus and murid host? Belgian Journal of
Zoology (Belgium), 127, 19–28.
Miranda G, Andrades-Miranda J, Oliveira L, Langguth A,
Mattevi M (2007) Geographic patterns of genetic variation
and conservation consequences in three South American
rodents. Biochemical Genetics, 45, 839–856.
Ogbu O, Ajuluchukwu E, Uneke CJ (2007) Lassa fever in West
African sub-region: an overview. Journal of vector borne
diseases, 44, 1–11.
Paga´n I, Firth C, Holmes EC (2010) Phylogenetic analysis
reveals rapid evolutionary dynamics in the plant RNA virus
genus tobamovirus. Journal of Molecular Evolution, 71, 298–
307.
Palacios G, Druce J, Du L et al. (2008) A new arenavirus in a
cluster of fatal transplant-associated diseases. New England
Journal of Medicine, 358, 991–998.
Paterson AM, Banks J (2001) Analytical approaches to
measuring cospeciation of host and parasites: through a glass,
darkly. International Journal for Parasitology, 31, 1012–1022.
Peters CJ, Buchmeier M, Rollin PE, Ksiazek TG (1996)
Arenaviruses. In: Fields Virology (eds Fields BN, Knipe DM,
Howley PM, Chanock RM, Melnick JL, Monath TP, Roizman
R and Straus SE), pp. 1521–1552. Lippincott-Raven
Publishers, Philadelphia.
Posada D, Crandall K (1998) Modeltest: testing the model of
DNA substitution. Bioinformatics, 14, 817–818.
Poulin R, Krasnov BR, Mouillot D (2011) Host specificity in
phylogenetic and geographic space. Trends in Parasitology, 27,
355–361.
R Development Core Team, 2.12 (2010) R: A Language and
Environment for Statistical Computing. R Foundation for
Statistical Computing, Vienna, Austria.
Radoshitzky S, Abraham J, Spiropoulou C et al. (2007)
Transferrin receptor 1 is a cellular receptor for New World
haemorrhagic fever arenaviruses. Nature, 446, 92–96.
Radoshitzky S, Kuhn J, Spiropoulou C et al. (2008) Receptor
determinants of zoonotic transmission of New World
hemorrhagic fever arenaviruses. Proceedings of the National
Academy of Sciences, 105, 2664.
Ramsden C, Holmes EC, Charleston MA (2009) Hantavirus
evolution in relation to Its rodent and insectivore hosts: no
evidence for codivergence. Molecular Biology and Evolution,
26, 143–153.
Reignier T, Oldenburg J, Flanagan ML et al. (2008) Receptor
use by the Whitewater Arroyo virus glycoprotein. Virology,
371, 439–446.
Ronquist F, Huelsenbeck JP (2003) MrBayes 3: bayesian
phylogenetic inference under mixed models. Bioinformatics,
19, 1572–1574.
Rousseau MC, Saron MF, Brouqui P, Bourgeade A (1997)
Lymphocytic choriomeningitis virus in southern France: four
case reports and a review of the literature. European Journal
of Epidemiology, 13, 817–823.
Salazar-Bravo J, Ruedas LA, Yates TL (2002) Mammalian
reservoirs of arenaviruses. Current Topics in Microbiology and
Immunology, 262, 25–63.
Skinner HH, Knight EH (1979) The potential role of Syrian
hamsters and other small animals as reservoirs of
lymphocytic choriomeningitis virus. The Journal of Small
Animal Practice, 20, 145–161.
Skinner HH, Knight EH, Grove R (1977) Murine lymphocytic
choriomeningitis: the history of a natural cross-infection
from wild to laboratory mice. Laboratory Animals, 11, 219–
222.
Smith AL, Paturzo FX, Gardner EP et al. (1984) Two epizootics
of lymphocytic choriomeningitis virus occurring in
laboratory mice despite intensive monitoring programs.
Canadian Journal of Comparative Medicine, 48, 335–337.
Spiropoulou CF, Kunz S, Rollin PE, Campbell KP, Oldstone
MBA (2002) New World arenavirus clade C, but not clade A
and B viruses, utilizes a-dystroglycan as its major receptor.
Journal of Virology, 76, 5140–5146.
Stamatakis A, Ludwig T, Meier H (2004) New fast and
accurate heuristics for inference of large phylogenetic
trees. Proceedings of 18th IEEE ⁄ ACM International Parallel
and Distributed Processing Symposium Santa Fe, NM, April
26–30.
Stamatakis A, Auch A, Meier-Kolthoff J, Go¨ker M (2007)
AxPcoords & parallel AxParafit: statistical co-phylogenetic
analyses on thousands of taxa. BMC Bioinformatics, 8, 405.
Weaver S, Salas R, de Manzione N et al. (2000) Guanarito virus
(Arenaviridae) isolates from endemic and outlying localities
in Venezuela: sequence comparisons among and within
strains isolated from Venezuelan hemorrhagic fever patients
and rodents. Virology, 266, 189–195.
Webb PA, Johnson KM, Hibbs JB, Kuns ML (1970) Parana, a
new Tacaribe complex virus from Paraguay. Arch Gesamte
Virusforsch, 32, 379–388.
Webb CO, Ackerly DD, , Kembel SW (2008) Phylocom:
software for the analysis of phylogenetic community
structure and character evolution. Bioinformatics, 24, 2098–
2100.
Worobey M, Telfer P, Souquie`re S et al. (2010) Island
biogeography reveals the deep history of SIV. Science, 329,
1487.
Xia X, Xie Z (2001) DAMBE: software package for data
analysis in molecular biology and evolution. Journal of
Heredity, 92, 371–373.
HOST SWITCHING IN NEW WORLD ARENAVIRUSES 4149
Ó 2012 Blackwell Publishing Ltd
212 Paper 2.3.1
N.I. is a molecular ecologist with interest in bat systematics,
phylogeography and their diseases. M.B. is interested in using
evolutionary techniques in areas of medical importance. O.M.
is interested in studying biodiversity at all scales (local to
global) and in statistical techniques that takes phylogenetic
information into account. N.M. works on phylogenetics and
evolutionary genetics with interest in infectious diseases in
wildlife.
Data accessibility
Alignments for all analyses and the R script used for the Mantel
tests have been made available: DRYAD entry doi:10.5061/
dryad.231g746f.
Supporting information
Additional supporting information may be found in the online
version of this article.
Table S1 Associations of NWA and their hosts. Junior synonyms
previously used in the literature are indicated below
the current taxonomic name.
Table S2 GenBank accession and strain numbers of NWA
sequences used in this study.
Table S3 GenBank accession numbers and rodents species
names with their synonyms.
Table S4 Country of occurrence for the American rodent species
used in this study and their overlap with each NWA
Clade.
Figure S1 Bayesian phylogenetic trees of New World arenaviruses
based on complete nucleotide sequences of glycoprotein
(A), nucleoprotein (B), L-protein (C) and Z-protein (D).
Figure S2 Reconciliation analysis of NWA based on nucleoprotein
(A), L-protein (B), and Z-protein (C) with known hosts.
Please note: Wiley-Blackwell are not responsible for the content
or functionality of any supporting information supplied by the
authors. Any queries (other than missing material) should be
directed to the corresponding author for the article.
4150 N. R. IRWIN ET AL.
Ó 2012 Blackwell Publishing Ltd
Paper 2.3.1 213
Paper 2.3.2
Pečnerová P., Moravec J. C., Martínková N. 2015. A skull might lie: Modeling ancestral
ranges and diet from genes and shape of tree squirrels. Systematic Biology 64: 1074-
1088.
– 215 –
Syst. Biol. 64(6):1074–1088, 2015
© The Author(s) 2015. Published by Oxford University Press, on behalf of the Society of Systematic Biologists. All rights reserved.
For Permissions, please email: journals.permissions@oup.com
DOI:10.1093/sysbio/syv054
Advance Access publication August 8, 2015
A Skull Might Lie: Modeling Ancestral Ranges and Diet from Genes
and Shape of Tree Squirrels
PATRÍCIA PE ˇCNEROVÁ1,2,3,4,∗, JI ˇRÍ C. MORAVEC2,5, AND NATÁLIA MARTÍNKOVÁ2,6
1Department of Botany and Zoology, Masaryk University, Kotláˇrská 2, 602 00 Brno, Czech Republic; 2Institute of Vertebrate Biology, Academy of Sciences
of the Czech Republic, v.v.i., Kvˇetná 8, 603 65 Brno, Czech Republic; 3Department of Zoology, Stockholm University, 10691 Stockholm, Sweden;
4Department of Bioinformatics and Genetics, Swedish Museum of Natural History, PO Box 50007, 10405 Stockholm, Sweden; 5Institute of Fundamental
Sciences, Massey University, Private Bag 11 222, Palmerston North, New Zealand; and 6Institute of Biostatistics and Analyses, Masaryk University,
Kamenice 3, 625 00 Brno, Czech Republic
∗Correspondence to be sent to: Department of Bioinformatics and Genetics, Swedish Museum of Natural History, PO Box 50007, 10405 Stockholm, Sweden;
E-mail: patricia.pecnerova@nrm.se
Received 8 January 2015; reviews returned 23 March 2015; accepted 29 July 2015
Associate Editor: Norm McLeod
Abstract.—Tropical forests of Central and South America represent hotspots of biological diversity. Tree squirrels of the tribe
Sciurini are an excellent model system for the study of tropical biodiversity as these squirrels disperse exceptional distances,
and after colonizing the tropics of the Central and South America, they have diversified rapidly. Here, we compare signals
from DNA sequences with morphological signals using pictures of skulls and computational simulations. Phylogenetic
analyses reveal step-wise geographic divergence across the Northern Hemisphere. In Central and South America, tree
squirrels form two separate clades, which split from a common ancestor. Simulations of ancestral distributions show western
Amazonia as the epicenter of speciation in South America. This finding suggests that wet tropical forests on the foothills
of Andes possibly served as refugia of squirrel diversification during Pleistocene climatic oscillations. Comparison of
phylogeny and morphology reveals one major discrepancy: Microsciurus species are a single clade morphologically but are
polyphyletic genetically. Modeling of morphology–diet relationships shows that the only group of species with a direct
link between skull shape and diet are the bark-gleaning insectivorous species of Microsciurus. This finding suggests that the
current designation of Microsciurus as a genus is based on convergent ecologically driven changes in morphology. [Ancestral
range reconstruction; diet modeling; geometric morphometry; multilocus phylogeny; Sciurini; speciation.]
Speciation in the tropics is a key area of inquiry in the
study of biological diversity and the latitudinal gradient
in species richness, one of the most intensively discussed
patterns in biology (Pianka 1966; Mittelbach et al. 2007).
While there is no general explanation for tropical species
richness (Mittelbach et al. 2007), studies of diverse
model organisms show that an array of mechanisms
and processes are responsible for high biodiversity
(Haffer 1997; Schluter 2001; Via 2001). In the Neotropics,
biodiversity is mostly explained by a combined effect
of Pliocene and Miocene orogenic processes and
Pleistocene climatic oscillations (Hooghiemstra and van
der Hammen 1998; Rull 2008; Turchetto-Zolet et al. 2013).
The different effects of orogenic processes and climatic
changes have created a mosaic of phylogeographic
patterns that can be detected in Neotropical forests
today (Turchetto-Zolet et al. 2013). Tree squirrels of the
tribe Sciurini represent an excellent model system for
studying these processes. Sciurini are a widespread and
diverse group with a broad food and habitat niche. These
squirrels inhabit much of Eurasia and North America
(Thorington et al. 2012), but they are more diverse in the
recently colonized tropical forests of Central and South
America.
According to molecular dating based on multilocus
data (Mercer and Roth 2003), the tribe Sciurini
originated in the Northern Hemisphere approximately
23 Ma, when the Sciurini last shared a common ancestor
with the Pteromyini. Within the tribe Sciurini, recent
phylogenetic analyses show basal divergence between
the genus Tamiasciurus, which inhabits North America,
and the Sciurus lineage (including Microsciurus,
Rheithrosciurus, and Syntheosciurus). The genus Sciurus
originated in Eurasia and colonized the Americas after
the formation of a land bridge in Beringia (Oshida et al.
2009; Peˇcnerová and Martínková 2012). This scenario
is supported by molecular dating (Mercer and Roth
2003) as well as by independent evidence gathered
from the paleobotanical record (Wolfe et al. 1966; Wing
1998), which shows that the Beringian land corridor
was forested when the transition occurred between 8.6
and 5.3 Ma (Mercer and Roth 2003). In the Americas,
divergence of Sciurus species proceeded from north
to south, until Sciurus colonized the tropical forests
of Central America (which is considered to include
Mexico throughout the study). After the formation
of the Isthmus of Panama, squirrels entered South
America approximately 2.8 Ma (Mercer and Roth
2003). In the tropical regions of Central and South
America, Sciurus squirrels reached high levels of species
richness and species of the genera Microsciurus and
Syntheosciurus diverged from Sciurus here (Mercer and
Roth 2003; Peˇcnerová and Martínková 2012). The genus
Rheithrosciurus, which is endemic to Borneo, probably
diverged from the genus Sciurus early after the origin of
Sciurus in Eurasia (Peˇcnerová and Martínková 2012).
Species richness in the tropical forests of Central and
South America is also reflected in the Sciurini tree
squirrels. Out of the 37 species of the tribe, 22 inhabit
tropical regions of Central and South America (Wilson
and Reeder 2005). The high Neotropical biodiversity
in the phylogenetically youngest group points to rapid
1074
byguestonDecember9,2015http://sysbio.oxfordjournals.org/Downloadedfrom
216 Paper 2.3.2
2015 PE ˇCNEROVÁ ET AL.—RECONCILING PHYLOGENY AND MORPHOLOGY IN SQUIRRELS 1075
speciation (Roth and Mercer 2008), but very little is
known about these species and the driver of their
intensive diversification remains unclear.
The taxonomy of Sciurini is based on two
morphological studies from the mid-20th century,
a qualitative analysis of cranial morphology by Moore
(1959) and a paleontological revision with emphasis on
osteological and cranial measurements by Black (1963).
The tribe Sciurini currently consists of five genera:
Microsciurus, Rheithrosciurus, Sciurus, Syntheosciurus,
and Tamiasciurus. However, molecular studies in the
last decade show that species of Microsciurus and
Syntheosciurus and possibly also Rheithrosciurus are
closely related to species of the genus Sciurus and
cluster into a single lineage (Mercer and Roth 2003;
Herron et al. 2004; Steppan et al. 2004). No explanation of
the discrepancy between molecular and morphological
data has yet been provided.
In this study, we hypothesize that Sciurini
differentiated while dispersing to new regions. We
address this hypothesis by extending the genetic record
of Sciurini and by modeling the possible areas of origin.
We also examine the conflicting signals of molecular
and morphological data. We consider two hypothetical
causes of morphological differentiation and their
inconsistency with the molecular phylogeny: allopatric
differentiation and ecological adaptation. We test these
hypotheses using a combined molecular-morphological
approach, accompanied by simulations in a Bayesian
framework and predictive modeling.
MATERIALS AND METHODS
Tissue Samples
The Natural History Museum (NHM) in London
provided 34 samples of soft tissue residues retrieved
from squirrel bones and the National Museum of
Natural History of the Smithsonian Institution in
Washington, DC, provided 13 samples of frozen soft
tissues (Supplementary Table S1, available on Dryad
at http://dx.doi.org/10.5061/dryad.v2t5n). Samples
represented 36 species, out of which 34 were species
of the tribe Sciurini and two species were used as an
outgroup (Glaucomys volans and Pteromys volans).
DNA Extraction and Amplification
We used DNeasy Blood and Tissue Kit by QIAGEN
(Hilden, Germany) to extract total genomic DNA. Due
to the characteristic features of museum specimens,
including chemical treatment, risk of contamination
and DNA damage affecting molecules after the death
of an organism (Pääbo et al. 2004; Martínková and
Searle 2006), we used a conservative approach to obtain
sequences from this material. We accepted results from
relatively long PCR products, indicating that the DNA
was preserved in sufficient quality and quantity (c.f.
Martínková and Searle 2006).
We amplified three mitochondrial loci (12S rRNA,
16S rRNA, and MT-CYB) and two nuclear loci
(MYC and IRBP). Sequences of MT-CYB for four
samples (Sciurus deppei, Sciurus flammifer, Sciurus
igniventris, and Sciurus pyrrhinus) were obtained
from overlapping fragments with newly designed
internal primers (Supplementary Table S2, available
on Dryad at http://dx.doi.org/10.5061/dryad.v2t5n).
Amplifications for sequencing were performed in 50 L
volumes and the reaction mixture contained 2 L of
DNA, 2 mM MgCl2, 0.2 mM dNTPs, 1.0 unit of Platinum
Taq DNA Polymerase (Invitrogen, Life Technologies,
Carlsbad, CA, USA), 1×PCR buffer, 0.2 M of each
primer,andddH2Ouptothetotalvolume.Amplification
conditions were adjusted for each primer pair (Table 1).
The common steps for all amplifications were initial
denaturation at 94°C for 3 min, followed by 35–40
cycles of 94°C for 1 min, 49–60°C for 30 s to 1 min
and 72°C for 45 s to 3 min, with the final extension
at 72°C for 3 min (Table 1). We were able to extract
and amplify DNA from 35 samples of 28 species. The
amplified products were purified and sequenced by
Macrogen Europe (Amsterdam, the Netherlands) with
both amplification primers. The sequence data set was
enriched with tree squirrel sequences from GenBank.
The complete data set for molecular analyses then
contained 30 species (Supplementary Table S3, available
on Dryad at http://dx.doi.org/10.5061/dryad.v2t5n).
We used CodonCode Aligner (CodonCode Corporation,
Centerville, MA, USA) to process chromatograms into
consensus sequences.
To verify that the generated sequences represent
endogenous DNA of Sciurini squirrels, all sequences
were checked using BLAST (Altschul et al. 1990). The
sequences were the most similar to respective squirrel
species expected to be the most closely related, except
for three samples (12S rRNA and 16S rRNA sequences
of Sciurus ignitus, Sciurus richmondi, and 12S rRNA
sequence of Sciurus stramineus) that matched human
DNA and these were excluded from further analyses.
Phylogenetic Analyses of the Supermatrix
The DNA sequences were aligned in ClustalW2
(Larkin et al. 2007; Goujon et al. 2010) and the rRNA
loci sequences were aligned according to their predicted
secondary structure in MAFFT 7.2 with the q-ins-i
algorithm (Katoh and Standley 2013). The alignments
were manually edited in BioEdit 7.1.3 (Hall 1999).
We used two concepts of phylogenetic inference,
supermatrix and supertree. The supermatrix approach
requires sequences of all loci and taxa to be concatenated
into a single matrix. The concatenated data set was
constructed in SequenceMatrix 1.7.8 (Vaidya et al. 2011)
and analyzed by maximum-likelihood (ML) method
in RAxML 8.1.15 (Stamatakis 2014) and by Bayesian
inference (BI) in MrBayes 3.2.1 (Huelsenbeck and
Ronquist 2001). Each analysis used a partitioning scheme
selected according to the Bayesian Information Criterion
in PartitionFinder 1.0 (Lanfear et al. 2012) with linked
byguestonDecember9,2015http://sysbio.oxfordjournals.org/Downloadedfrom
Paper 2.3.2 217
1076 SYSTEMATIC BIOLOGY VOL. 64
TABLE 1. Characteristics of sequence alignments analyzed in this study
Locus PCR Alignment Partitioned substitution
amplification model
Number of Length Gene BI from ML from Species
sequences trees supermatrix supermatrix tree
12S rRNA (mt) Ta =60, te =3 25 867 GTR+ GTR+ 1 GTR+ 1 GTR+ 1
16S rRNA (mt) Ta =49, te =1.5 18 1077 GTR+ GTR+ 1 GTR+ 1 GTR+ 1
MT-CYB (mt) Ta =55, te =2 22 1140 GTR+ , GTR+ 1, GTR+ 1, GTR+ 2
HKY+I, GTR+ HKY+I8, GTR+ 9 GTR+ 4, GTR+ 5
D-loop (mt) n/a 8 1228 GTR+ GTR+ 10 GTR+ 6 GTR+ 3
MYC exon n/a 5 674 JC, HKY JC6, JC6, GTR+ 4, HKY4
2 (nuc) (3rd) HKY7 GTR+ 4, GTR+ 2
MYC exon Ta =60, te =0.75 14 967 K80, K80 HKY+I3, GTR+ 3, GTR+I5
3 (nuc) (3rd) HKY+I4, K805 GTR+ 4, GTR+ 3
IRBP (nuc) Ta =58, te =3 19 1253 HKY+ , GTR+ 2, GTR+ 2, GTR+I5
HKY, HKY+I HKY+I3, HKY+I4 GTR+ 3, GTR+ 4
RAG1 (nuc) n/a 6 2141 HKY, HKY HKY+I3, GTR+ 3, GTR+I5
(3rd) HKY+I4, GTR+ 2 GTR+ 4, GTR+ 2
Total 117 9347
Notes: mt, mitochondrial; nuc, nuclear; Ta annealing temperature in °C; te, extension time in minutes; n/a, not available; 1−10 partitions with
jointly estimated model parameters in the given analysis, model order corresponds to codon positions unless specified as the third codon position.
branches and a greedy algorithm that considered
non-coding genes and codon positions of the coding
genes. The tested models were: JC, K80, HKY, GTR
for individual genes; JC, HKY, GTR for BI of the
concatenated data set and predefined RAxML model set
for ML tree (see Table 1). Models for individual genes
and the concatenated data set could include proportion
of invariable sites (I) or rate heterogeneity modeled with
a distribution model ( ).
In the ML method implemented in RAxML 8.1.15
(Stamatakis 2014), the concatenated data set was
analyzed with six partitions (Table 1). The analysis
was conducted with the rapid bootstrapping algorithm,
which combines ML search and bootstrapping, and
was run with 1000 bootstrap searches. In the BI of
the concatenated data set conducted in MrBayes 3.2.1
(Huelsenbeck and Ronquist 2001), a substitution model
with 10 partitions was applied (Table 1). Convergence
and chain mixing were facilitated by running five
Markov Chain Monte Carlo (MCMC) chains for 3 million
generations, with trees sampled every 1000th generation.
One chain swap was attempted every third generation
and the temperature was adjusted to 0.08 after the
initial run when the default temperature (0.1) did not
converge. Successful neighboring chain swaps between
30% and 70% showed that chain mixing was effective
with this setting. A burn-in fraction of 30% was used
for all analyses. We verified that the runs converged and
reached stationary distribution by observing that after
3 million generations the average standard deviation of
split frequencies reached a value below 0.01 (0.0087) and
all potential scale reduction factors approached 1.0 as the
run converged.
Phylogenetic Analyses of Supertrees
BI of the gene trees was conducted in MrBayes
3.2.1 (Huelsenbeck and Ronquist 2001) with appropriate
partitioning schemes and substitution models estimated
for each locus separately (Table 1). The BI was conducted
with four (for RNA loci) or five (for the other loci) MCMC
chains run for 2 million generations, with trees sampled
every 1000th generation, a chain swap attempted every
third generation of the run and temperature set to
0.1. MCMC convergence was assessed as described
above.
In this study, gene trees were analyzed by two methods
of supertree reconstruction, SuperTriplets (Ranwez et al.
2010) and a veto method implemented in PhySIC_IST
(Scornavacca et al. 2008). The SuperTriplets method
uses a polynomial algorithm to search for a median
tree (a tree minimizing the sum of distances to the
source trees) based on triplets (rooted, binary, threeleaf
topologies) (Ranwez et al. 2010). The SuperTriplets
analysis was executed with the default settings
through the online version of the program available
at http://www.supertriplets.univ- montp2.fr/, (last
accessed14August,2015).Vetomethodsshowtopologies
supported by all source (gene) trees and eliminate
taxa that produce conflicting topologies. Hence, these
methods are more conservative and usually result
in less resolved supertrees. We performed a veto
analysis in the PhySIC_IST program available online
at http://www.atgc-montpellier.fr/physic_ist/, (last
accessed 14 August, 2015), with the bootstrap threshold
for source clade selection set to default value (0) and
the correction threshold used by source tree correction
(preprocessing the data to detect anomalies) set to
0.3. Four taxa were omitted from the analysis due to
conflicting topologies.
Species Trees Reconstruction
We used two methods that model the coalescent
process to infer species trees. SVDquartets (Chifman
and Kubatko 2014) uses a score based on decomposition
byguestonDecember9,2015http://sysbio.oxfordjournals.org/Downloadedfrom
218 Paper 2.3.2
2015 PE ˇCNEROVÁ ET AL.—RECONCILING PHYLOGENY AND MORPHOLOGY IN SQUIRRELS 1077
of a matrix to assess a supported topology for taxa
quartets. The matrix consists of frequencies of splits
that characterize quartets of investigated taxa based on
available sequence data. We used the method embedded
in PAUP 4.0a142 (Swofford 2015), utilizing exhaustive
quartet sampling (27,405 quartets) and 100 bootstrap
replicates for the multispecies coalescent.
BI of species trees in *BEAST 2.2 (Heled and
Drummond 2010) jointly estimates gene trees, where
the gene trees must be embedded in the species tree,
and divergence times and population sizes of present
and ancestral taxa using MCMC sampling from the
posterior. Substitution model partitioning scheme was
estimated without distinguishing codon positions to
retain gene partitioning in the species tree reconstruction
(Table 1). MCMC was run for 10 million generations
sampled every 1000th step with a birth–death speciation
tree prior and strict clock. Convergence was checked in
Tracer 1.6 (Rambaut et al. 2013), where trace of model
likelihood and parameter values oscillated around a
plateau value, their posterior densities showed roughly
unimodal distribution and estimated sample sizes (ESSs)
were >100.
Analysis of Phylogenetic Stability
The impact of missing data was evaluated in
Phyloterraces (Sanderson et al. 2011). The algorithm
breaks the supertree into subtrees representing taxa sets
as they appear in the individual gene trees. The subtrees
are further subdivided into triplets that appear in the
final, binary tree. The triplets are used to construct all
possible parent trees. Given the amount of missing data
and its location in the alignment, the parent trees will
have identical tree likelihood. For the purpose of this
analysis, the BI tree was summarized with resolved all
compatible relationships.
Testing Microsciurus Monophyly
Support for Microsciurus monophyly was estimated
implicitly from the posterior tree sample and explicitly
by comparing likelihoods of trees sampled from
constrained and negatively constrained analyses.
Following Bergsten et al. (2013) and Suchard et al.
(2005), support for a certain topology could be estimated
from the posterior tree sample. MCMC samples tree
space and the posterior probability of a tree is computed
as the frequency of posterior trees conforming to a
hypothesis. We explored our posterior tree samples from
the BI based on the supermatrix and Bayesian coalescent
inference of species trees.
We tested Microsciurus monophyly by comparing
model likelihoods from two BI analyses: a constraint
analysis in which monophyly of Microsciurus alfari and
Microsciurus flaviventer was enforced, and a negative
constraint in which monophyly of this pair was
disallowed. Analytical parameters were as described
above. Model likelihoods were estimated with the
harmonic mean estimator and compared with Bayes
factors (BFs).
Molecular Dating
Divergence dates were inferred with Bayesian
coalescent analysis (Bouckaert et al. 2014) using multiple
time constraint priors on internal nodes. The root of
Palearctic Sciurini prior was constrained with a uniform
distribution between 3 and 36 Ma, assuming that Sciurus
was present in Europe since late Pliocene, and the earliest
known squirrel fossil, Douglassciurus jeffersoni, was dated
to late Eocene (Thorington et al. 2012). The time
constraints on root of Nearctic Sciurus between 7.4 and 36
Ma were based on a combination of our results with the
fossilrecord.Ourdatashowthat SciuruscolonizedNorth
America from Asia, but the oldest fossil attributable to
Sciurus dates back to late Miocene in Nevada (Emry et al.
2005). The colonization therefore must have occurred
before the earliest estimated opening of the Bering Strait
7.4 Ma (Marincovich and Gladenkov 1999). Expecting
that Sciurini arrived to South America via the Isthmus
of Panama, their root prior was set to log-normal
distribution with mean of tested prior parameter values
equal to 3 Ma (mean=eμ+σ2/2, where μ=1.0186, σ=0.4)
(Keigwin1982).Twotimeconstraintswithin Tamiasciurus
are based on their first known fossils. Tamiasciurus
hudsonicus was first reported from the Irvingtonian
(Steele 1998), implemented as exponential prior with
mean = 0.5, offset by 0.24 Ma, for the Tamiasciurus root.
The split between Tamiasciurus douglasii and Tamiasciurus
mearnsi is believed to be complete by the end of the
last glaciation (Steele 1999) and was modeled with
exponential prior with mean = 0.5, offset by 0.015 Ma.
Tree and clock models were tested with BF, comparing
marginal likelihoods from analyses with strict clock
and log-normal relaxed clock and with birth–death and
Yule speciation tree priors. The MCMC algorithm ran
for 100 million generations and was sampled every
1000th generation. The substitution model partitioning
scheme was identical to that used in the BI from the
supermatrix. Convergence was assessed as described
above.
Morphological Analyses of Skulls
Morphological data were based on digital
photographs of skulls of 525 specimens that represented
25 species, out of which 24 were species of the tribe
Sciurini and one species was used as an outgroup
(Supplementary Table S5, available on Dryad at
http://dx.doi.org/10.5061/dryad.v2t5n). All species
were represented by 10–27 specimens, and samples
of both sexes and various localities within the species
range were included (Supplementary Table S6, available
on Dryad at http://dx.doi.org/10.5061/dryad.v2t5n).
Ventral and dorsal photographs of the skulls were
captured with a digital camera Nikon D90. Two different
tripods were used to secure the camera due to data
byguestonDecember9,2015http://sysbio.oxfordjournals.org/Downloadedfrom
Paper 2.3.2 219
1078 SYSTEMATIC BIOLOGY VOL. 64
collection in different museums. As a result, specimens
from the NHM in London were photographed from a
larger distance. To overcome the influence of different
distance, a ruler was used in all photographs and size
was recalibrated during scoring landmarks in tpsDig
2.16 (Rohlf 2005). The use of plasticine was not allowed;
therefore, for capturing the ventral side, a skull was
horizontally aligned by supporting the frontal part of
the skull with a ruler, so that the occlusal surface was
parallel to the underlying base. The alignment was
assessed by eye. For capturing the dorsal side, skulls
were aligned by positioning on incisors and auditory
bullae. Some specimens were missing incisors (~2%)
and there was a slight variability in the preservation of
incisors, which might have subtly influenced the angle
under which a skull was captured. However, including
at least 10 specimens per species should minimize the
differences. We also repeated all analyses including
only species with at least 20 specimens to verify that
the sample size was sufficient and the results were
consistent.
Skull shape was captured with two-dimensional
landmarks in tpsDig 2.16 (Rohlf 2005). 2D methodology
is routinely applied to the study of rodent cranial
morphology (Fadda and Corti 2000; Cardini and
O’Higgins 2004; Cardini et al. 2005; Macholán et al.
2008) because of its “cost-effectiveness, rapidity of data
collection and analytical simplicity” (Cardini 2014).
Cardini (2014) studied the influence of missing the
third dimension in 2D analyses and concluded that
the congruence of 2D and 3D data is modest, but
there is a level of inaccuracy when 2D landmarks
are applied to a 3D object and these studies
require cautious interpretations. For that reason, we
refrained from making conclusions regarding specieslevel
differences in morphology and we concentrated
on morphological distinction between Microsciurus,
Sciurus, and Tamiasciurus. Twelve landmarks were
digitized on the dorsal side and 21 on the ventral
side (Fig. 1, Table 2). Landmarks were digitized on
the left part of the skull to avoid the influence of
asymmetry(CardiniandO’Higgins2004;Macholánetal.
2008). Information on size, location, and orientation of
specimens were removed by generalized least-squares
Procrustes superimposition (Gower 1975; Rohlf and Slice
1990) in tpsRelw 1.49 (Rohlf 2010). The shape space
in geometric morphometry (Kendall’s shape space) is
a curved surface and has a non-Euclidean geometry.
Approximation by the linear tangent space is used (Rohlf
1998), but validity of such an approximation needs to be
tested by estimating correlation between the Procrustes
distances of Kendall’s shape space and Euclidean
distances of the linear tangent space. This approximation
was evaluated in tpsSmall 1.20 (Rohlf 2003).
Shape change was visualized by the thin-plate spline
method (Bookstein 1989), which is based on the
deformation of a regular grid by bending energy, as an
analogy of energy required for shape change. Bending
energy is expressed by variables called principal warps.
These indicate the directions of shape change. Other
FIGURE 1. Positions of landmarks digitized on photographs of
squirrel skulls on their dorsal (a) and ventral (b) side. The skull of S.
vulgaris was borrowed from the collections of the Institute of Vertebrate
Biology of the Academy of Sciences of the Czech Republic.
variables, called partial warps, are used to illustrate
the proportion of shape change each principal warp
accounts for (Bookstein 1989).
To reduce the dimensionality of the data, we used
relative warp analysis (RWA; Rohlf 1993), which is a
version of principal component analysis (Pearson 1901;
Hotelling 1933) implemented in tpsRelw. Landmark
coordinates were used to perform linear discriminant
function analysis (DFA) to determine variables that can
be used to divide samples into groups. Multivariate
analysis of variance (MANOVA) was performed on
Procrustes distances to test the statistical significance
of differences between groups, applying Goodall’s
F-test with 999 permutations. Unweighted pair-group
method using arithmetic averages (UPGMA) was used
as the clustering method to construct trees based on
Mahalanobis distances between predicted DFA species
averages. MANOVA, DFA, and UPGMA were performed
in R (Paradis et al. 2004; Adams and Otarola-Castillo
2013; McFerrin 2013; R Core Team 2013).
The composition of species differed between
molecular and morphological analyses, with 19
species being examined with both approaches
(Supplementary Table S7, available on Dryad at
http://dx.doi.org/10.5061/dryad.v2t5n). The results
byguestonDecember9,2015http://sysbio.oxfordjournals.org/Downloadedfrom
220 Paper 2.3.2
2015 PE ˇCNEROVÁ ET AL.—RECONCILING PHYLOGENY AND MORPHOLOGY IN SQUIRRELS 1079
TABLE 2. Description of landmark positions on dorsal and ventral sides of squirrel skulls
Side Number Description
Dorsal 1 Posterior end of the curvature of the occipital
2 Intersection of the coronal and sagittal sutures
3 Intersection of the naso-frontal suture in the midline
4 Tip of the nasals at their anterior suture
5 Anterior tip of suture between nasal and premaxilla
6 Anterior tip of the suture between premaxilla and maxilla
7 Lateral end of the naso-frontal suture in the dorsal projection
8 Point on frontal at greatest interorbital constriction
9 Posterior base of the postorbital process
10 Rostral end of the zygomatic process of the temporal bone
11 Most posterior meeting point between jugal and squamosal process of the zygomatic arch
12 Posterio-lateral end of the interparietal–occipital suture
Ventral 1 Posterior limit of the foramen magnum
2 Anterior limit of the foramen magnum
3 Posterior midline suture of palatines
4 Intersection of the maxillo-palatine suture in the midline
5 Posterior end of the incisive foramen
6 Anterior end of the incisive foramen
7 Rostralmost point of the upper incisor in the midline
8 Tip of the nasals at their anterior suture
9 Rostral end of the zygomatic plate
10 Most anterior point of the orbit (in the ventral view)
11 Maximal curvature on internal zygomatic bar
12 Inner extreme curvature point of the zygomatic bar at the squaniosus process
13 Posterior tip of the zygomatic arch
14–15 Anterior and posterior tips of the external auditory meatus
16–17 Lateral tips of the occipital condyle
18 Lateral end of the basisphenoid–basioccipital suture in the dorsal projection
19 Posterior extremity of the foramen ovale
20 Anterior extremity of the toothrow
21 Posterior extremity of the toothrow
Notes: Landmarks were chosen on the basis of similar morphometric studies on rodents (Fadda and Corti 2000; Cardini and O’Higgins 2004;
Cardini et al. 2005; Macholán et al. 2008).
from the molecular genetic data are classified
with Roman numerals while the results from the
morphological data are classified with Arabic numbers.
Spatial and Ecological Hypotheses Testing
We evaluated Pearson’s correlation between genetic
and geographic distances with Mantel’s test in R
(Oksanen et al. 2013; R Core Team 2013), constructing the
genetic matrix from phylogenetic distances computed
from the chronogram and the geographic matrix
from range centroids in the International Union for
Conservation of Nature (IUCN) map data set (IUCN
2013) using the Geographic Distance Matrix Generator
software (Ersts 2014). To evaluate statistical significance,
we used 9999 permutations.
Probability of ancestral states of species ranges was
calculated with respect to the topological uncertainty of
the phylogeny in SIMMAP 1.5 (Bollback 2006). Species
were scored with multistate characters according to
their occurrence in Eurasia, North, Central and South
America, and ancestral ranges were evaluated from 100
trees randomly sampled from the posterior.
To test for the ancestral area of the South American
clade,wediscretizedrangepolygonsfromtheIUCNdata
set for South American species ranges by creating a grid
overSouthAmericawithacellsizeequaltothreedegrees
longitude by three degrees latitude. Grid cells were
coded as presence/absence data based on overlapping
species range polygons. Ancestral area simulations in
BayArea 1.0.2 (Landis et al. 2013) were run for 600
million MCMC generations on a pruned chronogram
that included only the South American species with
truncated geographic distances setting and with 20%
burnin. Ancestral ranges were visualized in BayArea-fig
(Landis et al. 2013).
To evaluate the relationship between dietary
preferences and skull shape, we used a generalized
linear model with a probit linkage function with relative
warps (RWs) as predictors and binarized dietary
preference (Supplementary Table S8, available on Dryad
at http://dx.doi.org/10.5061/dryad.v2t5n) as the
response variable. The dietary preferences of Sciurini
were estimated on the basis of Thorington et al. (2012).
A species’ diet was coded as follows: 0—food item not
mentioned/not consumed, 1—food item consumed,
2—food item highlighted as a common or preferred diet.
We used three sets of models, with dorsal, ventral, and
combined RW values. Each binarized food preference
was estimated separately and backward stepwise
regression using AIC was performed to simplify the
byguestonDecember9,2015http://sysbio.oxfordjournals.org/Downloadedfrom
Paper 2.3.2 221
1080 SYSTEMATIC BIOLOGY VOL. 64
model. Using these models, response variables (dietary
components) were predicted and compared with the
original values.
To further investigate the relationships between skull
shapes and dietary preferences, we constructed a
neighbor-joining (NJ) tree from Manhattan distances
based on the dietary preference table (Supplementary
Table S8, available on Dryad at http://dx.doi.org/
10.5061/dryad.v2t5n). The diets of tree squirrels are
generally variable, but due to the usage of a matrix of
all consumed items, clusters in the NJ tree avoid the
problem of a misleading classification into predefined
dietary categories. Sciurus richmondi was excluded from
all analyses of the relationship between diet and
morphology as no information about its diet was
available.
RESULTS
We obtained 42 sequences, including nine previously
unsampled species. With sequences from GenBank, our
data set consisted of 117 sequences distributed between
eight loci—12S rRNA, 16S rRNA, MT-CYB, D-loop, MYC
exon 2, MYC exon 3, IRBP, and RAG1. Number of
sequences per locus ranged between 5 and 25. The
concatenated data set consisted of 9347 base pairs (bp).
Gene alignments ranged from 674 to 2141 bp (Table 1).
The missing data represented 64.27% of the data set.
Phylogenetic Relationships
The ML and BI trees based on the supermatrix yielded
almost identical results (Fig. 2) and were consistent
with the SuperTriplets and veto trees reconstructed
from gene trees and species trees based on the
coalescent (Supplementary Fig. S1, available on Dryad
at http://dx.doi.org/10.5061/dryad.v2t5n). All trees
recognizedthreelineages.First,therewasasplitbetween
the genus Tamiasciurus (lineage I) and the other taxa.
Afterwards, the latter lineage evolved in the direction
of Rheithrosciurus (lineage II) and the remaining genera
(Microsciurus, Sciurus, and Syntheosciurus; lineage III).
The likelihood terrace in the tree space with the final
tree contained three alternative topologies in total,
meaning that the amount of missing data resulted in
the inability of likelihood-based tree reconstruction to
distinguish unequivocally between these topologies. The
consensus tree from all topologies sharing a terrace
showed that the position of Sciurus anomalus and
Rheithrosciurus macrotis varied on the alternative tree
topologies (Supplementary Fig. S4, available on Dryad
at http://dx.doi.org/10.5061/dryad.v2t5n).
In lineage III, no structure representative of the
current taxonomy was observed; instead, species
were grouped according to biogeography. After the
split from Rheithrosciurus, all Palearctic species of
Sciurus diversified gradually (Sciurus lis + Sciurus
vulgaris and S. anomalus), followed by gradual
speciation of Nearctic Sciurini (Sciurus niger, Sciurus
carolinensis with undetermined relationship to Sciurus
aberti + Sciurus griseus) and at last the divergence of the
Neotropical clade (that included genera Microsciurus
and Syntheosciurus) from the Nearctic ancestor.
Statistical support for the monophyletic origin of
Neotropical species was marginally significant in the
BI tree, but not significant in the ML tree (taking into
account the significance level of 70% for bootstrap in
ML and 0.95 for posterior probability in BI throughout
G. volans
M. alfari
M. flaviventer
M. mimulus
S. aberti
S. aestuans
S. alleni
S. arizonensis
S. aureogaster
S. carolinensis
S. colliaei
S. deppei
S. granatensis
S. griseus
S. igniventris
S. lis
S. nayaritensis
S. niger
S. oculatus
S. richmondi
S. variegatiodes
S. vulgaris
S. yucatanensis
T. douglasii
T. hudsonicus
0.03
Sciurus spadiceus
Sciurus aberti
Sciurus colliaei
Tamiasciurus mearnsi
Pteromys volans
Sciurus oculatus
Microsciurus flaviventer
Sciurus gilvigularis
Sciurus alleni
Sciurus niger
Sciurus flammifer
Sciurus ignitus
Sciurus aestuans
Rheithrosciurus macrotis
Tamiasciurus douglasii
Sciurus variegatoides
Sciurus anomalus
Sciurus stramineus
Syntheosciurus brochus
Microsciurus alfari
Sciurus pyrrhinus
Sciurus carolinensis
Sciurus granatensis
Sciurus deppei
Sciurus vulgaris
Tamiasciurus hudsonicus
Glaucomys volans
Sciurus griseus
Sciurus igniventris
Sciurus lis
0.95/87
0.99/86
1/100
0.56/55
0.97/100
0.98/93
0.98/89
1/100
0.79/89
0.97/85
0.94/55
0.98/98
0.95/99
1/100
0.98/75
0.89/55
0.77/58
0.92/77
1/89
0.67/-
Group
III-B
Group
III-A
LineageIIILineageII
LineageI
FIGURE 2. Phylogenetic tree of Sciurini tree squirrels calculated from supermatrix DNA sequence data using BI analysis (left) and UPGMA
dendrogram of morphological shape (right). Branch labels indicate Bayesian posterior probabilities and bootstrap support from the ML analysis.
Pie charts at basal nodes represent probability of ancestral distributions for the node, given the topological uncertainty of the phylogeny. Inset
map shows respective continents of origin. Tanglegram depicts disparity between molecular phylogeny and morphological similarity.
byguestonDecember9,2015http://sysbio.oxfordjournals.org/Downloadedfrom
222 Paper 2.3.2
2015 PE ˇCNEROVÁ ET AL.—RECONCILING PHYLOGENY AND MORPHOLOGY IN SQUIRRELS 1081
the study). Two reciprocally monophyletic groups
with significant support were recognized inside the
Neotropical clade. The first group (III-A) consisted of
Central American taxa with M. alfari as the sister branch
to other taxa, and an unresolved relationship of S.
deppei, Sciurus variegatoides, Sciurus colliaei + Sciurus
granatensis and Syntheosciurus brochus. The second group
(III-B) included South American taxa, but relationships
of taxa after divergence from the common ancestor were
unresolved, in agreement with a rapid diversification.
However, three groupings were significantly supported
in group III-B. First, S. pyrrhinus and S. flammifer + S.
igniventris, second, Sciurus aestuans + Sciurus gilvigularis
and third, S. ignitus + Sciurus spadiceus.
The SuperTriplets and veto trees constructed from
gene trees were similar (Supplementary Fig. S1, available
on Dryad at http://dx.doi.org/10.5061/dryad.v2t5n).
They recognized the same three clades as retrieved
from the supermatrix, but the mutual relationships
were not clear. The veto method takes into account
only uncontradicted relationships, therefore, several
taxa with conflicting topologies in the gene trees
were excluded from the analysis (R. macrotis, S.
aestuans, S. deppei, and S. ignitus). The main difference
compared with the supermatrix-based trees (ML and
BI) was that the Central American taxa did not form
a monophyletic group and almost all relationships
inside the Neotropical clade were unresolved.
Comparing the SuperTriplets tree to the ML and
BI trees, S. griseus occurred in a different position
than the previously suggested close relationship
with S. aberti (Supplementary Fig. S1b, available on
Dryad at http://dx.doi.org/10.5061/dryad.v2t5n).
Two monophyletic groups inside the Neotropical
clade were also recognized, but the relationships
inside these groups slightly differed (Supplementary
Fig. S1b, available on Dryad at http://dx.doi.org/
10.5061/dryad.v2t5n).
In species trees based on the coalescent, the basic
tree structure was recognizable, but not supported
(Supplementary Fig. S1c,d, available on Dryad
at http://dx.doi.org/10.5061/dryad.v2t5n). Both
species trees placed North American S. griseus within
the South American group. The SVDquartets tree
(Supplementary Fig. S1c, available on Dryad at
http://dx.doi.org/10.5061/dryad.v2t5n) additionally
placed R. macrotis at the root of Sciurini.
Phylogenetic–biogeographic discongruency of
Sciurus alleni and Sciurus oculatus
Two species (S. alleni and S. oculatus) inhabiting
Central America were placed close to the Neotropical
clade. We considered this placement as an analysis
artifact, because these two species were represented
only by sequences of the 12S rRNA locus that provided
no information about the relationships of Neotropical
squirrels (Supplementary Fig. S2, available on Dryad at
http://dx.doi.org/10.5061/dryad.v2t5n).
Nonmonophyly of Microsciurus
No trees from the posterior of either the supermatrix
BI or species tree BI contained trees where M. alfari
and M. flaviventer were sisters. The probability of this
happening by chance alone, if prior distribution is
considered, is 5.61 × 10−39 for the BI from a supermatrix
posterior sample and 3.15 × 10−62 for the BI of
species trees posterior sample. The constraint test of
Microsciurus monophyly showed very strong support for
nonmonophyly of the genus (2 log(BF) =−119.82).
Morphology of Skulls
The RWA showed that the first two RWs captured
62.5% of the variation in the data for the dorsal side
and 47.6% for the ventral side of the tree squirrel skulls
(Fig. 3; further plots in Supplementary Fig. S2, available
on Dryad at http://dx.doi.org/10.5061/dryad.v2t5n).
The correlation between Procrustes and Euclidean
distances was high (both sides: r=1.000, slope b=
0.999, intercept a=0.000), which allowed us to proceed
with subsequent analyses. DFA successfully assigned
specimens into defined species with an average rate
72.9% (Goodall’s F=47.083, P=0.001) for the dorsal
side and 76.3% (Goodall’s F=32.557, P=0.001) for
the ventral side (Supplementary Table S9, available
on Dryad at http://dx.doi.org/10.5061/dryad.v2t5n).
The most successfully assigned specimens belonged to
the outgroup G. volans with 100% for both sides, and
the least successfully S. granatensis for the dorsal side
(40.9%). Most misclassifications occurred among closely
related species. For example, the genus Microsciurus
had 100% success in assigning measured skulls into the
appropriate genus, but the assignment into particular
species of the genus was less successful (58.3–83.3%).
Goodall’s F-test showed that affiliation to a species
explained 69.3% of differences in the dorsal shape of
the skull (F=47.083; df = 24,500; P=0.001) and 61% of
differences in the ventral shape of the skull (F=32.557;
df = 24,500; P=0.001).
According to the RWA (Fig. 3) and the cluster
analysis (Supplementary Fig. S3, available on Dryad
at http://dx.doi.org/10.5061/dryad.v2t5n), three
morphological groups were recognized in the tribe
Sciurini. The most distinct skull shape was observed
in the genus Microsciurus, which formed a separate
cluster against all other squirrels. The second group
was represented by the genus Tamiasciurus and the
third group included all Sciurus squirrels (Figs. 2
and 3 and Supplementary Fig. S3, available on
Dryad at http://dx.doi.org/10.5061/dryad.v2t5n). The
Sciurus group was divided into three subgroups
(Supplementary Fig. S3, available on Dryad at
http://dx.doi.org/10.5061/dryad.v2t5n). The first
subgroup (3.2) contained squirrels of North America
and northern Mexico (except for the Neotropical S.
igniventris), the second (3.1) consisted of Neotropical
squirrels, and the third subgroup (3.3) contained both
Neotropical and Palearctic squirrels.
byguestonDecember9,2015http://sysbio.oxfordjournals.org/Downloadedfrom
Paper 2.3.2 223
1082 SYSTEMATIC BIOLOGY VOL. 64
−0.06 −0.04 −0.02 0.00 0.02 0.04 0.06 0.08
−0.04−0.020.000.020.040.06
RW 1
−0.10 −0.05 0.00 0.05
−0.10−0.050.000.05
RW 1
RW2RW2
Glaucomys volans
Microsciurus alfari
Microsciurus flaviventer
Microsciurus mimulus
Sciurus aberti
Sciurus aestuans
Sciurus alleni
Sciurus arizonensis
Sciurus aureogaster
Sciurus carolinensis
Sciurus colliaei
Sciurus deppei
Sciurus granatensis
Sciurus griseus
Sciurus igniventris
Sciurus lis
Sciurus nayaritensis
Sciurus niger
Sciurus oculatus
Sciurus richmondi
Sciurus variegatoides
Sciurus vulgaris
Sciurus yucatanensis
Tamiasciurus douglasii
Tamiasciurus hudsonicus
a)
b)
FIGURE 3. Plots of the first two relative warps (RW1, RW2) for dorsal (a) and ventral (b) sides of tree squirrel skulls. The first two RWs
accounted for 62.5% and 47.6% of variability in the skull sides, respectively. Deformation grids show the shape change along the first two RW
axes.
The most distinct features of the skull of Microsciurus
in comparison to Sciurus and Tamiasciurus (except for
their smaller and broader shape) were their more convex
shape, short zygomatic arches, small orbits, short nasals,
and wide interorbital region. In detail, the deformation
grids for dorsal RW1 showed how with increasing
values the interorbital region gets broader, that is, the
interorbital constriction gets narrower, and the width
byguestonDecember9,2015http://sysbio.oxfordjournals.org/Downloadedfrom
224 Paper 2.3.2
2015 PE ˇCNEROVÁ ET AL.—RECONCILING PHYLOGENY AND MORPHOLOGY IN SQUIRRELS 1083
of a skull gets more uniform. The most negative values
(the narrowest interorbital region and relatively broad
posterior part of a skull) represented the outgroup taxon
G. volans. Tamiasciurus had an intermediate form and
Microsciurus and Sciurus species had broader interorbital
space. The axis for dorsal RW2 showed size change
of nasals and zygomatic arches that were shorter and
wider in Microsciurus (negative values) and longer and
narrower for, for example, S. griseus (positive values).
The axis for ventral RW1 showed differences in relative
length of nasals and the anterior part of a skull that
were shorter for more positive values (Glaucomys and
Microsciurus) and longer for more negative values (e.g.,
S. igniventris). The axis for ventral RW2 showed that
the relative width of the anterior part of a skull and
zygomatic arches became narrower with more positive
values (Glaucomys) and wider with more negative values
(Microsciurus).
Spatio-Temporal and Ecological Hypotheses of Molecular
and Morphometric Variation
We found a significant correlation between
phylogenetic distances of taxa and distances of their
species ranges (Mantel’s test: r=0.3574, P=0.0226),
indicating coupled dispersal and genetic diversification
in Sciurini tree squirrels. The probability of occurrence
of the ancestor of Sciurini in the Palearctic region
is 0.70 and in the Nearctic region is 0.26. We found
strong support for a single southward colonization with
diversification of lineage III in the Americas (Fig. 2).
To test stability of the ancestral range of South
American Sciurini, the number of generations of MCMC
sampling was iteratively increased from 100 to 600
million. In each run, Western Amazonia was identified
as the ancestral state (Fig. 4). However, in runs with
<150 million generations, other putatively ancestral
areas showed similar posterior probability. Estimated
posterior distributions for parameters for all runs were
long tailed, although estimated parameter values were
similar. At 600 million generations, ESSs were around
400, posterior distribution was unimodal and trace
showed no apparent spike.
In the molecular dating analysis, the log-normal
relaxed clock outperformed a strict clock (BF >20), and
there was no difference between performance of the
tested speciation tree priors (BF <5). A birth–death
speciation prior was chosen as a generalization of the
Yule process that allows lineage speciation as well
as extinction. The molecular dating showed that the
diversification of Sciurini in South America occurred
3.7–6.03 Ma (Supplementary Fig. S5, available on Dryad
at http://dx.doi.org/10.5061/dryad.v2t5n).
The dietary preferences of Sciurini were modeled with
the probit regression where the final model contained
those morphological shape variables expressed as the
RWs that provided the lowest AIC values for the model.
The evaluated food items showed a strong correlation
with several RWs with the exception of feeding on seeds,
which did not significantly correlate with any RWs at
=0.05. Although residual variance remained high for
multiplemodels,themisclassificationratewas8%forthe
FIGURE 4. Ancestral ranges of South American squirrels reconstructed from the pruned chronogram containing South American species and
their current distribution (maps at tips). The putative ancestral area was in the western Amazonia, on the east side of the Andes. Circle shading
reflects Bayesian posterior probability that the ancestor occupied the grid cell.
byguestonDecember9,2015http://sysbio.oxfordjournals.org/Downloadedfrom
Paper 2.3.2 225
1084 SYSTEMATIC BIOLOGY VOL. 64
combined RW data set, meaning that at the individual
level, 8% of animals could not be correctly predicted
to feed on a specific food source based on their skull
morphology (Supplementary Table S10, available on
Dryad at http://dx.doi.org/10.5061/dryad.v2t5n). The
models based on the dorsal and ventral data had a
16% and 13% misclassification rate, respectively. The
application of the majority rule over multiple samples
per species yielded no improvement.
Our set of models most accurately predicted seeds,
fruits, leaves, and bark as dietary preferences, while
affinities for flowers and fungi were erroneously
predicted in 4 out of 24 species (Supplementary
Table S10, available on Dryad at http://dx.doi.org/
10.5061/dryad.v2t5n). Dietary preferences of 18 species
were predicted accurately with zero or one incorrect
prediction.
The NJ tree (Fig. 5) based on the dietary preferences
divided taxa according to differences in the composition
of their diet. In general, the dietary groups did not
reflect association with morphology, as species with
similar skull morphology (according to the UPGMA
analyses) were distributed in different dietary groups,
indicating that the type of food consumed is not
mirrored by skull shape. The only exceptions were
species that feed on insects, bark, and exudates, which
represent the exclusive diet of all Microsciurus species
(Fig. 5).
DISCUSSION
WeshowedthatthephylogenyofSciuriniisdecoupled
from morphological differentiation in the Neotropics.
Where the phylogeny reflected geographic occurrence of
the species, the morphological cluster analyses showed
three groups, which correspond with the current
taxonomy of the tree squirrels.
Changes in the Skull Shape
Geometric morphometry recognized three
morphologicalgroupsofSciurinisquirrels(Microsciurus,
Sciurus, and Tamiasciurus). While species of
Microsciurus and Sciurus form a clade on a phylogeny,
morphologically they represent valid genera. The main
characteristic distinguishing the skulls of Microsciurus
and Sciurus is a shorter, broader, and more convex
shape. The skulls of Microsciurus are characterized
by small orbits with wider interorbital space, short
zygomatic arches, short palatines and short nasals,
more ventral position of foramen magnum. These
findings are consistent with morphological description
of the genus Microsciurus by Allen (1914).
Our results on cranial morphology of Microsciurus
filled the knowledge gap on the morphology of
Microsciurus as only data on postcranial and mandibular
morphology were previously available. The comparison
with other dwarf squirrels with similar morphology and
behavior patterns suggests that the changes of skull
shape in Microsciurus are associated with morphological
convergence.
Morphological Convergence
The sharp conflict between molecular and
morphological data was in the position of Microsciurus.
In the molecular analyses, species of the genus
Microsciurus belonged to different monophyletic
groups and were closely related to species of the
genus Sciurus. However, the morphological data
recognized Microsciurus species as clearly distinct
from Sciurus and Tamiasciurus species. The skulls of
Microsciurus are smaller, broader, and more convex,
with shorter nasals and wider interorbital regions
compared with Sciurus and Tamiasciurus as already
described by Allen (1914). We tested the role of feeding
habits on skull morphology, and the probit model
categorizing diet based on morphology as well as
the comparison of morphological and dietary clusters
showed an important influence of bark gleaning and
insectivory in explaining skull shape in Microsciurus. The
overall cranial morphology of Microsciurus resembles
the cranial morphology of other dwarf squirrels
(Neotropical S. pusillus, Afrotropical M. pumilio, and
Oriental Nannosciurus and Exilisciurus species). The
cranial features in dwarf squirrels are accompanied by
lengthened limbs and other morphological adaptations
for vertical climbing and claw clinging (Thorington
and Thorington 1989; Thorington et al. 1997), along
with deep mandibular bodies providing support to jaw
muscles for a strong incisor bite, needed for bark feeding
(Thorington and Darrow 1996; Velhagen and Roth 1997;
Casanovas-Vilar and van Dam 2013). The morphological
divergence of mandibles in dwarf squirrels is associated
with allometry (Roth 1996; Velhagen and Roth 1997).
However, each group of dwarf squirrels has a different
evolutionary allometric trajectory influenced by, for
example, adaptation (Hautier et al. 2009). Morphological
similarities of Microsciurus with other dwarf squirrels,
together with behavioral observations of tearing off
bark, claw clinging, and active searching for arthropods
in Microsciurus species (Youlatos 1999; Thorington et al.
2012) suggest that the morphological differentiation
from other Sciurini is related to the reduced body size
and adaptations to bark gleaning.
Dietary Preference
Even with relatively simple models, we have been
able to achieve high accuracy in predicting dietary
preferences (84% for models from combined dorsal
and ventral information). However, response was coded
as a binary variable that lacked information on
proportion of dietary preferences and could not weigh
evolutionary pressure on certain skull-shapes. Thus,
although generalist squirrels that change diet depending
on availability could have different skull shape than
byguestonDecember9,2015http://sysbio.oxfordjournals.org/Downloadedfrom
226 Paper 2.3.2
2015 PE ˇCNEROVÁ ET AL.—RECONCILING PHYLOGENY AND MORPHOLOGY IN SQUIRRELS 1085
Glaucomys volans
Microsciurus alfari
Microsciurus flaviventer
Microsciurus mimulus
Sciurus aberti
Sciurus aestuans
Sciurus alleni
Sciurus arizonensis
Sciurus aureogaster
Sciurus carolinensis
Sciurus colliaei
Sciurus deppei
Sciurus granatensis
Sciurus griseus
Sciurus igniventris
Sciurus lis
Sciurus nayaritensis
Sciurus niger
Sciurus oculatus
Sciurus variegatoides
Sciurus vulgaris
Sciurus yucatanensis
Tamiasciurus douglasii
Tamiasciurus hudosnicus
Morphological clusters - dorsal side
Group 3.3
Group 3.1
Group 2
Group 1
Group 3.2
FIGURE 5. NJ tree built from Manhattan distances based on the dietary preference table of Sciurini. Clusters reflect species with
similar diet composition. Tip labels indicate affiliation to morphological clusters (inset per Supplementary Fig. S3, available on Dryad
at http://dx.doi.org/10.5061/dryad.v2t5n). In general, dietary groups do not overlap with morphological clusters, except for species of
Microsciurus.
specialists that continuously eat similar diet, our models
could not express this relationship.
For those reasons, we have kept from
overinterpretation of the dietary preferences models
apart from bark gleaners, as the results for this group
were supported from two additional analyses.
Molecular Phylogeny Conflicts with Taxonomy
Evolutionary history estimated from phylogenies
supported previous results (Mercer and Roth 2003;
Herron et al. 2004; Steppan et al. 2004; Peˇcnerová and
Martínková 2012) suggesting that the genus Tamiasciurus
(lineage I) is sister to all other taxa clustered in a separate
clade (II and III). Compared with earlier research, we
obtained more specific results for the monotypic genus
Rheithrosciurus, which was recognized as a sister lineage
(II) to the other taxa of lineage III (Microsciurus, Sciurus,
and Syntheosciurus), suggesting an early divergence
from a common ancestor with lineage III. Our results
confirm that two crossings of Beringia occurred in the
evolutionary history of Sciurini: an early crossing that
caused divergence of the Tamiasciurus lineage and the
other two lineages, and later an eastbound crossing
after the split of Rheithrosciurus ancestors (Peˇcnerová
and Martínková 2012). Additionally, the monophyly of
lineage III provides additional support for the need
of a taxonomic revision of the tribe Sciurini suggested
already by Mercer and Roth (2003).
Tropical Mountain Forests as Epicenters of Squirrel
Speciation
The colonization of the Neotropics by Sciurini was
accompanied by a split between the Central American
(III-A) and the South American clade (III-B) and
extensive speciation within these clades. Timing of
the diversification of the two clades predates the
full formation of the Isthmus of Panama around 3
Ma. Sciurini were probably early colonizers of South
America,accidentallytransportedduringshoalingofthe
Isthmus of Panama. However, this date is dependent on
the assumption that Sciurus arrived to North America
earlier than 7.4 Ma (Supplementary Figure S5, available
on Dryad at http://dx.doi.org/10.5061/dryad.v2t5n;
Marincovich and Gladenkov 1999; Emry et al. 2005). The
ranges of the majority of Neotropical Sciurini species
are anchored around the eastern slopes of the Andes,
suggesting that this area played a significant role in the
evolutionary history of Sciurini. We used a simulation
approach to reconstruct the ancestral range of South
American tree squirrels and our results revealed an
epicenter of tree squirrel speciation in the western
Amazonia.
The center of diversification in wet tropical mountain
forests demonstrates the importance of this habitat in
the process of speciation. Villalobos (2013) studied the
biogeography of Mesoamerican squirrels and concluded
that a combination of vicariant barriers and dispersal
events generated modern levels of Mesoamerican
squirrel diversity and proposed that vicariant events
byguestonDecember9,2015http://sysbio.oxfordjournals.org/Downloadedfrom
Paper 2.3.2 227
1086 SYSTEMATIC BIOLOGY VOL. 64
might be explained by ice age cycles. During the
Pleistocene ice ages, tropical areas were drier, had less
forest and more grassland (Hewitt 2000). According to
the refuge hypothesis (Haffer 1997), patches of moist
mountain forests served as refugia during the arid
periods of ice ages and forests and forest species
spread from the refugia in interglacials. However,
multiple studies (e.g., Colinvaux et al. 2000; Bush and
de Oliveira 2006) showed evidence that the extent of
spreading savannas was not of a sufficient magnitude
to generate forest refugia. A complex scenario, taking
into consideration both approaches has been suggested
(Turchetto-Zolet et al. 2013). Our data are consistent
with the refuge hypothesis. Moist forest areas of western
Amazonia, estimated as the ancestral area of the
Sciurini distribution, correspond to the long-recognized
Quaternary refugia of Napo and Inambari (Haffer 1969;
Cracraft 1985). These refugia are defined by the Andes to
the west and they provided a stable forest habitat during
ice age cycles (Hewitt 2000). Our findings place tree
squirrels of the tribe Sciurini among other Neotropical
taxa with a Pleistocene origin of biodiversity, including
monkeys (Chiou et al. 2011), birds (Sousa-Neves et al.
2013), butterflies (Garzón-Orduña et al. 2014), and trees
(Richardson et al. 2001).
CONCLUSIONS
The integrated approach of molecular and
morphological data and computational simulations
revealed two key findings in the speciation of Sciurini.
First, the rapid speciation of tree squirrels in South
America was anchored in the area recognized as a
Quaternary refugium in the western Amazonia in
South America. Mountain forests probably served
as areas where squirrels were relegated during the
climatic oscillations in the Pleistocene. Second, the
ecological diversity of tropical forests, in particular
the availability of varied food resources, stimulated
ecological adaptation. While molecular analyses
considered the genus Microsciurus to be polyphyletic,
with its species being nested deep inside the clade
of the genus Sciurus, morphologically, the species of
Microsciurus form a well-recognized cluster. Geometric
morphometrics and analysis of dietary preferences
linked the change in skull shape to bark gleaning
insectivory, indicating that the present recognition of
Microsciurus as a genus is based on morphological
convergence and not on evolutionary history. By using a
set of varied analytical tools, we were able to provide a
picture of the processes leading to the rapid speciation
in the Neotropics. Our study shows that by studying
the evolutionary history of particular model groups we
can gain better insight into the complex origin of one of
the world’s biodiversity hotspots.
SUPPLEMENTARY MATERIAL
Data available from the Dryad Digital Repository:
http://dx.doi.org/10.5061/dryad.v2t5n.
ACKNOWLEDGMENTS
TheauthorswouldliketoacknowledgeOndˇrejMikula
for lending a digital camera to obtain photographs
for morphological analyses and his valuable advice on
geometric morphometrics. They thank Louise Tomsett
from the NHM in London and Darrin Lunde from the
National Museum of Natural History of the Smithsonian
Institution in Washington, DC, for tissue samples and
access to museum collections.
FUNDING
This work was supported by several grants
and scholarships: Program rektora, Masaryk
University [MUNI/C/0772/2011; Short visit grant,
Frontiers of Speciation Research, European Science
Foundation [4650]; PROVAZ: PROpojení Vzdˇelávání
A nových pˇrístup˚u v Zoologicko-ekologickém
výzkumu—od teorie k praxi [in the frame of project
CZ.1.07/2.4.00/17.0138; Scholarship for support of
creative activity, Department of Botany and Zoology,
Masaryk University [30144/2012].
REFERENCES
Adams D.C., Otarola-Castillo E. 2013. geomorph: An R package for
the collection and analysis of geometric morphometric shape data.
Methods Ecol. Evol. 4:393–399.
Allen J.A. 1914. Review of the genus Microsciurus. Bull. Am. Mus. Nat.
Hist. 33:145–165.
Altschul S.F., Gish W., Miller W., Myers E.W., Lipman D.J. 1990. Basic
local alignment search tool. J. Mol. Biol. 215:403–410.
Bergsten J., Nilsson A.N., Ronquist F. 2013. Bayesian tests of topology
hypotheses with an example from diving beetles. Syst. Biol. 62:
660–673.
Black C.C. 1963. A review of the North American Tertiary Sciuridae.
Bull. Mus. Comp. Zool. 130:109–248.
Bollback J.P. 2006. SIMMAP: Stochastic character mapping of discrete
traits on phylogenies. BMC Bioinformatics 7:88.
Bookstein F.L. 1989. Principal warps—thin-plate splines and the
decomposition of deformations. IEEE Trans. PAMI 11:567–585.
Bouckaert R., Heled J., Kühnert D., Vaughan T.G., Wu C.-H., Xie
D., Suchard M.A., Rambaut A., Drummond A.J. 2014. BEAST2:
A software platform for Bayesian evolutionary analysis. PLoS
Comput. Biol. 10:e1003537.
Bush M.B., de Oliveira P.E. 2006. The rise and fall of the refugial
hypothesis of Amazonian speciation: A paleoecological perspective.
Biota Neotrop. 6:1–17.
Cardini A. 2014. Missing the third dimension in geometric
morphometrics: How to assess if 2D images really are a good proxy
for 3D structures? Hystrix 25:73–81.
Cardini A., Hoffmann R.S., Thorington R.W. Jr. 2005. Morphological
evolution in marmots (Rodentia, Sciuridae): Size and shape of the
dorsal and lateral surfaces of the cranium. J. Zool. Syst. Evol. Res.
43:258–268.
Cardini A., O’Higgins P. 2004. Patterns of morphological evolution
in Marmota (Rodentia, Sciuridae): Geometric morphometrics of
the cranium in the context of marmot phylogeny, ecology and
conservation. Biol. J. Linn. Soc. 82:385–407.
Casanovas-Vilar I., van Dam J. 2013. Conservatism and adaptability
during squirrel radiation: What is mandible shape telling us? PLoS
One 8:e61298.
Chifman J., Kubatko L. 2014. Quartet inference from SNP data under
the coalescent model. Bioinformatics 30:3317–3324.
byguestonDecember9,2015http://sysbio.oxfordjournals.org/Downloadedfrom
228 Paper 2.3.2
2015 PE ˇCNEROVÁ ET AL.—RECONCILING PHYLOGENY AND MORPHOLOGY IN SQUIRRELS 1087
Chiou K.L., Pozzi L., Alfaro J.W.L., Di Fiore A. 2011. Pleistocene
diversification of living squirrel monkeys (Saimiri spp.) inferred
from complete mitochondrial genome sequences. Mol. Phylogenet.
Evol. 59:736–745.
Colinvaux P.A., De Oliveira P.E., Bush M.B. 2000. Amazonian and
neotropical plant communities on glacial time-scales: The failure
of the aridity and refuge hypotheses. Quat. Sci. Rev. 19:141–169.
Cracraft J. 1985. Historical biogeography and pattern of differentiation
within the South American avifauna: Areas of endemism. Ornithol.
Monogr. 36:49–84.
Emry R.J., Korth W.W., Bell M.A. 2005. A tree squirrel (Rodentia,
Sciuridae, Sciurini) from the late Miocene (Clarendonian) of
Nevada. J. Vertebr. Paleontol. 25:228–235.
Ersts P.J. 2014. Geographic distance matrix generator. American
Museum of Natural History, Center for Biodiversity and
Conservation. Available from: URL http://biodiversityinformatics.
amnh.org/open_source/gdmg, last accessed 14 August, 2015.
Fadda C., Corti M. 2000. Three dimensional morphometric study of the
Ethiopian Myomys-Stenocephalemys complex (Murinae, Rodentia).
Hystrix 10:131–143.
Garzón-Orduña I.J., Benetti-Longhini J.E., Brower A.V.Z. 2014. Timing
the diversification of the Amazonian biota: Butterfly divergences
are consistent with Pleistocene refugia. J. Biogeogr. 41:1631–1638.
Goujon M., McWilliam H., Li W., Valentin F., Squizzato S., Paern J.,
Lopez R. 2010. A new bioinformatics analysis tools framework at
EMBL–EBI. Nucleic Acids Res. 38:W695–W699.
Gower J.C. 1975. Generalized procrustes analysis. Psychometrika
40:33–51.
Haffer J. 1969. Speciation in Amazonian forest birds. Science 165:
131–137.
Haffer J. 1997. Alternative models of vertebrate speciation in Amazonia:
An overview. Biodivers. Conserv. 6:451–476.
Hall T.A. 1999. BioEdit: A user-friendly biological sequence alignment
editor and analysis program for Windows 95/98/NT. Nucleic Acids
Symp. Ser. 41:95–98.
Hautier L., Fabre P.-H., Michaux J. 2009. Mandible shape and dwarfism
in squirrels (Mammalia, Rodentia): Interaction of allometry and
adaptation. Naturwissenschaften 96:725–730.
Heled J., Drummond A.J. 2010. Bayesian inference of species trees from
multilocus data. Mol. Biol. Evol. 27:570–580.
Herron M.D., Castoe T.A., Parkinson C.L. 2004. Sciurid phylogeny and
the paraphyly of Holarctic ground squirrels (Spermophilus). Mol.
Phylogenet. Evol. 31:1015–1030.
Hewitt G. 2000. The genetic legacy of the Quaternary ice ages. Nature
405:907–913.
Hooghiemstra H., van der Hammen T. 1998. Neogene and Quaternary
development of the neotropical rain forest: The forest refugia
hypothesis, and a literature overview. Earth-Sci. Rev. 44:147–183.
Hotelling H. 1933. Analysis of a complex of statistical variables into
principal components. J. Educ. Psychol. 24:417–441.
Huelsenbeck J.P., Ronquist F. 2001. MRBAYES: Bayesian inference of
phylogenetic trees. Bioinformatics 17:754–755.
IUCN. 2013. IUCN Red List of Threatened Species. Available from:
URL www.iucnredlist.org (downloaded 19 January 2014).
Katoh K., Standley D.M. 2013. MAFFT multiple sequence alignment
software version 7: Improvements in performance and usability.
Mol. Biol. Evol. 30:772–780.
Keigwin L. 1982. Isotopic paleoceanography of the Caribbean and East
Pacific: Role of Panama uplift in late Neogene time. Science 217:
350–353.
Landis M.J., Matzke N.J., Moore B.R., Huelsenbeck J.P. 2013. Bayesian
analysis of biogeography when the number of areas is large. Syst.
Biol. 62:789–804.
Lanfear R., Calcott B., Ho S.Y.W., Guindon S. 2012. PartitionFinder:
Combined selection of partitioning schemes and substitution
models for phylogenetic analyses. Mol. Biol. Evol. 29:1695–1701.
Larkin M.A., Blackshields G., Brown N.P., Chenna R., McGettigan
P.A., McWilliam H., Valentin F., Wallace I.M., Wilm A., Lopez R.,
ThompsonJ.D.,GissonT.J.,HigginsD.G.2007.ClustalWandClustal
X version 2.0. Bioinformatics 23:2947–2948.
Macholán M., Mikula O., Vohralík V. 2008. Geographic phenetic
variation of two eastern-Mediterranean non-commensal mouse
species, Mus macedonicus and M. cypriacus (Rodentia: Muridae)
based on traditional and geometric approaches to morphometrics.
Zool. Anz. 247:67–80.
Marincovich L. Jr., Gladenkov A.Yu. 1999. Evidence for an early
opening of the Bering Strait. Nature 397:149–151.
MartínkováN.,SearleJ.B.2006.AmplificationsuccessrateofDNAfrom
museum skin collections: A case study of stoats from 18 museums.
Mol. Ecol. Notes 6:1014–1017.
McFerrin L. 2013. HDMD: Statistical analysis tools for high dimension
molecular data (HDMD). R package version 1.2. Available from:
URL http://CRAN.R-project.org/package=HDMD, last accessed
14 August, 2015.
Mercer J.M., Roth V.L. 2003. The effects of Cenozoic global change on
squirrel phylogeny. Science 299:1568–1572.
Mittelbach G.G., Schemske D.W., Cornell H.V., Allen A.P., Brown
J.M., Bush M.B., Harrison S.P., Hurlbert A.H., Knowlton N., Lessios
H.A., McCain C.M., McCune A.R., McDade L.A., McPeek M.A.,
Near T.J., Price T.D., Ricklefs R.E., Roy K., Sax D.F., Schluter D.,
Sobel J.M., Turelli M 2007. Evolution and the latitudinal diversity
gradient: Speciation, extinction and biogeography. Ecol. Lett.
10:315–331.
Moore J.C. 1959. Relationships among the living squirrels of the
Sciurinae. Bull. Am. Mus. Nat. Hist. 118:153–206.
Oksanen J., Blanchet F.G., Kindt R., Legendre P., Minchin P.R., O’Hara
R.B., Simpson G.L., Solymos P., Stevens M.H.H., Wagner H. 2013.
vegan: Community ecology package. R package version 2.0-7.
Available from: URL http://CRAN.R-project.org/package=vegan,
last accessed 14 August, 2015.
Oshida T., Arslan A., Noda M. 2009. Phylogenetic relationships among
the Old World Sciurus squirrels. Folia Zool. 58:14–25.
Paradis E., Claude J., Strimmer K. 2004. APE: Analyses of phylogenetics
and evolution in R language. Bioinformatics 20:289–290.
Pearson K. 1901. On lines and planes of closest fit to systems of points
in space. Phil. Mag. 2:559–572.
Peˇcnerová P., Martínková N. 2012. Evolutionary history of tree squirrels
(Rodentia, Sciurini) based on multilocus phylogeny reconstruction.
Zool. Scr. 41:211–219.
Pianka E.R. 1966. Latitudinal gradients in species diversity—a review
of concepts. Am. Nat. 100:33–46.
PääboS.,PoinarH.,SerreD.,Jaenicke-DesprésV.,HeblerJ.,RohlandN.,
Kuch M., Krause J., Vigilant L., Hofreiter M. 2004. Genetic analyses
from ancient DNA. Annu. Rev. Genet. 38:645–679.
R Core Team. 2013. R: A language and environment for statistical
computing. Vienna, Austria: R Foundation for Statistical
Computing. Available from: URL http://www.R-project.org/,
last accessed 14 August, 2015.
Rambaut A., Suchard M.A., Xie D., Drummond A.J. 2013. Tracer
v1.5. Available from: URL http://beast.bio.ed.ac.uk/Tracer, last
accessed 14 August, 2015.
Ranwez V., Criscuolo A., Douzery E.J. 2010. SuperTriplets: A
triplet-based supertree approach to phylogenomics. Bioinformatics
26:i115–i123.
Richardson J.E., Pennington R.T., Pennington T.D., Hollingsworth P.M.
2001. Rapid diversification of a species-rich genus of neotropical rain
forest trees. Science 293:2242–2245.
Rohlf F.J. 1993. Relative warp analysis and example of its application
to mosquito wings. In: Marcus LF, Bello E, Garcia-Valdecasas A,
editors. Contribution to morphometrics. Madrid: Museo Nacional
de Ciencias Naturales. p. 131–159.
Rohlf F.J. 1998. On applications of geometric morphometrics to studies
of ontogeny and phylogeny. Syst. Biol. 47:147–158.
Rohlf F.J. 2003. tpsSmall. New York: Department of Ecology and
Evolution, State University of New York at Stony Brook. Available
from: URL http://life.bio.sunysb.edu/morph/index.html.
Rohlf F.J. 2005. tpsDig2. New York: Department of Ecology and
Evolution, State University of New York at Stony Brook. Available
from: URL http://life.bio.sunysb.edu/morph/index.html, last
accessed 14 August, 2015.
Rohlf F.J. 2010. tpsRelw. New York: Department of Ecology and
Evolution, State University of New York at Stony Brook. Available
from: URL http://life.bio.sunysb.edu/morph/index.html, last
accessed 14 August, 2015.
Rohlf F.J., Slice D. 1990. Extensions of the procrustes method for the
optimal superimposition of landmarks. Syst. Zool. 39:40–59.
byguestonDecember9,2015http://sysbio.oxfordjournals.org/Downloadedfrom
Paper 2.3.2 229
1088 SYSTEMATIC BIOLOGY VOL. 64
Roth V.L. 1996. Cranial integration in the Sciuridae. Amer. Zool. 36:
14–23.
Roth V.L., Mercer J.M. 2008. Differing rates of macroevolutionary
diversification in arboreal squirrels. Curr. Sci. 95:857–861.
Rull V. 2008. Speciation timing and neotropical biodiversity: The
Tertiary-Quaternary debate in the light of molecular phylogenetic
evidence. Mol. Ecol. 17:2722–2729.
Sanderson M.J., McMahon M.M., Steel M. 2011. Terraces in
phylogenetic tree space. Science 333:448–450.
Schluter D. 2001. Ecology and the origin of species. Trends Ecol. Evol.
16:372–380.
Scornavacca C., Berry V., Lefort V., Douzery E.J., Ranwez V. 2008.
PhySIC_IST: Cleaning source trees to infer more informative
supertrees. BMC Bioinformatics 9:413.
Sousa-Neves T., Aleixo A., Sequeira F. 2013. Cryptic patterns
of diversification of a widespread Amazonian Woodcreeper
species complex (Aves: Dendrocolaptidae) inferred from multilocus
phylogenetic analysis: Implications for historical biogeography and
taxonomy. Mol. Phylogenet. Evol. 68:410–424.
Stamatakis A. 2014. RAxML version 8: A tool for phylogenetic analysis
and post-analysis of large phylogenies. Bioinformatics 30:1312–1313.
Steele M.A. 1998. Tamiasciurus hudsonicus. Mammal. Species 586:1–9.
Steele M.A. 1999. Tamiasciurus douglasii. Mammal. Species 630:1–8.
Steppan S.J., Storz B.L., Hoffmann R.S. 2004. Nuclear DNA phylogeny
of the squirrels (Mammalia: Rodentia) and the evolution of
arboreality from c-myc and RAG1. Mol. Phylogenet. Evol. 30:
703–719.
Suchard M.A., Weiss R.E., Sinsheimer J.S. 2005. Models for estimating
Bayes factors with applications to phylogeny and tests of
monophyly. Biometrics 61:665–673.
Swofford D. 2015. PAUP 4.0a142. Available from: URL http://
people.sc.fsu.edu/∼dswofford/paup_test/ (24 February 2015).
Thorington R.W. Jr., Darrow K. 1996. Jaw muscles of old world
squirrels. J. Morphol. 230:145–165.
Thorington R.W. Jr., Darrow K., Betts A.D.K. 1997. Comparative
myology of the forelimb of squirrels (Sciuridae). J. Morphol.
234:155–182.
Thorington R.W. Jr., Koprowski J.L., Steele M.A., Whatton J.F. 2012.
Squirrels of the World. Baltimore: The Johns Hopkins University
Press.
Thorington R.W. Jr., Thorington E.M. 1989. Postcranial
proportion of Microsciurus and Sciurillus, the American pygmy
squirrels. In: Redfor KH, Eisenberg JF, editors. Advances in
neotropical mammalogy. Gainesville, FL: Sandhill Crane Press,
p. 125–136.
Turchetto-Zolet A.C., Pinheiro F., Salgueiro F., Palma-Silva C. 2013.
Phylogeographical patterns shed light on evolutionary process in
South America. Mol. Ecol. 22:1193–1213.
Vaidya G., Lohman D.J., Meier R. 2011. SequenceMatrix:
Concatenation software for the fast assembly of multi-gene
datasets with character set and codon information. Cladistics 27:
171–180.
Velhagen W.A., Roth V.L. 1997 Scaling of the mandible in squirrels. J.
Morphol. 232:107–132.
Via S. 2001. Sympatric speciation in animals: The ugly duckling grows
up. Trends Ecol. Evol. 16:381–390.
Villalobos F. 2013. Tree squirrels: A key to understand the historic
biogeography of Mesoamerica? Mamm. Biol. 78:258–266.
Wilson D.E., Reeder D.M. 2005. Mammal species of the world. A
taxonomic and geographic reference. Baltimore: The Johns Hopkins
University Press.
Wing S.L. 1998. Tertiary vegetational history of North America as a
context for mammalian evolution. In: Janis C, Jacobs L, Scott K,
editors. Evolution of tertiary mammals of North America: Vol.
I: Terrestrial carnivores, ungulates, and ungulate-like mammals.
Cambridge: Cambridge University Press. p. 37–65.
Wolfe J.A., Hopkins D.M., Leopold E.B. 1966. Tertiary stratigraphy
and paleobotany of the Cook Inlet region. Alaska: Discussion of
stratigraphic significance of fossil plants from the Chickaloon,
Kenai, and Tsadaka formations. U.S. Geological Survey Professional
Paper 398-A.
Youlatos D. 1999. Locomotor and postural behavior of Sciurus
igniventris and Microsciurus flaviventer (Rodentia, Sciuridae) in
eastern Ecuador. Mammalia 63:405–416.
byguestonDecember9,2015http://sysbio.oxfordjournals.org/Downloadedfrom
230 Paper 2.3.2
Paper 2.4.1
Zukal J., Bandouchova H., Bartonicka T., Berkova H., Brack V., Brichta J., Dolinay M.,
Jaron K. S., Kovacova V., Kovarik M., Martínková N., Ondracek K., Rehak Z., Turner
G. G., Pikula J. 2014. White-nose syndrome fungus: a generalist pathogen of hibernating
bats. PLoS ONE 9: e97224.
– 231 –
White-Nose Syndrome Fungus: A Generalist Pathogen of
Hibernating Bats
Jan Zukal1,2
, Hana Bandouchova3
, Tomas Bartonicka2
, Hana Berkova1
, Virgil Brack4
, Jiri Brichta3
,
Matej Dolinay2
, Kamil S. Jaron5
, Veronika Kovacova3
, Miroslav Kovarik6
, Nata´lia Martı´nkova´1,5
,
Karel Ondracek3
, Zdenek Rehak2
, Gregory G. Turner7
, Jiri Pikula3
*
1 Institute of Vertebrate Biology, Academy of Sciences of the Czech Republic, Brno, Czech Republic, 2 Department of Botany and Zoology, Masaryk University, Brno, Czech
Republic, 3 Department of Ecology and Diseases of Game, Fish and Bees, University of Veterinary and Pharmaceutical Sciences, Brno, Czech Republic, 4 Environmental
Solutions & Innovations Inc., Cincinnati, Ohio, United States of America, 5 Institute of Biostatistics and Analysis, Masaryk University, Brno, Czech Republic, 6 Administration
of the Moravian Karst Protected Landscape Area, Nature Conservation Agency of the Czech Republic, Blansko, Czech Republic, 7 Pennsylvania Game Commission,
Harrisburg, Pennsylvania, United States of America
Abstract
Host traits and phylogeny can determine infection risk by driving pathogen transmission and its ability to infect new hosts.
Predicting such risks is critical when designing disease mitigation strategies, and especially as regards wildlife, where
intensive management is often advocated or prevented by economic and/or practical reasons. We investigated
Pseudogymnoascus [Geomyces] destructans infection, the cause of white-nose syndrome (WNS), in relation to chiropteran
ecology, behaviour and phylogenetics. While this fungus has caused devastating declines in North American bat
populations, there have been no apparent population changes attributable to the disease in Europe. We screened 276 bats
of 15 species from hibernacula in the Czech Republic over 2012 and 2013, and provided histopathological evidence for 11
European species positive for WNS. With the exception of Myotis myotis, the other ten species are all new reports for WNS in
Europe. Of these, M. emarginatus, Eptesicus nilssonii, Rhinolophus hipposideros, Barbastella barbastellus and Plecotus auritus
are new to the list of P. destructans-infected bat species. While the infected species are all statistically phylogenetically
related, WNS affects bats from two suborders. These are ecologically diverse and adopt a wide range of hibernating
strategies. Occurrence of WNS in distantly related bat species with diverse ecology suggests that the pathogen may be a
generalist and that all bats hibernating within the distribution range of P. destructans may be at risk of infection.
Citation: Zukal J, Bandouchova H, Bartonicka T, Berkova H, Brack V, et al. (2014) White-Nose Syndrome Fungus: A Generalist Pathogen of Hibernating Bats. PLoS
ONE 9(5): e97224. doi:10.1371/journal.pone.0097224
Editor: Justin G. Boyles, Southern Illinois University, United States of America
Received January 5, 2014; Accepted April 16, 2014; Published May 12, 2014
This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for
any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.
Funding: This study was supported by the Grant Agency of the Czech Republic (Project No. P506/12/1064) and by the National Speleological Society of the USA.
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing Interests: V.B. is currently affiliated with a commercial company Environmental Solutions & Innovations Inc. He provided expertise in bat ecology in
this paper. This does not alter the authors’ adherence to PLOS ONE policies on sharing data and materials.
* E-mail: pikulaj@vfu.cz
Introduction
Host-pathogen dynamics represent a balance between the
pathogen’s ability to infect and the host’s ability to resist, with
an intensive arms race between the two reflected in coevolutionary
adaptations. With host switching, pathogens temporarily
escape the arms race. New, naive hosts may show lower
resistance and other characteristics favourable to the pathogen.
Overlapping distribution of a pathogen and its potential host(s) is
key to host switching driven by opportunity [1]. The spread of
emerging wildlife pathogens may have economic consequences,
even in species indirectly linked to humans [2]. Fungal infections
in amphibians and bats that result in population declines [3], for
example, can lead to increased agricultural costs where humans
chemically compensate for ecosystem services provided by these
organisms in terms of insect control.
White-nose syndrome (WNS) is an emerging disease of
hibernating bats associated with skin infection by Pseudogymnoascus
[Geomyces] destructans, a recently recognised fungal pathogen [4–7].
Severe skin damage results in disruption of torpor pattern,
premature depletion of fat reserves and mortality in affected bats
in North America [8]. High mortality rates at affected localities
and the rapid spread of the infection since 2006 continues to
threaten bat diversity [9–11].
While WNS has characteristics of an epizootic, gradually
expanding through North American hibernacula from its original
detection site [12,13], P. destructans is pan-European in distribution
[14,15]. Aside from seasonality in the appearance of white fungal
growth [15], detailed spatio-temporal data for P. destructans
infection in Europe are lacking. Fortunately, mass mortality has
not been observed in European bats to date [14].
WNS can be transmitted either directly through bat-to-bat
contact or indirectly through contact with pathogen propagules in
the environment [6,16] and the infection’s spread is assumed to be
both density- and frequency-dependent [10]. Multiple factors,
such as hibernation in large assemblages or length of hibernation
season, play a role in the epidemiology of this fungal disease and
ecological and behavioural characteristics of bat species may affect
the risk of infection [3]. Traits such as selection of hibernaculum
roost sites with differing microclimatic conditions and solitary
PLOS ONE | www.plosone.org 1 May 2014 | Volume 9 | Issue 5 | e97224
232 Paper 2.4.1
versus gregarious hibernation behaviour may also influence the
impact of WNS [10]. Other risk factors associated with
hibernating bat mortality in North America include distance from
the first WNS-affected site, cluster size, species diversity and
composition and type of hibernaculum [13].
Prior to 2012, bat species positive for P. destructans in North
America included Myotis austroriparius, M. grisescens, M. leibii, M.
lucifugus, M. septentrionalis, M. sodalis, M. velifer, Perimyotis subflavus and
Eptesicus fuscus [11,16,17]. Dermatohistopathology has revealed
fungal infection with cupping erosions and skin invasion diagnostic
for WNS in a M. myotis specimen hibernating in the Moravian
Karst, Czech Republic [18] and eight species (M. myotis, M.
oxygnathus/M. blythii, M. brandtii, M. daubentonii, M. dasycneme, M.
mystacinus, M. nattereri and M. bechsteinii) have been reported positive
for the WNS fungus in Europe based on direct microscopy of
characteristic P. destructans conidia, fungal culture and genetic
analysis [14,15,19–22]. Photographic evidence of fungal growth
suggests that M. emarginatus, E. nilssonii and Rhinolophus hipposideros
may also prove positive for P. destructans [14].
In summary, a total of 17 vespertilionid bat species had been
reported positive for the WNS fungus in North America and
Europe prior to 2012 and, as the epizootic spreads through North
America and surveillance continues in Europe, it is expected that
the number of infected species will increase. Hereinafter, the term
P. destructans-infected or -positive relates to those species for which
the fungal pathogen has been confirmed by laboratory methods
such as fungal culture and genetic analysis. WNS-positive
represents those species where the infection has been diagnosed
through characteristic histopathological lesions, such as fungal
hyphae densely packed in so-called cupping erosions and/or
invasion of the dermis [23].
Little is known about P. destructans infection in European bat
species less abundant or less commonly observed in hibernacula.
Knowledge of pathological effects associated with the WNS fungus
in European bat species is even poorer. While it is a commonly
held view that European bats are more resistant or resilient than
those in North America [17], our monitoring revealed three
further species positive for P. destructans infection in 2012.
Differences in their hibernation behaviour and taxonomy inspired
us to examine the ecological and behavioural traits and phylogeny
of European and North American species reported positive for P.
destructans in order to identify any similarities in behaviour and
habitat use and to identify any other species that may be at risk.
First, we examined the hypothesis that more bat species are
positive for WNS in Europe than currently reported via
histopathology, considered as the ‘gold standard’ for diagnosing
WNS [23,24]. Second, we hypothesised that ecology, behaviour
and phylogenetic relationships of hibernating bat species influence
risk of infection by P. destructans. Aside from ecological similarities,
those species most often found positive for P. destructans and WNS
belong to the genus Myotis, indicating that phylogenetic relatedness
of hosts may facilitate invasion by the fungus. To test this, we
compared the ecological and behavioural traits of hibernating bats
from Europe and North America. We grouped species with similar
behaviour and habitat use and used confirmed positive species to
propose possible additional species susceptible to infection. We
constructed a phylogeny of vespertilionids and rhinolophids from
Europe and North America, to test the hypothesis that infected
bats are phylogenetically closely related. Finally, we screened
species of unknown infection status in Czech hibernacula to test
the validity of our models and predictions.
Here, we provide histopathological evidence of multiple
European species positive for WNS. We found that infected bat
species are ecologically diverse, utilising a range of hibernating and
feeding strategies. Although bat species previously described as
being P. destructans-positive have been phylogenetically related, the
pattern begins to break down with the newly diagnosed taxa; the
data presented herein demonstrating that the host range for this
fungal pathogen is more diverse than previously realized.
Materials and Methods
Ethics statement
The Czech Academy of Sciences’ Ethics Committee has
reviewed and approved Animal Use Protocol No. 169/2011 in
compliance with Law No. 312/2008 on Protection of Animals
against Cruelty, as adopted by the Parliament of the Czech
Republic. Bats were monitored for WNS and presence of the
causative agent P. destructans in the spring of 2012 and 2013 in
caves of the Moravian Karst, mines near Mala Moravka in the
Jeseniky Mountains, and in the Podyji National Park, all in the
Czech Republic. Non-lethal sampling was in compliance with Law
No. 114/1992 on Nature and Landscape Protection and was
based on permits 01662/MK/2012S/00775/MK/2012, 866/JS/
2012 and 00356/KK/2008/AOPK issued by the Nature Conservation
Agency of the Czech Republic. Bats were handled so as
to minimise stress and duration of sampling procedures between
capture and release. Numbered aluminium rings were attached
around the forearm for long-term identification prior to release at
the site.
Screening bat species in Czech hibernacula for P.
destructans infection
When screening bats for WNS and P. destructans we 1. captured
bats emerging from hibernacula at the end of the hibernation
season using mist nets, 2. swabbed the wing membrane for fungal
culture using the Fungi-Quick transport system (Copan Innovation,
Italy), 3. briefly illuminated the bats with a flashlight to detect
any visible fungal growth, 4. trans-illuminated the wing membrane
using ultraviolet light (UV; wavelength 368 nm) to detect any
WNS lesions, 5. photographed wing membranes of each bat under
both visible and UV light, 6. took a wing punch biopsy from all
WNS-suspected skin lesions (i.e. areas of orange-yellow fluorescence)
using a sterile and disposable 4 mm skin biopsy punch
(Kruuse, Denmark), 7. used polymerase chain reaction (PCR) to
confirm P. destructans from fungal cultures or skin swabs using the
FLOQSwabs system (CopanFlock Technologies, Italy), and 8.
undertook complete histopathological examinations of skin sam-
ples.
A total of 276 bats were screened for WNS and P. destructans and
123 skin biopsies were taken for histological examination from 15
bat species (Table 1).
Formalin-fixed punch biopsy samples were embedded in
paraffin and serial 5 mm tissue sections were prepared and stained
for fungi with periodic acid–Schiff stain. Histopathological findings
were classified as WNS based on previously described diagnostic
criteria [23]. Samples collected to cultivate fungi were transferred
onto Petri dishes containing Sabouraud agar, sealed with parafilm,
inverted and incubated in the dark at 10 uC. Pure fungal cultures
were established from fungal growth developing at 14 days or
later. Pseudogymnoascus destructans was confirmed through characteristic
asymmetrically curved conidia via direct microscopy [5].
Fungal isolates or skin swabs were further identified using PCR
and follow-up sequencing of amplicons [25] and real-time PCR
[24]. A pure culture of P. destructans isolate grown at 10 uC on
Sabouraud agar and genetically confirmed (EMBL-Bank accession
number: HE588133; [18]) served as a PCR control.
Bat Taxa at Infection Risk from WNS
PLOS ONE | www.plosone.org 2 May 2014 | Volume 9 | Issue 5 | e97224
Paper 2.4.1 233
Analysis of European and North American bat ecological
traits
The list of European and North American bat species was
prepared according to Simmons [26]; those species included in the
study being those with complete or partial distribution in
continental Europe or North America for which data are
available. In total, we reviewed ecological and behavioural
variables for 87 species. Of these, 47 were assessed for all variables
and were subjected to ecological modelling analysis.
Eleven traits were chosen to describe bat species: 1. Infection
status (P. destructans-positive/P. destructans-negative), 2. Cave or
non-cave hibernation, 3. Region (Palearctic/Nearctic distribution),
4. Clustering during hibernation (clustering/non-clustering; i.e.
hibernating in groups where multiple individuals touch), 5.
Temperature preference (thermophilic/cryophilic; mainly according
to Webb et al. [27]), 6. Preferred roost type during hibernation
(exposed/hidden/both), 7. Size of clusters during hibernation (no
clustering/small clusters , 50 bats/medium clusters of 51 to 500
bats/large clusters . 500 bats), 8. Distribution range (very large
area/large area of approximately half a continent/moderate size/
small area/very small area; according to Hora´cˇek et al. [28]), 9.
Food (dominant insect food group represented by Diptera/
Lepidoptera/Coleoptera/generalists), 10. Foraging habitat
(open/edge/closed), 11. Body size (small - up to 5 g/medium - 5
to10 g/large - over 10 g).
These traits, which were assessed based on a primary literature
review and expert evaluation, were chosen as those most likely to
influence susceptibility of bats to WNS, the spread of P. destructans
infection or survival rate during hibernation [10]. As the disease
can inflict long-term damage to affected bats surviving the
hibernation season, other medically relevant factors may also
influence survival and reproduction in the active season [29].
Categorical variables were coded as n – 1 binary, dummy
variables, where n is the number of categories. Data for the traits of
each bat species are provided in Table S1. Grouping of
ecologically similar species was performed via neighbour-joining
clustering of squared Euclidean distances using ape in R language
[30,31].
Phylogenetic reconstruction of European and North
American bats
The phylogeny of bats from Europe and North America was
extracted from a maximum likelihood phylogenetic tree using
Phylocom version 4.2 [32]. The complete phylogeny of the
Vespertilionidae, Miniopteridae and Cistugidae families (from
which the tree used here was pruned) was reconstructed from a
concatenated DNA sequence matrix of 13 mitochondrial and
nuclear genes with 64% of missing data (Table S2). Three
Rhinolophus species were used as an outgroup. Phylogeny was
reconstructed using the partitioning scheme suggested by the
greedy algorithm, utilising the Bayesian Information Criterion
assessment in PartitionFinder [33]. The tree space was searched
using maximum likelihood analysis with automatic majority-rule
bootstopping option [34]. By extracting the target species’
phylogeny from a comprehensive tree, we were able to obtain a
phylogeny that exploits currently available diversity to optimise
relationships and branch lengths, and thus mitigate possible
analysis artefacts.
Statistical analysis and hypothesis testing of P.
destructans infection occurrence
We explored the distribution pattern of P. destructans infection on
a tree based on bat trait variation and molecular phylogeny.
Occurrence of P. destructans infection represents a presence/
absence variable, rather than a continuous trait, meaning that it is
suitable for community structure analysis. The phylogenetic signal
for explanatory variables and for P. destructans infection was
calculated in Phylocom using the comstruct function. In order to
Table 1. Bats examined for white-nose syndrome and Pseudogymnoascus destructans infection in Czech hibernacula (Europe).
Species Screened Biopsied Histo+ WNS prevalence (%) St. error
Myotis myotisa
67 56 37 55.22 6.08
Myotis daubentoniib
25 13 4 16.00 5.76
Myotis bechsteiniib
21 7 2 9.52 6.78
Myotis nattererib
20 8 3 15.00 7.10
Myotis brandtiib
17 1 1 5.88 8.23
Myotis alcathoeb
8 7 0 0 15.94
Myotis emarginatusb
39 7 5 12.82 5.35
Rhinolophus hipposiderosb
28 5 1 3.57 5.18
Eptesicus nilssoniib
4 1 1 25.00 26.89
Plecotus auritusc
23 11 5 21.73 8.60
Barbastella barbastellusc
17 3 3 17.64 8.24
Plecotus austriacus 3 1 0 0 32.22
Eptesicus serotinus 2 1 0 0 39.61
Pipistrellus pipistrellus 1 1 0 0 48.47
Myotis dasycnemeb
1 1 1 100 48.47
Total 276 123 63 22.82 2.53
a
= species reported positive for WNS fungus prior to 2012, b
= species recognised as positive in 2012, c
= bat species recognised as positive in 2013.
Screened = numbers of bats captured and examined using UV light trans-illumination to detect WNS lesions, biopsied = numbers of bats biopsied due to WNSsuspected
skin lesions viewed under UV light, histo+ = specimens positive for WNS diagnostic features under histopathological examination (i.e. cupping erosions and
fungal invasion of dermis), WNS prevalence = percentage of bats positive on histopathology out of the total number screened.
doi:10.1371/journal.pone.0097224.t001
Bat Taxa at Infection Risk from WNS
PLOS ONE | www.plosone.org 3 May 2014 | Volume 9 | Issue 5 | e97224
234 Paper 2.4.1
assess relatedness of species that share a specific characteristic,
mean phylogenetic distance (MPD) and mean nearest phylogenetic
taxon distance (MNTD) were compared to the null model,
which assumes random dispersal of the trait on the tree. MPD
measures the mean branch length between two randomly selected
taxa from a sample, and is calculated as the sum of branch lengths
to the node representing their most recent common ancestor.
MNTD is the mean branch length between a taxon within the
sample and its nearest relative. The null model randomised
samples across phylogeny in 9,999 replicates. The distribution of
the trait on a tree is clustered if values of MPD and MNTD
obtained are higher than 95% of values obtained from the null
samples standardised by the standard deviation of the null
samples. The comparison is expressed as net relatedness index
(NRI) and nearest taxon index (NTI) greater than zero [32].
Clustered distribution of a trait or phylogenetic signal means that
species that share the trait are more closely related to one another
than to a random taxon sampled from the tree.
Species’ ecological and behavioural characteristics often show a
heritable component such that close relatives have similar traits
[35]. Such characteristics might then be adaptive and their
evolution further decoupled from the assumption of sample
independence needed for general statistical approaches. The
evolutionary relationships of traits in our dataset were removed
from comparisons by using phylogenetic generalised least squares
(PGLS) in the Caper package of R [36]. We used a variancecovariance
matrix calculated from the phylogeny with branch
lengths transformed according to the Ornstein-Uhlenbeck model
in geiger [37]. The PGLS model was developed via a step-down
procedure, using the Akaike Information Criterion (AIC) to
compare alternative models.
Results
Screening species with unknown infection status in
Czech hibernacula
We tested a broad diversity of European hibernating bats for P.
destructans infection and skin lesions pathognomonic for WNS.
Analysis of 123 skin biopsy samples collected in 2012 and 2013
revealed histopathological findings matching criteria used for
diagnosis of WNS in 63 bats (22.82% prevalence; Table 1) of 11
species, i.e. M. myotis, M. daubentonii, M. bechsteinii, M. nattereri, M.
brandtii, M. emarginatus, M. dasycneme, E. nilssonii, R. hipposideros, B.
barbastellus and P. auritus (Figure 1). With the exception of M. myotis,
the other ten species are all new reports of WNS in Europe. Of
these, M. emarginatus, E. nilssonii, R. hipposideros, B. barbastellus and P.
auritus are new to the list of P. destructans-infected bat species.
Fungus isolates or skin swabs from histopathologically positive bats
were identified as P. destructans using PCR.
Risk of P. destructans infection in bat species of unknown
infection status
Testing the hypothesis of phylogenetic relatedness. P.
destructans-infected species were clustered together by molecular
phylogeny (MPD = 0.212, NRI = 2.913, p , 0.001), meaning
that pairs of infected species were, on average, more closely related
than random species pairs from Europe and North America.
When sister species or nearest relatives were considered, however,
our results indicated that infection of both, either or neither was
random (MNTD = 0.111, NTI = 1.556, p = 0.06; Table 2,
Figure 2). Nine explanatory variables showed NRI and/or NTI
not equal to zero, indicating that phylogenetic comparative
methods were needed due to relatedness of taxa with shared traits
(Table 2).
Figure 1. Histopathological skin lesions consistent with whitenose
syndrome in ten European bat species. (A) Myotis
emarginatus, (B) Eptesicus nilssonii, (C) Rhinolophus hipposideros, (D)
Plecotus auritus, (E) Barbastella barbastellus, (F) M. dasycneme, (G) M.
nattereri, (H) M. daubentonii, (I) M. bechsteinii, (J) M. brandtii. The
photographs illustrate i) extensive infection of the wing membrane and
cup-shaped epidermal erosions (A, E, H, J; long black arrow); ii) cup-like
epidermal erosions in the pinna (B; long black arrow), iii) Pseudogymnoascus
destructans hyphae obscuring the basement membrane and
invading the dermis (A, B, C, E, H; black arrow); iv) a single cupping
erosion packed with fungal hyphae in the wing membrane (C, D, G, I;
long black arrow); v) colonisation of a hair follicle by P. destructans,
fungal hyphae present in the associated sebaceous gland and regional
connective tissue (F; black arrow); vi) marked signs of inflammation (B,
F, J); and vii) a cellular inflammatory crust that sequesters fungal hyphae
(A, J). White arrows within each photograph indicate the interface
between epidermis and dermis. Periodic acid-Schiff stain; scale bar =
50 mm. M. myotis not shown because WNS lesions in this species have
already been documented elsewhere [18].
doi:10.1371/journal.pone.0097224.g001
Bat Taxa at Infection Risk from WNS
PLOS ONE | www.plosone.org 4 May 2014 | Volume 9 | Issue 5 | e97224
Paper 2.4.1 235
Figure 2. Phylogenetic reconstruction of bats from Europe and North America. The reconstruction was based on a concatenated DNA
sequence matrix from 13 loci, purged from a maximum likelihood vespertilionid phylogeny rooted on Rhinolophus. Blue = species reported positive
for WNS fungus prior to 2012, red = species recognised positive in this study, bat image = bat species diagnosed as WNS positive in this study.
doi:10.1371/journal.pone.0097224.g002
Bat Taxa at Infection Risk from WNS
PLOS ONE | www.plosone.org 5 May 2014 | Volume 9 | Issue 5 | e97224
236 Paper 2.4.1
Table2.Pseudogymnoascusdestructansinfectioninrelationtochiropteranphylogenyandecologicalsimilarity.
VariableNMPDNRIpNRIMNTDNTIpNTI
Explanatory
CAVE420.2701.3000.1050.0911.9840.025
REGION230.33221.6060.9440.1141.1250.135
CLUSTER270.2412.1510.0120.1091.1800.124
TEMPERATURE210.31821.0200.8410.1280.5690.292
SHELTERhidden310.2462.1470.0130.1210.1150.450
SHELTERexposed210.31620.9290.8140.1191.0040.165
SHELTERbotha
50.2241.0510.1060.1700.7540.229
CSIZEno200.33721.6390.9470.1250.7360.240
CSIZEsmall70.1263.0250.0010.0822.4740.004
CSIZEmedium90.2800.2010.4740.1620.4040.358
CSIZElarge110.2650.5920.3060.1660.0640.488
RANGEverylarge150.33321.2170.8750.18821.3280.902
RANGElarge130.31120.5650.7080.21321.8720.970
RANGEmoderate140.2461.1890.1120.1330.8590.203
RANGEsmall50.0982.8650.0010.0592.6750.001
FOODcolleoptera50.2580.4820.3300.1690.7060.249
FOODdiptera90.2371.0930.1170.1440.9020.188
FOODgeneralist180.31320.7630.7700.1280.7720.232
FOODlepidoptera140.2860.1170.4750.1470.3550.368
FOODother1n/a
HABITATclosed110.2800.2280.4580.1420.8020.220
HABITATopen130.31420.6230.7240.17820.6520.739
HABITATedge230.2670.8550.2110.0932.3210.007
BODYsmall100.29220.0970.5760.1420.8550.204
BODYmedium210.2112.8290.0010.1051.6980.048
BODYlarge160.36822.2590.9890.1490.0700.475
Response
Pd+onphylogeny(Figure2)200.2122.9140.0010.1111.5550.058
Pd+on’eco’tree(Figure3)207.8390.8620.2014.3560.2770.384
a
=speciesusingbothtypesofsheltersarealsoincludedinthepreviouscategories.
PhylogeneticsignalofexplanatoryvariablesonaphylogenyofbatsfromEuropeandNorthAmericaandofP.destructansinfectiononbothphylogenyandaneighbour-joiningtreebasedonsquaredEuclideandistancesof
ecologicalandbehaviouraltraitsofhibernatingbatspecies.Valuesinboldindicatesignificantclusteringorover-dispersionofP.destructansinfectiononthetree.Notethatallcategoriesofexplanatoryvariablesweretestedhere,
butn-1dummyvariableswereincludedinthePGLSmodel.N=numberofspeciesscoredpositiveforthegivenvariable,MPD=meanphylogeneticdistance,NRI=netrelatednessindex,MNTD=meannearesttaxon
phylogeneticdistance,NTI=nearesttaxonindex.
doi:10.1371/journal.pone.0097224.t002
Bat Taxa at Infection Risk from WNS
PLOS ONE | www.plosone.org 6 May 2014 | Volume 9 | Issue 5 | e97224
Paper 2.4.1 237
Testing the hypothesis of ecological similarity. The
ecological similarity tree for European and North America bats
was constructed using the squared Euclidean distances of the traits
dataset (Figure 3). Analysis of P. destructans-infected species
distribution indicated that infected species were randomly
distributed (MPD = 7.839, NRI = 0.862, p = 0.201) across
the ecological diversity of bats from these two continents. The
most ecologically similar species were also infected randomly
(MNTD = 4.356, NTI = 0.277, p = 0.384; Table 2, Figure 3).
Predicting species at risk from P. destructans
infection. We explored relationships between ecological traits
after removing the effects of bat species relatedness using PGLS.
The final model, displaying lowest AIC (AIC = 69.08, F-statistic
= 8.98, df = 7 and 39, p , 0.001, adjusted R2
= 0.55), differed
from two more complex models by DAIC , 3 (Table S3). The
addition of the variables did not markedly alter results of the
analysis as reported below, and therefore we used the model with
the lowest AIC. It describes the relationship between P. destructans
infection in bat species and Temperature preference during
hibernation (b = 20.207, SE = 0.085); Roost Shelter during
hibernation: Hidden (b = 20.109, SE = 0.211), Exposed (b =
0.560, SE = 0.192); Cluster Size during hibernation: Small (b =
20.386, SE = 0.130), Medium (b = 20.309, SE = 0.194);
Distribution range size: Moderate (b = 20.201, SE = 0.089); and
Feeding habitat: Closed (b = 0.319, SE = 0.113). Shapiro-Wilks’
normality test indicated that model residuals were normally
distributed (W = 0.981, p = 0.63).
The fitted values of P. destructans infection in bats based on the
PGLS model showed overlap between the P. destructans-positive
and -negative bats (Figure 4). Bat species currently recognised as P.
destructans-negative with highest fitted PGLS values were (in
descending order): Corynorhinus townsendii, Lasiurus cinereus, Plecotus
austriacus, Rhinolophus ferrumequinum, and Miniopterus schreibersii.
Discussion
At the end of the hibernation seasons of 2012 and 2013, we
screened 276 bats of unknown infection status and biopsied all bats
with WNS-suspected skin lesions. Both the number of bats and the
number of taxa (n = 15) examined make this the most extensive
Figure 3. Neighbour-joining tree based on ecological and behavioural traits of bats from Europe and North America (rooted at
midpoint). See Figure 2 for a description of the colour scheme.
doi:10.1371/journal.pone.0097224.g003
Bat Taxa at Infection Risk from WNS
PLOS ONE | www.plosone.org 7 May 2014 | Volume 9 | Issue 5 | e97224
238 Paper 2.4.1
and species-rich study of P. destructans infection in Europe to date.
Earlier European studies have provided data from bats originally
sampled for fungal microscopy, culture and genetic analysis when
they exhibited obvious fungal growth during hibernation, numbers
of species examined ranging from 1 to 12 and numbers of
specimens from 1 to 107 [14,15,19–22].
This study documented five additional bat species as positive for
P. destructans infection and added ten species from the genera
Myotis, Eptesicus, Plecotus, Barbastella and Rhinolophus to the list of
European bat species showing histopathological findings consistent
with WNS [23]. Species-specific prevalence of WNS-diagnostic
skin lesions ranged from 0 to 100% (4–55% for species with n .
20; Table 1). Highest prevalence of WNS lesions was observed in
M. myotis (after excluding species for which just one specimen was
caught, i.e. M. dasycneme). Note, however, that with prevalence
differing by an order of magnitude, detection of positive specimens
may have been biased by small sample size in rarer species and
species less frequently visible in hibernacula. With our experimental
design (i.e. trapping of winter survivors leaving hibernacula
and non-destructive UV fluorescence screening), we were able to
confirm that M. myotis had highest prevalence of WNS lesions
(based on histopathology), with the possible exception of M.
dasycneme. Our results confirmed WNS skin lesions in 11 bat
species, which is in contrast to a previous study that found no
invasive growth of P. destructans in European bats [38]. The latter
study, however, examined a low number of bats and used skin
biopsy methods lacking in sensitivity and this may explain the
failure to detect lesions diagnostic of WNS.
Bat species with P. destructans infection exhibited diverse
hibernation behaviours. Among these, R. hipposideros was special
as it is an exclusively solitary hibernator and hangs free in exposed
places with the wing membranes covering the body. It is capable of
hibernation at higher temperatures but requires the microclimate
stability ensured by using the inner parts of caves [39]. Rhinolophus
hipposideros is also the most abundant species in winter-monitoring
counts, amounting to more than 50% of all bats registered in
Czech hibernacula (unpublished data, Czech Bat Conservation
Trust). Moreover, as a member of the suborder Pteropodiformes,
it is phylogenetically distantly related to all other species infected
by P. destructans (Figure 2). As documented by the low infection
prevalence (3.57%), environmentally mediated indirect and
density-dependent transmission probably does not result in higher
risk for this rhinolophid species [10]. Similarly, E. fuscus and M.
leibii, both solitary hibernators from North America, were the least
impacted species. In comparison, higher declines were observed in
large winter colonies of two species that roost solitarily or in small
groups, i.e. P. subflavus and M. septentrionalis [10]. Disease risk,
however, was not related to conspecific transmission only. When
multiple co-occurring species can host the pathogen, densitydependent
transmission can be amplified [40,41].
We confirm here that bat species previously known to be
positive for P. destructans may later show as WNS-positive based on
histopathology. As the lists of bats positive for P. destructans or WNS
lesions in North American and European species are nearly equal,
conclusions drawn from analysis of infection or disease risk should
be similar. Traits describing ecological and behavioural characteristics
of bats occurring within the known distribution of P.
destructans indicate that species belonging to additional genera may
also be found positive for the infection in the future. Our screening
of Czech underground hibernacula, however, demonstrates that
the initial, relatively low, number of bat species positive for P.
destructans infection is more likely the result of sampling bias than a
biological phenomenon. Currently, affected species are ecologically
diverse, to the point where predictions for infected and noninfected
species overlap. We therefore assume that more species
will be revealed as WNS-positive with increased sensitivity of
detection methods [42].
Phylogenetic representation of P. destructans infection indicates
that closely related species are most likely to be infected. In terms
of field surveys, therefore, some Myotis bats are universally likely to
be WNS-positive and most effort should be devoted to these
species. Interestingly, M. alcathoe was free of WNS-positive skin
lesions in the present study (but note the low sample size). Myotis
species typically form clusters during hibernation and such
behaviour promotes frequency-dependent transmission of the
infection, independent of population size, and may yet drive the
species to extinction [9] unless they change their social behaviour,
as documented in M. lucifugus and M. sodalis [10]. In light of our
new data, clustering behaviour is not a descriptor common to
infected species. Rather, a suite of characteristics, including cluster
size, type of shelter during hibernation, temperature at hibernation,
as well as size of the distribution range and feeding in closed
habitats, play a role in characterising bats with P. destructans
infection.
Based on the list of species currently known to be affected by
WNS or P. destructans, it is clear that the fungus is neither species-,
genus- nor family-specific. The multi-host occurrence of the
pathogen might make the disease less predictable using ecologically-
and phylogenetically-based analysis [43]; however, this is
likely to change in the future as additional species are revealed as
susceptible. Hibernation in contaminated caves and mines under
conditions favourable for fungal growth [5,44] appears to be the
main risk factor [4,12]. Distribution of P. destructans is also
correlated with disease in hibernating bats [45]. Importantly, 25
species of insectivorous bats presently hibernate in the United
States and Canada, all of which represent possible hosts of the
fungal pathogen should the disease spread to their geographic
range [46]. This scenario is predicted to happen in most counties
with caves in the contiguous United States by the winter of 2105-
2106 [47].
The reason bats in North America have been so hard-hit, with
millions dying, while bats in Europe apparently cope better with
the infection, has not yet been explained. Likewise, the pathogenFigure
4. Boxplot of fitted values from the phylogenetic
generalised least squares model of P. destructans infection.
Predictions for infected and non-infected species overlap.
doi:10.1371/journal.pone.0097224.g004
Bat Taxa at Infection Risk from WNS
PLOS ONE | www.plosone.org 8 May 2014 | Volume 9 | Issue 5 | e97224
Paper 2.4.1 239
esis of WNS still remains unclear [12]. Behavioural aberrations,
physiological disruption and immunosuppression during hibernation
are, however, considered key pathomechanisms [4,48,49]. On
the other hand, restoration of immune responses in WNS-positive
bats early in post-hibernation may result in immune-mediated
destruction of infected tissues and death [50].
The fact that our samples were mostly collected from bats
emerging from hibernacula at the end of the hibernation season
indicates that European bat species can survive P. destructans
infection and highlights the need for a comparison of European
and North American bat population responses to this fungal
disease. As all European bat species are strictly protected and any
thorough pathological study of P. destructans infection would be
controversial [18,22], implementation of non-lethal sampling
methods is necessary, such as the wing membrane biopsy used
in this study [42]. While a detailed comparison of histopathological
findings in European and North America bats represents a
valid approach to the better understanding of WNS mortality
[17,48], it was outside the objectives of this ecological study. We
are however, planning a comprehensive study to investigate the
extent of WNS wing lesions in hibernating bats from the two
continents.
Conclusions
This hypothesis-driven study explored clustering of P. destructans
infection in relation to chiropteran ecological and behavioural trait
variation and phylogeny and supported this with field data.
Extension of the surveillance to a broader number of species to test
the study’s hypotheses identified multiple European species
positive for WNS. The increased number of positive bat species
resulted in random dispersion of P. destructans infection across trait
trees and weakened the pattern of phylogenetic clustering of P.
destructans infection. Distantly related bat species, characterised by
diverse life histories, were infected and all hibernating bats may,
therefore, be at risk from P. destructans infection. Ecological and
evolutionary constraints on hibernating bats do not pose a barrier
to this generalist fungal pathogen, with WNS occurring in both
suborders of bats. Our findings indicate that a wider focus is
needed in studying the ecology and epidemiology of this fungal
disease of major conservation concern.
Supporting Information
Table S1 Ecological and behavioural traits of European
and North American bat species. The dataset includes the
most common behavioural or trait value for each bat species.
(XLSX)
Table S2 Accession numbers for phylogeny reconstruction.
A total of 13 available mitochondrial and nuclear genes were
used for phylogeny reconstruction of bat species from Europe and
North America.
(XLSX)
Table S3 Phylogenetic generalized least squares model
selection. The model predicting P. destructans infection based on
ecological and behavioural characteristics of bats was selected with
the step-down procedure, where the full model is given on the first
line and removed variables are listed subsequently.
(PDF)
Acknowledgments
The authors would like to thank Kevin Roche for his correction of the
English text. Comments from Justin Boyles and two anonymous reviewers
helped to improve the manuscript.
Author Contributions
Conceived and designed the experiments: JZ H. Bandouchova NM JP.
Performed the experiments: H. Bandouchova JP JZ TB H. Berkova JB MD
KSJ VK MK KO ZR. Analyzed the data: JZ MD NM KSJ VB GGT.
Contributed reagents/materials/analysis tools: JZ H. Bandouchova NM JP
GGT. Wrote the paper: JZ H. Bandouchova NM JP. Compiled the
ecological and behavioural traits of bats: JZ. Reconstructed the phylogeny
of European and North American bats: MD NM. Performed the statistical
analysis: JZ KSJ NM. Examined the histopathological findings: H.
Bandouchova JP. Collected field data: JZ H. Bandouchova TB H. Berkova
JB MD KSJ VK MK KO ZR. Cultured and diagnosed the fungus: JB VK
KO. Reviewed compiled data on North American bats: VB GGT.
Provided a new method to detect WNS lesions: GGT.
References
1. Irwin NR, Bayerlova´ M, Missa O, Martı´nkova´ N (2012) Complex patterns of
host switching in New World arenaviruses. Mol Ecol 21: 4137–4150.
2. Boyles JG, Cryan PM, McCracken GF, Kunz TH (2011) Economic importance
of bats in agriculture. Science 332: 41–42.
3. Eskew EA, Todd BD (2013) Parallels in amphibian and bat declines from
pathogenic fungi. Emerg Infect Dis doi: http://dx.doi.org/10.3201/eid1903.
120707.
4. Blehert DS, Hicks AC, Behr M, Meteyer CU, Berlowski-Zier BM, et al. (2009)
Bat white-nose syndrome: An emerging fungal pathogen? Science 323: 227–227.
5. Gargas A, Trest MT, Christensen M, Volk TJ, Blehert DS (2009) Geomyces
destructans sp nov associated with bat white-nose syndrome. Mycotaxon 108:
147–154.
6. Lorch JM, Meteyer CU, Behr MJ, Boyles JG, Cryan PM, et al. (2011)
Experimental infection of bats with Geomyces destructans causes white-nose
syndrome. Nature 480: 376–378.
7. Minnis AM, Lindner DL (2013) Phylogenetic evaluation of Geomyces and allies
reveals no close relatives of Pseudogymnoascus destructans, comb. nov., in bat
hibernacula of eastern North America. Fungal Biol 117: 638–649.
8. Warnecke L, Turner JM, Bollinger TK, Lorch JM, Misra V, et al. (2012)
Inoculation of bats with European Geomyces destructans supports the novel
pathogen hypothesis for the origin of white-nose syndrome. Proc Natl Acad Sci
USA 109: 6999–7003.
9. Frick WF, Pollock JF, Hicks AC, Langwig KE, Reynolds DS, et al. (2010) An
emerging disease causes regional population collapse of a common North
American bat species. Science 329: 679–682.
10. Langwig KE, Frick WF, Bried JT, Hicks AC, Kunz TH (2012) Sociality, densitydependence
and microclimates determine the persistence of populations
suffering from a novel fungal disease, white-nose syndrome. Ecol Lett 15:
1050–1057.
11. Turner GG, Reeder DM (2009) Update of white-nose syndrome in bats,
September 2009. Bat Research News 50(3): 47–53.
12. Blehert DS (2012) Fungal disease and the developing story of bat white-nose
syndrome. PLoS Pathog 8: e1002779.
13. Wilder AP, Frick WF, Langwig KE, Kunz TH (2011) Risk factors associated
with mortality from white-nose syndrome among hibernating bat colonies. Biol
Lett 7: 950–953.
14. Martı´nkova´ N, Bacˇkor P, Bartonicˇka T, Blazˇkova´ P, Cˇ erveny´ J, et al. (2010)
Increasing incidence of Geomyces destructans fungus in bats from the Czech
Republic and Slovakia. PLoS One 5: e13853.
15. Puechmaille SJ, Wibbelt G, Korn V, Fuller H, Forget F, et al. (2011) PanEuropean
distribution of white-nose syndrome fungus (Geomyces destructans) not
associated with mass mortality. PLoS One 6: e19167.
16. Foley J, Clifford D, Castle K, Cryan P, Ostfeld RS (2010) Investigating and
managing the rapid emergence of white-nose syndrome, a novel, fatal, infectious
disease of hibernating bats. Conserv Biol 25: 223–231.
17. Puechmaille SJ, Frick WF, Kunz TH, Racey PA, Voigt CC, et al. (2011) Whitenose
syndrome: is this emerging disease a threat to European bats? Trends Ecol
Evol 26: 570–576.
18. Pikula J, Bandouchova H, Novotny L, Meteyer CU, Zukal J, et al. (2012)
Histopathology confirms white-nose syndrome in bats in Europe. J Wildl Dis 48:
207–211.
19. Kuba´tova´ A, Koukol O, Nova´kova´ A (2011) Geomyces destructans, phenotypic
features of some Czech isolates. Czech Mycol 63: 65–75.
20. Puechmaille SJ, Verdeyroux P, Fuller H, Gouilh MA, Bekaert M, et al. (2010)
White-nose syndrome fungus (Geomyces destructans) in bat, France. Emerg Infect
Dis 16: 290–293.
Bat Taxa at Infection Risk from WNS
PLOS ONE | www.plosone.org 9 May 2014 | Volume 9 | Issue 5 | e97224
240 Paper 2.4.1
21. Sˇimonovicˇova´ A, Pangallo D, Chovanova´ K, Lehotska´ B (2011) Geomyces
destructans associated with bat disease WNS detected in Slovakia. Biologia 66: 562–
564.
22. Wibbelt G, Kurth A, Hellmann D, Weishaar M, Barlow A, et al. (2010) Whitenose
syndrome fungus (Geomyces destructans) in bats, Europe. Emerg Infect Dis 16:
1237–1243.
23. Meteyer CU, Buckles EL, Blehert DS, Hicks AC, Green DE, et al. (2009)
Histopathologic criteria to confirm white-nose syndrome in bats. J Vet Diagn
Invest 21: 411–414.
24. Muller LK, Lorch JM, Lindner DL, O’Connor M, Gargas A, et al. (2013) Bat
white-nose syndrome: a real-time TaqMan polymerase chain reaction test
targeting the intergenic spacer region of Geomyces destructans. Mycologia 105: 253–
259.
25. Lorch JM, Gargas A, Meteyer CU, Berlowski-Zier BM, Green DE, et al. (2010)
Rapid polymerase chain reaction diagnosis of white-nose syndrome in bats. J Vet
Diagn Invest 22: 224–230.
26. Simmons NB (2005) Chiroptera. In: Wilson DE, Reeder DM, editors. Mammal
species of the world: A taxonomic and geographic reference. Smithsonian series
in comparative evolutionary biology, 3rd edition. Washington DC: Smithsonian
Inst. Press. pp. 312–529.
27. Webb PI, Speakman JR, Racey PA (1996) How hot is a hibernaculum? A review
of the temperatures at which bats hibernate. Can J Zool 74: 761–765.
28. Hora´cˇek I, Hana´k V, Gaisler J (2000) Bats of Palearctic region. In: Woloszyn
BW, editor. Proceedings of the VIIIth European bat research symposium: 23–27
August 1999. Krakow, Poland: Institute of Systematics and Evolution of Animals
PAS. pp. 11–157.
29. Reichard JD, Kunz TH (2009) White-nose syndrome inflicts lasting injuries to
the wings of little brown myotis (Myotis lucifugus). Acta Chiropt 11: 457–464.
30. Paradis E, Claude J, Strimmer K (2004) APE: Analyses of phylogenetics and
evolution in R language. Bioinformatics 20: 289–290.
31. R Core Team: R: a language and environment for statistical computing. Vienna,
Austria: R Foundation for Statistical Computing. Available: http://www.Rproject.org.
Accessed 2013 Oct 10.
32. Webb CO, Ackerly DD, Kembel SW (2008) Phylocom: software for the analysis
of phylogenetic community structure and character evolution. Bioinformatics 24:
2098–2100.
33. Lanfear R, Calcott B, Ho SYW, Guindon S (2012) PartitionFinder: Combined
selection of partitioning schemes and substitution models for phylogenetic
analyses. Mol Biol Evol 29: 1695–1701.
34. Pattengale ND, Alipour M, Bininda-Emonds ORP, Moret BME, Stamatakis A
(2010) How many bootstrap replicates are necessary? J Comput Biol 17: 337–
354.
35. Felsenstein J (1985) Phylogenies and the comparative method. Am Nat 125: 1–
15.
36. Orme D, Freckleton R, Thomas G, Petzoldt T, Fritz S, et al. (2013) Caper:
Comparative analyses of phylogenetics and evolution in R. R package version
0.5.2. Available: http://CRAN.R-project.org/package = caper. Accessed 2014
Feb 24.
37. Harmon LJ, Weir JT, Brock CD, Glor RE, Challenger W (2008) GEIGER:
investigating evolutionary radiations. Bioinformatics 24: 129–131.
38. Wibbelt G, Puechmaille SJ, Ohlendorf B, Mu¨hldorfer K, Bosch T, et al. (2013)
Skin lesions in European hibernating bats associated with Geomyces destructans, the
etiologic agent of white-nose syndrome. PLoS ONE 8: e74105.
39. Zukal J, Berkova´ H, Rˇ eha´k Z (2005) Activity and shelter selection by Myotis
myotis and Rhinolophus hipposideros hibernating in the Katerˇinska´ cave (Czech
Republic). Mamm Biol 70: 271–281.
40. Keesing F, Holt RD, Ostfeld RS (2006) Effects of species diversity on disease
risk. Ecol Lett 9: 485–498.
41. Ostfeld RS, Keesing F (2012) Effects of host diversity on infectious disease. Annu
Rev Ecol Evol Syst 43: 157–182.
42. Turner GG, Meteyer CU, Barton H, Gumbs JF, Reeder DM, et al. (2014)
Nonlethal screening of bat-wing skin with the use of ultraviolet fluorescence to
detect lesions indicative of white-nose syndrome. J Wildl Dis, doi: 10.7589/2014-
03-058.
43. Woolhouse MEJ, Taylor LH, Haydon DT (2001) Population biology of
multihost pathogens. Science 292: 1109–1112.
44. Verant ML, Boyles JG, Waldrep W, Wibbelt G, Blehert DS (2012)
Temperature-dependent growth of Geomyces destructans, the fungus that causes
bat white-nose syndrome. PLoS ONE 7: e46280.
45. Lorch JM, Muller LK, Russell RE, O9Connor M, Lindner DL, et al. (2013)
Distribution and environmental persistence of the causative agent of white-nose
syndrome, Geomyces destructans, in bat hibernacula of the Eastern United States.
Appl Environ Microbiol 79: 1293–1301.
46. U.S. Geological Survey (2012) White-Nose Syndrome Threatens the Survival of
Hibernating Bats in North America. Available: http://www.fort.usgs.gov/
WNS/Default.asp. Accessed 2014 Jan 05.
47. Maher SP, Kramer AM, Pulliam JT, Zokan MA, Bowden SE, et al. (2012)
Spread of white-nose syndrome on a network regulated by geography and
climate. Nat Commun 3: 1306.
48. Cryan PM, Meteyer CU, Boyles JG, Blehert DS (2010) Wing pathology of
white-nose syndrome in bats suggests life-threatening disruption of physiology.
BMC Biology 8: 135.
49. Cryan PM, Meteyer CU, Blehert DS, Lorch JM, Reeder DM, et al. (2013)
Electrolyte depletion in white-nose syndrome bats. J Wildl Dis 49: 398–402.
50. Meteyer CU, Barber D, Mandl JN (2012) Pathology in euthermic bats with
white nose syndrome suggests a natural manifestation of immune reconstitution
inflammatory syndrome. Virulence 3: 583–588.
Bat Taxa at Infection Risk from WNS
PLOS ONE | www.plosone.org 10 May 2014 | Volume 9 | Issue 5 | e97224
Paper 2.4.1 241
Paper 2.4.2
Zukal J., Bandouchova H., Brichta J., Cmokova A., Jaron K. S., Kolarik M., Kovacova
V., Kubátová A., Nováková A., Orlov O., Pikula J., Presetnik P., Šuba J., Zahradníková
A. Jr., Martínková N. 2016. White-nose syndrome without borders: Pseudogymnoascus
destructans infection confirmed in Asia. Scientific Reports 6: 19829.
– 243 –
1Scientific Reports | 6:19829 | DOI: 10.1038/srep19829
www.nature.com/scientificreports
White-nose syndrome without
borders: Pseudogymnoascus
destructans infection tolerated in
Europe and PalearcticAsia but not
in NorthAmerica
Jan Zukal1,2,*
, Hana Bandouchova3,*
, Jiri Brichta3
,Adela Cmokova4
, Kamil S. Jaron1
,
Miroslav Kolarik4
,Veronika Kovacova3
,Alena Kubátová5
,Alena Nováková4
,Oleg Orlov6,7
,
Jiri Pikula3
, Primož Presetnik8
, Jurģis Šuba9
,Alexandra Zahradníková Jr.10
&
Natália Martínková1,11
A striking feature of white-nose syndrome, a fungal infection of hibernating bats, is the difference
in infection outcome between NorthAmerica and Europe. Here we show highWNS prevalence both
in Europe and on theWest Siberian Plain inAsia. Palearctic bat communities tolerate similar fungal
loads of Pseudogymnoascus destructans infection as their Nearctic counterparts and histopathology
indicates equal focal skin tissue invasiveness pathognomonic forWNS lesions. Fungal load positively
correlates with disease intensity and it reaches highest values at intermediate latitudes. Prevalence
and fungal load dynamics in Palearctic bats remained persistent and high between 2012 and 2014.
Dominant haplotypes of five genes are widespread in NorthAmerica, Europe andAsia, expanding the
source region of white-nose syndrome to non-European hibernacula.Our data provides evidence for
both endemicity and tolerance to this persistent virulent fungus in the Palearctic, suggesting that hostpathogen
interaction equilibrium has been established.
Emerging wildlife infections are a threat to global biodiversity1
, yet the emergence and transmission of such
infectious diseases may also be a function of ecosystem quality and biodiversity2
. Moreover, anthropogenic disturbance
and introductions have been highlighted as contributing considerably to such disease outbreaks3,4
. The
majority of emerging wildlife pathogens are of viral origin but fungal infections have also been recognised across
a diverse range of taxa, including plants and poikilothermic animals5
. Relatively few fungal species cause severe
diseases in mammals due to their high body temperature and effective immune system6,7
.
Prevalence of pathogens may be high in bat populations8,9
. Despite this, bats have been reported to survive
infections that are lethal to other taxa10–12
. Bat’s capacity to carry pathogens may result from its ability to tolerate
damage caused by the agent or the associated immune response13–15
. Understanding the mechanisms that
1
Institute ofVertebrate Biology,Academy of Sciences of theCzech Republic, Květná 8, 603 65 Brno,Czech Republic.
2
Department of Botany andZoology, MasarykUniversity, Kotlářská 2, 611 37 Brno,Czech Republic. 3
Department of
Ecology and Diseases ofGame, Fish and Bees,University ofVeterinary and PharmaceuticalSciences Brno, Palackeho
1/3, 612 42 Brno, Czech Republic. 4
Laboratory of Fungal Genetics and Metabolism, Institute of Microbiology,
Academy of Sciences of the Czech Republic, Vídeňská 1083, 142 20 Prague 4, Czech Republic. 5
Department of
Botany, Faculty of Science, Charles University in Prague, Benátská 2, 128 01 Prague, Czech Republic. 6
Ural State
Pedagogical University, Kosmonavtov str. 26, 620017Yekaterinburg, Russia. 7
Ural State Medical University, Repina
str. 3, 620028Yekaterinburg, Russia. 8
Centre for Cartography of Fauna and Flora, Antoličičeva 1, SI-2204 Miklavž
na Dravskem polju, Slovenia. 9
Latvian State Forest Research Institute “Silava”, 111 Rigas str., LV-2169 Salaspils,
Latvia. 10
Institute of Molecular Physiology and Genetics, Slovak Academy of Sciences,Vlárska 5, 83334 Bratislava,
Slovakia. 11
Institute of Biostatistics and Analysis, Masaryk University, Kamenice 3, 625 00 Brno, Czech Republic.
*
These authors contributed equally to this work.Correspondence and requests for materials should be addressed to
J.P. (email: pikulaj@vfu.cz) or N.M. (email: martinkova@ivb.cz)
received: 08 July 2015
accepted: 15 December 2015
Published: 29 January 2016
OPEN
244 Paper 2.4.2
www.nature.com/scientificreports/
2Scientific Reports | 6:19829 | DOI: 10.1038/srep19829
underlie any trade-off between tolerance and resistance to infectious agents in reservoir hosts is of utmost importance
for effective disease control16
.
Most bats in temperate regions enter seasonal hibernation, at which time they reduce their metabolic rate and
decrease body temperature until it approaches the ambient temperature of the hibernaculum, which may alter
immune response to pathogens17,18
. During hibernation, bats are vulnerable to infection by the psychrophilic
fungus Pseudogymnoascus destructans [formerly Geomyces destructans]19
, which emerged as a novel pathogen
in eastern North America in 200620
. The P. destructans fungus is the causative agent of white-nose syndrome
(WNS)20–23
. Considered an epidemic of major conservation concern in North America, WNS is responsible for
an unprecedented decline in bat populations24–26
. WNS combines some of the worst possible epidemiological
characteristics, including a highly virulent pathogen23,27
with density- and frequency-dependent transmission28
,
an environmental reservoir19
, long-term persistence in hibernacula29,30
and susceptibility of multiple hosts19,20,31
.
The origin of WNS in North America remains unknown and current research focuses on identifying the
infectious agent’s source and identifying whether the agent is an introduced pathogen. Six main lines of evidence
support introduction of WNS into North America: 1) Fungal communities associated with bats and their hibernacula
in eastern North America comprise a diverse range of P. destructans allies22,32
; phylogenetic evaluation,
however, indicates that none of these is closely related to P. destructans22
. 2) Previous studies have demonstrated
clonal dispersal of a single P. destructans genotype among WNS-infected bats in the United States33,34
. 3) Spread
of WNS in North America follows a clear invasion front with varying survival rate since first appearance of the
disease24,35–37
. 4) Presence of P. destructans has been confirmed in many European countries38–41
but with no
reports of mass mortality38
. 5) European P. destructans isolates are pathogenic for North American bats23
. Further,
virulent skin infections producing focally severe lesions pathognomonic for WNS have been documented in
European bats under natural infection conditions42
. 6) While only one heterothallic fungal mating type has been
recorded in North American P. destructans populations, two types have been found coexisting in European hibernacula.
Effective recombination during sexual reproduction results in genetic variability and may be linked with
virulence43
. These indirect sources of evidence tend to support the introduced pathogen hypothesis, suggesting
WNS may have originated outside of North America22,23,43,44
. Comparative studies between North America and
Europe have been proposed to explain the origin and differential manifestation of the fungal infection4,40,42,45
.
Although some authors speculate on a non-European origin4,25
, Europe is believed to be the likely source
of WNS44
. Supporting evidence for long exposure to the WNS fungus based on old European photographs,
which appear to show a white cutaneous fungal infection38,40
, lacks specificity for WNS. Another dermatophyte,
Trichophyton redellii, produces a similar gross appearance in hibernating bats46
and the photographs might represent
either. On the contrary, presence of WNS in multiple and diverse hosts31
indicates that the WNS fungus
could persist across their ranges in the Palearctic.
To date, WNS diagnostic skin lesions have only been reported from the Czech Republic, Europe, with almost
half of all species being WNS positive and prevalence based on histopathology reaching 55% in bats emerging
from hibernacula in spring31,42
. Occurrence of WNS in distantly related bat species with diverse ecologies,
however, suggests the pathogen is a generalist and that all bats hibernating within the distribution range of P.
destructans may be at risk of infection31
. If we assume that P. destructans range expansion and its concurrent WNS
epidemic occurred unnoticed in Europe in the past, we expect that WNS should be found at any site in Europe
with conditions favourable for the pathogen and could extend to non-European hibernacula of the Palearctic
region. We tested the following four hypotheses. First, P. destructans is present in European and non-European
Palearctic hibernacula and, if the fungus is present in a hibernaculum, bats would test positive for WNS under
histopathology. Second, an endemic steady-state of P. destructans infection in the Palearctic would be reflected in
a persistent high prevalence among bats with no associated population declines. Third, given that the differences
in population size changes in WNS-positive regions24,38
are explained as WNS resistance in Europe, Palearctic
bats will display a lower pathogen load and smaller size of cupping erosions diagnostic for WNS than Nearctic
species. Fourth, in regions with endemic pathogen occurrence, P. destructans load will be influenced by geographic
location, sampling date and bat community structure.
Results
WNS in the Palearctic.  We sampled and examined bats from sites in geographically distant regions (the
Czech Republic, Slovenia, Latvia and Russia; n =  481) to assess occurrence of WNS in the Palearctic (Table 1,
Supplementary Table S1). We found WNS-positive bats from multiple species at all sites, including the West
Siberian Plain in Russia. At each site, all of the following criteria categorised the site as WNS positive: suspected
fungal growth (Fig. 1), characteristic curved conidia of P. destructans observed on an adhesive tape imprint
(Fig. 2A), P. destructans confirmed by DNA sequencing and qPCR using a species-specific probe, and definitive
diagnosis confirmed through histopathology of biopsy punches from wing membrane lesions targeted by UV
trans-illumination (Fig. 2B).
Histopathological findings matching WNS diagnostic criteria were present in 100 bats of 13 different species
(Table 1). Two species were newly identified as positive for WNS, Miniopterus schreibersii and Rhinolophus
euryale. Species-specific prevalence, using qPCR to detect P. destructans and UV trans-illumination to detect
lesions indicative of WNS, ranged from 64 to 100% (overall prevalence =  83.1%) and 14 to 100% (overall preva-
lence =  51.9%), respectively (Table 1).
In order to evaluate P. destructans genetic variability, we sequenced six loci resulting in 254 new sequences.
Two alleles were found in TUB2, MAT1-1-1, MAT1-2-1 and CAM loci, four alleles in TEF1α and three in ITS
(Fig. 3). The most frequent alleles from all loci were found in both European and Asian P. destructans isolates. In
ITS and TEF1α from Asia, we sampled isolates with one substitution difference from the most frequent allele.
The dominant haplotype from the concatenated sequence of four genes (TUB2, CAM, TEF1α, ITS) is widespread
in North America, Europe and Asia (Fig. 3E). Three divergent concatenated haplotypes were found in the Czech
Paper 2.4.2 245
www.nature.com/scientificreports/
3Scientific Reports | 6:19829 | DOI: 10.1038/srep19829
Republic and one local to Asian Russia. The P. destructans mating type proportion was 27 (MAT1-1-1) to 19
(MAT1-2-1) (χ 2
 =  1.391, P =  0.238).
Quantitative comparison of WNS on bats.  The fungal load on qPCR-positive bats ranged from 0.21
pg to 3.41 μ g across the surface of the left wing (Supplementary Fig. S1, see Table 1 for sample sizes). This range
included fungal loads reported from Nearctic bat species (Supplementary Fig. S1). Log-transformed fungal load
was not lower on Palearctic bats than on Nearctic bats (Wilcoxon W =  68030, P =  1, n =  413 and 247, respectively),
even assuming that fungal load on Nearctic bats could vary up to ten-times due to the difference in sample
collection method between continents (Wilcoxon W =  50910, P =  0.484).
In order to account for body size differences between Palearctic bat species, we used log-transformed fungal
load per cm2
of wing area (henceforth referred to as fungal load, unless specified otherwise). Fungal load differed
significantly between regions (Kruskal-Wallis χ2
 =  94.2, P <  0.001; Fig. 4) and species (Kruskal-Wallis χ2
 =  221.2,
P <  0.001; Fig. 5), with highest values recorded in the Czech Republic and two euryvalent bat species, Myotis myotis
and Myotis nattereri, respectively. In frequently captured M. myotis, fungal load did not change between 2012
and 2014 (ANOVA: F2,146 =  1.209, P =  0.301). Prevalence dynamics of nine species captured in multiple years
was statistically equal between years (χ2
test: P >  0.05; Czech Republic). The sampled regions contained multiple
species with varying fungal loads (Fig. 5), and some species were present in multiple regions (Table 1). The
observed difference of fungal load can be caused by a different set of species or different environmental conditions
Region Bat species
Screened WNS qPCR+ WNS UV+ WNS histo+
n n
qPCR+
(%) ± s.e. n
UV+
(%) ± s.e. n
histo+
(%) ± s.e.
Slovenia
Miniopterus schreibersii 21 20 100 ±  7.1 21 47.62 ±  10.9 6 16.67 ±  20.05
Myotis emarginatus 3 0 NA ±  NA 3 100 ±  32.22 3 100 ±  32.22
Myotis myotis 11 10 100 ±  13.21 11 63.64 ±  12.16 4 0 ±  26.89
Rhinolophus euryale 14 14 64.29 ±  9.83 14 28.57 ±  9.83 1 100 ±  48.47
Total 49 44 88.1  ±  4.88 49 59.96  ±  7 14 54.17  ±  13.32
Czech Republic
Barbastella barbastellus 18 17 64.71 ±  11.59 18 50 ±  11.79 4 75 ±  26.89
Eptesicus nilssonii 3 2 50 ±  39.61 2 0 ±  39.61 1 100 ±  48.47
Myotis alcathoe 7 7 71.43 ±  17.76 7 14.29 ±  17.76 6 0 ±  20.05
Myotis bechsteinii 23 23 86.96 ± 6.23 23 39.13 ± 10.18 9 44.44 ± 14.45
Myotis brandtii 17 16 68.75 ± 8.71 17 23.53 ± 8.24 1 100 ± 48.47
Myotis dasycneme 1 1 100 ± 48.47 1 100 ± 48.47 1 100 ± 48.47
Myotis daubentonii 33 30 86.67 ± 4.85 31 51.61 ± 8.98 14 35.71± 12.81
Myotis emarginatus 33 32 96.88 ± 4.56 32 43.75 ± 8.77 5 60 ± 23
Myotis myotis 156 150 99.33 ± 1.01 108 96.3 ± 1.4 61 73.77 ±  5.63
Myotis nattereri 22 21 95.24 ± 6.78 22 54.55 ± 10.62 9 33.33 ± 14.45
Plecotus auritus 23 23 91.3 ±  6.23 22 68.18 ±  9.93 11 54.55 ±  12.16
Plecotus austriacus 3 3 66.67 ±  32.22 3 33.33 ±  32.22 1 0 ±  48.47
Rhinolophus hipposideros 35 29 79.31 ±  7.52 35 34.29 ±  8.02 8 50 ±  15.94
Total 374 354 81.33  ±  2.07 321 46.84  ±  2.79 131 55.91 ±  4.34
Latvia
Eptesicus nilssonii 4 4 25 ±  26.89 4 25 ±  26.89 1 0 ±  48.47
Myotis brandtii 4 4 50 ±  26.89 4 50 ±  26.89 0 NA ±  NA
Myotis dasycneme 8 8 100 ±  15.94 8 100 ±  15.94 6 50 ±  20.05
Myotis daubentonii 9 9 100 ±  14.45 9 88.89 ±  14.45 6 66.67 ±  20.05
Total 25 25 68.75  ±  9.27 25 65.97  ±  9.48 13 38.89 ±  13.52
Russia
Eptesicus nilssonii 5 5 100 ±  23 5 100 ±  23 2 100 ±  39.61
Myotis brandtii 9 9 100 ± 14.45 9 100 ±  14.45 2 0 ±  39.61
Myotis dasycneme 17 17 100 ± 8.24 17 100 ± 8.24 12 75 ± 11.27
Myotis daubentonii 1 1 100 ±  48.47 1 100 ± 48.47 0 NA ±  NA
Plecotus auritus 1 1 100 ±  48.47 1 100 ± 48.47 1 100 ±  48.47
Total 33 33 100 ±  4.43 33 100 ±  4.43 17 68.75 ±  11.24
USA Myotis lucifugus 10 0 NA± NA 10 100 ± 13.21 10 100 ± 13.21
Total 10 0 NA ±  NA 10 100  ±  13.21 10 100  ±   13.21
Table 1.  Prevalence of white-nose syndrome and Pseudogymnoascus destructans infection in bat species
from the Holarctic region. Screened =  number of bats captured and examined by PCR (to detect the
pathogen), UV light trans-illumination or histopathology (to detect WNS lesions). +  =  percentage of positive
bats from the number screened by the method. NA =  not available. Samples subjected for histopathology
examination were suspect lesions selected under field conditions based on UV trans-illumination and thus
prevalence on histopathology is not based on a randomized sample. It rather reflects qualitative information
that WNS was confirmed in the species with histopathology.
246 Paper 2.4.2
www.nature.com/scientificreports/
4Scientific Reports | 6:19829 | DOI: 10.1038/srep19829
in sampled regions. Therefore, we separated the effect of the two variables on fungal load by correcting for the
random effect of species and region. This had no impact on significance of the comparison between regions or
species. The phylogenetic signal of mean species fungal load was significant (Blomberg’s K =  0.892, P =  0.002),
meaning that closely related taxa had more similar mean fungal loads in the Palearctic than expected by chance.
Fungal load was significantly correlated with the number of WNS lesions (Spearman’s rank correlation:
r =  0.61, P <  0.001). Phylogenetic generalised least-squares, which accounts for intra-specific variability and
Figure 1.  Fungal growth on hibernating bats from Russia. (a) A hibernating cluster of pond bats Myotis
dasycneme in a cave near Yekaterinburg, Russia, in May 2014. Black and white arrows indicate fungal growth
on the muzzle and forearm, respectively. (b) A pond bat from the same hibernaculum showing visible fungal
growths on the uropatagium, pelvic limb toes, plagio- and pro-patagium and the ears and muzzle (white
arrows). Photo: Jiri Pikula.
Figure 2.  White-nose syndrome on a pond bat Myotis dasycneme near Yekaterinburg, Russia, in May 2014.
(A) Microscopic identification of the characteristic curved conidia of Pseudogymnoascus destructans (black
arrow). (B) Invasive fungal growth penetrating the full-thickness of the wing membrane (black arrows), with
several inflammatory cells (neutrophils) situated at both margins of the lesion (white arrows). (C) Packed fungal
hyphae of cupping erosions (black arrows) sequestered by neutrophils (white arrows). (D) Histopathological
finding from a WNS-positive Nearctic Myotis lucifugus, identical to that found in Palearctic Asia. Skin sections
stained with periodic acid-Schiff stain.
Paper 2.4.2 247
www.nature.com/scientificreports/
5Scientific Reports | 6:19829 | DOI: 10.1038/srep19829
species phylogenetic relationships, indicated that the number of WNS lesions increased with increasing fungal
load across bat diversity in this study (intercept =  –0.058, slope =  0.177; Fig. 6). No significant phylogenetic signal
was detected in the mean number of WNS lesions by species (Blomberg’s K =  0.511, P =  0.073).
Microscopic WNS cupping erosion width ranged between 28.6 and 397.2 μ m (median =  86.34 μ m, n =  105)
and depth between 11.3 and 91.8 μ m (median =  31.29 μ m), this also being the WNS-lesion size range detectable
photographically as individual spots using UV trans-illumination. Size of WNS lesion (log-transformed) did
not differ significantly between species (Kruskal-Wallis χ2
 =  20.7, P =  0.079, n =  14 species across the Holarctic;
GF
D E
B CA
1 bp
1 45 185
Figure 3.  Median-joining networks for Pseudogymnoascus destructans DNA sequences. Node centroid
distances correspond to the number of substitutions between haplotypes and the node size to the number of
isolates sharing the haplotype. Isolate origin is signified through different colours based on the inset map. The
map was modified from http://www.amcharts.com/visited_countries/ (last accessed on 9 October, 2015).
(A) TUB2 (n =  46), (B) ITS (n =  203), (C) CAM (n =  47), (D) TEF1α (n =  51), (E) sequences concatenated with
TUB2, ITS, CAM and TEF1α (n =  34); haplotypes pooled according to the available sequence and gaps treated
as missing data, (F) MAT1-1-1 (n =  27), (G) MAT1-2-1 (n =  19).
8
6
4
2
Fungalloadpercm2
(log10(µg))
Measured values
Predicted values
Slovenia Czech Republic Latvia Russia
2.0
1.5
1.0
0.5
0.0
0.5
1.0
NumberofWNSlesionspercm2
(log10)
Figure 4.  Fungal load and number of WNS lesions in Europe (Czech Republic, Latvia, Slovenia) and
in Asia (Russia). Fungal load is quantified as P. destructans-specific DNA per cm2
of wing area (established
through quantitative PCR) and number of WNS lesions is counted on the same wing (using UV light trans-
illumination).
248 Paper 2.4.2
www.nature.com/scientificreports/
6Scientific Reports | 6:19829 | DOI: 10.1038/srep19829
Fig. 7, see Table 1 for sample sizes). Phylogenetic signal for mean species-specific WNS lesion size was also not
significant (Blomberg’s K =  0.448, P =  0.177) and there was no significant difference observed between Palearctic
and Nearctic bats (phylogenetic ANOVA: F =  0.43, P =  0.292).
The number of WNS lesions in animals positive over UV ranged from 1 to 805 (median =  13) and differed
significantly between Palearctic regions (log-transformed per cm2
; Kruskal-Wallis χ2
 =  33.6, P <  0.001; Fig. 4)
and species (Kruskal-Wallis χ2
 =  76.7, P <  0.001; Fig. 5; correction did not influence significance). The fungal
load from UV-negative individuals (median =  3.78 ×  10−5
 μ g) overlapped that from UV-positive individuals
(median =  7.46 ×  10−4
 μ g; Fig. 8).
Comparison of AIC values indicated that the best fitting model for fungal load across regions included geographic
coordinates (95% confidence interval of parameter estimates: longitude: –0.07–(–0.03), latitude: 0.16–
0.36), sampling date (0.002–0.02) and structure of common bat species community found in the respective region
(–3.86− (−2.24)) (Supplementary Table S2). On the other hand, disease intensity, quantified as number of WNS
lesions, was best modelled using fungal load (0.18–0.28) and sampling date (–0.007–0.0003), though addition or
removal of other variables did not change the relationship between WNS-lesion number and fungal load significantly
(0.18–0.27). In both cases, species was treated as a random effect. Values predicted from the optimal model
for both fungal load and number of WNS lesions were highest in the Czech Republic (Fig. 4). Lowest values were
predicted in Slovenia and intermediate values in Latvia and Russia.
Discussion
Here we show WNS infection in Palearctic bat communities within and beyond the borders of Europe, greatly
extending the distribution range of P. destructans and confirming its generalist nature. Furthermore, we confirm
WNS skin lesions in two more bat species, belonging to families Miniopteridae and Rhinolophidae.
Histopathology of skin sections is presently considered the ‘gold standard’ for diagnosing WNS47
. The microscopically
identified WNS in bat species sampled in Russia was consistent with WNS histopathological criteria42,47
.
These included characteristic cup-shaped epidermal erosions, P. destructans-infected hair follicles, associated
sebaceous glands and regional connective tissues, and invasive fungal growth throughout the wing-membrane
thickness (Fig. 2). Our results demonstrate that P. destructans is virulent for Palearctic bats under natural infectious
conditions, as previously shown in the Czech Republic31,42
. Local invasiveness and WNS-lesion severity in
wing membranes of bats collected in the Palearctic was comparable with that of Nearctic bats (Figs 2 and 8), with
8
6
4
2
P.destructansloadpercm2
(log10(µg))
Barbastellabarbastellus
Eptesicusnilssonii
Miniopterusschreibersii
Myotisalcathoe
Myotisbechsteinii
Myotisbrandtii
Myotisdasycneme
Myotisdaubentonii
Myotisemarginatus
Myotismyotis
Myotisnattereri
Plecotusauritus
Plecotusaustriacus
Rhinolophuseuryale
Rhinolophushipposideros
2.0
1.5
1.0
0.5
0.0
0.5
1.0
NumberofWNSlesionspercm2
(log10)
Figure 5.  Fungal load and number of WNS lesions on bats from Europe (Czech Republic, Latvia, Slovenia)
and Asia (Russia). Fungal load is quantified as P. destructans-specific DNA per cm2
of wing area (established
through quantitative PCR) and number of WNS lesions is counted on the same wing (using UV light trans-
illumination).
Paper 2.4.2 249
www.nature.com/scientificreports/
7Scientific Reports | 6:19829 | DOI: 10.1038/srep19829
diagnostic features ranging from cupping erosions to full-thickness fungal invasion found on all continents. In
our experience, detection of WNS-positive bats was greatly enhanced by the use of a UV lamp combined with
non-lethal punch biopsies of suspected skin lesions48
. The effectiveness of this method enabled us to identify
seven species as WNS-positive in multiple regions, i.e. Myotis brandtii, Myotis dasycneme, Myotis daubentonii,
Myotis emarginatus, M. myotis, Eptesicus nilssonii and Plecotus auritus (Table 1). As predicted31
, M. schreibersii
and R. euryale were confirmed positive for skin lesions. These are thermophilic species that only hibernate in
-8 -6 -4 -2
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
Barbastella barbastellus
Eptesicus nilssonii
Miniopterus schreibersii
Myotis alcathoe
Myotis bechsteinii
Myotis brandtii
Myotis dasycneme
Myotis daubentonii
Myotis emarginatus
Myotis myotis
Myotis nattereri
Plecotus auritus
Plecotus austriacus
Rhinolophus euryale
Rhinolophus hipposideros
Barbastella barbastellus
Eptesicus nilssonii
Miniopterus schreibersii
Myotis alcathoe
Myotis bechsteinii
Myotis brandtii
Myotis dasycneme
Myotis daubentonii
Myotis emarginatus
Myotis myotis
Myotis nattereri
Plecotus auritus
Plecotus austriacus
Rhinolophus euryale
Rhinolophus hipposideros
average per species
P. destructans load per cm2
(log10(µg))
NumberofWNSlesionspercm2
(log10)
Figure 6.  Phylogenetic generalised least-squares for number of WNS lesions, dependent on fungal load
and accounting for within-species variation and relatedness. The dotted line represents the linear regression
without phylogenetic correction.
Barbastellabarbastellus
Eptesicusnilssonii
Miniopterusschreibersii
Myotislucifugus
Myotisbechsteinii
Myotisbrandtii
Myotisdasycneme
Myotisdaubentonii
Myotisemarginatus
Myotismyotis
Myotisnattereri
Plecotusauritus
Rhinolophuseuryale
Rhinolophushipposideros
3.0
3.5
4.0
4.5
Cuppingerosionsize(log10(µm2
))
Palearctic bat species Nearctic
Figure 7.  Variation in focal tissue invasiveness among hibernating bat species infected by
Pseudogymnoascus destructans. A series of 80 periodic acid–Schiff stained sections from each bat’s wing
membrane biopsy were used to determine mean maximum size of WNS-diagnostic cupping erosions
(log10(μ m2
)). See Table 1 for sample sizes.
250 Paper 2.4.2
www.nature.com/scientificreports/
8Scientific Reports | 6:19829 | DOI: 10.1038/srep19829
the northern part of their ranges. Miniopterus schreibersii, a long-distance migrant, forms large mixed colonies
with other cave-dwelling bat species49
and such characteristics make the species another effective candidate for
pathogen dispersal.
Theoretically, WNS spread is limited by conditions prevalent in underground hibernacula being favourable
to P. destructans and presence of susceptible hibernating bats50
. Pseudogymnoascus destructans, being a generalist
pathogen, could infect any bat species hibernating under the right microclimatic conditions and ecological
and evolutionary differences in the hibernating bats would not pose a barrier31
. Distribution ranges of multiple
WNS-positive bat species (E. nilssonii, M. dasycneme, M. daubentonii, M. brandtii, P. auritus) extend across
Palearctic Asia and overlap with their sister species or ecological counterparts (e.g. Myotis petax, Myotis sibiricus).
As bats can form multi-species clusters at hibernation sites, there is a good chance that P. destructans could switch
host species to presently WNS-negative taxa (e.g. Myotis ikonnikovi, M. petax, M. sibiricus). This suggests that
WNS may be present throughout the Palearctic where suitable environmental conditions occur and in all bat
species that utilise such sites. In fact, a paper published ahead of print during review of this article corroborates
our conclusions with one M. petax found positive for WNS in North China51
.
Given the tragic impact of WNS on bat biodiversity in North America24
, it is important to identify the source
of possible introduction4
. However, P. destructans exhibits notoriously low genetic diversity in both the clonal
North American populations33,34,43
and the sexually reproducing populations of Europe44
, effectively thwarting
efforts to pinpoint the source region. Our study further exacerbates the enigma of where the North American
P. destructans strain originates as the North American haplotype was found in all non-mating gene types in
Palearctic Asia, thereby expanding the putative source region (Fig. 3). Additionally, the search for a source region
should shift to markers with a faster mutation rate and better resolution for P. destructans phylogeography, as the
low genetic variability in markers routinely used in fungal population studies indicates a relatively recent origin
or expansion of the Palearctic population.
Theory suggests that if a disease is detectable at high prevalence it is probably mild and unlikely to be a major
problem52
. Two methods were used to establish prevalence in this study: qPCR, used to quantify the pathogen
on wing skin53
, and UV trans-illumination of the wing membrane, which enables detection of fluorescing
lesions associated with WNS skin infection48
. Both methods provided comparable results of high prevalence, and,
together with stable or increasing host population sizes54
, our data strongly suggest endemicity of P. destructans
within Palearctic bat populations. Persistent high prevalence and pathogen load with absence of population
declines in the Palearctic are in sharp contrast to the situation in the Nearctic. In Midwestern United States, high
prevalence of P. destructans infection was followed by a decrease in number of hibernating bats55
. Some form of
host-pathogen equilibrium, analogous to the amphibian chytrid fungus observed in post-decline frog communi-
ties56
, may already have occurred in the Palearctic but not in the Nearctic.
Pathogen load indicates both exposure to the infectious agent and suitability of the host for pathogen replica-
tion57
. Pseudogymnoascus destructans fungal load can be evaluated using swab samples from wing membranes53
.
Although swabs do not sample fungus invasion deep in the wing tissue, our results show that fungal load sampled
in this way is positively correlated with the number of WNS lesions (Fig. 6). Hence, fungal load from swabs may
be assumed to approximate the total load associated with the skin infection. Fungal loads determined across the
Palearctic were similar to those observed in Nearctic species (Supplementary Fig. S1). The swabbing technique
used in the North American studies19,58,59
, however, does not allow standardisation of P. destructans load per cm2
Fungal load per cm (log (µg))
Density
8 6 4 2 0
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
UV positive, n = 255
UV negative, n = 151
Figure 8.  Frequency of Pseudogymnoascus destructans fungal load per cm2
of wing area. Bats were
identified as positive (grey; n =  255) and negative (orange; n =  151) using UV trans-illumination.
Paper 2.4.2 251
www.nature.com/scientificreports/
9Scientific Reports | 6:19829 | DOI: 10.1038/srep19829
and thus detailed statistical comparisons cannot be made. Nevertheless, fungal loads in Nearctic bats are similar
to the observed fungal load of Palearctic bats (Supplementary Fig. S1), even if the former were underestimated
ten-fold with the different swabbing technique.
In the Palearctic study area, fungal load differed regionally (Fig. 4) and between species (Fig. 5) with no
reported mass decline in bat populations attributable to WNS38,54
. Fungal load increased with increasing northing
and westing, sampling day (load increasing later in the year) and in regions where phylogenetically closely-related
hibernating bats predominated (Fig. 4). The driving factor that best modelled the number of WNS lesions
observed using UV trans-illumination in the Palearctic was the fungal load. Increasing fungal load positively
correlated with disease intensity across species diversity indicates that hyphae are more likely to invade deep
tissues and cause WNS lesions with heavy fungal growth on bat wings (Fig. 5). However, with overlap of fungal
loads in UV-negative and UV-positive bats, no clear fungal load threshold defines the pathogen pressure where
development of WNS lesions starts.
With persistent and similar fungal load (Supplementary Fig. S1) and cupping erosion size (Fig. 7) on bats
from the Palearctic and the Nearctic, the difference in population size response is striking. While population
sizes dropped dramatically under the WNS epidemic in the Nearctic24
, population size changes remained within
normal inter-annual fluctuation levels in the Palearctic38,54,60
. Our data suggest that, once the hibernacula are contaminated
by the fungus, Palearctic bats are exposed to high pathogen pressure and the infection is continuously
present to a high level, i.e. the so-called hyperendemic condition. The difference in infection outcome is therefore
a function of adaptation to pathogen pressure.
While reduction of pathogen load is a function of host resistance, the ability to limit the harm caused by a
given load is a result of tolerance61,62
. These two alternative and complementary forms of defence may have profound
effects on the epidemiology of infectious diseases and on host-pathogen coevolution62
. Resistance protects
the host at the expense of the pathogen and tolerance saves the host from harm without direct negative effects
on the infectious agent. Evolution of resistance, therefore, should also reduce the prevalence of the pathogen in
host populations. On the other hand, tolerance is expected to have a neutral, or even positive, effect on prevalence
of the pathogen. Lack of resistance to WNS infection in hibernating European bats has been demonstrated in
13 species31,42
that exhibit stable or increasing population sizes54
. Yet, prevalence of P. destructans-positive bats
reached 100% in the Palearctic. In light of the above arguments, it would appear that mechanisms promoting
tolerance to P. destructans infection are in operation in the area studied. Palearctic bat species tolerate comparable
fungal loads and WNS-lesion size to their Nearctic counterparts suffering population declines. We hypothesise
that the balance between tolerance and resistance mechanisms changes with transition of bats from hibernation
to euthermia.
Our results provide evidence of P. destructans and pathognomonic WNS skin lesions in hibernating bats
sampled from the West Siberian Plain of Russian Asia and indicate endemicity of this virulent fungus in the
Palearctic region. Data suggesting bat tolerance imply lowered risk following establishment of equilibrium in the
host-pathogen interaction. The extensive spatial distribution of the agent may pose a threat, however, representing
a continued source of introduction to other regions with naïve bats not yet exposed to the pathogen. Classical
models employing the disease triangle concept suggest that the epidemiological outcome of an infection depends
on determinants of the pathogen, host(s) and the environment. Alterations in any of these determinants may trigger
shifts in the complex host-pathogen system, as seen here by infection tolerance in hibernating Palearctic bats.
Methods
Material collection.  Between 2012 and 2014, we sampled 481 bats (15 species) at 20 sites in Slovenia, the
Czech Republic, Latvia and Russia (West Siberian Plain, Asia; Table 1, Fig. 3). Samples were taken as late in the
hibernation or as early in the post-hibernation season as possible (February-May) to minimise the impact of
disturbance to specific bat species. Following capture, the wings, ears and muzzle were swabbed with a nylon
(FLOQ Swabs, Copan Flock Technologies srl, Brescia, Italy) or cotton swab (Plain swab sterile plastic applicator,
Copan) in a standardised manner in order to collect fungal biomass from the whole skin area for fungal detection
(dorsal side of left wing only for qPCR). The bats were photographed over a 368 nm wavelength UV lamp and
wing punch biopsies of suspect tissue collected in 10% formalin for histopathological examination. All bats were
then released at the site.
WNS was diagnosed according to current standards47
, i.e. P. destructans presence was confirmed with qPCR53
and selected orange-yellow spots observed over UV were sampled. A series of 80 periodic acid-Schiff stained sections
embedded in paraffin were obtained from each wing membrane biopsy. These were then observed under an
Olympus BX51 light microscope (Olympus Corporation, Tokyo, Japan). Fungal cell walls were stained magenta
under the periodic acid-Schiff stain, which allowed measurement of cupping erosions packed with hyphae. Using
cellSense Software tools (Olympus Soft Imaging, GmbH, Münster, Germany), we measured total area (size) of
cupping erosions. Trans-illuminated photographs of the left wing membrane stretched over a UV lamp were used
to manually enumerate yellow-orange fluorescing pinpoints indicative of WNS lesions48
, using the individual
object counting tool of ImageJ63
.
Bats were considered WNS-positive if qPCR confirmed P. destructans infection, wings exhibited characteristic
UV fluorescence48
and cupping erosions packed with fungal hyphae or a full thickness fungal invasion of the
wing membrane were observed under histopathology47
. Additional swabs taken from the wings and muzzle after
sampling for qPCR were used for cultivation on Sabouraud dextrose agar plates and isolation of the fungus in
pure cultures38
. Representative isolates were deposited at the Culture Collection of Fungi, Charles University in
Prague, Czech Republic.
Field work and sampling in the Czech Republic was performed in accordance with Czech Law No. 114/1992
on Nature and Landscape Protection, based on permits 01662/MK/2012S/00775/MK/2012, 866/JS/2012 and
00356/KK/2008/AOPK issued by the Agency for Nature Conservation and Landscape Protection of the Czech
252 Paper 2.4.2
www.nature.com/scientificreports/
10Scientific Reports | 6:19829 | DOI: 10.1038/srep19829
Republic. Experimental procedures were approved by the Ethical Committee of the Academy of Sciences of the
Czech Republic (No. 169/2011). Sampling in Latvia was approved by the Nature Conservation Agency (No.
3.15/146/2014-N), in Slovenia by the Ministry of Environment and Spatial Planning of the Slovenian Republic,
Slovenian Environment Agency (No. 35601-35/2010-6) and in Russia by the Institute of Plant and Animal
Ecology, Ural Division of the Russian Academy of Sciences (No. 16353–2115/325). The authors were authorised
to handle free-living bats according to the Czech Certificate of Competency (No. CZ01341; §17, Act No. 246/1992
Coll.) and a permit approved by the Latvian Nature Conservation Agency (No. 05/2014).
DNA isolation.  Total DNA was isolated from swabs using the QIAamp DNA Mini Kit (Qiagen, Halden,
Germany). Swabs were placed in tissue lysis buffer with proteinase K and incubated at 56 °C for two hours. Lysis
buffer was added and the samples incubated for a further 10 min. After this step, we followed the Qiagen Buccal
Swab Spin Protocol according to the manufacturer’s recommendations.
Quantitative PCR.  We performed quantitative PCR53
(qPCR) using TaqMan® Universal Master Mix II with
UNG (Life Technologies, Foster City, CA, USA). To optimise the PCR reaction, we added bovine serum albumin
and Platinum® Taq DNA Polymerase (Life Technologies) in final concentrations of 0.05 mg/μ l and 0.025 U/μ l,
respectively. We used both forward and reverse primers (0.3 μ M each) and species-specific and genus-specific
fluorescently-labelled custom probes were used for quantification of the PCR product. The reaction mix was
prepared on ice and three replicates were prepared for each DNA sample. We performed real-time PCR reaction
on a LightCycler 480 PCR platform (Life Technologies) with initial inactivation at 50 °C for 2 min and a hot start
at 96 °C for 10 min. Nine cycles with a denaturation step at 95 °C for 15 sec and annealing at 62 °C for 1 min were
followed by 43 identical cycles with quantification detection. qPCR was finalised by dissociation at 95-60-95 °C
for 15 sec each and cooling to 40 °C for 10 min. DNA isolated from CCF3937 culture38
and water were used as
positive and negative controls and as concentration references for each run.
Fungal load on Palearctic bats.  We calculated a DNA concentration calibration curve using a CCF3937
dilution series. Exact DNA concentrations (ng μ l−1
) in the dilution series were determined using Qubit HS fluorometry
via the manufacturer’s protocol. Effectivity of qPCR was 1.96. We used this dilution series to estimate the
relationship between fungal load and qPCR cycle, calculating P. destructans DNA concentration in the sample
using custom scripts in R64
with the equation log(qPdDNA) =  3.194 – 0.287 Cp (R2
 =  0.9719), where q is the DNA
concentration and Cp the cycle. Each result was converted to fungal load based on the positive control and overall
elution of DNA.
Fungal load on Nearctic bats.  We downloaded previously published P. destructans loads from natural
infections19,58
or recalculated loads from mean Cp values according to the authors’ equation59
. In one case, we
plotted means as individual data points as the authors19
included species means per site (plus standard error) but
fungal loads per individual were not published. The sample sizes for Nearctic bat species sampled within the dates
used in this study were: Corynorhinus rafinesquii (n =  2), Eptesicus fuscus (n =  6), Lasiurus borealis (n =  2), Myotis
grisescens (n =  11), M. leibii (n =  1), M. lucifugus (n =  36), M. septentrionalis (n =  4), M. sodalis (n =  13) and
Perimyotis fuscus (n =  172). None of the North American studies specified sampled area on the bat with sufficient
precision to enable standardisation to cm2
or universal statistical comparison.
Molecular genetic variability of P. destructans.  To assess genetic variability, we used a set of isolates collected
between 2009 and 2014 in the Czech Republic38,65
and isolates from this study. The isolates were characterised
using ITS and five other nuclear gene sequences using previously published primers (Supplementary Table S3).
Chromatograms were assembled to DNA sequences in Geneious 6 (Biomatters, Auckland, New Zealand) and
submitted to the European Nucleotide Archive (LN871244-LN871428). These were checked with blast and P.
destructans was confirmed at all sites with a sequence identity to previously sequenced strains ≥ 99% in all markers.
We then downloaded previously published P. destructans sequences of the respective genes from GenBank.
Sequence haplotypes were identified based on available nucleotide residues, with gaps and unresolved bases
treated as missing data. We estimated relationships between haplotypes using median-joining networks66
.
Individual-based linear mixed models.  We modelled intensity of WNS infection using fungal load on
the dorsal side of the left wing and the number of WNS lesions on the same wing visualised over UV in R64
.
UV-detectable lesions correspond with cupping erosions diagnostic for WNS in Europe and North America,
thereby reflecting disease intensity47,48
. Our previous experience in the Czech Republic has shown that wing damage
tends to be localised and that lesions do not usually merge31
, allowing them to be counted. The yellow-orange
fluorescent spots were counted by a researcher with no knowledge of the qPCR and histopathology results for the
samples. We scaled variables to 1 cm2
of wing area67
and log-transformed them in order to account for body-size
differences between species. We included latitude and longitude, sampling day (from the beginning of the calendar
year) and scores for the first two principal components (characterising dominant hibernating bat community
in each region; Supplementary Fig. S2) as fixed-effect variables and species as a random effect. Microclimate at a
specific bat roost might influence fungal load and disease intensity on an individual through a compound effect
of optimizing growth conditions for the pathogen and hibernation conditions for the host. We used geographic
coordinates as a proxy for possible general site microclimate (mean annual temperature, elevation, humidity)
under the assumption that fungal load and number of WNS lesions would increase at higher latitudes and
with longitudinal shift from oceanic to more continental climate. Similarly, we expected day of sampling to be
reflected in the dependent variables as increased time for fungus propagation might increase its loading. As such,
Paper 2.4.2 253
www.nature.com/scientificreports/
11Scientific Reports | 6:19829 | DOI: 10.1038/srep19829
geographic location, combined with sampling date, can be understood as a proxy for cave microclimate, which
would influence growth of the WNS fungus50,68
.
Both relatedness and species assemblage differences at hibernacula in a given region could skew regional fungal
load and disease intensity in favour of those areas where severely infested species co-occur. We estimated bat
communities present at hibernacula from local surveys. Relatedness of species recorded was assessed using mean
phylogenetic distance and mean nearest taxon distance (both with weighted abundance) and UniFrac metric
measuring the percentage of phylogeny shared by a given community. We used these metrics for principal components
analysis. We used backward stepwise variable selection to develop the model, which was assessed using
the Akaike Information Criterion (AIC).
Species-level phylogenetically-informed model.  Previously published phylogenetic trees for European
bats31
based on multilocus sequence data were rescaled as an ultrametric tree using penalised likelihood69,70
with λ and root height =  1. We used R70,71
to calculate phylogenetic signal in species means and phylogenetic
ANOVA71,72
. We used the phylogenetic generalised least-squares method72,73
to explain the number of visible
WNS lesions on a UV trans-illuminated wing based on P. destructans fungal load.
References
1.	 Daszak, P., Cunningham, A. A. & Hyatt, A. D. Emerging infectious diseases of wildlife - Threats to biodiversity and human health.
Science 287, 443–449 (2000).
2.	 Keesing, F. et al. Impacts of biodiversity on the emergence and transmission of infectious diseases. Nature 468, 647–652 (2010).
3.	 Dobson, A. & Foufopoulos, J. Emerging infectious pathogens of wildlife. Philos. Trans. R. Soc. Lond. B Biol. Sci. 356, 1001–1012
(2001).
4.	 Puechmaille, S. J. et al. White-nose syndrome: is this emerging disease a threat to European bats? Trends Ecol. Evol. 26, 570–576
(2011).
5.	 Fisher, M. C. et al. Emerging fungal threats to animal, plant and ecosystem health. Nature 484, 186–194 (2012).
6.	 Bergman, A. & Casadevall, A. Mammalian endothermy optimally restricts fungi and metabolic costs. mBio 1, e00212–00210 (2010).
7.	 Garcia-Solache, M. A. & Casadevall, A. Global warming will bring new fungal diseases for mammals. mBio 1, e00061–00010 (2010).
8.	 González-González, A. E. et al. Histoplasma capsulatum and Pneumocystis spp. co-infection in wild bats from Argentina, French
Guyana, and Mexico. BMC Microbiology 14, 23 (2014).
9.	 Johara, M. Y. et al. Nipah virus infection in bats (order Chiroptera) in peninsular Malaysia. Emerg. Infect. Dis. 7, 439–441 (2001).
10.	 Hayman, D. T. S. et al. Long-term survival of an urban fruit bat seropositive for Ebola and Lagos bat viruses. PLoS ONE 5, e11978
(2010).
11.	 Middleton, D. J. et al. Experimental Nipah virus infection in Pteropid bats (Pteropus poliocephalus). J. Comp. Pathol. 136, 266–272
(2007).
12.	 Olival, K. J. & Hayman, D. T. S. Filoviruses in bats: Current knowledge and future directions. Viruses 6, 1759–1788 (2014).
13.	 Calisher, C. H., Childs, J. E., Field, H. E., Holmes, K. V. & Schountz, T. Bats: Important reservoir hosts of emerging viruses. Clin.
Microbiol. Rev. 19, 531–545 (2006).
14.	 Moratelli, R. & Calisher, C. H. Bats and zoonotic viruses: can we confidently link bats with emerging deadly viruses? Mem. Inst.
Oswaldo Cruz. 110, 1–22 (2015).
15.	 O’Shea, T. J. et al. Bat flight and zoonotic viruses. Emerg. Infect. Dis. 20, 741–745 (2014).
16.	 Mandl, J. N. et al. Reservoir host immune responses to emerging zoonotic viruses. Cell 160, 20–35 (2015).
17.	 Meteyer, C. U., Barber, D. & Mandl, J. N. Pathology in euthermic bats with white nose syndrome suggests a natural manifestation of
immune reconstitution inflammatory syndrome. Virulence 3, 583–588 (2012).
18.	 Bouma, H. R., Carey, H. V. & Kroese, F. G. M. Hibernation: the immune system at rest? J. Leukoc. Biol. 88, 619–624 (2010).
19.	 Langwig, K. E. et al. Host and pathogen ecology drive the seasonal dynamics of a fungal disease, white-nose syndrome. Proc. R. Soc.
B Biol. Sci. 282, 20142335 (2015).
20.	 Blehert, D. S. et al. Bat white-nose syndrome: An emerging fungal pathogen? Science 323, 227 (2009).
21.	 Gargas, A., Trest, M. T., Christensen, M., Volk, T. J. & Blehert, D. S. Geomyces destructans sp. nov. associated with bat white-nose
syndrome. Mycotaxon 108, 147–154 (2009).
22.	 Minnis, A. M. & Lindner, D. L. Phylogenetic evaluation of Geomyces and allies reveals no close relatives of Pseudogymnoascus
destructans, comb. nov., in bat hibernacula of eastern North America. Fungal Biol. 117, 638–649 (2013).
23.	 Warnecke, L. et al. Inoculation of bats with European Geomyces destructans supports the novel pathogen hypothesis for the origin of
white-nose syndrome. Proc. Nat. Acad. Sci. USA 109, 6999–7003 (2012).
24.	 Frick, W. F. et al. An emerging disease causes regional population collapse of a common North American bat species. Science 329,
679–682 (2010).
25.	 Coleman, J. T. H. & Reichard, J. D. Bat white-nose syndrome in 2014: A brief assessment seven years after discovery of a virulent
fungal pathogen in North America. Outlooks in Pest Management 25, 374–377 (2014).
26.	 Turner, G. G., Reeder, D. M. & Coleman, J. T. H. A five-year assessment of mortality and geographic spread of white-nose syndrome
in North American bats, with a look at the future. Update of white-nose syndrome in bats. Bat Research News 52, 13–27 (2011).
27.	 Lorch, J. M. et al. Experimental infection of bats with Geomyces destructans causes white-nose syndrome. Nature 480, 376–378
(2011).
28.	 Langwig, K. E. et al. Sociality, density-dependence and microclimates determine the persistence of populations suffering from a
novel fungal disease, white-nose syndrome. Ecol. Lett. 15, 1050–1057 (2012).
29.	 Lorch, J. M. et al. Distribution and environmental persistence of the causative agent of white-nose syndrome, Geomyces destructans,
in bat hibernacula of the Eastern United States. Appl. Environ. Microbiol. 79, 1293–1301 (2013).
30.	 Hoyt, J. R. et al. Long-term persistence of Pseudogymnoascus destructans, the causative agent of white-nose syndrome, in the absence
of bats. EcoHealth, doi: 10.1007/s10393-014-0981-4 (2014).
31.	 Zukal, J. et al. White-nose syndrome fungus: A generalist pathogen of hibernating bats. PLoS ONE 9, e97224 (2014).
32.	 Lorch, J. M. et al. A culture-based survey of fungi in soil from bat hibernacula in the eastern United States and its implications for
detection of Geomyces destructans, the causal agent of bat white-nose syndrome. Mycologia 105, 237–252 (2013).
33.	 Ren, P. et al. Clonal spread of Geomyces destructans among bats, midwestern and southern United States. Emerg. Infect. Dis. 18,
883–885 (2012).
34.	 Rajkumar, S. S. et al. Clonal genotype of Geomyces destructans among bats with white nose syndrome, New York, USA. Emerg. Infect.
Dis. 17, 1273–1276 (2011).
35.	 Escobar, L. E., Lira-Noriega, A., Medina-Vogel, G. & Townsend Peterson, A. Potential for spread of the white-nose fungus
(Pseudogymnoascus destructans) in the Americas: use of Maxent and NicheA to assure strict model transference. Geospat. Health 9,
221–229 (2014).
254 Paper 2.4.2
www.nature.com/scientificreports/
12Scientific Reports | 6:19829 | DOI: 10.1038/srep19829
36.	 Maher, S. P. et al. Spread of white-nose syndrome on a network regulated by geography and climate. Nat. Commun. 3, 1306 (2012).
37.	 Maslo, B. & Fefferman, N. H. A case study of bats and white-nose syndrome demonstrating how to model population viability with
evolutionary effects. Conserv. Biol. 29, 1176–1185 (2015).
38.	 Martínková, N. et al. Increasing incidence of Geomyces destructans fungus in bats from the Czech Republic and Slovakia. PLoS ONE
5, e13853 (2010).
39.	 Puechmaille, S. J. et al. Pan-European distribution of white-nose syndrome fungus (Geomyces destructans) not associated with mass
mortality. PLoS ONE 6, e19167 (2011).
40.	 Wibbelt, G. et al. White-nose syndrome fungus (Geomyces destructans) in bats, Europe. Emerg. Infect. Dis. 16, 1237–1243 (2010).
41.	 Pavlinić, I., Đaković, M. & Lojkić, I. Pseudogymnoascus destructans in Croatia confirmed. Eur. J. Wildl. Res. 61, 325–328 (2015).
42.	 Bandouchova, H. et al. Pseudogymnoascus destructans: evidence of virulent skin invasion for bats under natural conditions, Europe.
Transbound. Emerg. Dis. 62, 1–5 (2015).
43.	 Palmer, J. M. et al. Molecular characterization of a heterothallic mating system in Pseudogymnoascus destructans, the fungus causing
white-nose syndrome of bats. G3 (Bethesda) 4, 1755–1763 (2014).
44.	 Leopardi, S., Blake, D. & Puechmaille, S. J. White-nose syndrome fungus introduced from Europe to North America. Curr. Biol. 25,
R217–219 (2015).
45.	 Blehert, D. S. Fungal disease and the developing story of bat white-nose syndrome. PLoS Pathog. 8, e1002779 (2012).
46.	 Lorch, J. M. et al. The fungus Trichophyton redellii sp. nov. causes skin infections that resemble white-nose syndrome of hibernating
bats. J. Wildl. Dis. 51, 36–47 (2015).
47.	 Meteyer, C. U. et al. Histopathologic criteria to confirm white-nose syndrome in bats. J. Vet. Diagn. Invest. 21, 411–414 (2009).
48.	 Turner, G. G. et al. Nonlethal screening of bat-wing skin with the use of ultraviolet fluorescence to detect lesions indicative of whitenose
syndrome. J. Wildl. Dis. 50, 566–573 (2014).
49.	 Hutterer, R., Ivanova, T., Meyer-Cords, C. & Rodrigues, L. Bat Migrations in Europe: A Review of Banding Data and Literature.
(Federal Agency for Nature Conservation, 2005).
50.	 Perry, R. W. A review of factors affecting cave climates for hibernating bats in temperate North America. Environ. Rev. 21, 28–39
(2013).
51.	 Hoyt, J. R. et al. Widespread bat white-nose syndrome fungus, Northeastern China. Emerg. Infect. Dis. 22, doi: 10.3201/
eid2201.151314 (2016).
52.	 McCallum, H. & Dobson, A. Detecting disease and parasite threats to endangered species and ecosystems. Trends Ecol. Evol. 10,
190–194 (1995).
53.	 Shuey, M. M., Drees, K. P., Lindner, D. L., Keim, P. & Foster, J. T. Highly sensitive quantitative PCR for the detection and
differentiation of Pseudogymnoascus destructans and other Pseudogymnoascus species. Appl. Environ. Microbiol. 80, 1726–1731
(2014).
54.	 Van der Meij, T. et al. Return of the bats? A prototype indicator of trends in European bat populations in underground hibernacula.
Mammalian Biology - Zeitschrift für Säugetierkunde 80, 170–177 (2015).
55.	 Langwig, K. E. et al. Invasion dynamics of white-nose syndrome fungus, Midwestern United States, 2012–2014. Emerg. Infect. Dis.
21, doi: 10.3201/eid2106.150123 (2015).
56.	 Retallick, R. W., McCallum, H. & Speare, R. Endemic infection of the amphibian chytrid fungus in a frog community post-decline.
PLoS Biol. 2, e351 (2004).
57.	 Horrocks, N. P., Matson, K. D. & Tieleman, B. I. Pathogen pressure puts immune defense into perspective. Integr. Comp. Biol. 51,
563–576 (2011).
58.	 Bernard, R. F., Foster, J. T., Willcox, E. V., Parise, K. L. & McCracken, G. F. Molecular detection of the causative agent of white-nose
syndrome on Rafinesque’s big-eared bats (Corynorhinus rafinesquii) and two species of migratory bats in the southeastern USA. J.
Wildl. Dis. 51, 519–522 (2015).
59.	 Janicki, A. F. et al. Efficacy of visual surveys for White-nose syndrome at bat hibernacula. PLoS ONE 10, e0133390, doi: 10.1371/
journal.pone.0133390 (2015).
60.	 Frick, W. F. et al. Disease alters macroecological patterns of North American bats. Glob. Ecol. Biogeogr. 24, 741–749, doi: 10.1111/
geb.12290 (2015).
61.	 Medzhitov, R., Schneider, D. S. & Soares, M. P. Disease tolerance as a defense strategy. Science 335, 936–941 (2012).
62.	 Råberg, L., Graham, A. L. & Read, A. F. Decomposing health: tolerance and resistance to parasites in animals. Philos. Trans. R. Soc.
Lond. B Biol. Sci. 364, 37–49 (2009).
63.	 Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. Nat. Meth. 9, 676–682 (2012).
64.	 R., Core Team. R: A language and environment for statistical computing, < http://www.R-project.org/>  (2013).
65.	 Kubátová, A., Koukol, O. & Nováková, A. Geomyces destructans, phenotypic features of some Czech isolates. Czech Mycol. 63, 65–75
(2011).
66.	 Bandelt, H. J., Forster, P. & Röhl, A. Median-joining networks for inferring intraspecific phylogenies. Mol. Biol. Evol. 16, 37–48
(1999).
67.	 Norberg, U. M. & Rayner, J. M. V. Ecological morphology and flight in bats (Mammalia, Chiroptera): Wing adaptations, flight
performance, foraging strategy and echolocation. Philos. Trans. R. Soc. Lond. B Biol. Sci. 316, 335–427 (1987).
68.	 Hallam, T. G. & Federico, P. The panzootic white-nose syndrome: an environmentally constrained disease? Transbound. Emerg. Dis.
59, 269–278 (2012).
69.	 Paradis, E., Claude, J. & Strimmer, K. APE: Analyses of phylogenetics and evolution in R language. Bioinformatics 20, 289–290
(2004).
70.	 Sanderson, M. J. Estimating absolute rates of molecular evolution and divergence times: a penalized likelihood approach. Mol. Biol.
Evol. 19, 101–109 (2002).
71.	 Harmon, L. J., Weir, J. T., Brock, C. D., Glor, R. E. & Challenger, W. GEIGER: Investigating evolutionary radiations. Bioinformatics
24, 129–131 (2008).
72.	 Revell, L. J. phytools: An R package for phylogenetic comparative biology (and other things). Methods Ecol. Evol. 3, 217–223 (2012).
73.	 Ives, A. R., Midford, P. E. & Garland Jr., T. Within-species measurement error in phylogenetic comparative methods. Syst. Biol. 56,
252–270 (2007).
Acknowledgements
This study was supported through a grant from the Czech Science Foundation (Grant No. P506-12-1064). We are
grateful to Tomáš Bartonička, Hana Berková and Masha Orlova for invaluable field assistance, to Matej Dolinay,
Jiří C. Moravec, Patrícia Pečnerová and Aneta Reichová for laboratory assistance and to Gregory G. Turner for
biopsy samples from Nearctic bats.
Paper 2.4.2 255
www.nature.com/scientificreports/
13Scientific Reports | 6:19829 | DOI: 10.1038/srep19829
AuthorContributions
J.Z., J.P. and N.M. designed the study and participated in field research, with help from H.B., J.B., V.K., O.O., P.P.
and J.S. H.B., A.C., A.K., A.N., J.P. and A.Z. performed the laboratory analyses. K.S.J., M.K., J.P. and N.M. analysed
the data. J.Z., J.P. and N.M. wrote the manuscript, with contributions from all authors.
Additional Information
Supplementary information accompanies this paper at http://www.nature.com/srep
Competing financial interests: The authors declare no competing financial interests.
How to cite this article: Zukal, J. et al. White-nose syndrome without borders: Pseudogymnoascus destructans
infection tolerated in Europe and Palearctic Asia but not in North America. Sci. Rep. 6, 19829; doi: 10.1038/
srep19829 (2016).
This work is licensed under a Creative Commons Attribution 4.0 International License. The images
or other third party material in this article are included in the article’s Creative Commons license,
unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license,
users will need to obtain permission from the license holder to reproduce the material. To view a copy of this
license, visit http://creativecommons.org/licenses/by/4.0/
256 Paper 2.4.2
Paper 2.4.3
Pikula J., Amelon S. K., Bandouchova H., Bartonička T., Berkova H., Brichta J., Hooper
S., Kokurewicz T., Kolarik M., Köllner B., Kovacova V., Linhart P., Piacek V., Turner G.
G., Zukal J., Martínková N. 2017. White-nose syndrome pathology grading in Nearctic
and Palearctic bats. PLoS ONE 12: e0180435.
– 257 –
RESEARCH ARTICLE
White-nose syndrome pathology grading in
Nearctic and Palearctic bats
Jiri Pikula1,2
*, Sybill K. Amelon3
, Hana Bandouchova1
, Toma´sˇ Bartonička4
,
Hana Berkova5
, Jiri Brichta1
, Sarah Hooper6
, Tomasz Kokurewicz7
, Miroslav Kolarik8
,
Bernd Ko¨llner9
, Veronika Kovacova1
, Petr Linhart1
, Vladimir Piacek1
, Gregory G. Turner10
,
Jan Zukal4,5
, Nata´lia Martı´nkova´5,11
1 Department of Ecology and Diseases of Game, Fish and Bees, University of Veterinary and
Pharmaceutical Sciences Brno, Brno, Czech Republic, 2 CEITEC—Central European Institute of
Technology, University of Veterinary and Pharmaceutical Sciences Brno, Brno, Czech Republic, 3 United
States Department of Agriculture Forest Service, Northern Research Station, Columbia, Missouri, United
States of America, 4 Department of Botany and Zoology, Masaryk University, Brno, Czech Republic,
5 Institute of Vertebrate Biology, Czech Academy of Sciences, Brno, Czech Republic, 6 Department of
Veterinary Pathobiology, University of Missouri, Columbia, Missouri, United States of America, 7 Institute of
Biology, Department of Vertebrate Ecology and Palaeontology, Wrocław University of Environmental and Life
Sciences, Wrocław, Poland, 8 Laboratory of Fungal Genetics and Metabolism, Institute of Microbiology,
Czech Academy of Sciences, Prague, Czech Republic, 9 Institute of Immunology, Friedrich-Loeffler-Institute,
Federal Research Institute for Animal Health, Greifswald-Insel Riems, Germany, 10 Pennsylvania Game
Commission, Harrisburg, Pennsylvania, United States of America, 11 Institute of Biostatistics and Analyses,
Masaryk University, Brno, Czech Republic
* pikulaj@vfu.cz
Abstract
While white-nose syndrome (WNS) has decimated hibernating bat populations in the Nearctic,
species from the Palearctic appear to cope better with the fungal skin infection causing
WNS. This has encouraged multiple hypotheses on the mechanisms leading to differential
survival of species exposed to the same pathogen. To facilitate intercontinental comparisons,
we proposed a novel pathogenesis-based grading scheme consistent with WNS diagnosis
histopathology criteria. UV light-guided collection was used to obtain single biopsies
from Nearctic and Palearctic bat wing membranes non-lethally. The proposed scheme
scores eleven grades associated with WNS on histopathology. Given weights reflective of
grade severity, the sum of findings from an individual results in weighted cumulative WNS
pathology score. The probability of finding fungal skin colonisation and single, multiple or
confluent cupping erosions increased with increase in Pseudogymnoascus destructans
load. Increasing fungal load mimicked progression of skin infection from epidermal surface
colonisation to deep dermal invasion. Similarly, the number of UV-fluorescent lesions
increased with increasing weighted cumulative WNS pathology score, demonstrating congruence
between WNS-associated tissue damage and extent of UV fluorescence. In a case
report, we demonstrated that UV-fluorescence disappears within two weeks of euthermy.
Change in fluorescence was coupled with a reduction in weighted cumulative WNS pathology
score, whereby both methods lost diagnostic utility. While weighted cumulative WNS
pathology scores were greater in the Nearctic than Palearctic, values for Nearctic bats were
within the range of those for Palearctic species. Accumulation of wing damage probably
influences mortality in affected bats, as demonstrated by a fatal case of Myotis daubentonii
PLOS ONE | https://doi.org/10.1371/journal.pone.0180435 August 2, 2017 1 / 21
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111
OPEN ACCESS
Citation: Pikula J, Amelon SK, Bandouchova H,
Bartonička T, Berkova H, Brichta J, et al. (2017)
White-nose syndrome pathology grading in
Nearctic and Palearctic bats. PLoS ONE 12(8):
e0180435. https://doi.org/10.1371/journal.
pone.0180435
Editor: Sharon Swartz, Brown University, UNITED
STATES
Received: December 2, 2016
Accepted: May 26, 2017
Published: August 2, 2017
Copyright: This is an open access article, free of all
copyright, and may be freely reproduced,
distributed, transmitted, modified, built upon, or
otherwise used by anyone for any lawful purpose.
The work is made available under the Creative
Commons CC0 public domain dedication.
Data Availability Statement: All relevant data are
within the paper and its Supporting Information
files.
Funding: This study was supported by the Czech
Science Foundation (Grant No. P506-12-1064 and
17-20286S). This research was carried out as part
of the CEITEC 2020 project (LQ1601), with further
financial support from the Ministry of Education,
Youth and Sports of the Czech Republic under
National Sustainability Programme II.
258 Paper 2.4.3
with natural WNS infection and healing in Myotis myotis. The proposed semi-quantitative
pathology score provided good agreement between experienced raters, showing it to be a
powerful and widely applicable tool for defining WNS severity.
Introduction
Wildlife conservation medicine is currently being challenged by a number of infectious and
non-infectious diseases [1] that can potentially induce mass mortality events [2]. Recently, the
health of temperate-region bats has been compromised by a generalist fungal agent that causes
white-nose syndrome, Pseudogymnoascus destructans [3–7]. White-nose syndrome (WNS)
emerged as a point-source epidemic. Its geographic spread since 2006 has been associated with
a major decline in Nearctic bat populations [8–10]. On the other hand, Palearctic bat communities
in Europe and Asia appear to tolerate hyperendemic exposure to this virulent pathogen
[11].
WNS is characterised by the invasive skin infection caused by P. destructans [3,5,12,13].
Extensive damage to flight membranes may alter the torpor pattern of hibernating bats by
increasing their arousal frequency and depleting their fat reserves prematurely [14,15]. Disruption
of the effective skin’s barrier function can, therefore, explain the pathophysiological mechanisms
underlying mortality in WNS-affected bats [16–20].
Pathognomonic skin lesions are the only reliable sign of this syndromic disease that are
easy to detect using laboratory methods. In combination with identification of the pathogen
[4,6], therefore, histopathology is seen as the gold standard for diagnosing WNS qualitatively
[13]. While histopathology has indicated equivalent focal skin-tissue invasiveness in multiple
bat species naturally infected with P. destructans throughout its known geographic range
[7,11–13,21], it does not explain the striking difference in infection outcome between Nearctic
and Palearctic bats [11]. Likewise, no significant difference in fungal load and cupping erosion
size has been found on bats in Europe, Palearctic Asia and North America [11]. In order to
better understand WNS progression and severity, quantification of histopathological findings
in bats sampled from different regions is needed.
Until recently, bats had to be dead or euthanised for laboratory testing procedures, collection
of dermato-histopathological samples from the wings, muzzle and ears and to optimise
detection of the disease [13]. What is more, the severity scoring system for WNS currently in
use utilises the whole membrane from one wing [14]. Both in Europe and elsewhere, bat species
are under strict protection; hence, the need for a non-lethal sampling method is imperative.
In response, our team recently validated a new non-lethal technique for identifying and
targeting WNS skin lesions for sampling [22]. Trans-illumination of a wing membrane with
366–385 nm ultraviolet (UV) light elicits a distinct orange-yellow fluorescence that corresponds
directly with the fungal cupping erosions in histological sections of the respective skin
area. The fluorescence emitted from these skin lesions is associated with hyperaccumulation of
riboflavin, a secondary fungal metabolite that may also represent a virulence factor leading to
skin damage [23]. When not being used in combination with infected wing membrane biopsies,
UV transillumination can also be used for non-invasive photographic surveillance of
infection intensity.
Based on the need for a standardised non-lethal tool for measuring skin pathology in bats
from different regions, the objective of the present study was to establish and validate a novel
grading system for defining WNS severity based on single biopsies. Here, we propose a
WNS pathology scoring
PLOS ONE | https://doi.org/10.1371/journal.pone.0180435 August 2, 2017 2 / 21
Competing interests: The authors have declared
that no competing interests exist.
Paper 2.4.3 259
weighted cumulative WNS pathology score based on UV trans-illumination guided biopsies
that allows for semi-quantitative comparison and shows high inter-pathologist reproducibility.
We use this grading system to describe and compare histopathological features of P. destructans
infection in both Nearctic and Palearctic bats previously qualitatively diagnosed with
WNS. While non-lethal diagnostic tools offer the opportunity to follow the progression of skin
pathology, prior experience with WNS pathology scores may be used to predict the outcome
of infection in a diseased bat. With intensive research for treatment of WNS in North America
[24], ability to predict patient prognosis becomes imperative. Lacking sufficient sample sizes
for statistical model evaluation, case report experience might provide valuable information.
We, therefore, used time series data on two bats receiving supportive care in a rehabilitation
facility to document the clinical outcome of WNS and to examine the diagnostic utility of the
proposed grading system in the early post-hibernation period.
Material and methods
Ethics statement
Collection of bat samples from hibernacula in the Czech Republic complied with Czech Law
No. 114/1992 on Nature and Landscape Protection. Collection was based on permits 01662/
MK/2012S/00775/MK/2012, 866/JS/2012 and 00356/KK/2008/AOPK issued by the Agency for
Nature Conservation and Landscape Protection of the Czech Republic. Approval of all experimental
procedures was provided by the Ethical Committee of the Czech Academy of Sciences
(No. 169/2011). Sampling at the “Nietopierek” Natura 2000 site (Poland) was approved by the
II Local Ethical Commission in Wrocław (No. 45/2015). Sampling in Latvia, Slovenia, Russia
and Poland was approved by the Latvian Nature Conservation Agency (No. 3.15/146/2014-N),
the Ministry of Environment and Spatial Planning of the Slovenian Republic, the Slovenian
Environment Agency (No. 35601-35/2010-6), the Institute of Plant and Animal Ecology—Ural
Division of the Russian Academy of Sciences (No. 16353–2115/325) and the Regional Directorate
for Environmental Protection in Gorzow Wielkopolski (No. WPN-I-6205.10.2015.AI).
Collection of samples from bats in Hannibal, MO complied with Missouri Department of
Conservation Scientific Wildlife Collector’s Permit (No. 15947/2014) and was performed
under a protocol approved by University of Missouri, Animal Care and Use Committee. The
authors were authorised to handle wild bats according to the Czech Certificate of Competency
(No. CZ01341; §17, Act No. 246/1992 Coll.) and a permit approved by the Latvian Nature
Conservation Agency (No. 05/2014).
Surveillance of bats for white-nose syndrome
The origin of Holarctic bat samples has been described previously [7,11,12]. Briefly, bats were
sampled at 20 sites in the Czech Republic, Slovenia, Latvia, Poland, Russia and the USA. Additional
Nearctic Myotis septentrionalis males were sampled in Hannibal, Missouri (39.70 N,
91.36 W; USA) from January to March 2015, two hibernation seasons after the first documentation
of P. destructans infection.
As described previously [7,11,12], bat wings were swabbed for laboratory examination of P.
destructans infection using culture and estimation of associated fungal load using quantitative
polymerase chain reaction (qPCR), and photographed over a 368 nm UV lamp for later enumeration
of lesions indicative of WNS [22]. One WNS-suspect spot from each bat was then
sampled under UV guidance. In this way, we collected wing membrane biopsies from 210
specimens of 21 different Nearctic and Palearctic bat species during the late hibernation and
early post-hibernation periods of 2012 to 2015 (S1 Table). All bats were handled in such a way
as to minimise any impact from disturbance and then quickly released at the capture site. Skin
WNS pathology scoring
PLOS ONE | https://doi.org/10.1371/journal.pone.0180435 August 2, 2017 3 / 21
260 Paper 2.4.3
samples collected with a 4 mm sterile punch (Kruuse, Denmark) were immediately fixed in
10% formalin, then dehydrated in the laboratory and embedded in paraffin. To obtain a representative
range of histopathology findings, a series of 80 5 μm sections were prepared from
each wing membrane biopsy and stained with periodic acid-Schiff (PAS). The slides were
examined with light microscopy with focus on invasive fungal growth and identification of the
WNS skin pathology grades described below.
Case reports of WNS progression
In the Czech Republic, single specimens of M. daubentonii and M. myotis with extensive WNS
infection were recognised at hibernacula in the Podyjı´ National Park and Jesenı´ky Mountains
and sent to the Rescue Centre at the University of Veterinary and Pharmaceutical Sciences
Brno (Czech Republic). While healing was being augmented with captive nutritional support,
skin disease progression under euthermic conditions was documented using photography and
histopathology.
A novel grading system of WNS skin pathology
The semi-quantitative grading system presented here is based on a single bat flight membrane
biopsy per animal guided by UV transillumination. Eleven binary encoded grades g, pathognomonic
for WNS [13] or associated with the early stages of P. destructans skin infection, were
selected as index signs for grading and were identified on Periodic acid-Schiff (PAS)-stained
histopathological slides. The grades were ordered with increasing invasion severity from presence
of the fungus on the wing’s surface to replacement of tissue by fungal hyphae throughout
the wing’s thickness. Skin surface colonisation by the fungus (g1) was followed by hair follicle
infection (g2), sebaceous gland infection (g3), single occurrences of cup-like lesions (g4), multiple
and/or confluent cupping erosions (g5), skin basement membrane breach by the fungus
(g6) and full-thickness invasion (g7). The bat’s immune response to infection was represented
by variable inflammatory response manifested as tissue infiltration with neutrophils (g8) and
fungal sequestration by extensive inflammatory response (g11). Findings of skin necrosis (g9),
characterised by loss of identifiable skin structures and skin infarction (g10), were also included
among the criteria associated with P. destructans skin infection.
In order to evaluate infection severity, we assign a weight w to each grade based on its invasiveness
and extensiveness. The weights were selected so that a cumulative score of lower
grade findings could not outweigh more severe grades of skin infection. A combination of fungal
skin colonisation (w1 = 1) and hair follicle (w2 = 2) and sebaceous gland infection (w3 = 2)
represent the least severe finding, with a cumulative score lower than that in samples with single
(w4 = 6) or multiple (w5 = 12) cupping erosions. Basement membrane breach (w6 = 13) and
full thickness infection (w7 = 19) represent severe disruptions of the deep layers of bat wing
membranes. We consider inflammation (w8 = 20), necrosis (w9 = 25) and infarction (w10 = 30)
to be the most severe WNS grades. When the immune response succeeds in fungal sequestration
(w11 = −20 to reflect potential healing), however, recovery from WNS appears more likely.
Inspection of 80 wing membrane sections from each biopsy could result in a weighted
cumulative WNS pathology score defined as:
histoSum ¼
X11
i¼1
giwi;
where g 2 {0,1}, with g = 0 meaning absence and g = 1 meaning presence of the respective
WNS pathology grade. The histoSum values may range from −20 (binary code corresponding
WNS pathology scoring
PLOS ONE | https://doi.org/10.1371/journal.pone.0180435 August 2, 2017 4 / 21
Paper 2.4.3 261
to the listed grades: 00000000001) to 130 (11111111110). As an example, a sample scored 32 if
the biopsy contained fungal skin colonisation (+1) together with multiple cupping erosions
(+12) and hyphae that breached the basal membrane (+13). Note that a single cupping erosion
(+6) is scored automatically when multiple or confluent cupping erosions are present.
Data used in this study is available in S2 Table.
Inter-pathologist reproducibility of the proposed grading system
In order to test whether the above-described index signs for WNS grading were universally
recognizable, we photographed 30 randomly selected PAS-stained sections prepared from batwing
biopsies. In a blind evaluation, five independent pathologists examined and scored the
photographs. Clarity of grading in naïve raters was evaluated in an experiment whereby 72
undergraduate students from the University of Veterinary and Pharmaceutical Sciences Brno
scored a subset of 10 photographs following 45-minutes training in WNS histopathology. The
participating students of veterinary medicine already received education in general pathology
during their veterinary study program and volunteered for the task of scoring P. destructans
skin infection pathology. These volunteers were recruited at a Wildlife Diseases lecture.
Assignment to three evaluating groups was performed to limit the time that the volunteers
spent reading WNS histopathology.
Estimation of fungal load
Fungal DNA was isolated from swabs of the dorsal side of bat left wing using QIAamp DNA
Mini Kit (Qiagen, Halden, Germany). DNA from P. destructans was quantified with a dualprobe
TaqMan (Life Technologies, Foster City, CA, USA) assay developed by Shuey et al. [25],
following the detailed protocol for the qPCR of Zukal et al. [11]. Each sample was run in triplicate
and each plate included positive and negative controls. Fungal load was estimated from
positive control dilution series calibration curve according to equation log (P. destructans
DNA concentration) = 3.194–0.287 Ã
cycle, and standardized to total fungal load in nanograms
per 1 cm2
of wing area to correct for species size differences [11].
Statistical analyses
Weighted cumulative WNS grading scores for the Palearctic and Nearctic zones were compared
using the Mann-Whitney U test. The influence of species-specific histoSum was assessed
by filtering the value for species random effect using the lme4 package in R [26,27].
The relationship between data provided by non-destructive diagnostic methods and initial
progression of WNS infection by histopathology was evaluated using logistic regression [27].
Two sets of three logistic regressions were fitted for skin surface colonisation by the fungus,
formation of single and multiple cupping erosions. Fungal load (log10(ng)) and number of
UV-fluorescent spots on the left wing recalculated to cm2
of wing membrane [11] were used as
independent variables in the respective models. Wing membrane area for M. septentrionalis
was not available in literature and thus it was calculated from UV photographs using a custom
R script and the splancs and jpeg packages [28,29]. The relationship between histoSum and
number of UV fluorescent spots was evaluated using phylogenetic generalised least-squares
[30] in the phytools package in R [31]. Previously published multilocus phylogeny was used to
correct for species relatedness [7].
Inter-rater agreement of pathologists and veterinary students was tested in a paired study
using the unweighted Cohen’s κ coefficient [32] against scores included in the results below,
where confidence was estimated with equations from [33]. The overall evaluation of κ for multiple
raters was used according to Fleiss et al.’s modification [34]. The κ coefficient corrects
WNS pathology scoring
PLOS ONE | https://doi.org/10.1371/journal.pone.0180435 August 2, 2017 5 / 21
262 Paper 2.4.3
grading category agreement frequency for an effect where raters achieve consensus by chance.
We used custom R scripts along with the irr package [35]. Confidence intervals for multi-rater
agreement were estimated from 1000 bootstrap replicates.
Results
Prevalence of each WNS pathology grade (Figs 1 and 2) induced by natural P. destructans skin
infection in Palearctic and Nearctic bat species varied from 0 to 100% in different bat species
(Table 1). Signs of fungal skin-surface colonisation not classified as WNS (in the absence of
other lesions) were present in all Nearctic bats and in 89% of Palearctic bats. Prevalence of
infection within hair follicles and associated glands was highly variable in all hibernating bat
species and for species with multiple investigated individuals ranged from 17 to 67% for the
hair follicle infection and from 15 to 67% for the sebaceous gland infection. While wing membrane
infection progressed to multiple and/or confluent cupping erosions in most bat species,
single cupping erosions only were observed in Miniopterus schreibersii (n = 1), M. bechsteinii
(n = 3) and Rhinolophus euryale (n = 1). Fungal hyphae regularly breached the epidermal/dermal
interface in M. lucifugus (100%, n = 10) and M. septentrionalis (86%, n = 7) as well as in
Palearctic bats (46–100%). Full thickness invasion of the wing membrane occurred more frequently
in Palearctic bats (34% on average) and M. septentrionalis (71%) than in M. lucifugus
(30%). A high percentage of both Palearctic and Nearctic bats (73 and 76% on average, respectively)
showed an inflammatory response to fungal invasion, though this mostly resulted in a
considerably lower occurrence of fungal sequestration (25 and 22% on average, respectively;
Table 1). While Eptesicus serotinus (n = 1), M. alcathoe (n = 7), Nyctalus noctula (n = 8), Pipistrellus
pipistrellus (n = 2), Pipistrellus pygmaeus (n = 2) and Plecotus austriacus (n = 1) were also
sampled for histopathology, all were confirmed negative for WNS. Among these, four M.
alcathoe were confirmed positive for P. destructans infection using qPCR (S2 Table).
Sum of scores and inter-pathologist agreement in the proposed grading
system
While the histoSum score can reach a theoretical maximum of 130, the maximum score in
practice was 126 for a fatal case of WNS in the M. daubentonii specimen described below
(binary code 10011111011). The species with highest average histoSum were Eptesicus nilssonii
from Asia (n = 2), M. brandtii (n = 1), Rhinolophus hipposideros (n = 4), Barbastella barbastellus
(n = 3) from Europe and M. septentrionalis from North America (n = 7; Fig 3). While the mean
histoSum for all sampled bats was 39.2 (std. error = 2.74), E. nilssonii (n = 2), M. schreibersii
(n = 2), M. bechsteinii (n = 7), M. dasycneme (n = 6), M. myotis (n = 57), M. nattereri (n = 8)
and R. euryale (n = 1) from Europe had a lower average weighted sum of concurrent findings
(Fig 3). Nearctic bats had a significantly higher histoSum score (mean ± std. error: 64.9 ± 6.4)
than Palearctic bats (35.9 ± 2.9; Mann-Whitney U = 1688, p < 0.001). Correction for random
effect of species had no influence on significance.
Paired scoring agreement corrected for chance between five experienced pathologists and
results presented herein ranged from 0.76 to 0.85 (Cohen’s κ, p < 0.001; 95% confidence intervals
(CI): {[0.75, 0.87], [0.8, 0.91], [0.72, 0.85], [0.77, 0.89], [0.7, 0.83]}) with an overall Fleiss’
κ5 of 0.78 (z = 57.49, p < 0.001, CI: [0.74, 0.82]), signifying good agreement in scoring WNS
severity. Inter-rater agreement evaluating each WNS pathology grade separately differed, with
basement membrane breach and inflammation being most difficult to score consistently
(Table 2). Naïve raters displayed lower inter-rater agreement with the reported scores (Cohen’s
κ 2 [0.5, 0.92]) and lower within-group agreement (Fleiss’ κ 2 [0.64, 0.71]).
WNS pathology scoring
PLOS ONE | https://doi.org/10.1371/journal.pone.0180435 August 2, 2017 6 / 21
Paper 2.4.3 263
Table1.PrevalenceofhistopathologyseveritygradesassociatedwithnaturalskininfectionbyPseudogymnoascusdestructans.Prevalencewascalculatedasthepercentage
ofpositivebiopsiesatthegivengradeinthedatasetofallbatspositiveforWNSbasedonhistopathology.Allinfectiongradesapplytothewingmembraneand,asidefromfungalskinsur-
facecolonisationintheabsenceofanyotherfindings,areclassifiedaswhite-nosesyndrome(WNS).Tested=numberofindividualsexaminedforhistopathology.WNShisto+=number
ofindividualsconfirmedpositiveforWNS.Std.error(%)=100
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pð1ÀpÞ=n
p
,wherenistheoverallnumberofscoredgradesforthegivenspeciesandpistheproportionofpositivelyscored
grades.TherespectiveseveritygradesareshowninFigs1and2.NumbersinbracketsrepresentWNSpathologygradeweighting—seetextfordetails.
Battaxon
data
WNS
histo
+
Std.
error
(%)
Prevalence(%)
Fungalskin
colonisation
(1)
Hair
follicle
infection
(2)
Sebaceous
gland
infection(2)
Single
cupping
erosion
(6)
Multiple
cupping
erosions
(12)
Basement
membrane
breach(13)
Full
thickness
infection
(19)
Inflammation
(20)
Fungal
sequestration
(−20)
Necrosis
(25)
Infarction
(30)
Barbastella
barbastellus
38.671000010010010066.6766.67066.670
Eptesicus
nilssonii
37.7510066.6766.6710010010066.6710033.3366.670
Miniopterus
schreibersii
114.51001001001000100010001000
Myotis
bechsteinii
38.2166.6733.330100066.67066.67033.330
Myotisbrandtii18.671001001001001001001001001001000
Myotis
dasycneme
134.1810030.7715.3810038.4661.5423.0884.6230.7738.460
Myotis
daubentonii
104.7690302010070705050107010
Myotis
emarginatus
66.1510016.6716.6710083.3366.67505016.67500
Myotismyotis502.0996463610046462346240
Myotis
nattereri
47.0210025251002550025000
Plecotus
auritus
75.6910042.8628.5710071.4371.4328.5771.4328.5742.860
Rhinolophus
euryale
114.50001000100010010000
Rhinolophus
hipposideros
47.25100505010075100501000750
Total/meanin
Palearctic
bats
10688.6741.6435.2510054.5679.4133.6172.9525.0351.310.77
Std.errorin
Palearctic
bats
2.064.754.460.04.854.753.874.854.710.943.29
Myotis
lucifugus
104.71003030100100100308030400
Myotis
septentrionalis
75.6210014.2914.2910010085.7171.4371.4314.2971.430
Total/meanin
Nearcticbats
1710022.1422.1410010092.8650.7175.7222.1455.720
Std.errorin
Nearcticbats
0.010.2910.290.00.05.7112.1110.2912.110.010.29
https://doi.org/10.1371/journal.pone.0180435.t001
WNS pathology scoring
PLOS ONE | https://doi.org/10.1371/journal.pone.0180435 August 2, 2017 7 / 21
264 Paper 2.4.3
Fig 1. Histopathology grades induced by natural Pseudogymnoascus destructans skin infection in Holarctic bats.
(A) Myotis myotis: fungal skin-surface colonisation with aerial hyphae developing conidia (g1, black arrow), not classified as
WNS pathology scoring
PLOS ONE | https://doi.org/10.1371/journal.pone.0180435 August 2, 2017 8 / 21
Paper 2.4.3 265
Progression of severe WNS skin infection resulting in extensive skin
necrosis and fatality in a Myotis daubentonii bat
Progression of severe WNS lesions on the wing membranes of a M. daubentonii specimen was
investigated at the University of Veterinary and Pharmaceutical Sciences Brno (Czech Republic)
Rescue Centre. A time series of UV transillumination and daylight photography images
spanning seven days from capture to death indicated that the wing membranes started to show
signs of dry necrosis within two days (Fig 4). Flight membrane areas with extensive infection
lost tone, elasticity and sheen as they contracted and tore around the WNS lesions in a proximal-to-distal
pattern. Histopathology demonstrated wing membrane necrosis associated with
confluent cupping-erosion from fungal infection, distended blood vessels with intraluminal
neutrophilic infiltration and haemorrhagic infarct consisting of red blood cell and fibrin clots
caused by blood flow obstruction (Fig 2).
Healing time series of WNS lesions in a Myotis myotis bat
A M. myotis bat captured at the end of the hibernation period was diagnosed with pathognomonic
WNS skin lesions using UV transillumination (Fig 5A) and histopathology (Fig 1B).
The fluorescent yellow-orange WNS lesions disappeared after two weeks at euthermy (Fig 5).
Healing progressed until cupping erosion-like structures packed with fungal hyphae were contained
within a scab covering the repaired wing membrane (Fig 6). Though this specimen had
a histoSum score of 31 (11110001000) on the day of capture, the score had decreased to -20
(00000000001) after 15 days of healing at euthermy.
WNS in absence of other findings; (B) M. myotis: a single cupping erosion (g4, black arrow) eroding to the epidermal/dermal
interface; (C) M. myotis: three confluent cupping erosions (g5, black arrows); (D) M. daubentonii: necrotic wing membrane
(witnessed as loss of dermal tissue stainability; g9, white asterisk) next to multiple cupping erosions packed with P.
destructans hyphae (g5, black arrows) that also breached the basement membrane (g6, white arrowheads); (E) M.
daubentonii: infection of hair follicle (g2, black arrow) and associated glands (g3, black arrowhead). Surface skin colonisation
(g1), multiple cupping erosions (g5), inflammatory cells (g8) and necrotic tissue (g9, white asterisk) are also present in the
section; (F) M. dasycneme: an outline of a cupping erosion (g4, black arrow) clearly visible together with full thickness fungal
invasion (g7) replacing the necrotic wing membrane (g9) and sporadic neutrophils (g8, white asterisk); (G) M. lucifugus:
marked inflammatory response (g8, black asterisks) to fungal invasion of several cupping erosions (g5, black arrows) on both
sides of the wing membrane; (H) M. dasycneme: fungal sequestration (g11, black arrow) with neutrophils (g8, black asterisk)
from the wing membrane. Periodic acid-Schiff stain. Scale bar– 50 μm.
https://doi.org/10.1371/journal.pone.0180435.g001
Fig 2. Skin infarction and necrosis associated with progressive white-nose syndrome lesions in Myotis daubentonii. Samples for histopathology
were collected on Day 7 of the time series documented using UV and daylight photography. Extensive Pseudogymnoascus destructans infection of the wing
membrane produced confluent cupping erosions (g5, black arrows) resulting in skin necrosis (g9, white asterisk), characterised as loss of identifiable skin
structures (A). Intraluminal neutrophilic infiltration (g8, black asterisk) in distended blood vessels was associated with the compromised wing membrane (B).
Other skin lesions in this bat included haemorrhagic infarcts (g10, black arrowhead) with stagnant blood and skin necrosis (g9, white asterisk) (C). Periodic
acid-Schiff stain.
https://doi.org/10.1371/journal.pone.0180435.g002
WNS pathology scoring
PLOS ONE | https://doi.org/10.1371/journal.pone.0180435 August 2, 2017 9 / 21
266 Paper 2.4.3
Weighted cumulative WNS pathology score
Europe
Eptesicus nilssonii
Myotis dasycneme
Plecotus auritus
Barbastella barbastellus
Eptesicus nilssonii
Miniopterus schreibersii
Myotis bechsteinii
Myotis brandtii
Myotis dasycneme
Myotis daubentonii
Myotis emarginatus
Myotis myotis
Myotis nattereri
Plecotus auritus
Rhinolophus euryale
Rhinolophus hipposideros
Myotis lucifugus
Myotis septentrionalis
0 10 20 30 40 50 60 70 80 90 100 110 120
AsiaNorthAmerica
Fig 3. Species-specific weighted cumulative white-nose syndrome pathology score (histoSum). Average sum of
weighted qualitative scoring for white-nose syndrome severity grades displayed in Figs 1 and 2 (± std. error). Animals with
WNS pathology scoring
PLOS ONE | https://doi.org/10.1371/journal.pone.0180435 August 2, 2017 10 / 21
Paper 2.4.3 267
Functional analysis of WNS pathology scores
Logistic regression (Fig 7A) revealed that the odds of observing surface skin colonization on
histopathology increase with increase in fungal load estimated from wing swabs of Palearctic
bats (odds ratio: OR = 2.1, CI: 1.6–2.8, p < 0.001; fungal load was not available for histologically
examined Nearctic bats, S2 Table). The OR for observing single cupping erosions dependent
on fungal load is lower, but significantly different from 1 (OR = 2.0, CI: 1.5–2.6,
p < 0.001), indicating that single cupping erosions occur in response to roughly one-order-ofmagnitude
higher fungal loads than that during fungal skin-surface colonisation (Fig 7). In
Palearctic bats, wing membrane infection progressed to multiple and/or confluent cupping
erosions with fungal loads randomly (OR = 1.3, CI: 1–1.6, p = 0.54). Investigating both Palearctic
and Nearctic bats (S2 Table), a similar pattern was observed in odds of recognizing surface
skin colonization (OR = 10.2, CI: 4.7–22.2, p < 0.001) and single cupping erosions (OR = 6.6,
CI: 3.5–12.5, p < 0.001) on histopathology dependent on the number of UV fluorescent lesions
(Fig 7B). The number of UV fluorescent lesions was a better predictor for multiple and/or confluent
cupping erosions with OR = 4.3 (CI: 2.6–7.2, p < 0.001) than fungal load.
Cumulative WNS pathology scores from single 4 mm biopsies increased with the number
of UV fluorescent lesions observed on the surface of the whole wing (β0 = 62.1, β1 = 16.41; Fig
8), indicating that species with more extensive UV fluorescence tend toward multiple concurrent
WNS findings on histopathology.
Discussion
Fuelled by no reports of mass die-offs from Europe [36,37], the most erroneous conclusion concerning
white-nose syndrome is the belief that P. destructans infection causes no harm to Palearctic
bats. Pseudogymnoascus destructans infection is known to induce complex physiological
histoSum = 1 not classified as positive for WNS on histopathology are included in the figure. Species sampled on multiple
continents are presented separately. See Table 1 for sample sizes.
https://doi.org/10.1371/journal.pone.0180435.g003
Table 2. Inter-rater agreement in scoring white-nose syndrome (WNS) pathology according to Fleiss’
κ. Experienced pathologists (n = 5) scored 30 photographs of randomly drawn histopathology slides, each
group of naïve raters (n 2{27, 20, 25}) scored a subset of 10 photographs. Standard errors for the given sample
sizes were 0.047, 0.017, 0.023 and 0.018, respectively. Negative κ values indicate no inter-rater
agreement.
Grade
index
WNS pathology grade Fleiss’ κ
experienced naïve group 1 naïve group 2 naïve group 3
1 Fungal skin colonisation 0.671 0.560 0.114 0.014
2 Hair follicle infection 0.931 0.690 0.772 0.850
3 Sebaceous gland infection 0.712 0.074 0.575 0.839
4 Single cupping erosion 0.790 0.521 0.200 0.655
5 Multiple cupping erosions 0.861 0.716 0.445 0.596
6 Basement membrane
breach
0.496 0.592 0.237 0.285
7 Full thickness infection 0.704 0.655 0.330 0.809
8 Inflammation 0.469 0.436 0.368 0.392
9 Skin necrosis 0.460 0.293 0.414 0.288
10 Skin infarction 0.851 -0.007 0.415 -0.008
11 Fungal sequestration 0.805 0.755 -0.032 0.159
https://doi.org/10.1371/journal.pone.0180435.t002
WNS pathology scoring
PLOS ONE | https://doi.org/10.1371/journal.pone.0180435 August 2, 2017 11 / 21
268 Paper 2.4.3
Fig 4. Time series of a Myotis daubentonii wing showing progression of white-nose syndrome lesions to fatality.
Extensive white-nose syndrome infection was recognised on a M. daubentonii bat at a hibernaculum in the Podyjı´ National
Park (Czech Republic). The bat was kept under euthermic conditions at a rescue centre, fed ad libitum and supplied with
drinking water. The wing was extended over a Wood’s lamp at 366 nm wavelength and photographed in a darkroom. (A) =
Day 0, (B) = Day 1, (C) = Day 2, (D) = Day 3, (E) = Day 5, (F) = Day 6, (G) = Day 7, white arrows indicate biopsy punch sites
(results presented in Fig 5). Day 7 was also documented using daylight photography (H). Scale bar = 1 cm. This time series
spanned seven days from capture at the hibernaculum to death in the rescue centre. Wing membrane areas with extensive
Pseudogymnoascus destructans infection became dry and necrotic within two days of euthermy, whereupon they contracted
WNS pathology scoring
PLOS ONE | https://doi.org/10.1371/journal.pone.0180435 August 2, 2017 12 / 21
Paper 2.4.3 269
and transcriptional effects in hibernating bats long before the onset of the clinical signs indicative
of late-stage WNS [19,38]. Palearctic and Nearctic bats are exposed to similar fungal loads,
resulting in equivalent focal skin-tissue invasiveness pathognomonic for WNS lesions [7,11–
13,39,40]. It would be reasonable, therefore, to expect both morbidity effects (not always recognisable
in the field) and mortality across the distribution range of P. destructans. Here, we show
the full range of WNS skin pathology in Palearctic and Nearctic bats. Moreover, we document a
case of P. destructans infection in a European M. daubentonii bat progressing to fatality due to
extensive skin necrosis and infarction. The differential outcome of WNS in Eurasia and North
America appears to be associated with some form of tolerance mechanism [11]. Our data, however,
suggests that Palearctic bat tolerance to WNS infection may be limited by severity of wing
damage.
In this paper, we presented a novel pathogenesis-based grading system for P. destructans
skin infection that utilises non-lethal biopsy sampling to examine the WNS lesions resulting
from host-pathogen interaction. The severity grades chosen are simple to differentiate on histopathology
and show good agreement amongst experienced raters. Identification of inflammation
and skin necrosis exhibited lowest agreement and requires careful consideration in
multi-rater comparisons. Naïve, minimally trained, raters showed similar agreement with
scoring most grades as did experienced raters, but fungal skin colonization, sebaceous gland
infection, skin infarction and fungal sequestration were most challenging grades for this group
(Table 2). Generally speaking, histopathology enables one to assign nominal diagnostic categories
to medical conditions observed in organs and tissues. Grading may provide additional
and tore around the white-nose syndrome lesions in a proximal-to-distal pattern. The animal displayed loss of skin tone,
elasticity and surface sheen, and ceased eating one day prior to death.
https://doi.org/10.1371/journal.pone.0180435.g004
Fig 5. UV transillumination time series of a Myotis myotis wing with decrease in fluorescence corresponding to white-nose syndrome
lesions over time. The bat was captured at the end of the hibernation period, kept in captivity at euthermy, fed ad libitum with cockroaches and
mealworms and supplied with drinking water, and released after the white-nose syndrome lesions had healed. The wing was extended over a Wood’s
lamp at 366 nm wavelength and photographed in a darkroom. A = Day 0, B = Day 3, C = Day 5, D = Day 7, E = Day 11, F = Day 15. The top-left corner
matches in each image, wing deformation is due to variable handling of the live animal during sampling. Scale bar = 1 cm.
https://doi.org/10.1371/journal.pone.0180435.g005
WNS pathology scoring
PLOS ONE | https://doi.org/10.1371/journal.pone.0180435 August 2, 2017 13 / 21
270 Paper 2.4.3
data allowing one to estimate severity, predict biological behaviour of the disease, predict survival
rate and make decisions regarding patient prognosis and treatment [41]. Recent studies
linking histopathology WNS severity scores with frequent arousal from hibernation and mortality
[14] and altered physiological homeostasis [18–20,42] have used destructive methods for
examining bats. While such methods for examining wing membrane damage score WNS
severity only at the moment of death or euthanasia [14,18–20], our grading system allows one
to follow temporal patterns of disease progression with consecutive multiple biopsies targeted
with UV transillumination. UV fluorescence-guided non-lethal biopsy diagnostics of WNS
skin showed 95.5% sensitivity and 100% specificity [22]. Collection of additional biopsies from
each specimen may further increase diagnostic and grading sensitivity.
Direct comparison between histological WNS severity scoring from the whole wing membrane,
which provides a reasonable representation of skin damage associated with P. destructans
infection [14,20], and our proposed single non-lethal biopsy method is complicated.
Instead, we suggest that the number of bats positive for WNS lesions on histopathology
(n = 123) and the number of taxa (n = 15) from the Palearctic and Nearctic regions examined,
together with the non-lethal nature of the methodology, make this grading system universally
applicable to all species exposed to the infection. The ability to select a biopsy site on the bat’s
wing displaying the most representative and extensive UV fluorescence [22] enables a reasonable
assessment of pathology while reducing the impact of WNS surveillance on the bat.
Previous WNS grading schemes have proposed four or five WNS-positive grades [14,20],
with presence, extent and distribution of skin lesions densely packed with fungal hyphae (cupping
erosions) used as the main criteria for assigning whole-wing-based scores [14]. As an
alternative, we propose a finer, semi-quantitative 11-grade scale. Analogous quantitative measures
of P. destructans skin infection in this alternative grading system include presence of
Fig 6. Healing of a white-nose syndrome lesion in Myotis myotis. Samples for histopathology were
collected on Day 15 of the time series described in Fig 5. A cupping erosion-like structure (black arrow)
packed with fungal hyphae within a scab covering the healing wing membrane of M. myotis. Periodic acidSchiff
stain.
https://doi.org/10.1371/journal.pone.0180435.g006
WNS pathology scoring
PLOS ONE | https://doi.org/10.1371/journal.pone.0180435 August 2, 2017 14 / 21
Paper 2.4.3 271
single, multiple and/or confluent cupping erosions within a 4 mm punch biopsy that can be
evaluated through image analysis to address invasiveness and size [11], then approximated for
the whole wing from UV photographs [22]. Expanding the previous systems towards more
severe pathology means that invasive fungal growth penetrating the full-thickness of the wing
membrane represents a higher grade of skin infection than cup-like fungal structures eroding
to the epidermal/dermal interface only [12,13]. In addition to cupping erosions, the ability of
P. destructans to invade the dermis and the full thickness of the wing membrane distinguishes
WNS from other dermatophytic infections.
As dysregulated immunity may contribute to severe post-emergent pathology [43], we also
included bat immune response to infection (inflammation and fungal sequestration) as index
criteria in our grading system. Inflammatory responses were frequently associated with fungal
invasion in both Palearctic and Nearctic bats, probably due to sampling in the late-hibernation
or early post-hibernation periods. However, the ability of bats to contain the infection through
sequestration of the fungus with neutrophils and skin debris was only observed in a quarter of
bats showing an inflammatory response. Warnecke et al. [20] scored bat tissue replaced by the
fungus as necrotic. Unlike their indirect indication, we considered skin necrosis as findings
characterised by loss of staining pattern, indistinct outlines of individual bat tissue cells and
hypereosinophilia (Figs 1D and 2A). Skin infarction, which has previously been recognised in
association with WNS [16], accounted for the most severe wing membrane damage (Fig 2C).
In future studies, therefore, we urge that WNS severity be assessed with respect to skin infarction,
as wing membrane damage in such cases extended from WNS skin lesions to terminal
wing areas affected with blood flow obstruction (cf. Fig 4).
Skincolonization/single/multiplecuppingerosions
7 6 5 4 3 2 1
0.0
0.2
0.4
0.6
0.8
1.0
Fungal load per cm2
(log10(ng))
A B
2 1 0 1 2
0.0
0.2
0.4
0.6
0.8
1.0
Number of UV fluorescent spots per cm2
(log10)
Fig 7. Relationship between data from non-destructive diagnostic methods and selected histopathology severity grades. Logistic regressions of
skin surface colonisation by the fungus (black circle, dotted line; coefficients: (A) β0 = 3.38, β1 = 0.74, (B) β0 = 2.45, β1 = 2.32), single cupping erosion
(orange cross, dashed line; coefficients: (A) β0 = 2.60, β1 = 0.68, (B) β0 = 1.49, β1 = 1.88) and multiple cupping erosions (red star, solid line; coefficients: (A)
β0 = -0.09, β1 = 0.24, (B) β0 = -0.06, β1 = 1.44) dependent on fungal load detected on qPCR in Palearctic bats (A) or number of UV fluorescent spots in
Holarctic bats (B). Shaded area represents 95% confidence interval on predicted probability.
https://doi.org/10.1371/journal.pone.0180435.g007
WNS pathology scoring
PLOS ONE | https://doi.org/10.1371/journal.pone.0180435 August 2, 2017 15 / 21
272 Paper 2.4.3
While some hibernating bats are able to recover from fungal infection [44,45], extensive
skin damage in WNS survivors may alter flight performance, foraging efficiency and metabolism
during the post-emergent season [16,43,46,47]. In line with the ecological and evolutionary
trade-off concept [48], mounting an immune response against P. destructans infection, and
the necessity to repair the damaged wing membrane, represent physiological costs that may
affect the bat’s fitness and survival. To address the series of healing from different grades of
skin damage associated with WNS, our histopathology findings in conjunction with the case
reports suggest that cupping erosions heal through marked neutrophilic inflammation (Fig
1G), sequestration of the agent from the skin (Fig 1H) and re-epithelialization (Figs 1H and 6).
As shown in Fig 6, the outcome of healing is regeneration of the skin, suggesting a return to
1 0 1 2
0
20
40
60
80
100
120
WeightedcumulativeWNSpathologyscore
Barbastella barbastellus
Eptesicus nilssonii
Miniopterus schreibersii
Myotis bechsteinii
Myotis brandtii
Myotis dasycneme
Myotis daubentonii
Myotis emarginatus
Myotis lucifugus
Myotis myotis
Myotis nattereri
Myotis septentrionalis
Plecotus auritus
Rhinolophus euryale
Rhinolophus hipposideros
Barbastella barbastellus
Eptesicus nilssonii
Miniopterus schreibersii
Myotis bechsteinii
Myotis brandtii
Myotis dasycneme
Myotis daubentonii
Myotis emarginatus
Myotis lucifugus
Myotis myotis
Myotis nattereri
Myotis septentrionalis
Plecotus auritus
Rhinolophus euryale
Rhinolophus hipposideros
average per species
Number of UV fluorescent spots per cm2
(log10)
Fig 8. Increase in weighted cumulative white-nose syndrome pathology score with progressing UV fluorescence. Phylogenetic
generalised least-squares accounts for phylogeny and intraspecific variability in evaluating the relationship between weighted cumulative
white-nose syndrome pathology score dependent on unit number of UV fluorescent lesions (β0 = 62.1, β1 = 16.41; solid line). Linear
regression without phylogenetic correction (α = 42.38, β = 11.59, F1,131 = 9.12, p = 0.003, r2
= 0.07; dotted line).
https://doi.org/10.1371/journal.pone.0180435.g008
WNS pathology scoring
PLOS ONE | https://doi.org/10.1371/journal.pone.0180435 August 2, 2017 16 / 21
Paper 2.4.3 273
full function. On the other hand, tissue necrosis (Figs 1D and 2A) and full-thickness skin infection
(Fig 1F) probably need more remodelling and may result in wing membrane scarring.
Haemorrhagic infarcts (Fig 2C) induce larger flight membrane defects (Fig 4) and may result
in tears that stretch to the wing margin (Fig 4E, 4F, 4G and 4H).
Case reports on infection time-series in M. daubentonii and M. myotis document the
dichotomy of development that may occur in European bats infected with P. destructans. Our
findings indicate that the severity of invasion and associated wing damage determine the clinical
outcome in terms of morbidity and/or mortality. Importantly, the M. daubentonii in our
case study died during the early post-hibernation period, similar to observations by Meteyer
et al. [43]. Sporadic cases of mortality in European bats due to P. destructans infection may go
undetected, therefore, outside of hibernacula. From a practical diagnostic point of view, it is
necessary to bear in mind that WNS-specific UV fluorescence [22] disappears within two to
three weeks of euthermy.
Quantified per-unit wing membrane area fungal load correlates with disease intensity measured
as the number of WNS lesions detected under UV transillumination [11]. It has been
hypothesised that P. destructans hyphae are more likely to invade deeper skin layers and produce
WNS lesions with heavy fungal growth on bat wings [11]. In fact, our data documents
progression of fungal infection to WNS pathognomonic single and multiple cupping erosions
in response to one- and several-orders-of-magnitude higher wing membrane fungal loads,
respectively, compared with the early stage of infection characterised by skin surface colonisation
(cf. Fig 7A). For this chronic skin infection, severity-grading scores were understandably
higher in bats with more extensive UV fluorescence (Fig 8). The UV fluorescence associated
with WNS is caused by riboflavin, a fungal secondary metabolite, that accumulates in skin
lesions and results in cytotoxic damage of adjacent tissues [23] (as reflected by the weighted
cumulative WNS pathology score). All three diagnostic methods (i.e. qPCR, UV transillumination
and histology), when used in a quantitative or semi-quantitative manner, proved useful
for discrimination between colonisation of bats exposed to the pathogen only and those with
the invasive skin infection known as WNS.
Comparative findings from Palearctic and Nearctic bats infected with P. destructans show
differences in prevalence, fungal load and qualitative [11] and semi-quantitative histopathology
(this study). To gain greater insight into WNS severity, it is imperative that complex data
are collected on infection intensity in each bat. Our proposed biopsy-based grading system for
WNS is suitable for this purpose as histopathology reveals degree of invasiveness in terms of
skin layers invaded, along with associated damage; qPCR provides data on pathogen load, and
image analysis of photographs taken under UV light quantifies total affected wing/body surface
area. The ease of diagnostic scoring using our system allows open, effective WNS surveillance
across Europe, as well as comparative studies across the Holarctic. The use of uniform
criteria and training sessions in WNS histopathology can substantially improve consensus rate
in diagnosis of the disease among pathologists and promote its universality. Undertaken
alongside UV trans-illumination, which enables targeted biopsy, our WNS severity-grading
system provides an attractive option for use with conservation-sensitive bat species.
Supporting information
S1 Table. Species sample sizes. n–number of individuals per species undergoing histopathological
examination, WNS histo+–number of individuals confirmed positive for white-nose
syndrome.
(PDF)
WNS pathology scoring
PLOS ONE | https://doi.org/10.1371/journal.pone.0180435 August 2, 2017 17 / 21
274 Paper 2.4.3
S2 Table. Data on fungal load, ultra-violet fluorescence and histopathology in Holarctic
bats. Fungal load and UV fluorescence are given on log10 scale per cm2
of wing area. Histopathology
is reported as a binary code for index signs of grades corresponding to those listed in
Methods and Table 2. NA–not available, CZ–Czech Republic, LV–Latvia, PL–Poland, RU–
Russia, SI–Slovenia, histoSum–weighted cumulative WNS pathology score.
(TXT)
Acknowledgments
This study was supported by the Czech Science Foundation (Grant No. P506-12-1064 and 17-
20286S). This research was carried out as part of the CEITEC 2020 project (LQ1601), with further
financial support from the Ministry of Education, Youth and Sports of the Czech Republic
under National Sustainability Programme II. We thank the following students of the Veterinary
Study Programme at the University of Veterinary and Pharmaceutical Sciences Brno for
volunteering to participate in the study: Martina Bala´zˇova´, Tereza Bartosˇova´, Tereza Beˇlkova´,
Martina Benesˇova´, Miriam Birosˇova´, Kateřina Bohata´, Denisa Bohusˇova´, Kla´ra Bory´skova´,
Petra Buchnı´čkova´, Anna Ča´sova´, Ondřej Daneˇk, Lenka Danielova´, Anna Drahoňovska´,
Marie Dvořa´kova´, Anna Frumarova´, Nella Fuchsova´, Lenka Hora´kova´, Va´clav Hůlka, Patricie
Janı´čkova´, Martina Jasanska´, Miroslava Juranova´, Lenka Kapustova´, Lenka Klimesˇova´, Magda
Kohoutova´, Katarı´na Kopa´lova´, Lucie Kosˇtˇa´lova´, Tereza Kova´řova´, Jan Kovol, Barbora Krejčı´,
Jan Krejčı´, Linda Kubasˇtova´, Marie Kuba´tova´, Veronika Kvakova´, Petra Lanı´kova´, Barbora
Lo¨fflerova´, Dobromila Malı´kova´, Lenka Malinova´, Jana Michalcova´, Alena Mı´čkova´, Kristı´na
Miklosˇovičova´, Kateřina Musa´lkova´, Michaela Petrı´kova´, Kateřina Podra´bska´, Michaela Prokesˇova´,
Polina Rapekta, Nelly Reisova´, Karel Sˇlajs, Jiřı´ Slavı´k, Marke´ta Sosňa´kova´, Ivana Sˇtˇa´hlavska´,
Veronika Stařecka´, Nata´lie Sˇtefaňa´kova´, Va´clav Sˇtellar, Soňa Struha´rova´, Monika
Sˇubrtova´, Ondřej Ta´borsky´, Ivana Timova´, Kla´ra Tlačbabova´, Anita Turanska´, Juraj Turňa,
Karel Tvrdoň, Gabriela Vackova´, Daniela Valvodova´, Martina Vavřincova´, Lenka Večerkova´,
Lucie Veselkova´, Lenka Vojteˇchovska´, Michaela Vojtkova´, Zuzana Vola´kova´, Jakub Za´lesky´,
Henrieta Zbořilova´ and Anna Zemanova´.
Author Contributions
Conceptualization: Jiri Pikula, Sybill K. Amelon.
Formal analysis: Nata´lia Martı´nkova´.
Funding acquisition: Jiri Pikula, Nata´lia Martı´nkova´.
Investigation: Jiri Pikula, Hana Bandouchova, Veronika Kovacova, Petr Linhart, Vladimir
Piacek, Nata´lia Martı´nkova´.
Methodology: Jiri Pikula, Hana Bandouchova, Nata´lia Martı´nkova´.
Resources: Jiri Pikula, Sybill K. Amelon, Hana Bandouchova, Toma´sˇ Bartonička, Hana Berkova,
Jiri Brichta, Sarah Hooper, Tomasz Kokurewicz, Miroslav Kolarik, Bernd Ko¨llner,
Veronika Kovacova, Petr Linhart, Vladimir Piacek, Gregory G. Turner, Jan Zukal, Nata´lia
Martı´nkova´.
Writing – original draft: Jiri Pikula, Nata´lia Martı´nkova´.
Writing – review & editing: Jiri Pikula, Sybill K. Amelon, Hana Bandouchova, Toma´sˇ Bartonička,
Jiri Brichta, Sarah Hooper, Tomasz Kokurewicz, Miroslav Kolarik, Veronika Kovacova,
Vladimir Piacek, Jan Zukal, Nata´lia Martı´nkova´.
WNS pathology scoring
PLOS ONE | https://doi.org/10.1371/journal.pone.0180435 August 2, 2017 18 / 21
Paper 2.4.3 275
References
1. Deem SL, Karesh WB, Weisman W. Putting theory into practice: Wildlife health in conservation. Conserv
Biol. 2001; 15: 1224–1233.
2. Fey SB, Siepielski AM, Nussle´ S, Cervantes-Yoshida K, Hwan JL, et al. Recent shifts in the occurrence,
cause, and magnitude of animal mass mortality events. Proc Natl Acad Sci USA. 2015; 112: 1083–
1088. https://doi.org/10.1073/pnas.1414894112 PMID: 25583498
3. Blehert DS, Hicks AC, Behr M, Meteyer CU, Berlowski-Zier BM, et al. Bat white-nose syndrome: An
emerging fungal pathogen? Science. 2009; 323: 227. https://doi.org/10.1126/science.1163874 PMID:
18974316
4. Gargas A, Trest MT, Christensen M, Volk TJ, Blehert DS. Geomyces destructans sp. nov. associated
with bat white-nose syndrome. Mycotaxon. 2009; 108: 147–154.
5. Lorch JM, Meteyer CU, Behr MJ, Boyles JG, Cryan PM, et al. Experimental infection of bats with Geomyces
destructans causes white-nose syndrome. Nature. 2011; 480: 376–378. https://doi.org/10.1038/
nature10590 PMID: 22031324
6. Minnis AM, Lindner DL. Phylogenetic evaluation of Geomyces and allies reveals no close relatives of
Pseudogymnoascus destructans, comb. nov., in bat hibernacula of eastern North America. Fungal Biol.
2013; 117: 638–649. https://doi.org/10.1016/j.funbio.2013.07.001 PMID: 24012303
7. Zukal J, Bandouchova H, Bartonicka T, Berkova H, Brack V, et al. White-nose syndrome fungus: A generalist
pathogen of hibernating bats. PLoS ONE. 2014; 9: e97224. https://doi.org/10.1371/journal.pone.
0097224 PMID: 24820101
8. Frick WF, Pollock JF, Hicks AC, Langwig KE, Reynolds DS, et al. An emerging disease causes regional
population collapse of a common North American bat species. Science. 2010; 329: 679–682. https://
doi.org/10.1126/science.1188594 PMID: 20689016
9. Turner GG, Reeder DM, Coleman JTH. A five-year assessment of mortality and geographic spread of
white-nose syndrome in North American bats and a look to the future. Bat Research News. 2011; 52:
13–27.
10. Coleman JTH, Reichard JD. Bat white-nose syndrome in 2014: A brief assessment seven years after
discovery of a virulent fungal pathogen in North America. Outlooks on Pest Management. 2014; 25:
374–377.
11. Zukal J, Bandouchova H, Brichta J, Cmokova A, Jaron KS, et al. White-nose syndrome without borders:
Pseudogymnoascus destructans infection tolerated in Europe and Palearctic Asia but not in North
America. Sci Reports. 2016; 6: 19829.
12. Bandouchova H, Bartonicka T, Berkova H, Brichta J, Cerny J, et al. Pseudogymnoascus destructans:
Evidence of virulent skin invasion for bats under natural conditions, Europe. Transbound Emerg Dis.
2015; 62: 1–5. https://doi.org/10.1111/tbed.12282 PMID: 25268034
13. Meteyer CU, Buckles EL, Blehert DS, Hicks AC, Green DE, et al. Histopathologic criteria to confirm
white-nose syndrome in bats. J Vet Diagn Invest. 2009; 21: 411–414. https://doi.org/10.1177/
104063870902100401 PMID: 19564488
14. Reeder DM, Frank CL, Turner GG, Meteyer CU, Kurta A, et al. Frequent arousal from hibernation linked
to severity of infection and mortality in bats with white-nose syndrome. PLoS ONE. 2012; 7: e38920.
https://doi.org/10.1371/journal.pone.0038920 PMID: 22745688
15. Warnecke L, Turner JM, Bollinger TK, Lorch JM, Misra V, et al. Inoculation of bats with European Geomyces
destructans supports the novel pathogen hypothesis for the origin of white-nose syndrome. Proc
Natl Acad Sci USA. 2012; 109: 6999–7003. https://doi.org/10.1073/pnas.1200374109 PMID: 22493237
16. Cryan P, Meteyer C, Boyles J, Blehert D. Wing pathology of white-nose syndrome in bats suggests lifethreatening
disruption of physiology. BMC Biol. 2010; 8: 135. https://doi.org/10.1186/1741-7007-8-135
PMID: 21070683
17. Cryan P, Meteyer C, Boyles J, Blehert D. White-nose syndrome in bats: illuminating the darkness. BMC
Biol. 2013; 11: 47. https://doi.org/10.1186/1741-7007-11-47 PMID: 23587401
18. Cryan PM, Meteyer CU, Blehert DS, Lorch JM, Reeder DM, et al. Electrolyte depletion in white-nose
syndrome bats. J Wildlife Dis. 2013; 49: 398–402.
19. Verant ML, Carol MU, Speakman JR, Cryan PM, Lorch JM, et al. White-nose syndrome initiates a cascade
of physiologic disturbances in the hibernating bat host. BMC Physiol. 2014; 14: 10. https://doi.org/
10.1186/s12899-014-0010-4 PMID: 25487871
20. Warnecke L, Turner JM, Bollinger TK, Misra V, Cryan PM, et al. Pathophysiology of white-nose syndrome
in bats: A mechanistic model linking wing damage to mortality. Biol Letters. 2013; 9: 20130177.
21. Hoyt JR, Sun K, Parise KL, Lu G, Langwig KE, et al. Widespread bat white-nose syndrome fungus,
Northeastern China. Emerg Inf Dis. 2016; 22: 140.
WNS pathology scoring
PLOS ONE | https://doi.org/10.1371/journal.pone.0180435 August 2, 2017 19 / 21
276 Paper 2.4.3
22. Turner GG, Meteyer CU, Barton H, Gumbs JF, Reeder DM, et al. Nonlethal screening of bat-wing skin
with the use of ultraviolet fluorescence to detect lesions indicative of white-nose syndrome. J Wildlife
Dis. 2014; 50: 566–573.
23. Flieger M, Bandouchova H, Cerny J, Chudı´čkova´ M, Kolarik M, et al. Vitamin B2 as a virulence factor in
Pseudogymnoascus destructans skin infection. Sci Reports. 2016; 6: 33200.
24. Court MH, Robbins AH, Whitford AM, Beck EV, Tseng FS, et al. Pharmacokinetics of terbinafine in little
brown myotis (Myotis lucifugus) infected with Pseudogymnoascus destructans. Am J Vet Res. 2017;
78: 90–99. https://doi.org/10.2460/ajvr.78.1.90 PMID: 28029293
25. Shuey MM, Drees KP, Lindner DL, Keim P, Foster JT. Highly sensitive quantitative PCR for the detection
and differentiation of Pseudogymnoascus destructans and other Pseudogymnoascus species.
Appl Environ Microb. 2014; 80: 1726–1731.
26. Bates D, Maechler M, Bolker B, Walker S. lme4: Linear mixed-effects models using Eigen and S4. R
package version 1.1–9. 2015. Available from: https://CRAN.R-project.org/package=lme4.
27. R Core Team. R: A language and environment for statistical computing. 2016. Available from: http://
www.R-project.org/.
28. Rowlingson B, Diggle P. splancs: Spatial and space-time point pattern analysis. R package version
2.01–38. 2015. Available from: https://CRAN.R-project.org/package=splancs.
29. Urbanek S. jpeg: Read and write JPEG images. R package version 0.1–8. 2014. Available from: https://
CRAN.R-project.org/package=jpeg.
30. Ives AR, Midford PE, Garland T Jr. Within-species variation and measurement error in phylogenetic
comparative methods. Syst Biol. 2007; 56: 252–270. https://doi.org/10.1080/10635150701313830
PMID: 17464881
31. Revell LJ. phytools: An R package for phylogenetic comparative biology (and other things). Methods
Ecol Evol. 2012; 3: 217–223.
32. Cohen J. A coefficient of agreement for nominal scales. Educ Psychol Meas. 1960; 20: 37–46.
33. McHugh ML. Interrater reliability: the kappa statistic. Biochem Medica. 2012; 22: 276–282.
34. Fleiss JL. Measuring nominal scale agreement among many raters. Psychol Bull. 1971; 76: 378–382.
35. Gamer M, Lemon J, Singh IFP. irr: Various coefficients of interrater reliability and agreement. R package
version 0.84. 2012. Available from: https://CRAN.R-project.org/package=irr.
36. Martı´nkova´ N, Bačkor P, Bartonička T, Blazˇkova´ P, Červeny´ J, et al. Increasing incidence of Geomyces
destructans fungus in bats from the Czech Republic and Slovakia. PLoS ONE. 2010; 5: e13853. https://
doi.org/10.1371/journal.pone.0013853 PMID: 21079781
37. Puechmaille SJ, Wibbelt G, Korn V, Fuller H, Forget F, et al. Pan-European distribution of white-nose
syndrome fungus (Geomyces destructans) not associated with mass mortality. PLoS ONE. 2011; 6:
e19167. https://doi.org/10.1371/journal.pone.0019167 PMID: 21556356
38. Field KA, Johnson JS, Lilley TM, Reeder SM, Rogers EJ, et al. The white-nose syndrome transcriptome:
Activation of anti-fungal host responses in wing tissue of hibernating little brown Myotis. PLoS
Pathog. 2015; 11: e1005168. https://doi.org/10.1371/journal.ppat.1005168 PMID: 26426272
39. Pikula J, Bandouchova H, Novotny´ L, Meteyer CU, Zukal J, et al. Histopathology confirms white-nose
syndrome in bats in Europe. J Wildlife Dis. 2012; 48: 207–211.
40. Lucan RK, Bandouchova H, Bartonicka T, Pikula J, Zahradnikova A Jr., et al. Ectoparasites may serve
as vectors for the white-nose syndrome fungus. Parasit Vectors 2016; 9: 16. https://doi.org/10.1186/
s13071-016-1302-2 PMID: 26762515
41. Cross SS, Benes K, Stephenson TJ, Harrison RF. Grading in histopathology. Diagnostic Histopathology.
2011; 17: 263–267.
42. McGuire LP, Turner JM, Warnecke L, McGregor G, Bollinger TK, et al. White-nose syndrome disease
severity and a comparison of diagnostic methods. EcoHealth. 2016; 13: 60–71. https://doi.org/10.1007/
s10393-016-1107-y PMID: 26957435
43. Meteyer CU, Barber D, Mandl JN. Pathology in euthermic bats with white nose syndrome suggests a
natural manifestation of immune reconstitution inflammatory syndrome. Virulence. 2012; 3: 583–588.
https://doi.org/10.4161/viru.22330 PMID: 23154286
44. Meteyer CU, Valent M, Kashmer J, Buckles EL, Lorch JM, et al. Recovery of little brown bats (Myotis
lucifugus) from natural infection with Geomyces destructans, white-nose syndrome. J Wildl Dis. 2011;
47: 618–626. https://doi.org/10.7589/0090-3558-47.3.618 PMID: 21719826
45. Fuller NW, Reichard JD, Nabhan ML, Fellows SR, Pepin LC, et al. Free-ranging little brown myotis
(Myotis lucifugus) heal from wing damage associated with white-nose syndrome. Ecohealth. 2011; 8:
154–162. https://doi.org/10.1007/s10393-011-0705-y PMID: 21922344
WNS pathology scoring
PLOS ONE | https://doi.org/10.1371/journal.pone.0180435 August 2, 2017 20 / 21
Paper 2.4.3 277
46. Reichard JD, Kunz TH. White-nose syndrome inflicts lasting injuries to the wings of little brown myotis
(Myotis lucifugus). Acta Chiropterol. 2009; 11: 457–464.
47. Voigt CC. Bat flight with bad wings: is flight metabolism affected by damaged wings? J Exp Biol. 2013;
216: 1516–1521. https://doi.org/10.1242/jeb.079509 PMID: 23348945
48. Lochmiller RL, Deerenberg C. Trade-offs in evolutionary immunology: just what is the cost of immunity?
Oikos. 2000; 88: 87–98.
WNS pathology scoring
PLOS ONE | https://doi.org/10.1371/journal.pone.0180435 August 2, 2017 21 / 21
278 Paper 2.4.3
Paper 2.5.1
Jaron K. S., Moravec J. C., Martínková N. 2014. SigHunt: Horizontal gene transfer finder
optimized for eukaryotic genomes. Bioinformatics 30: 1081-1086.
– 279 –
Vol. 30 no. 8 2014, pages 1081–1086
BIOINFORMATICS ORIGINAL PAPER doi:10.1093/bioinformatics/btt727
Sequence analysis Advance Access publication December 25, 2013
SigHunt: horizontal gene transfer finder optimized for eukaryotic
genomes
Kamil S. Jaron1,
*, Jirˇı´ C. Moravec1
and Nata´ lia Martı´nkova´ 1,2,
*
1
Institute of Biostatistics and Analyses, Masaryk University and 2
Institute of Vertebrate Biology, Academy of Sciences of
the Czech Republic, Brno, Czech Republic
Associate Editor: John Hancock
ABSTRACT
Motivation: Genomic islands (GIs) are DNA fragments incorporated
into a genome through horizontal gene transfer (also called lateral
gene transfer), often with functions novel for a given organism. While
methods for their detection are well researched in prokaryotes, the
complexity of eukaryotic genomes makes direct utilization of these
methods unreliable, and so labour-intensive phylogenetic searches
are used instead.
Results: We present a surrogate method that investigates nucleotide
base composition of the DNA sequence in a eukaryotic genome and
identifies putative GIs. We calculate a genomic signature as a vector of
tetranucleotide (4-mer) frequencies using a sliding window approach.
Extending the neighbourhood of the sliding window, we establish a
local kernel density estimate of the 4-mer frequency. We score the
number of 4-mer frequencies in the sliding window that deviate from
the credibility interval of their local genomic density using a newly
developed discrete interval accumulative score (DIAS). To further improve
the effectiveness of DIAS, we select informative 4-mers in a
range of organisms using the tetranucleotide quality score developed
herein. We show that the SigHunt method is computationally efficient
and able to detect GIs in eukaryotic genomes that represent nonameliorated
integration. Thus, it is suited to scanning for change in
organisms with different DNA composition.
Availability and implementation: Source code and scripts freely
available for download at http://www.iba.muni.cz/index-en.php?pg=
research–data-analysis-tools–sighunt are implemented in C and R
and are platform-independent.
Contact: 376090@mail.muni.cz or martinkova@ivb.cz
Received on August 9, 2013; revised on November 18, 2013;
accepted on December 9, 2013
1 INTRODUCTION
Horizontal gene transfer (HGT) occurs when a DNA sequence
passes between organisms otherwise than by reproductive descent.
It results in a relationship of orthologous sequences that is
not tree-like and contains reticulations. Notorious examples include
antibiotic resistance plasmids transferred between bacterial
strains (Freeman, 1951), pathogenicity islands (Friesen et al.,
2006), incorporation of retroviruses (Jern and Coffin, 2008)
and artificial HGT in the forms of genetically modified organisms
(Wolfenbarger and Phifer, 2000). When a horizontally
transferred gene becomes fixed in a population, it is termed a
genomic island (GI). Relatively frequent HGT occurs between
organisms of similar complexity, such as between prokaryotes,
but successful HGT between domains and kingdoms is also
known. Incorporation of the alien sequence into a recipient
genome must be compatible with survival of the cell; it should
not, for example, knock out an essential gene. In eukaryotes,
distortion of the open reading frame with HGT is less likely
due to the sparseness of coding sequences; yet alien genes face
molecular biological limitations relating to metabolism in the
recipient organism. When genes are transferred from prokaryotes
to eukaryotes, the genetic code difference might hinder correct
protein translation. In cases of HGT between eukaryotes,
incorrect intron splicing would render the gene product altered
and potentially dysfunctional. Nevertheless, successful implementation
of HGT in a suitable place within the genome could
result in expression of the relevant protein. Proteins encoded in a
GI that could be expressed in the recipient organism might provide
a novel, highly adaptive function (Casacuberta and Gonzlez,
2013; Scho¨ nknecht et al., 2013). To find such a GI is to discover
exciting information that is often transformative for the given
research field.
HGT detection is well studied in prokaryotes, having started
from measuring variability of oligonucleotide frequency along
the genome (Karlin and Burge, 1995). Eukaryotic genomes, however,
are comparatively heterogeneous in their composition and
much more extensive. That situation complicates the HGT
search. Therefore, those methods developed for prokaryotes
fail either due to their inability to handle the sequence heterogeneity
or because their computational requirements skyrocket.
Researchers studying HGT in eukaryotes use two types of methods:
surrogate and comparative. Surrogate methods use nucleotide
base composition of the DNA sequence. They have been
applied in the form of chaos game representation clustering
based on tetranucleotide composition (Mallet et al., 2010).
Comparative methods are computationally intensive because
they compute phylogenetic comparisons between large numbers
of identified genes or they use local alignment comparisons
against a reference sequence database. They require prior annotation
of a genome, an extensive database of comparable
orthologues and computationally intensive phylogenetic analyses.
Despite these limitations, the results from comparative
methods are considered most reliable, as they enable identification
of HGT and the donor organism in the form of testable
hypotheses. Therefore, methods have been developed to reduce
the candidate dataset for GIs while balancing the numbers of
false positives and false negatives (Boc and Makarenkov, 2011;*To whom correspondence should be addressed.
ß The Author 2013. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com 1081
byguestonMay22,2014http://bioinformatics.oxfordjournals.org/Downloadedfrom
280 Paper 2.5.1
Mallet et al., 2010; Podell and Gaasterland, 2007). To date, the
effort in eukaryotic genomes has been demonstrated consistently
to fail on these criteria due to the genome heterogeneity in eukaryotes
(Mallet et al., 2010), thus leading to the need for an
expanding reference database for comparative analyses.
We show here, however, that genome complexity may be overcome
by carefully examining the pattern of genomic signature
along a sequence. We use both the selection of tetranucleotides
with greatest interspecific variability in their frequencies and
local composition shifts that take into account natural variation
of tetranucleotide frequency changes along a chromosomal arm
to locate regions that differ and might thus represent recent acquisitions
via HGT.
2 ALGORITHM
2.1 Calculation of 4-mer sliding density
The principle of the SigHunt (genomic signature hunter) method
is to use a sliding window approach both to detect the HGT and
to take into account DNA sequence composition change along
chromosomes. The genomic signature is calculated as a frequency
vector of tetranucleotides (4-mer) within the window of
a genomic sequence. Within each sliding window, frequency of
every 4-mer is calculated as
FmðSÞ ¼
CmðSÞ
NS À 3
where Fm(S) is the frequency of the 4-mer m within a sequence in
the sliding window (S), Cm(S) is the count of m in S and NS is the
length of the sequence in the window S. Only fully resolved sites
are counted towards Fm(S). Tetranucleotides that contain an unresolved
base are omitted. Unresolved sites are skipped, and the
length of the window, but not the length of the sequence, is
increased by the given number of nucleotides until it reaches
20% of the window size. This is the maximum extension of the
window. There are two reasons for choosing oligonucleotide
length. First, 4-mers provide a vector with a number of dimensions
sufficient for complex comparisons. Second, the 4-mers
frequency vector is representative even for relatively short sliding
windows, which is of interest in cases when short alien genes
could be expected.
Along a chromosomal sequence, DNA base composition
changes with functional regions, such as coding and noncoding
regions, repetitive elements, telomeres, centromeres or
other structural elements that stabilize a chromosomal arm.
Differences between repetitive and coding regions in particular
are prone to variable frequency of a few 4-mers that could either
distort or inform a signal from alien fragments. To account for
this, we develop a novel scoring system and introduce a slidingdensity
concept. This takes into account genomic regions D that
are directly adjacent to the sliding window S, offset at the 50
and
30
ends of the sequence. Within the long sliding window of D, we
calculate a kernel density estimate for each 4-mer. The measured
4-mer frequency in the short sliding window of sequence S is
tested for whether or not it is located outside of the credibility
interval (CI) of the 4-mer density in D
FmðSÞ 2 0, ÈÀ1
D

2
  
[ ÈÀ1
D 1 À

2
 
, 1
 
where ÈD is a cumulative distribution function of Fm(S) in D and
 is a confidence level. Values found to be outside of the CI are
scored in three intervals for  2 f0:05, 0:025, 0:01g, adding 1, 2
and 3, respectively, to the discrete interval accumulative score
(DIAS). To avoid autocorrelation in measuring Fm(S) on D, D
is selected in such a way that S and a specified number of its
surrounding sliding windows (x) are excluded from the 4-mer
density calculation (eye-of-the-storm approach). Thus, we compare
the S sequence to its context but not to its immediate surroundings.
To compare the sliding window approach to previous
genome-wide signature studies, we also show global 4-mer
density.
DIAS measures how many 4-mers deviate in their frequency
from the local background of the genomic sequence and by how
much. While this is more stable along a chromosome than any
other compositional measure known to us, it is nevertheless sensitive
to local changes.
2.2 Selection of informative 4-mers
To further improve computational speed of the SigHunt method,
informative 4-mers can be selected to reduce the signal-to-noise
ratio. Multiple genomes are used to train the procedure for 4-mer
selection based on their intra- and inter-genomic variability.
Fm(S) values for all consecutive windows are used to calculate
the 4-mer density in a given chromosome. All chromosomes are
used for the training genomes. Informative 4-mers are selected as
those where means of Fm in organisms are distinctive from the
overall estimates and within-genome Fm variance is small. We
score the 4-mers using the tetranucleotide quality score (TES) for
each 4-mer
TESm ¼
Xn
k¼1
Kk À "K
À Á2
À
Xn
k¼1
Ak þ Bk þ Dkð Þ þ E
K ¼
1
c
Xc
i¼1
i
A ¼
Xc
i¼1
1
c
i À Kð Þ2
B ¼
Xc
i¼1
1
c
i À cð Þ2
D ¼ "2
c
E ¼
Xn
k¼1
Kk À Keð Þ2
À Ae þ Be þ Deð Þ
where n is the number of all organisms used for training, c is the
number of chromosomes in the given organism, i is mean 4-mer
frequency on the given chromosome i ¼ FmðSÞ
À Á
i
 Ã
, "K is the
average of all mean frequencies of the given 4-mer on all tested
chromosomes in all organisms within the dataset,  denotes respective
variances and e is the estimated organism intended for
the SigHunt search.
Using TES, we are interested to learn the extent to which
4-mer frequencies vary between organisms with respect to their
variability within a genome. To achieve this, K estimates average
frequency of a 4-mer that is found in the given organism. It is
calculated as a mean of means to avoid weighting of the value
according to the number and size of chromosomes. A sums
1082
K.S.Jaron et al.
byguestonMay22,2014http://bioinformatics.oxfordjournals.org/Downloadedfrom
Paper 2.5.1 281
squared differences between each typical 4-mer frequency on a
chromosome compared with the whole genome; B similarly penalizes
4-mers that have deviant frequency variance in a chromosome
compared with the background genome. By including the
D component into the equation, we ensure that the frequency
variance in an organism is small to facilitate finding and interpreting
4-mer frequencies outside of the confidence interval of
their density. E provides a measure that would stress usefulness
of the given 4-mer for discrimination of the home sequence. TES
increases where 4-mer frequencies differ between organisms, they
are stable for a given organism and they exhibit little variation
within a genome. We demonstrate below that 4-mers with high
TES scores will be informative in recognizing putative HGT. GIs
identified with SigHunt should subsequently be verified using
comparative methods (Fig. 1).
3 METHODS
The sensitivity and specificity of the SigHunt method was tested using a
receiver operating characteristic (ROC) curve on simulated data. We
introduced alien sequences into the recipient sequence by randomly selecting
and replacing DNA fragments between 10 organisms with complete
genomic sequences. Eukaryotes were represented by the fungi
Aspergillus fumigatus (Nierman et al., 2005), Encephalitozoon cuniculi
(Katinka et al., 2001) and Saccharomyces cerevisiae (Goffeau et al.,
1996); the red alga Cyanidioschyzon merolae (Matsuzaki et al., 2004);
the chromalveolates Cryptosporidium parvum (Abrahamsen et al., 2004),
Plasmodium falciparum (Hall et al., 2002) and Thalassiosira pseudonana
(Armbrust et al., 2004); and an animal, Drosophila melanogaster (Adams
et al., 2000). SigHunt’s performance in prokaryotes was demonstrated in
Buchnera sp. (Shigenobu et al., 2000) and Escherichia coli (Welch et al.,
2002). In each of 500 replicates of the procedure, we replaced three genomic
fragments with alien DNA from other organisms. The length of
each introduced fragment was 2, 5 and 15kb in each chromosome,
where the origin of each fragment was randomly drawn from the pool
of analysed organisms and chromosomes. We scored the sequence according
to DIAS for 5 kb windows and 1 kb sliding windows. Those
regions with known alien sequences were scored as the only GIs present.
We calculated the area under the curve (AUC) from the ROC curve in R
(R Development Core Team, 2011; Robin et al., 2011; Sing et al., 2005)
and estimated the optimal threshold while maximizing AUC. The same
analysis was conducted using INDeGeNIUS, which is a recent surrogate
method that uses oligonucleotide frequencies (Shrivastava et al., 2010),
and with Alien_Hunter, which uses interpolated variable order motifs of
DNA composition (Vernikos and Parkhill, 2006).
3.1 Case studies
To test the SigHunt method on real biological data, we used genomic
sequences of Aspergillus, Cryptosporidium and Saccharomyces, as listed
above, and added genomic sequences not yet assembled to chromosomes
for organisms with known GIs. The latter were the red algae Galdieria
sulphuraria, wherein the horizontally transferred genes provide multiple
environmental adaptations (Scho¨ nknecht et al., 2013), and the fungus
Pyrenophora tritici-repentis, which recently acquired a pathogenicity
island (Friesen et al., 2006) and had additional proteins originating
from HGT (Sun et al., 2013). We used organisms where GIs had been
identified previously and their location was specific in the available genomic
sequence (Hall et al., 2005; Huang et al., 2004; Mallet et al., 2010;
Scho¨ nknecht et al., 2013; Sun et al., 2013). Contigs at least 200 kb
were used from these organisms. This enabled us to cross-check
SigHunt against previous studies and thereby to demonstrate its utility.
The optimal threshold value for the DIAS as tested with ROC on simulated
data was 6.04, and we relaxed this value further to account for
sequence amelioration. The cut-off value used here was DIAS! 5. Two
windows adjacent to the previously identified GI were assessed to compensate
for the fact that most GIs were identified as a coding gene sequence
and the transferred region could likely include flanking regions
(Friesen et al., 2006).
4 RESULTS
We estimated 4-mer density and its variance in the 10 reference
genomes. Comparing the 4-mers using TES, we selected the 16
most informative 4-mers. These were used for all subsequent
analyses.
4.1 Sensitivity and specificity of SigHunt
SigHunt showed average AUC values equal to 0.77 for global
density, 0.72 for sliding density and 0.77 for the eye-of-the-storm
approach, meaning that sensitivity and specificity of detecting
GIs in simulated data were high (Table 1). We assigned a GI
only where it had been artificially introduced, and the remaining
home sequence still contained its natural GIs, which were disregarded
for the purpose of this test (Table 2). This could have
lowered the performance indicators. The analyses of individual
chromosomes required 8–60 min to calculate DIAS for all three
approaches presented here. INDeGeNIUS and Alien_Hunter
performed in a similar way with respect to their ability to correctly
score the introduced GIs, and no differences between the
methods were significant. INDeGeNIUS showed the highest
values of AUC from the tested methods. However, those analyses
took 20 min–20 h per chromosome, and Drosophila could
not be analysed due to extensive memory requirements. The
speed of Alien_Hunter was 20–120 min per chromosome.
4.2 HGT detection with SigHunt
We estimated that the studied genomes exhibit regions with
deviant genomic signature that could be considered alien.
These provide a genome-wide assessment of candidate regions
Calculate genomic signatures
Genomic sequence
in fasta format
Calculate TES for selection
of informative tetranucleotides
Calculate 4-mer density
and DIAS
Establish putative genomic islands
Verify whether the regions
represent HGT with comparative methodsStreamlined SigHunt
Optional steps
SigHunt
Fig. 1. Flow diagram of GIs analyses using the SigHunt method
1083
SigHunt
byguestonMay22,2014http://bioinformatics.oxfordjournals.org/Downloadedfrom
282 Paper 2.5.1
for HGT. We searched for consistency of specific sequences in
Aspergillus, Saccharomyces, Pyrenophora, Galdieria and
Cryptosporidium between SigHunt and published results.
Compared with the extent of HGT identified in previous studies
that used predominantly comparative methods on annotated
genes, SigHunt found from 30% (Cryptosporidium) to 80%
(Aspergillus) of previously identified GIs (Table 2). In
Aspergillus, we were able to find the majority of published GIs
as per Mallet et al. (2010), which can be expected given that those
authors used a surrogate method verified with phylogenetic comparison
to localize GIs. We identified 5 of 10 putative GIs recognized
in Saccharomyces by Hall et al. (2005). The missed GIs
were protein-coding genes51.2 kb, for which our chosen window
size might not be optimal. In Pyrenophora, we searched for 17
genes and 6 were retrieved by SigHunt. HGT in Galdieria can be
attributed to $9% of the genomic sequence in the longest scaffolds,
which we analysed (Scho¨ nknecht et al., 2013). SigHunt was
able to recognize the genomic signatures of the published GIs as
being alien in 61% of the cases. In Cryptosporidium, we found 9
of 30 previously identified GIs (Huang et al., 2004). As in previous
cases with low success, the GIs in Cryptosporidium that
were missed consisted predominantly of short genes. In
these cases, experimentation with window size would be beneficial.
INDeGeNIUS found more known GIs in Aspergillus,
Pyrenophora and Cryptosporidium, but the analyses took an
order of magnitude longer than in SigHunt. SigHunt outperformed
Alien_Hunter in recognizing the previously established
HGT events in most tested organisms. The exception
Table 1. Average area under a ROC curve for SigHunt, INDeGeNIUS and Alien_Hunter analyses on 500 random replacements of three GIs into
reference chromosomal sequences from the model organisms
Organism Global density Sliding density Eye of the storm INDeGeNIUS Alien_Hunter
Fungi
Aspergillus 0.75 (0.11) 0.72 (0.11) 0.75 (0.12) 0.84 (0.07) 0.65 (0.17)
Encephalitozoon 0.71 (0.09) 0.70 (0.08) 0.74 (0.10) 0.88 (0.07) 0.88 (0.11)
Saccharomyces 0.81 (0.07) 0.72 (0.07) 0.78 (0.06) 0.90 (0.08) 0.83 (0.12)
Red alga
Cyanidioschyzon 0.83 (0.08) 0.81 (0.07) 0.86 (0.07) 0.91 (0.07) 0.84 (0.15)
Animal
Drosophila 0.74 (0.11) 0.68 (0.09) 0.72 (0.09) n/a 0.75 (0.13)
Chromalveolates
Cryptosporidium 0.77 (0.07) 0.68 (0.08) 0.74 (0.07) 0.86 (0.07) 0.78 (0.17)
Plasmodium 0.67 (0.12) 0.63 (0.09) 0.70 (0.12) 0.87 (0.07) 0.82 (0.17)
Thalassiosira 0.87 (0.09) 0.81 (0.07) 0.85 (0.08) 0.89 (0.06) 0.78 (0.18)
Prokaryotes
Buchnera 0.83 (0.06) 0.73 (0.07) 0.81 (0.06) 0.91 (0.06) 0.85 (0.16)
Escherichia 0.76 (0.09) 0.68 (0.09) 0.72 (0.09) 0.85 (0.07) 0.82 (0.16)
Note: Standard deviation is given in parentheses.
n/a, not available.
Table 2. Number of correctly assigned GIs in model organisms from those identified previously as retrieved by SigHunt eye-of-the-storm variant,
INDeGeNIUS and Alien_Hunter
Organism Number of
chromosomes
or contigsa
Sequence
length (Mb)
Previously
established GIs
SigHunt INDeGeNIUS Alien_
Hunter
References
Fungi
Aspergillus 8 29.4 189 150 189 54 (Mallet et al., 2010)
Pyrenophora 19a
33.9 17b
6 11 0 (Friesen et al., 2006; Sun et al., 2013)
Saccharomyces 16 12 10b
5 2 5 (Hall et al., 2005)
Red algae
Galdieria 17a
4.4 79b,c
48 15 33 (Scho¨ nknecht et al., 2013)
Chromalveolates
Cryptosporidium 8 9.1 30b
9 12 11 (Huang et al., 2004)
Note: In SigHunt, a selection of 16 4-mers identified by TES was used. DIAS! 5 was used as a cut-off value and two windows adjacent to the island borders were considered.
Sequence length ¼ size of the analysed genomic sequence.
a
Number of assessed contigs.
b
In annotated genes.
c
GIs found in 13.7 Mb of the genomic sequence, but only the longest contigs were analysed here.
1084
K.S.Jaron et al.
byguestonMay22,2014http://bioinformatics.oxfordjournals.org/Downloadedfrom
Paper 2.5.1 283
was Cryptosporidium, where Alien_Hunter found 11 of 30
GIs (Table 2).
5 DISCUSSION
We present here a tool for identifying genomic regions as candidates
for HGT assessment in eukaryotes. To our knowledge, this
is the first surrogate method primarily optimized for eukaryotic
genomes. It detects non-ameliorated HGT in large genomic sequences,
it is computationally efficient and its implementation
provides step-wise user access to results that enables data exploration
and analytical optimization.
We demonstrated good success in using SigHunt to find introduced
GIs across kingdoms and GIs in real genomic sequences
(particularly in some fungi). Considering from a biological perspective
reproduction within this group, HGT might be more
common in fungi than in other eukaryotes (Rosewich and
Kistler, 2000). With HGT events being relatively common
within the group, one could speculate that some of these will
be non-ameliorated and thus easily detectable using surrogate
methods. The further example of Galdieria, within which GIs
were plentiful across the genome, seems to corroborate this.
5.1 SigHunt’s advantages
The advantage of SigHunt lies in its utilization of informative
4-mers. Selecting only parts of the genomic signature that are
most informative according to TES reduces noise in the data
and computational requirements, thereby speeding up the analysis.
Computational demands are not negligible in eukaryotic
genomics. For example, INDeGeNIUS analysis of the largest
chromosome in Drosophila would require %70 TB of memory,
which is beyond the capacity available to many researchers,
including our group, and the current version of the program
does not allow for changes in memory use. By contrast,
SigHunt analysed the same problem using 900 MB of maximum
allocated memory. At a given time, SigHunt stores in memory
only that chromosome sequence needed to calculate the signatures,
or, once the corresponding signatures are calculated and
the chromosome sequence deleted from memory, the signatures
and densities themselves. Such orderly memory utilization reduces
the memory requirements and thus allows computations
of even large datasets on regular office computers.
Contrary to other recent methods that use a genomic signature
for HGT detection (Elhai et al., 2012; Shrivastava et al., 2010;
but see Mallet et al., 2010), SigHunt does not assume a point
estimate of the genomic signature. Instead, it uses a density distribution
and thus acknowledges the natural variability of DNA
composition and distinguishes only those regions that deviate
from the broad ‘norm’ for an organism. We recognize that the
density distributions must contain tails even for home signatures.
Due to this, DIAS measures accumulations of deviant 4-mer
frequencies rather than their mere occurrence. With sliding density
and its eye-of-the-storm variant that avoids autocorrelation,
SigHunt is able to find putative GIs in any region of the chromosome.
This includes regions rich in repetitive DNA because
SigHunt assumes a differing genomic signature typical for a genomic
region rather than for the whole chromosome.
SigHunt makes it unnecessary to have knowledge as to the
exact position on a chromosome of the examined sequence
(Podell and Gaasterland, 2007). It can successfully analyse unassembled
genomes, provided that the supercontigs are sufficiently
long. It also does not require information about gene locations.
By filtering nucleotide positions in a sequence that are not fully
resolved, we limit the amount of information while increasing the
accuracy. In case of a long eukaryotic genome, a trade-off in
favour of accuracy is paramount for SigHunt.
5.2 SigHunt’s disadvantages
Unfortunately, SigHunt is not a universal black box solution for
all HGT problems. Its very principle denies universality, as it
rises and falls on the assumption that there are differences in
oligonucleotide frequencies between organisms (Karlin and
Burge, 1995). This is not always sufficiently true, as shown by
our analyses on manipulated and real data. Some random islands
were undifferentiated from the home signature. For others, the
GI size in real datasets might have been too small to accurately
estimate the 4-mer frequency density for the DIAS calculation. In
addition to reasons of there being similar genomic signatures
among the organisms involved, SigHunt is prone to false negatives
due to amelioration over time of the compositional bias in
the horizontally transferred region compared with the host
genome. The false-positive rate might also be increased. In regions
with strong selection bias and functional restrictions, home
signature might vary locally. The extent to which this is the case
remains to be tested.
SigHunt expands the search for GIs across the genome without
annotation limitations, yet it is able to guide the comparative
search more effectively than do other similar methods. We have
shown that SigHunt provides a fair basis of target regions for
comparative assessment that consists of true GIs as confirmed by
phylogenetic analyses in the published data (Friesen et al., 2006;
Hall et al., 2005; Huang et al., 2004; Mallet et al., 2010;
Scho¨ nknecht et al., 2013; Sun et al., 2013).
5.3 Global density paradox
We claim that the advantage of SigHunt lies in its ability to
account for variation of the genomic signature along a chromosomal
sequence. Yet, in Table 1, the sensitivity and specificity test
shows the highest (albeit not statistically significant) AUC values
to be for the global density estimate in six cases. This is probably
caused by the fact that we introduced HGT directly between
organisms that spanned kingdoms and our islands were thus
devoid of any amelioration. In other words, while one would
rarely encounter such an event in practice, it is one that is relatively
easy to capture by means of genomic signatures. Analysing
real biological data would require a more subtle approach.
Therefore, both the sliding density and its eye-of-the-storm variant
provide room for fine-tuning the method. These are parameterized
for window size, sliding window size, sliding-density
window size, and eye-of-the-storm size. All these parameters
might be optimized to further improve SigHunt for any specific
target organism. On the other hand, the global density reached
the height of its performance in this study. We nevertheless consider
global-density DIAS calculation a useful approach in view
1085
SigHunt
byguestonMay22,2014http://bioinformatics.oxfordjournals.org/Downloadedfrom
284 Paper 2.5.1
of the fact that it is computationally effective and has low
memory demands.
5.4 Usefulness of SigHunt
As a method for investigating genomic signature in complex eukaryotic
genomes, SigHunt can provide a rapid analysis tool for
ongoing sequencing projects. In particular, it will be sensitive to
recent HGT, such as in cases of emergent pathogens that have
acquired novel genes. The choice of informative 4-mers could
increase resolution and success for binning of metagenomic
DNA fragments (Saeed and Halgamuge, 2009). With the
recent finding of bacterial horizontally transferred genes in
human tumour cells (Riley et al., 2013), SigHunt shows promise
to be used in rapid screening of such events in genomic assemblies
of specific cell lines. We assume that further research will
reveal more fields within which sliding density of DNA composition
in eukaryotic genomes and the selection of informative
oligonucleotides will prove advantageous.
ACKNOWLEDGEMENT
The bioinformatic analyses were conducted at the MetaCentrum
computing facility of Masaryk University.
Funding: Czech Science Foundation (grant number P506/12/
1064). The access to the MetaCentrum was funded by the
Ministry of Education, Youth, and Sports of the Czech
Republic (grant number LM2010005).
Conflict of Interest: none declared.
REFERENCES
Abrahamsen,M.S. et al. (2004) Complete genome sequence of the apicomplexan,
Cryptosporidium parvum. Science, 304, 441–445.
Adams,M.D. et al. (2000) The genome sequence of Drosophila melanogaster.
Science, 287, 2185–2195.
Armbrust,E.V. et al. (2004) The genome of the diatom Thalassiosira pseudonana:
ecology, evolution, and metabolism. Science, 306, 79–86.
Boc,A. and Makarenkov,V. (2011) Towards an accurate identification of mosaic
genes and partial horizontal gene transfers. Nucleic Acids Res., 39, e144.
Casacuberta,E. and Gonzlez,J. (2013) The impact of transposable elements in environmental
adaptation. Mol. Ecol., 22, 1503–1517.
Elhai,J. et al. (2012) Detection of horizontal transfer of individual genes by anomalous
oligomer frequencies. BMC Genomics, 13, 245.
Freeman,V.J. (1951) Studies on the virulence of bacteriophage-infected strains of
Corynebacterium diphtheriae. J. Bacteriol., 61, 675.
Friesen,T.L. et al. (2006) Emergence of a new disease as a result of interspecific
virulence gene transfer. Nat. Genet., 38, 953–956.
Goffeau,A. et al. (1996) Life with 6000 genes. Science, 274, 546–567.
Hall,C. et al. (2005) Contribution of horizontal gene transfer to the evolution of
Saccharomyces cerevisiae. Eukaryot. Cell, 4, 1102–1115.
Hall,N. et al. (2002) Sequence of Plasmodium falciparum chromosomes 1, 3–9 and
13. Nature, 419, 527–531.
Huang,J. et al. (2004) Phylogenomic evidence supports past endosymbiosis, intracellular
and horizontal gene transfer in Cryptosporidium parvum. Genome Biol.,
5, R88.
Jern,P. and Coffin,J.M. (2008) Effects of retroviruses on host genome function.
Annu. Rev. Genet., 42, 709–732.
Karlin,S. and Burge,C. (1995) Dinucleotide relative abundance extremes: a genomic
signature. Trends Genet., 11, 283–290.
Katinka,M.D. et al. (2001) Genome sequence and gene compaction of the eukaryote
parasite Encephalitozoon cuniculi. Nature, 414, 450–453.
Mallet,L. et al. (2010) Whole genome evaluation of horizontal transfers in the
pathogenic fungus Aspergillus fumigatus. BMC Genomics, 11, 171.
Matsuzaki,M. et al. (2004) Genome sequence of the ultrasmall unicellular red alga
Cyanidioschyzon merolae 10d. Nature, 428, 653–657.
Nierman,W.C. et al. (2005) Genomic sequence of the pathogenic and allergenic
filamentous fungus Aspergillus fumigatus. Nature, 438, 1151–1156.
Podell,S. and Gaasterland,T. (2007) DarkHorse: a method for genome-wide prediction
of horizontal gene transfer. Genome Biol., 8, R16.
R Development Core Team. (2011) R: A Language and Environment for Statistical
Computing. R Foundation for Statistical Computing, Vienna, Austria.
Riley,D.R. et al. (2013) Bacteria-human somatic cell lateral gene transfer is enriched
in cancer samples. PLoS Comput. Biol., 9, e1003107.
Robin,X. et al. (2011) pROC: an open-source package for R and Sþ to analyze and
compare ROC curves. BMC Bioinformatics, 12, 77.
Rosewich,U.L. and Kistler,H.C. (2000) Role of horizontal gene transfer in the
evolution of fungi. Annu. Rev. Phytopathol., 38, 325–363.
Saeed,I. and Halgamuge,S. (2009) The oligonucleotide frequency derived error gradient
and its application to the binning of metagenome fragments. BMC
Genomics, 10, S10.
Scho¨ nknecht,G. et al. (2013) Gene transfer from bacteria and archaea facilitated
evolution of an extremophilic eukaryote. Science, 339, 1207–1210.
Shigenobu,S. et al. (2000) Genome sequence of the endocellular bacterial symbiont
of aphids Buchnera sp. aps. Nature, 407, 81–86.
Shrivastava,S. et al. (2010) INDeGenIUS, a new method for high-throughput identification
of specialized functional islands in completely sequenced organisms.
J. Biosci., 35, 351–364.
Sing,T. et al. (2005) ROCR: visualizing classifier performance in R. Bioinformatics,
21, 3940–3941.
Sun,B.F. et al. (2013) Multiple interkingdom horizontal gene transfers in
Pyrenophora and closely related species and their contributions to phytopathogenic
lifestyles. PLoS One, 8, e60029.
Vernikos,G.S. and Parkhill,J. (2006) Interpolated variable order motifs for identification
of horizontally acquired DNA: revisiting the Salmonella pathogenicity
islands. Bioinformatics, 22, 2196–2203.
Welch,R.A. et al. (2002) Extensive mosaic structure revealed by the complete
genome sequence of uropathogenic Escherichia coli. Proc. Natl Acad. Sci.
USA, 99, 17020–17024.
Wolfenbarger,L.L. and Phifer,P.R. (2000) The ecological risks and benefits of genetically
engineered plants. Science, 290, 2088–2093.
1086
K.S.Jaron et al.
byguestonMay22,2014http://bioinformatics.oxfordjournals.org/Downloadedfrom
Paper 2.5.1 285
Paper 2.5.2
Škrabánek P., Martínková N. 2017. Extraction of outliers from imbalanced sets. In:
Martínez de Pisón F., Urraca R., Quintián H., Corchado E. (Eds.) Hybrid Artificial
Intelligent Systems. HAIS 2017. Lecture Notes in Computer Science, vol. 10334. pp.
402-412. Springer, Cham, DOI: 10.1007/978-3-319-59650-1 34.
– 287 –
Extraction of Outliers from Imbalanced Sets
Pavel ˇSkrab´anek1(B)
and Nat´alia Mart´ınkov´a2,3
1
Faculty of Electrical Engineering and Informatics, University of Pardubice,
Studentsk´a 95, 532 10 Pardubice, Czech Republic
pavel.skrabanek@upce.cz
2
Institute of Vertebrate Biology, Czech Academy of Sciences,
Kvˇetn´a 8, 603 65 Brno, Czech Republic
martinkova@ivb.cz
3
Institute of Biostatistics and Analyses, Masaryk University,
Kamenice 3, 625 00 Brno, Czech Republic
Abstract. In this paper, we presented an outlier detection method,
designed for small datasets, such as datasets in animal group behaviour
research. The method was aimed at detection of global outliers in
unlabelled datasets where inliers form one predominant cluster and the
outliers are at distances from the centre of the cluster. Simultaneously,
the number of inliers was much higher than the number of outliers. The
extraction of exceptional observations (EEO) method was based on the
Mahalanobis distance with one tuning parameter. We proposed a visualization
method, which allows expert estimation of the tuning parameter
value. The method was tested and evaluated on 44 datasets. Excellent
results, fully comparable with other methods, were obtained on datasets
satisfying the method requirements. For large datasets, the higher computational
requirement of this method might be prohibitive. This drawback
can be partially suppressed with an alternative distance measure.
We proposed to use Euclidean distance in combination with standard
deviation normalization as a reliable alternative.
Keywords: Outlier analysis · Distance based method · Global outlier ·
Single cluster · Mahalanobis distance · Biology
1 Introduction
Data mining reveals new, valuable and non-trivial information in large datasets
[14]. It is a process of discovering interesting patterns and knowledge in the
data that is not immediately apparent. Various data mining approaches help to
specify the patterns in the data mining tasks. Examples include characterization
and discrimination, mining of frequent patterns, associations and correlations,
classification and regression, clustering analysis, and outlier analysis [10].
The outlier analysis has an important position among data mining
approaches. Hawkins specified an intuitive definition of the term outlier as:
‘Within a given dataset, the outlier is an observation which deviates so much
c Springer International Publishing AG 2017
F.J. Mart´ınez de Pis´on et al. (Eds.): HAIS 2017, LNAI 10334, pp. 402–412, 2017.
DOI: 10.1007/978-3-319-59650-1 34
288 Paper 2.5.2
Extraction of Outliers from Imbalanced Sets 403
from other observations as to arouse suspicions that it was generated by a different
mechanism’ [11]. The other observations are usually called inliers, normal
data or normal observations. Throughout the text, a predetermined battery of
features characterizes an observation.
The outlier analysis is used in a wide variety of domains such as the financial
industry, quality control, fault diagnosis, intrusion detection, web analytics, and
medical diagnosis [2]. The most typical application of the outlier analysis is
data cleaning. However, in many applications, outliers are more interesting than
inliers. Fraud detection is a classic example, where attention focuses on the
outliers, because these more likely represent cases of fraudulent behaviour [12].
The outlier analysis distinguishes three categories of outliers that require
specific analytical approaches: global outliers, contextual outliers (known also as
conditional outliers), and collective outliers [10]. A global outlier is an observation
that deviates significantly from the rest of the dataset, whereas a contextual
outlier deviates from inliers only with respect to a specific context. The term
collective outlier is used for a subset of observations. A subset of observations
forms a collective outlier if the subset as a whole deviates significantly from the
entire dataset.
To identify outliers, the outlier detection methods create models of normal
patterns in the data (so called data model or simply model), and then compute
an outlier score of a given observation on the basis of the deviations from the
normal patterns [2]. The outlier detection methods utilize clustering models,
distance-based models, density-based models, probabilistic and statistical models,
classification models, and information-theoretic models [2,10].
The selection of the model and outlier score calculation is data-specific and
relies on assumptions of information contained in the data. For example, classification
models require datasets of labelled observations. Methods based on other
models, e.g. statistical models or distance-based models, can be applied to both
labelled as well as unlabelled datasets.
The correct choice of the method from the perspective of the data model
determines results of the outlier analysis [2]. For example, application of a
method based on a statistical model, which expects a uniform distribution of
inliers, would be inappropriate for a dataset with the zipf distribution.
In biology, animal group behaviour studies generate specific datasets of
observable variables pre-selected in the experimental design [5,18]. Typically,
such datasets are unlabelled and may contain numerical as well as categorical
data. Given the complexity of animal behaviour, feature space of the observed
variables will not be exhaustive on an individual level and determinants of group
behaviour will exhibit subtle trends. From amongst the data mining approaches,
the outlier analysis provides functionality to identify observations putatively
generated by an alternative mechanism, which makes the analysis suitable for
application in animal group behaviour research. In order to ensure a simple and
reliable recognition of the outliers in such a dataset, we developed an outlier
detection method. Our method detects global outliers using a distance-based
model. Here, we introduce the method for numerical data.
Paper 2.5.2 289
404 P. ˇSkrab´anek and N. Mart´ınkov´a
2 Methods
2.1 Analysis of the Problem
A dataset considered for application with the proposed method contains inliers
that form one multidimensional cluster, while the outliers span at a distance
from the cluster centre. The outliers may or may not form small clusters. The
total number of observations in the dataset range from tens to hundreds of
observations. Further, distribution of inliers may significantly vary among various
datasets. The data contains no prior knowledge about the outliers, and the
information embodied in the outliers is the object of interest. These datasets may
include both numerical and categorical data; however, the proposed method is
intended for datasets composed of numerical data.
The first step in developing a new outlier detection method is identification
of the outlier category. Following the above stated setup, the proposed method
detects global outliers. The second step, selection of the model for inliers, delineates
the direction of the development process. Herein, we use backward selection
to select the proper model. Information-theoretic models are impropriate
for the defined datasets, because of the expected type of features. Without prior
knowledge about the outliers, the new detection method cannot be based on
a classification model. As different datasets may have different distributions of
inliers, usage of a probabilistic, density-based or a statistical model is inadvisable.
Consequently, the method has to be based on one of the remaining model
types; clustering or distance-based models.
Both clustering models and distance-based models represent appropriate
choices for the new outlier detection method given the data. Between them,
distance-based methods enable a higher granularity of analysis as compared to
clustering methods. This property of distance-based methods provides a more
refined ability to distinguish between weak and strong outliers in noisy data sets
[2]. Hence, the presented method has been developed on a distance-based model.
2.2 Description of the Method
Let X = {x1, . . . , xn} be a set of n observations x. The i-th observation xi,
where i ∈ I and I = {1, . . . , n}, is a d-dimensional real vector xi = (xi1, . . . , xid)
of features x ∈ F, where xik is the k-th feature of the i-th observation, and F
is a feature space. Let us expect that the majority of the observations, say m,
belongs to inliers. The remaining p observations correspond to outliers. A subset
of all outliers in the set X will be denoted as O.
The presented method belongs to the group of outlier detection methods
based on a distance model. It assumes two attributes in the observations x ∈ X:
(I) inliers form one predominant cluster and the outliers are at distances from
the centre of the cluster,
(II) the number of inliers is much higher than the number of outliers (m >> p).
290 Paper 2.5.2
Extraction of Outliers from Imbalanced Sets 405
In order to design the separation method, the outlier score had to be properly
formulized. For this purpose, an appropriate similarity measure had to be chosen.
Similarity of two observations, say xi, xj ∈ X, was assessed using a distance
measure. In order to ensure a comparable level of impact for all the features
x ∈ F, the observations should be compared with normalized data or the measure
should be unitless and scale-invariant. In our solution, we used Mahalanobis
distance [4,7]. This measure is unitless and scale-invariant. For the observations
xi, xj, the Mahalanobis distance is defined as
d (xi, xj) = (xi − xj)S−1
(xi − xj) , (1)
where S is a covariance matrix, and symbolizes transposition.
Considering the properties of the datasets expressed via the assumptions (I)
and (II), we proposed to formulate the outlier score J for the i-th observation
as the sum of distances between the i-th observation and the others, i.e.
Ji =
∀j∈I
d (xi, xj). (2)
The distance-based methods usually take into account distances between an
evaluated observation and its k nearest neighbours. Nevertheless, the outlier
score (2) considers all n distances. An example demonstrates rationalization for
the formulation of the score. In Fig. 1, the number of distances is identical regardless
of whether an inlier (Fig. 1a) or an outlier (Fig. 1b) is evaluated; however,
distributions of their values differ. For the outliers, longer distances appear more
frequently than for inliers. This holds for an arbitrarily chosen inlier and outlier,
since inliers form a single cluster and m >> p.
The specific properties of the dataset lead to the conclusion that the greater
the number of nearest neighbours which are included in the analysis, the larger
the difference between scores of inliers and outliers. Consequently, inclusion of
all observations in the comparison results in higher sensitivity of the method.
The associated increase in computational complexity of the method is irrelevant
for the expected dataset sizes. For larger datasets, a GPU optimized variant of
the method may be developed [3].
Observations evaluated using the outlier score (2) can be easily classified as
outliers or inliers using a threshold value t. In our case, the unusual structure of
the dataset X inspired the analytical expression of t. Indeed, values of the score
for inliers are markedly smaller than for outliers. Considering this fact and the
fact that m >> p, median of the score’s values ˆJ adequately describes inliers.
On the basis of the median and the smallest score values, the range of score
values of inliers can be estimated. Thus, the threshold value can be expressed as
t = ε.[ ˆJ − min
∀i∈I
Ji] + ˆJ, (3)
where the parameter ε is used as a tuning parameter. Each observation with a
score equal to or greater than the threshold value t is expected to be an outlier.
Paper 2.5.2 291
406 P. ˇSkrab´anek and N. Mart´ınkov´a
Fig. 1. Demonstration of the idea behind the outlier score by evaluation of: (a) an
inlier, and (b) an outlier. In both figure panels, three outliers (diamonds) and seven
inliers (circles) are plotted on a two-dimensional centred, rotated and standardized
feature space ˜x1, ˜x2 ∈ ˜F. The distance between two observations is symbolized using
a red line. (Color figure online)
Fig. 2. Visualization of the score J as a function j where j are indexes of the observations
sorted according to J in the ascending order. Threshold for outlier classification
t is placed at a point with rapid change in score trend
The presented method has one tuning parameter ε, and its setting significantly
predetermines the output of the method. We proposed a visualization
method in order to estimate the accurate value for ε. The visualization displays
a continuous line connecting scores J for ∀x ∈ X, where the scores are sorted
in ascending order. The line is approximately exponential. The initial lag phase
with gradual increase in J, includes inliers, and the subsequent exponential phase
includes outliers. Using the graph, an expert can estimate the boundary between
inliers and outliers and accordingly the threshold value determining ε (Fig. 2).
For datasets satisfying the assumptions, the right placement of the auxiliary
line is straightforward. However, the more the dataset X deviates from the ideal,
the deeper understanding of the data is necessary for the appropriate placement.
2.3 Algorithmic Expression of the Method
The proposed method can be realized as a function, here presented as a
pseudocode (Algorithm 1). The function has two inputs and two outputs. The
inputs are the set of observations X and the tuning parameter ε. The outputs
are a set of all outliers O and a set of outliers indexes Io in the original set X.
292 Paper 2.5.2
Extraction of Outliers from Imbalanced Sets 407
Algorithm 1. Extraction of Exceptional Observations
1: function EEO(X, ε)
Input: Set of n observations X = {x1, . . . , xn}, constant ε specifying the limit for
outliers
Output: Set O of all outliers and set of their indexes Io in X
2: Ji ← ∀j∈I d (xi, xj), ∀i ∈ I Evaluation of observations using the criterion
3: t ← ε.[ ˆJ − min∀i∈I Ji] + ˆJ Threshold value for exceptional observations
4: Io ← {i : i ∈ I where Ji ≥ t} Indexes of exceptional observations
5: O ← {xi : i ∈ Io} Set of exceptional observations
6: return O, Io
7: end function
2.4 Experimental Evaluation of the Method
We used 44 previously published datasets for evaluation of the proposed method
[8]. The datasets originated from areas such as biology, medicine, criminology
or astronautics. They contained three types of features: R - real numbers, I integers,
and N - nominal values. The datasets consisted of labelled samples
with two classes. All datasets were imbalanced with an imbalance ratio IR =
m/p, where IR ∈ [1.82, 129.44]. We expected that the minority class represented
outliers, while the majority class consisted of inliers.
We adapted three performance measures used in binary classification to evaluate
results obtained from our method. Namely, we considered sensitivity (Se),
specificity (Sp), and their geometric mean (G) [8,15]. For the outlier analysis,
they can be expressed as
Se =
|TO|
|TO| + |FI|
, Sp =
|TI|
|TI| + |FO|
, G = Se · Sp, (4)
where |TO| is the number of correctly recognized outliers (true outliers), |FO| is
the number of inliers labelled as outliers (false outliers), |TI| is the number of
correctly recognized inliers (true inliers), |FI| is the number of outliers labelled
as inliers (false inliers).
We evaluated our method (extraction of exceptional observations, EEO) for
two values of ε. Within the first experiment, we estimated the value of ε from the
graph (as per Fig. 2). This value was denoted as ˆε. In the second experiment, we
searched for an optimal setting (ε∗
) using genetic algorithms [16]. The genetic
algorithms used the objective function as max G(ε). We applied the MATLAB
function ga, with no constraints and default settings [1].
3 Results
Due to the presence of nominal variables, EEO could not be applied on datasets
‘abalone19’ and ‘abalone9–18’. Further, it was unsuccessful on datasets ‘ecoli-
0 vs 1’ and ‘segment0’, in which some features had constant values for all observations.
The obtained results are summarized in Table 1. In general, sensitivity of
Paper 2.5.2 293
408 P. ˇSkrab´anek and N. Mart´ınkov´a
EEO with manual estimation of ˆε was lower than for ε∗
, established from labelled
data with genetic algorithms in lieu of higher specificity. This was accompanied
by higher, and thus more conservative, values of ˆε.
To estimate the performance of EEO amongst existing outlier detection methods,
we compared our method with Chi et al.’s method with 3 and 5 labels (Chi-3
and Chi-5) [6], Ishibuchi et al.’s method (Ish05) [13], E-Algorithm (E-Alg) [19],
Fern´andez et al.’s method (HFRBCS) [8], and C4.5 decision tree (C4.5) [17]. We
adopted the evaluation results published in [8]. The evaluation results using G
are summarized for all expected methods, including the EEO with optimal and
manual setting of ε, in Table 2.
4 Discussion
The proposed EEO method was tested on 44 datasets previously used for algorithm
testing [8]. The datasets differed in the imbalance ratio, in the number
of features and their type (Table 1). However, from the viewpoint of EEO testing,
many of these datasets did not meet the assumptions that the inliers form
one multidimensional cluster (I) and the number of inliers is much higher than
the number of inliers (II). This is apparent when displaying the first two principal
components of observation scores with their class labels [9]. The dataset
‘shuttle-c0-vs-c4’ fully met the assumptions (Fig. 3b), and EEO was successful
in outlier detection (G ≈ 98). The datasets ‘ecoli-0-1-3-7 vs 2-6’, ‘shuttle-c2-vsc4’,
and ‘Wisconsin’ similarly showed nearly ideal class assignment with respect
to inliers (data not shown). On these datasets, EEO exhibited excellent results
according to all three measures (4) both for ˆε and ε∗
. The performance of EEO
is fully comparable to all evaluated methods. In fact, EEO provides considerably
better separation on ‘ecoli-0-1-3-7 vs 2-6’ dataset than any other considered
method (Table 2).
Fig. 3. Example of (a) inappropriate and (b) ideal datasets. The outliers (red diamonds)
and inliers (black circles) are plotted in a two-dimensional centred, rotated
and standardized feature space using the first two principal components. (Color figure
online)
294 Paper 2.5.2
Extraction of Outliers from Imbalanced Sets 409
Table 1. Evaluation of EEO on test datasets using sensitivity Se, specificity Sp, and
their geometric mean G. In first four columns, basic information about datasets is
listed. It includes name, feature type (R - real numbers, I - integers, and N - nominal
values), number of observations n, and imbalance ratio IR. The remaining columns
consist of evaluation results for estimated and suboptimal setting of ε, respectively.
Information about datasets EEO with ˆε EEO with ε∗
Name R/I/N n IR ε Se Sp G ε Se Sp G
abalone19 7/0/1 4174 129.44 - - - - - - - abalone9-18
7/0/1 472 16.4 - - - - - - - ecoli-0
vs 1 7/0/0 220 1.86 - - - - - - - -
ecoli-0-1-3-
7 vs 2-6
7/0/0 281 39.14 0.60 71.43% 86.13% 78.44% 2.47 71.43% 98.91% 84.05%
ecoli1 7/0/0 336 3.36 0.70 20.78% 83.78% 41.72% 0.04 59.74% 55.98% 57.83%
ecoli2 7/0/0 336 5.46 0.90 9.62% 86.97% 28.92% 0.26 57.69% 64.79% 61.14%
ecoli3 7/0/0 336 8.6 0.90 8.57% 87.04% 27.31% 0.06 65.71% 55.15% 60.20%
ecoli4 7/0/0 336 15.8 0.90 65.00% 90.82% 76.83% 0.75 70.00% 87.34% 78.19%
glass0 9/0/0 214 2.06 1.30 10.00% 68.75% 26.22% −0.30 51.43% 25.69% 36.35%
glass-0-1-2-
3 vs 4-5-6
9/0/0 214 3.2 1.30 52.94% 84.66% 66.95% 0.88 86.27% 83.44% 84.84%
glass-0-1-
6 vs 2
9/0/0 192 10.29 1.60 17.65% 77.71% 37.03% −0.21 88.24% 37.71% 57.69%
glass-0-1-
6 vs 5
9/0/0 184 19.44 1.40 22.22% 78.86% 41.86% 0.85 55.56% 69.71% 62.23%
glass1 9/0/0 214 1.82 1.30 19.74% 73.19% 38.01% −0.22 56.58% 34.78% 44.36%
glass2 9/0/0 214 11.59 1.00 11.76% 85.28% 31.67% 0.12 41.18% 55.84% 47.95%
glass4 9/0/0 214 15.47 1.30 46.15% 77.11% 59.66% 0.59 92.31% 66.67% 78.45%
glass5 9/0/0 214 22.78 1.30 22.22% 75.61% 40.99% 0.88 55.56% 67.80% 61.38%
glass6 9/0/0 214 6.38 1.30 65.52% 82.16% 73.37% 0.87 96.55% 76.76% 86.09%
haberman 0/3/0 306 2.78 0.60 14.81% 95.11% 37.54% 0.13 46.91% 66.22% 55.74%
iris0 4/0/0 150 2 0.80 10.00% 81.00% 28.46% −0.22 86.00% 30.00% 50.79%
new-thyroid1 4/1/0 215 5.14 0.70 65.71% 83.89% 74.25% 0.31 80.00% 74.44% 77.17%
new-thyroid2 4/1/0 215 5.14 0.70 65.71% 83.89% 74.25% 0.23 82.86% 71.11% 76.76%
page-blocks0 4/6/0 5472 8.79 0.70 85.87% 83.88% 84.87% 0.69 86.05% 83.68% 84.85%
page-blocks-
1-3 vs 4
4/6/0 472 15.86 0.70 53.57% 78.60% 64.89% 0.46 71.43% 71.40% 71.41%
pima 8/0/0 768 1.87 0.90 20.15% 88.40% 42.20% 0.10 61.57% 67.20% 64.32%
segment0 19/0/0 2308 6.02 - - - - - - - -
shuttle-c0-vs-
c4
0/9/0 1829 13.87 0.70 100.00% 95.96% 97.96% 0.77 100.00% 97.01% 98.49%
shuttle-c2-vs-
c4
0/9/0 129 20.5 0.50 100.00% 88.62% 94.14% 0.97 100.00% 96.75% 98.36%
vehicle0 0/18/0 846 3.25 0.50 16.08% 90.42% 38.13% 0.05 53.27% 60.59% 56.81%
vehicle1 0/18/0 846 2.9 0.40 9.68% 82.83% 28.31% −0.02 48.85% 46.10% 47.46%
vehicle2 0/18/0 846 2.88 0.40 18.35% 85.83% 39.68% 0.14 31.65% 65.61% 45.57%
vehicle3 0/18/0 846 2.99 0.40 10.85% 83.28% 30.06% −0.12 70.75% 37.54% 51.54%
vowel0 10/3/0 988 9.98 0.40 48.89% 86.64% 65.08% 0.20 68.89% 72.72% 70.78%
wisconsin 0/9/0 683 1.86 0.50 97.07% 88.51% 92.69% 1.15 93.72% 94.37% 94.05%
yeast-0-5-6-7-
9 vs 4
8/0/0 528 9.35 0.50 37.25% 81.34% 55.05% 0.30 54.90% 72.54% 63.11%
yeast1 8/0/0 1484 2.46 0.50 18.41% 78.58% 38.04% −0.03 48.48% 45.31% 46.87%
yeast-1 vs 7 7/0/0 459 14.3 0.70 40.00% 85.31% 58.42% 0.24 53.33% 66.67% 59.63%
yeast-1-2-8-
9 vs 7
8/0/0 947 30.57 0.70 40.00% 83.32% 57.73% 0.39 53.33% 72.30% 62.10%
yeast-1-4-5-
8 vs 7
8/0/0 693 22.1 0.70 10.00% 82.50% 28.72% −0.13 73.33% 40.57% 54.55%
yeast-2 vs 4 8/0/0 514 9.08 0.60 50.98% 84.45% 65.61% 0.18 84.31% 69.76% 76.69%
yeast-2 vs 8 8/0/0 482 23.1 0.60 70.00% 81.82% 75.68% 1.20 65.00% 93.94% 78.14%
yeast3 8/0/0 1484 8.1 0.50 25.15% 80.02% 44.86% −0.01 69.33% 51.40% 59.69%
yeast4 8/0/0 1484 28.1 0.50 45.10% 80.32% 60.19% 0.16 78.43% 62.11% 69.79%
yeast5 8/0/0 1484 32.73 0.70 34.09% 86.74% 54.38% 0.06 100.00% 55.69% 74.63%
yeast6 8/0/0 1484 41.4 0.70 22.86% 86.34% 44.42% −0.04 91.43% 47.83% 66.13%
Paper 2.5.2 295
410 P. ˇSkrab´anek and N. Mart´ınkov´a
Table 2. Comparison of EEO with other approaches for outlier detection. The geometric
mean G of sensitivity and specificity was used as an overall comparison value.
Results obtained by EEO are in bold on relevant datasets that meet designed criteria
(inliers form a predominant cluster with outliers spanned from it and the number of
inliers is greater than the number of outliers)
Dataset Chi-3 Chi-5 Ish05 E-Alg HFRBCS C4.5 EEO
ˆε ε∗
abalone19 62.69% 66.71% 66.09% 0.00% 70.19% 15.58% - abalone9-18
63.93% 66.47% 65.78% 32.29% 67.56% 53.19% - ecoli-0
vs 1 92.27% 95.56% 96.70% 95.25% 93.63% 67.95% - ecoli-0-1-3-7
vs 2-6 71.04% 49.57% 71.31% 73.65% 71.48% 71.21% 78.44% 84.05%
ecoli1 85.28% 86.05% 85.71% 77.81% 84.18% 76.10% 41.72% 57.83%
ecoli2 88.01% 87.64% 87.00% 70.35% 87.62% 91.60% 28.92% 61.14%
ecoli3 87.58% 91.61% 85.39% 78.54% 90.81% 88.77% 27.31% 60.20%
ecoli4 91.27% 92.11% 86.92% 92.43% 93.02% 81.28% 76.83% 78.19%
glass0 64.06% 63.69% 69.39% 0.00% 76.57% 78.14% 26.22% 36.35%
glass-0-1-2-3 vs 4-5-6 85.83% 85.94% 88.56% 82.09% 88.37% 90.13% 66.95% 84.84%
glass-0-1-6 vs 2 40.84% 56.17% 41.18% 0.00% 58.37% 48.91% 37.03% 57.69%
glass-0-1-6 vs 5 71.48% 75.59% 88.77% 65.14% 77.96% 72.08% 41.86% 62.23%
glass1 64.90% 64.91% 59.29% 0.00% 73.66% 75.11% 38.01% 44.36%
glass2 47.67% 49.24% 43.55% 9.87% 54.84% 33.86% 31.67% 47.95%
glass4 84.96% 81.75% 78.27% 83.38% 70.39% 83.71% 59.66% 78.45%
glass5 81.56% 64.33% 89.96% 50.61% 68.73% 86.70% 40.99% 61.38%
glass6 83.87% 78.13% 86.27% 90.23% 86.95% 83.00% 73.37% 86.09%
haberman 58.91% 60.40% 62.65% 4.94% 57.08% 61.32% 37.54% 55.74%
iris0 100.00% 98.97% 100.00% 100.00% 100.00% 98.97% 28.46% 50.79%
new-thyroid1 87.44% 95.38% 89.02% 88.52% 95.58% 97.98% 74.25% 77.17%
new-thyroid2 89.81% 96.34% 94.21% 88.57% 99.72% 96.51% 74.25% 76.76%
page-blocks0 79.91% 87.25% 32.16% 64.51% 91.40% 94.84% 84.87% 84.85%
page-blocks-1-3 vs 4 91.92% 92.93% 94.53% 94.12% 98.64% 99.55% 64.89% 71.41%
pima 66.80% 66.78% 71.10% 55.01% 68.72% 71.26% 42.20% 64.32%
segment0 94.99% 95.88% 42.47% 95.33% 97.51% 99.26% - shuttle-c0-vs-c4
99.12% 98.71% 99.16% 98.40% 99.12% 99.97% 97.96% 98.49%
shuttle-c2-vs-c4 89.99% 78.34% 99.17% 100.00% 97.49% 99.15% 94.14% 98.36%
vehicle0 86.41% 84.93% 75.94% 39.07% 88.92% 91.10% 38.13% 56.81%
vehicle1 70.92% 71.88% 64.89% 3.09% 71.76% 69.28% 28.31% 47.46%
vehicle2 85.54% 87.19% 67.82% 43.83% 90.61% 94.85% 39.68% 45.57%
vehicle3 69.22% 63.13% 63.12% 0.00% 66.80% 74.34% 30.06% 51.54%
vowel0 98.37% 97.87% 89.03% 89.63% 98.82% 94.74% 65.08% 70.78%
wisconsin 88.91% 43.58% 95.78% 96.01% 88.24% 95.44% 92.69% 94.05%
yeast-0-5-6-7-9 vs 4 78.91% 75.99% 79.49% 59.99% 73.18% 74.88% 55.05% 63.11%
yeast1 67.69% 69.66% 51.41% 0.00% 71.71% 70.86% 38.04% 46.87%
yeast-1 vs 7 80.05% 63.02% 53.15% 27.55% 70.74% 67.73% 58.42% 59.63%
yeast-1-2-8-9 vs 7 76.12% 69.26% 48.55% 50.00% 69.37% 64.13% 57.73% 62.10%
yeast-1-4-5-8 vs 7 62.40% 58.76% 40.80% 0.00% 62.49% 41.19% 28.72% 54.55%
yeast-2 vs 4 86.80% 86.39% 70.85% 80.92% 89.32% 85.09% 65.61% 76.69%
yeast-2 vs 8 72.75% 78.76% 72.83% 72.83% 72.47% 78.23% 75.68% 78.14%
yeast3 90.13% 89.33% 77.06% 81.99% 90.41% 88.50% 44.86% 59.69%
yeast4 82.99% 83.07% 71.36% 32.16% 82.64% 65.00% 60.19% 69.79%
yeast5 93.41% 93.64% 94.94% 88.17% 94.20% 92.04% 54.38% 74.63%
yeast6 87.50% 87.73% 88.42% 51.72% 84.92% 80.38% 44.42% 66.13%
296 Paper 2.5.2
Extraction of Outliers from Imbalanced Sets 411
Good results were obtained also on other datasets, e.g. on ‘glass6’ (Fig. 3a),
‘new-thyroid1’, or ‘yeast-2 vs 8’. Here, a majority of the inliers were concentrated
near the cluster center; however, many inliers (their number was similar
to the total number of outliers) were interspersed with the outliers. In such cases,
estimation of ε became vague and perfect separation was not possible. Thus, the
good EEO results on these datasets were coincidental and the presented method
was not suited for them.
While the threshold values t for outlier detection can be directly set from the
sorted distance visualization, estimating ε will represent a good practice in data
reporting. The ε value defines the position of the outliers relative to the median,
providing a data-independent approximation on outlier distribution comparable
between studies.
The presented method was based on the Mahalanobis distance (1). While
the distance was efficient for the proposed problem, we found the method to be
computationally extravagant. Thus, we suggest an alternative approach based
on the Euclidean distance for application where computational intensity would
be of concern. The Euclidean distance in combination with standard deviation
normalization [14] might provide equally good results while its time-complexity
would be considerably lower.
5 Conclusion
The outlier analysis has the potential to mine valuable information from a complex
dataset, but its sensitivity and specificity is dependent on both suitability
of the method and the model, to the data. We designed EEO for the specifics of
animal group behaviour observations, where the outliers could reveal alternative
mechanisms determining group behaviour. Our testing on varied imbalanced sets
demonstrated that the utility of the method is wider. The EEO was able to correctly
classify outliers in datasets from engineering, microbiology or medicine.
We therefore conclude that global outliers may be detected with EEO based on
the threshold estimated from sums of pairwise Mahalanobis distances in datasets
across fields that form one predominant multidimensional cluster with outliers
distanced from it.
Acknowledgments. The work was supported by the University of Pardubice (PˇS)
and the Czech Science Foundation grant number 17-20286S (NM).
References
1. MATLAB: Global optimization toolbox (R2016a) (2016). https://www.
mathworks.com/help/gads/index.html
2. Aggarwal, C.C.: Outlier Analysis. Springer, New York (2013)
3. Angiulli, F., Basta, S., Lodi, S., Sartori, C.: GPU strategies for distance-based
outlier detection. IEEE Trans. Parallel Distrib. Syst. 27(11), 3256–3268 (2016)
4. Brereton, R.G.: The Mahalanobis distance and its relationship to principal component
scores. J. Chemometr. 29(3), 143–145 (2015)
Paper 2.5.2 297
412 P. ˇSkrab´anek and N. Mart´ınkov´a
5. Broom, D.M., Fraser, A.F.: Domestic Animal Behaviour and Welfare, 4th edn.
CABI, Wallingford (2015)
6. Chi, Z., Yan, H., Pham, T.: Fuzzy Algorithms: With Applications to Image Processing
and Pattern Recognition, vol. 10. World Scientific, Singapore (1996)
7. Deza, M.M., Deza, E.: Encyclopedia of Distances, 3rd edn. Springer, Heidelberg
(2014)
8. Fernndez, A., del Jesus, M.J., Herrera, F.: Hierarchical fuzzy rule based classification
systems with genetic rule selection for imbalanced data-sets. Int. J. Approximate
Reasoning 50(3), 561–577 (2009)
9. Gower, J., Lubbe, S., Roux, N.: Understanding Biplots. Wiley, New York (2010)
10. Han, J., Kamber, M., Pei, J.: Data Mining, 3rd edn. Morgan Kaufmann, San
Francisco (2012)
11. Hawkins, D.M.: Identification of Outliers. Springer, Netherlands (1980)
12. Hawkins, S., He, H., Williams, G., Baxter, R.: Outlier detection using replicator
neural networks. In: Kambayashi, Y., Winiwarter, W., Arikawa, M. (eds.) DaWaK
2002. LNCS, vol. 2454, pp. 170–180. Springer, Heidelberg (2002). doi:10.1007/
3-540-46145-0 17
13. Ishibuchi, H., Yamamoto, T.: Rule weight specification in fuzzy rule-based classification
systems. IEEE Trans. Fuzzy Syst. 13(4), 428–435 (2005)
14. Kantardzic, M.: Data Mining: Concepts, Models, Methods, and Algorithms, 2nd
edn. Wiley, Hoboken (2011)
15. Kohl, M.: Performance measures in binary classification. Int. J. Stat. Med. Res.
1(1), 79–81 (2012)
16. Reeves, C.R., Rowe, J.E.: Genetic Algorithms: Principles and Perspectives: A
Guide to GA Theory. Kluwer Academic Publishers, Norwell (2002)
17. Salzberg, S.L.: C4.5: programs for machine learning by J. Ross Quinlan. Morgan
Kaufmann Publishers, Inc., 1993. Mach. Learn. 16(3), 235–240 (1994)
18. Ward, A., Webster, M.: Sociality: The Behaviour of Group-Living Animals.
Springer International Publishing, Heidelberg (2016)
19. Xu, L., Chow, M.Y., Taylor, L.S.: Using the data mining based fuzzy classification
algorithm for power distribution fault cause identification with imbalanced data.
In: 2006 IEEE PES Power Systems Conference and Exposition. pp. 1228–1233,
October 2006
298 Paper 2.5.2