03/05/2016
1
Genetic structure of populations,
drift, mutations
• Drift
→ differentiation of populations
random changes in allele
frequencies (may lead to fixation
of alternative alleles)
• Mutations & selection
increase differentiation
AA
AA
AA
AA
AA
AA
aa
Aa
AA
aa
Aa
AA
aa
Aa
Aa
aa
AA
AA
aa
aa
aa
aa
aa
aa
Aa
AaAa
Aa
drift
Gene flow
- acts against differentiation of subpopulations
AB
ac
GENE FLOW
= the transfer of genes/alleles from one
population to another
→ CHANGE OF ALLELIC FREQUENCIES
03/05/2016
2
MIGRATION versus GENE FLOW
• movement of individuals between
pops
• immigrants may not be
reproducing in a new pop! (even
a strong migration/dispersal does
not mean necessarily any gene
flow)
• detectable (with substatntial
difficulties) by direct ecological
methods
• movement of alelles (genes) between
pops
• via dispersion of individuals,
propagules (gametes – pollen, seeds)
• passive in plants, mostly active in
animals
• if strong → homogenization of allele
frequencies between the pops
• prevents pop differentiation,
divergence of pos, establishment of
pop structure, and ultimately to
speciation ---- by mixing the genepools
• prevents decrease of abilitiy to
survive due to inbreeding
• estimable from genetic data
Quantifying gene flow
1. Direct methods:
• observation
• Capture-Mark-Recapture sampling
• telemetry
2. Indirect methods – methods of population genetics
 we have information about pop structure (expected
subpopulations or estimated from genetic data)
 based on distribution of genetic variation
 based on deviations from Hardy-Weinberg equilibrium
 estimation based on FST
 model-based methods based on the coalescent theory (eg.
MIGRATE software)
03/05/2016
3
http://popgen.sc.fsu.edu/Migrate/Migrate-n.html
IBD
03/05/2016
4
Models of gene flow
• island model
(Wright 1931)
assume same size of subpops
assume symetrical flow of genes
assume equal probability of gene
exchange between subpops
• stepping stone model
(Kimura 1953)
exchange only between adjacent
subpops
• Private alleles (Slatkin 1985) – useful for highly polymorphic markers
= alleles occuring only in a single subpopulation
p(1) - frequency of private alleles
lnp(1) = -0,505 ln(Nem) - 2.44
• F statistics
mN
F
e
ST
41
1


Nem = number of adult, reproducing migrants between subpops per a
generation (island model assumed!)
It is just a rough estimation at a scale of „few“ and „a lot“
(only for Fst > 0.05-0.10)
Assumptions for using Nem:
• island model (= infinite number of subpops, no natural selection,
equal size of all subpops, equal probability of migrant exchange
between all subpops)
• migration-drift equilibrium (= no population expansion, no habitat
fragmentation, no population bottleneck)
03/05/2016
5
but be aware!!!
• even in a case of two very very distant populations
• FST → will never be equal to zero, Nem → there had been exchange
of individuals in the past
• even pops which have never exchanged any migrants will have
never Nem equals to zero
• extreme case of a complete genetic
isolation: Nm = 0, Fst = 1
• 1 migrant every forth generation: Nm
= 0.25, Fst = 0.5
• 1 migrant every second generation:
Nm = 0.5, Fst = 0.33
• 1 migrant every generation: Nm = 1,
Fst = 0.2
• 2 migrants every generation: Nm =
2, Fst = 0.11
Inference of Recent Migration
• BayesAss: Bayesian Inference of Recent Migration
Using Multilocus Genotypes
• Reference: G.A. Wilson and B. Rannala 2003. Bayesian
inference of recent migration rates using multilocus
genotypes. Genetics 163: 1177-1191.
• http://www.rannala.org/?page_id=245
03/05/2016
6
Assignment tests
• assign individuals to their most likely population of origin
• done by comparison of individual genotypes to the genetic profiles of
various populations
• vs Nem based indirect methods: not comparing overall genetic
similarities between pops, but a maximum likelihood method to
estimate probabilities that a given genotypes arose from alternative
pops (Paetkau et al. 1995)
• all pops are assumed to be in HWE and the loci not in LD
03/05/2016
7
Population assignment tests
• program GeneClass (Piry et al. 2004)
• estimates probabilities of a certain genotype being from a certain predefined
population – identification of recent migrants or samples of
unknown origin (fight against poaching)
• may combine data of various genetic markers
Depends on the level of genetic difference
between populations
5 microsatellite loci
Fst = 0.14
99.9% assigned correctly
5 microsatellite loci
Fst = 0.04
90.2% assigned correctly
03/05/2016
8
Subspecies identification of
chimpanzees in Czech ZOOs
• chimpanzees in ZOOs often of
unclear origin
• genetic data from natural
populations are available (300
msats, Becquet et al. 2007)
• 30 most informative microsatellites
– genotypization of all
chimpanzees in CZ
• GeneClass: assignment to the
subspecies/populations
Mapua et al. (2011)
• some individuals are genetically clearly assigned to ESU
(Evolutionary Significant Units = subspecies) – Zoo in
Liberec, Dvůr Králové
• but also quite a few of hybrids (mainly Ostrava, Brno,
etc.)
Mapua et al. (2011)
03/05/2016
9
BayesASS
GeneClass 2
• Giant panda
-Bayesian estimates of gene flow over few last generations
-identification of two possible first-generation migrants
-recommendations for conservation management – migration corridor construction
Zhu et al. 2011 Mol Ecol
Models of gene flow
• Island model
(Wright 1931)
assume same size of subpops
assume symetrical flow of genes
assume equal probability of gene exchange
between subpops
• Stepping stone model
(Kimura 1953)
exchange only between adjacent subpops
• Isolation by distance
Gene flow rate dicreases with increasing
distance between subpops
03/05/2016
10
Isolation by distance (IBD)
= the amount of gene flow between pops is inversely
proportional to the geographic distances between them
• Sewall G. Wright (1943)
• regression of log-transformed gene flow estimate (eg.
FST) and appropriate log-transformed geographic
distances
• significance of correlation tested by Mantel test (does not
assume independent population pairwise comparisons)
• relevant geographical scale (depends on dispersal abilities)
• migration-drift equilibrium must occur
• IBD (isolation-by-distance) is not
– in very recently isolated populations
– in completely isolated populations
– in case of high amount of migration
IBD detection
• correlation between matrices of genetic
and geographic distances
• Mantel test
• e.g. Genepop
03/05/2016
11
Isolation by distance
Crotaphytus collaris
Hutchinson & Templeton 1999
no barriers for
tens of thousands
years
equilibrium
between drift
and migration
postglacially
fragmentation
incluence of drift
postglacially
no barriers
influence of
migration
postglacially
increasing
fragmentation
influence of
drift
at big scales
equilibrium
at small scales
Wellenreuther et al. 2010
03/05/2016
12
Population assignments
Classical problems of population genetics
• Populations are defined, individuals a priori
assigned to populations, we are interested in
population characteristics (F-statistics) → i.e. pop
genetic diversity and structure
• Populations are defined, but we want to assign
individuals of unknown origin to them
• Cryptic population structure = nothing is known at
the beginning → we want to estimate clusters (i.e.
natural homogeneous populations) and assign the
individuals to the clusters (population
assignments)
A. Direct methods
• morphological variation (geographical races)
• leg-bands or similar markers (ex. over one million Ficedula hypoleuca have
been ringed in UK and Sweden – only six recaptured on wintering grounds
in Africa
• satellite telemetry – expensive, not useful for small animals
B. Biogeochemical approaches
• ratios of stable isotopes of naturally occurring elements (C, H, N, Sr) vary
across the landscape
• determined by the relative frequency of C3 and C4 plants, climate, and
bedrock
(1) geographical structure of isotopic ratio distributions
(2) knowledge about where animals incorporate isotopes
(3) tissue samples from individuals at different parts of their annual cycle
03/05/2016
13
C. Genetic approaches
• « few birds have rings, but everybody has genotype »
• genetic data about population structure
• problems: (1) low genetic differentiation between pops
(intense dispersal), (2) low differentiation in temperate
zone – recent postglacial colonization
• Solutions: (1) use more genetic markers, (2) study of
parasite DNA (e.g. avian malaria) – parasites have
quicker evolution, are more differentiated
• cryptic population structure
• unknown number of clusters
• level of an individual
• identify clusters and assign individuals to them simultaneously
• we have individual genotypes (sometimes also geographical
coordinates)
• Data: msats (other codominant loci, SINE), AFLP
Individual-based assignments
03/05/2016
14
Dendrogram based on
microsatellite genotype
distances between
individuals
(Cavali-Sforza
distances)
May be biased in
case of too few
markers
03/05/2016
15
LANDSCAPE GENETICS
• approach combining population genetics, spatial statistics (GIS) and
landscape ecology
• aiming to quantify the influence of landscape features and
environmental variables on the distribution of allele
frequencies among populations
= to understand the relationship between habitats and gene flow
• „landscape“ – the area that the organism of interest is utilizing (ie.
number of various habitats of varying suitability)
• homogeneous vs. heterogeneous landscape ???
• homogeneous: panmictic population
• homogeneous, but larger than the dispersal distance of an
individual: IBD
• heterogeneous (ie. various habitats): gene flow in not equal
throughout the landscape
Spatially explicit analyses = spatial
genetics = landscape genetics
• based on Bayesian clustering approach (of
STRUCTURE type) – individual-based models
• for modelling is added information of both genetic
data and geographical coordinates
• e.g. programs BAPS, TESS, Geneland (the „best“
number of clusters – K – is estimated
automatically)
Bayesian spatial clustering
03/05/2016
16
Spatial models use Voronoi diagrams
Voronoi polygons, Dirichlet tessellation
- type of decomposition of metric space defined by distances to a given
discrete set of objects in space, e.g. a discrete set of points
- separation of plane according to a given set of points M
- Voronoi diagram is a separation of plane in such a way that each point b
from M is provided by an area V(b) whose all points are closer to the point
b than to any other point of M
G. F. Voronoi (1868-1908)
http://is.muni.cz/th/143320/fi_b_a2/animace/voroneho_diagram.html
http://ivankuckir.blogspot.cz/2011/03/voroneho-diagram-v-as3.html
The example of very fragmented populations: the
best model in BAPS for Central and Southern
Dinaromys populations (spatial clustering of
groups of individuals): K=13 (i.e. evidence of
very high structuration) Best Partition:
Cluster 1: {C9, C13}
Cluster 2: {S6}
Cluster 3: {C8, C14}
Cluster 4: {C4}
Cluster 5: {C1, C2}
Cluster 6: {S1, S2, S3, S4}
Cluster 7: {C6}
Cluster 8: {C3, C15}
Cluster 9: {C5, C7}
Cluster 10: {C10}
Cluster 11: {C11, C12}
Cluster 12: {S5}
Cluster 13: {C16}
program BAPS
software for Bayesian Analysis of genetic
Population Structure
http://www.helsinki.fi/bsg/software/
03/05/2016
17
GENELAND
Population genetic and
morphometric data analysis
using R and the Geneland
program
03/05/2016
18
R platform
Posterior
probability maps
Spatial population genetics Fontaine et al. 2007
Phocoena phocoena
03/05/2016
19
Comparison of features of various
„individual-based assignment“ programs
STRUCTURE vs. BAPs
Robust support of the population structure
K = 7
03/05/2016
20
Population structure - summary
Connected populations
(gene flow)
Isolated populations
(no gene flow)
Ne  
Genetic drift  
Genetic diversity  
Population
differentiation
 