LETTER Geographic range size is predicted by plant mating system
Dena Grossenbacher,1
* Ryan
Briscoe Runquist,1
Emma E.
Goldberg,2
and Yaniv Brandvain1
1
Department of Plant Biology University
of Minnesota St. Paul, MN 55108,
USA
2
Department of Ecology Evolution
and Behavior University of Minnesota
St. Paul, MN 55108, USA
*Correspondence:
E-mail: dgrossen@umn.edu
Abstract
Species’ geographic ranges vary enormously, and even closest relatives may differ in range size by
several orders of magnitude. With data from hundreds of species spanning 20 genera in 15 families,
we show that plant species that autonomously reproduce via self-pollination consistently have
larger geographic ranges than their close relatives that generally require two parents for reproduction.
Further analyses strongly implicate autonomous self-fertilisation in causing this relationship,
as it is not driven by traits such as polyploidy or annual life history whose evolution is sometimes
correlated with selﬁng. Furthermore, we ﬁnd that selfers occur at higher maximum latitudes and
that disparity in range size between selfers and outcrossers increases with time since their evolutionary
divergence. Together, these results show that autonomous reproduction—a critical biological
trait that eliminates mate limitation and thus potentially increases the probability of
establishment—increases range size.
Keywords
Baker’s Law, outcrossing, phylogeny, range size, selﬁng.
Ecology Letters (2015) 18: 706–713
INTRODUCTION
Why does one species range across an entire continent while
its close relative is narrowly distributed (Darwin 1859, p6;
Brown et al. 1996; Gaston 2003; ch3)? There are many potential
ecological and evolutionary explanations for the enormous
variation in species’ geographic distributions. These
explanations include species’ geographic location, age, niche
breadth, environmental tolerance, competitive ability and key
life history traits such as body size, dispersal ability and mating
system (reviewed by Brown et al. 1996; Gaston 2003).
Most of these potential explanations have been the subject of
intensive research across a wide range of taxa, but none are
universally supported. We show that plant species possessing
one key trait—the ability to autonomously reproduce via
self-pollination—consistently have larger geographic ranges
than their close relatives that generally require two parents
for reproduction.
To understand the potential impacts of autonomous reproduction,
or any trait, on species’ range size, we consider the
mechanistic processes underlying a species’ range. We envision
a species’ range as a set of occupied cells on a grid. Over time, a
local population (a cell) might go extinct because individuals
cannot tolerate the local environment, because they are outcompeted
by another species, or because of demographic stochasticity
(Sexton et al. 2009). As some local populations
disappear, new populations can emerge by range expansion—
the process by which species spread to new geographic locations.
To achieve this, individuals must ﬁrst disperse to new
sites. Upon arrival, individuals must tolerate potentially novel
biotic and abiotic environmental conditions (Sexton et al. 2009)
and, for most sexual species, individuals must ﬁnd a suitable
mate for reproduction and establishment (Baker 1955; Stebbins
1957; Pannell & Barrett 1998). Together, the action of local
extinction, recolonisation and range expansion generates a species’
range.
Given equivalent amounts of time for range expansion and
contraction, consistent differences in geographic range size
among species (beyond those expected by stochasticity) are
likely due to traits that inﬂuence one or more key factors: dispersal,
environmental tolerance and mate availability. For
example, increased body size in birds and mammals could
increase species’ environmental tolerance by allowing individuals
to maintain homeostasis over a wider range of environmental
conditions, and body size may increase dispersal
ability and home-range size (Gaston 2003; p106-113). Indeed,
many studies report that body size correlates positively with
species’ range size, although it generally explains only a small
fraction of the overall variation in range size (Gaston 2003;
Agosta et al. 2013). Similarly, self-fertilisation and other
forms of autonomous reproduction may affect the basic biological
factors that shape range size. Autonomous reproduction
directly allows species to overcome limits to range
expansion enforced by mate limitation, and it may inﬂuence
range size via indirect effects on dispersal mode (Cheptou &
Massol 2009; Hargreaves & Eckert 2014) and environmental
tolerance.
There are two primary and opposing hypotheses regarding
the effect of selﬁng, like other forms of autonomous reproduction,
on species’ geographic distributions. First, the reproductive
assurance provided by selﬁng is predicted to increase
successful colonisation and establishment (Baker 1955).
Selfers, unlike outcrossing species, are not mate-limited at low
population densities and may successfully reproduce when
even a single individual lands in a new habitat (Baker 1955;
Stebbins 1957; Pannell & Barrett 1998). Under this scenario,
species that are already selﬁng or that evolve selﬁng will be
able to rapidly expand their ranges and occupy habitats that
support only small populations or have unpredictable pollinators.
Selfers may also be more likely to establish upon colonising
new locations because autonomous reproduction can
shield plants from Allee effects (reviewed in Goodwillie et al.
© 2015 John Wiley & Sons Ltd/CNRS
Ecology Letters, (2015) 18: 706–713 doi: 10.1111/ele.12449
2005), and because the history of selﬁng can purge deleterious
recessive alleles that would otherwise be exposed in small, isolated
populations (Pujol et al. 2009; see Haag & Ebert 2004
for an exploration of this hypothesis in asexual populations).
The logical extension of this argument is that selfers will have
larger geographic ranges than outcrossers, given similar initial
range sizes and amounts of time for range expansion (Henslow
1879; p391; Lowry & Lester 2006; Randle et al. 2009).
An opposing hypothesis is that the limited genetic variation
within selﬁng species may constrain their ability to adapt to
many habitats and therefore result in more limited ranges compared
to their outcrossing relatives. For instance, self-pollinating
populations may have reduced genetic diversity relative to
outcrossers (Hamrick & Godt 1996; Crawford et al. 2008). This
lack of diversity may limit the niche breadth of selﬁng populations
and prevent them from colonising and adapting to novel
environments (Stebbins 1957; Hamrick & Godt 1996; Crawford
& Whitney 2010; Sheth & Angert 2014). Furthermore, although
species that reproduce autonomously may initially purge deleterious
recessive alleles, they cannot effectively eliminate mildly
deleterious mutations, which may accumulate via a ratchet-like
process (Heller & Smith 1978; Wright et al. 2013). Competitively
superior outcrossing relatives may therefore further
diminish the realised niche of selﬁng species. Ultimately, the
evolutionary genetic consequences of self-fertilisation may constrain
the geographic ranges of selfers relative to outcrossers
(Lowry & Lester 2006).
Here, we test these alternative hypotheses by asking
whether, across pairs of sister species of ﬂowering plants, selﬁng
or outcrossing plants have larger ranges. Data from 194
sister species across 20 genera and generic sections consistently
show that selﬁng species have larger ranges than their outcrossing
relatives. Further analyses strongly implicate autonomous
fertilisation in causing this strong relationship, as it is
not driven by traits such as polyploidy or annual life history
whose evolution is correlated with the transition to autonomous
self-fertilisation (Barrett et al. 1996; Barringer 2007;
Robertson et al. 2011). Together, our results show that autonomous
reproduction—a critical biological trait that inﬂuences
the probability of establishment—has a major inﬂuence on
range size.
To identify potential biological drivers of this pattern, we
consider two additional questions. First, we ask whether species’
latitudinal distributions are explained by mating system.
If selfers attain larger ranges by colonising and establishing
at extreme latitudes during interglacial periods (e.g. Jordaens
et al. 2000; Grifﬁn & Willi 2014), we expect selfers to be
found at higher-latitude locations than their outcrossing relatives.
Recently melted glaciers open new habitat for colonisation,
and the unpredictable pollinator environments at high
latitudes may favour establishment of species with autonomous
reproduction (Baker 1966; Lloyd 1980). In addition,
the lack of pathogens and competitors at high latitudes relative
to tropical regions may permit selﬁng (for a similar
hypothesis in asexual species distributions, see Glesener &
Tilman 1978; Bell 1982). We ﬁnd that selfers do indeed occur
at higher maximum latitudes, although we cannot rule out
the possibility that a shift in ploidy or life history contributes
to this ﬁnding.
Second, we ask how differences in range size between selfers
and their outcrossing relatives change with the time since their
divergence. Continuing range expansion or contraction in selfers
will generate an increasing disparity in range size between
selfers and outcrossers over time. We ﬁnd that with increasing
divergence time, the range size of selfers increases relative to
that of their closest outcrossing relatives, suggesting that
range expansion in selfers is ongoing.
MATERIALS AND METHODS
We identiﬁed taxa with a published, species-level phylogeny
containing at least one predominantly selﬁng or functionally
selﬁng species and one predominantly outcrossing species, and
with DNA sequence data for at least 50% of the species
within the clade available on GenBank (http://www.ncbi.nlm.
nih.gov/genbank/) to be used for constructing time-calibrated
phylogenies. After removing Leavenworthia, a small North
American genus in which our phylogenetic model did not
converge, we had 20 clades from 15 families whose combined
native distributions spanned every continent except Antarctica
(see Figure S1). On average, clades contained 35 Æ 7 (Æ 1 SE)
extant species, 80 Æ 4.6 percent of which were included in our
phylogenies. These time-calibrated, species-level phylogenies
across a diverse set of plant taxa allow us to test whether
mating system inﬂuences species’ range size, while controlling
for shared evolutionary history.
For the methods and analyses described below, all data and
R scripts are available on the Dryad Digital Repository, and
software used for each analysis is detailed in Table S1.
Estimating phylogenies
We reconstructed time-calibrated phylogenies for all 20 clades
because most previously published phylogenies were not time
calibrated and consisted of only a single topology or consensus
tree, making it difﬁcult to incorporate uncertainty into our
analysis. Prior to estimating the phylogenies, for each clade
separately we downloaded and aligned nrITS sequences for
species within the clade from GenBank (Table S1). We simultaneously
estimated phylogenetic relationships and absolute
divergence times among species in a Bayesian framework
(Table S1). Because fossils are not known for our focal clades,
we estimated absolute divergence times from the substitution
rate for herbaceous and woody plants at the nrITSlocus (Kay
et al. 2006). The substitution rate was set to a normally distributed
prior for herbaceous lineages with mean of 4.13 9 10À9
subs/site/yr and standard deviation of 1.81 9 10À9
, and for
woody lineages with mean of 2.15 9 10À9
subs/site/yr and
standard deviation of 1.85 9 10À9
.
To accommodate heterogeneity in the molecular evolutionary
rate among branches, we used an uncorrelated log-normal
relaxed clock model. The prior model on branch lengths was
set to a Yule process of speciation. The prior model on substitutions
and the number of Markov chain Monte Carlo
(MCMC) generations varied by clade (Table S2). Posterior
samples of parameter values were summarised and assessed
for convergence and mixing (Table S1). After removing Leavenworthia
(for which the MCMC did not converge, and which
© 2015 John Wiley & Sons Ltd/CNRS
Letter Range size is predicted by mating system 707
we excluded for all analyses) all MCMC chains for phylogenies
of our 20 clades had minimum estimated sum of squares
(ESS) for the posterior > 1100, and minimum ESS across all
other parameters > 600 (Table S2).
For all ensuing analyses, we identiﬁed sister species in a
subset of 9000 trees from the posterior distribution for each
clade. For each sister pair, we recorded the average divergence
time and the proportion of trees in which the two species were
sister, as a measure of phylogenetic uncertainty. Since our
phylogenies sampled, on average, only 80% of extant taxa,
these sister pairs may not represent “true” extant sisters, but
they are recently diverged species representing independent
evolutionary replicates.
Estimating mating system, ploidy and life history
We collated 54 studies describing mating systems of species
from the clades identiﬁed above. Most published studies classiﬁed
species as predominantly outcrossing, variable mating
or predominantly selﬁng. Species were classiﬁed as variable
mating when outcrossing rates were between 0.2 and 0.8, or
when there was extensive among-population variation in outcrossing
rates and traits associated with outcrossing. Exceptions
to this classiﬁcation scheme were species in Oenothera
sect. oenothera, which were classiﬁed as either sexual or functionally
selﬁng asexual, due to permanent translocations
whereby plants self-fertilise but do not undergo segregation
and recombination (Johnson et al. 2009). Sexual Oenothera
sect. oenothera species are partially or wholly self-incompatible
and are assumed to be outcrossing. Different traits are
more reliable indicators of mating system in different taxa,
and so methods for mating system classiﬁcation varied
among clades but were generally consistent within clades
(described in Table S3). To extend our data set, we occasionally
classiﬁed taxa that were missing from the primary
studies using the same traits and metrics as those used for
other species within that clade (Table S3). Only sister pairs
with one selﬁng and one outcrossing species were included in
the ensuing analyses, hereafter termed “selﬁng-outcrossing
sister pairs”.
Although we focus on mating system, correlated traits such
as polyploidy (Stebbins 1950; Barringer 2007; Robertson
et al. 2011) and perennial or annual life history (Barrett et al.
1996) may coevolve with mating system. To test whether
these traits drive a relationship between mating system and
range size, we gathered published information on ploidy and
life history when possible. For ploidy, we recorded chromosome
counts and classiﬁed each species (relative to the base
ploidy reported for each genus in the literature) as diploid,
polyploid or mixed when both diploid and polyploid individuals
were known. Species’ life histories were classiﬁed as
annual, perennial, or mixed when both annual and perennial
individuals were known. See Table S4 for species’ classiﬁcations
and sources.
Estimating geographic range size and latitudinal distributions
We downloaded all known occurrence records for the species
in our study from the Global Biodiversity Information Facility
(http://www.gbif.org) and ﬁltered for quality by excluding
records with coordinate accuracy < 100 km, coordinates failing
to match the locality description, and taxonomic misidentiﬁcations
(veriﬁed by the authors and taxonomic specialists
of each clade). We checked species’ epithets against the most
recently published taxonomies and corrected synonyms and
spelling errors. To ensure that recent range expansion potentially
aided by anthropogenic effects did not inﬂuence our
results, we included only coordinates from the native range of
species. We identiﬁed coordinates outside the native species
range with published monographs and online databases that
report native and invaded ranges (e.g. GRIN database, http://
www.ars-grin.gov/).
We used the cleaned occurrence data to estimate species’
range size by dividing the world into a series of rectangular
cells by grid lines that follow longitude and latitude (Table
S1). We calculated range size as the summed area of occupied
grid cells for a given species. To assess whether the ensuing
analyses were sensitive to the spatial scale at which species’
ranges are estimated, we calculated range size across cell sizes
of 0.05, 0.1, 0.5 and 1 decimal degrees, representing grid cells
of roughly 25, 100, 2500 and 10 000 km2
respectively (exact
area varies by latitude).
In addition to range size, we quantiﬁed three components
of species’ latitudinal distributions from the ﬁltered occurrence
data: absolute minimum latitude, absolute midpoint latitude
(midpoint between minimum and maximum latitude) and
absolute maximum latitude. None of the species in our data
set span the equator, so minimum latitude was always > 0.
Analyses
We performed linear mixed effects models to determine
whether selfers and outcrossers differ in range size (Table
S1). We used natural log-transformed range size as the
dependent variable to improve normality of residuals and
homogeneity of variance. We treated mating system as a
ﬁxed effect and included random intercepts for clades
(genus and generic sections) and sister pairs, as well as byclade
random slopes. To incorporate phylogenetic uncertainty
into this model, we included a weighting factor for
each sister pair equal to the proportion of phylogenetic
trees that contained it (i.e. the posterior probability of this
pair). This weighting allowed us to include all selﬁngoutcrossing
sister pairs, while accounting for the uncertainty
associated with each sister pair. In addition, the weighting
factor accounts for species occurring in multiple sister pairs
across the posterior distribution of trees: the weighting
scores of all sister pairs containing a given species never
sum to more than 1.
Because the number of sister pairs was highly variable
among the 20 clades, we also performed a sign test (Table
S1), with each clade as a single datum. For this test, we took
the average sister pair difference in range size (selfer minus
outcrosser) for each clade and asked whether that value was,
on average, positive or negative; the null hypothesis was that
either was equally likely.
We replicated all analyses across the four range estimates
(based on different grid sizes described above) to ensure that
© 2015 John Wiley & Sons Ltd/CNRS
708 D. Grossenbacher et al. Letter
our results were robust to the spatial scale at which range size
was determined. We also ran all analyses including only selﬁng-outcrossing
sister pairs that did not differ in ploidy
(N = 127 sister pairs) or life history (N = 112 sister pairs) to
test whether correlations between mating system and these
traits confound our results. We calculated marginal R2
(proportion
of variance explained by the ﬁxed factors alone) and
conditional R2
(proportion of variance explained by both the
ﬁxed and random factors) values, and we determined signiﬁcance
using likelihood ratio tests with single-term deletions
(Table S1).
To determine whether selfers occupy different latitudes than
their outcrossing sister species, we used the same basic model
described above with absolute minimum, midpoint and maximum
latitude treated as response variables in three separate
models.
To determine whether range size and latitude were affected
by divergence time and whether this varied by mating system,
we added divergence time and its interaction with mating system
as ﬁxed effects to the above model. We allowed clades to
have varying slopes by including a by-clade random slope
term for divergence time. To meet model assumptions, we
log-transformed divergence time.
RESULTS
The phylogenetic analysis identiﬁed 194 sister species that differed
in mating system. Within clades, the number of selferoutcrosser
sister pairs ranged from 1 to 68, and their posterior
probabilities ranged from < 0.01 to 1.0.
Selﬁng-outcrossing sister species displayed tremendous variation
in range size, from having nearly equivalent range areas
to differing by > 3 orders of magnitude. Mating system shifts
explained a signiﬁcant proportion of this variation (16–21%)
with selfers having, on average, 1.5 – 2 times larger ranges
than their outcrossing sisters (Table 1; Fig. 1). This effect
was signiﬁcant across all spatial scales, and it was robust to
excluding sister pairs that differed in ploidy and annual/
perennial life history (P ≤ 0.01 in all cases; see Table S5).
Furthermore, a sign test revealed that the predicted cladeaverage
difference in range size (selﬁng minus outcrossing
member of sister pair) was positive (P < 0.001 in all cases;
see Table S6).
Mating system also explained the northerly latitudinal distributions
of sister species. Selfers had higher maximum latitudes
than their outcrossing sister species by about 1 decimal
degree or 110 km on average (Table 1; Fig. 2a). In contrast,
midpoint and minimum latitudes did not vary by mating system
(Table 1; Fig. 2b and c). Sign tests revealed that the predicted
clade-average difference in maximum and midpoint
latitudes (selﬁng minus outcrossing member of sister pair)
was, on average, positive (P = 0.042 in both cases; Table S6).
In contrast, minimum latitude differences did not differ on
average from zero (P = 0.503; Table S6). When we excluded
sister pairs that differed in ploidy or life history, the direction
and magnitude of the effect was similar but was no longer signiﬁcant
for maximum latitude (P > 0.22, see Tables S4 and
S5). This loss of signiﬁcance likely reﬂects decreased power,
but it could also be attributable to polyploidy and annual/
perennial life history directly inﬂuencing species’ latitudinal
distributions.
With increasing time since divergence, selﬁng species tended
to increase their ranges more than their outcrossing relatives
(Table 2; Fig. 3). In contrast, neither divergence time, nor the
interaction between divergence time and mating system signiﬁcantly
inﬂuenced the latitudinal distributions of species
(Table 2).
Table 1 Results of 7 separate linear mixed models analysing the effect of
mating system on species’ range size (estimated at four spatial scales or
grid cell sizes) and latitudinal distributions
Response LR P Marg.
R2
Condit.
R2
Predicted value
Selfer Outcrosser
Natural log range size
~ 25 km2
12.18 < 0.001 0.208 0.950 1979 940
~ 100 km2
11.85 < 0.001 0.199 0.947 6494 3224
~ 2500 km2
10.90 0.001 0.182 0.936 71 232 40 921
~ 10 000 km2
11.15 < 0.001 0.159 0.930 174 312 108 597
Absolute latitude
Minimum 0.20 0.65 0.002 0.976 31.95 32.31
Midpoint 1.61 0.20 0.007 0.990 37.49 36.75
Maximum 3.86 0.05 0.024 0.989 43.32 41.56
Signiﬁcance of ﬁxed effects was assessed by likelihood ratio tests (LR)
using single term deletions. Marg. R2
values are the proportion of variance
explained by mating system (ﬁxed effect). Condit. R2
values are the
variance explained by mating system and the random effects of clade and
sister pair. Predicted values for range size are back-transformed.
Rangesize(km2)
Selfer Outcrosser
2001000500024000140000
Amsinckia
Arabidopsis
Capsella
Clarkia
Collinsia
Dalechampia
Downingia
Erodium
Lasthenia
Leptosiphon
Limnanthes
Medicago
Mimulus
Oenothera sect. anogra
Oenothera sect. oenothera
Polemonium
Primula
Saltugilia
Schiedea
Schizanthus
Difference:
selfer-outcrosser
–1012
Figure 1 Top panel: Box plots of predicted range size of selﬁng and
outcrossing sister species at the c. 100 km2
grid cell spatial scale.
Coloured line segments indicate predicted slopes for each of 20 clades.
Vertical axis is natural logarithmic scale (back-transformed km2
). Bottom
bar charts: predicted average sister species difference in range size [ln
(selfer) – ln (outcrosser)] for each of 20 clades, with vertical lines
representing standard errors. See Table 1 for statistical results.
© 2015 John Wiley & Sons Ltd/CNRS
Letter Range size is predicted by mating system 709
DISCUSSION
Our analysis across 20 clades (from 15 families) reveals that
selfers have geographic ranges on average twice the size of
their outcrossing sister species. This difference increases with
divergence time between selﬁng-outcrossing pairs, and it may
be partially attributable to the colonisation of high latitude
regions by selfers. Together, our results suggest that the
increased colonisation ability in selfers associated with the
escape from mate limitation (Baker 1955; Stebbins 1957;
Pannell & Barrett 1998) allows them to expand and extend
their geographic ranges relative to outcrossers (Randle et al.
2009). This effect overwhelms potentially opposing forces that
may limit the geographic ranges of selfers.
Henslow (1879) observed a relationship between mating system
and geographic range size when he noted that most widespread
members of the British ﬂora tended to be selfers.
However, it was more than a century until this pattern was
systematically tested in four North American plant genera,
with mixed results. In Collinsia, Randle et al. (2009) found
that automatically selﬁng species had larger ranges than their
outcrossing sister species. The same trend was found in Oenothera
sect. oenothera, but it was not statistically signiﬁcant
and relied on different methods (Johnson et al. 2009). In Clarkia,
the trend was variable, possibly due to poor phylogenetic
resolution or to confounding correlated traits such as ploidy
(Lowry & Lester 2006). In Fragaria, selﬁng species have larger
ranges than outcrossing species (Johnson et al. 2014). The former
three genera were included in this study and supported
the overall pattern of greater range size for selﬁng than outcrossing
sister species (Fig. 1). It is worth noting that most
species examined in past studies have temperate rather than
tropical distributions, and 18 of 20 clades in this study are
temperate. Nonetheless, the two clades containing species with
largely tropical distributions included here (Dalechampia and
Schiedea) support the pattern of selfers having larger ranges
than their outcrossing sisters (Fig. 1). As more data become
available, it will be valuable to test globally the relationship
between mating system and range size.
Our analyses are all correlative, and therefore we cannot
claim a causative link between mating system and range size.
However, we can exclude numerous alternative explanations
of this correlation. With additional analyses, we exclude the
possibility that life history and polyploidy (traits correlated
with selﬁng) are driving greater range size in selﬁng species.
We also largely rule out two other traits that could drive this
correlation: habitat afﬁnity and dispersal ability. If selﬁng species
occupy widely distributed ruderal habitats (Baker 1955;
Stebbins 1957) or have greater dispersal abilities than outcrossing
species, this could lead to their larger ranges. We
account for the former by excluding populations outside the
known native ranges of each species, and by restricting the
analysis to species pairs with the same life histories (because
species that occupy ruderal habitat tend to have annual rather
than perennial life histories; Baker 1974). Theory suggests predominant
selfers have smaller dispersal distances (Cheptou &
Massol 2009; Hargreaves & Eckert 2014), opposing the results
observed here, and therefore dispersal abilities that differ with
mating system are unlikely to explain our ﬁndings.
We can also largely rule out two potential artifactual explanations
for our results. First, because there is more landmass
at higher latitudes (particularly in the northern hemisphere),
and because selfers occupy higher maximum latitudes, selfers
may simply have greater opportunity to achieve larger ranges.
However, if this landmass effect drove our observations, it
should also apply to the larger-ranged member of outcrossing–outcrossing
sister pairs. In a separate analysis, we did not
ﬁnd that the larger-ranged member of outcrossing-outcrossing
sister pairs occurs at higher maximum latitudes (see Appendix
S1). Second, range size estimates may be biased if sampling
10203040506070
Maximumlatitude
Selfer Outcrosser
*
10203040506070
10203040506070
Midpointlatitude
Selfer Outcrosser
n.s.
10203040506070
10203040506070
Minimumlatitude
Selfer Outcrosser
n.s.
10203040506070
Difference:
selfer-outcrosser
–6–226
–6–226
–6–226
(a) (b) (c)
Figure 2 Box plots of predicted absolute latitudinal distributions of selﬁng and outcrossing sister species: (a) maximum latitude, (b) midpoint latitude, (c)
minimum latitude. Coloured line segments indicate predicted slopes for each of 20 clades. Bottom bar charts: predicted average sister species difference in
latitude (selfer - outcrosser) for each of 20 clades, with vertical lines representing standard errors. *P < 0.05. n.s. indicates not signiﬁcant. See Table 1 for
full statistical results.
© 2015 John Wiley & Sons Ltd/CNRS
710 D. Grossenbacher et al. Letter
effort varies with mating system. This is not the case—the
average number of voucher specimens did not differ by mating
system (Appendix S2). Therefore, ascribing the larger
ranges of selﬁng species to the escape from mate limitation is
consistent with all subsequent analyses, whereas alternative
explanations that we examined are not.
The larger geographic ranges of selfers could, however,
reﬂect either greater individual environmental tolerances or
species-wide niche breadth. In contrast to the idea that
reduced genetic variation in selfers will prevent them from
establishing and adapting in new environments (Stebbins
1957; Hamrick & Godt 1996; Crawford & Whitney 2010), we
found that selfers had larger geographic ranges, achieved in
part by occupation of higher latitudes. Latitude is among the
most extreme environmental gradients, with high latitudes
experiencing high seasonality and low biotic diversity relative
to low latitudes (reviewed in Mittelbach et al. 2007). The ability
to persist across a range of latitudes suggests a greater realised
species-wide niche breadth for selfers, which may be
partially attributable to ploidy and/or life history, or to
greater environmental tolerance of selfers versus outcrossers
(e.g. Sheth & Angert 2014). Further exploration of the environmental
breadth occupied by selfers and outcrossers may
uncover axes along which selfers have expanded their realised
environmental niches and generate hypotheses concerning
their environmental tolerances.
Like self-fertilisation, other forms of uniparental reproduction
may allow plants to evade Allee effects during range
expansion and achieve larger ranges. Indeed, the ranges of
asexual plants are sometimes larger than those of their sexual
relatives (Bierzychudek 1985), and asexual species sometimes
occur at extreme latitudes (Bell 1982; Bierzychudek 1985).
Numerous hypotheses have been put forth to explain the
correlation between asexuality and latitude. For example,
asexuals may have ‘generalist genotypes’ and broader environmental
tolerance (Lynch 1984), or ﬂuctuating biotic interactions
with pathogens and competitors in tropical regions may
maintain sex in low but not high latitudes (Glesener & Tilman
1978; Bell 1982). Excluding these alternative explanations
requires both additional experimental and correlational studies.
For example, large ranges in pseudogamous apomicts
[plants that, for their own asexual propagation, require pollen
which is often not their own (H€orandl 2010)] would argue
against our hypothesis that reproductive assurance acts to
increase range size.
The age-range relationship, speciation and extinction
The observation that range size increases with time since speciation
has been observed previously, although this effect varies
widely across taxa (reviewed in Pigot et al. 2012). We
found a similar effect, but only for selﬁng species (Fig. 3).
Although it is tempting to interpret this as evidence of strong
directional range size evolution for selfers relative to outcrossers,
we caution that the geography of speciation and ﬁltering
effects of extinction could also contribute to this pattern. For
instance, species that inherit large ranges across speciation
events are free to shrink and expand their ranges considerably;
in contrast, species that inherit small ranges during speciation
will go extinct if their range shrinks substantially. This
bias will cause an apparent increase in range size with age, at
least for relatively young ages like we consider, because species
with decreased ranges were lost to extinction. Thus the
geography of speciation and the pace of range size evolution
can introduce trends in range size evolution (Pigot et al.
2012). This process could be relevant for speciation events
involving selﬁng-outcrossing pairs (e.g., if selfers commonly
arise in small populations via budding speciation); however,
these ideas have not been modelled in this context, and therefore
the impacts remain unclear.
Our ﬁnding that selfers have larger ranges than outcrossers,
and that ranges of selﬁng species increase with age, seems at
odds with the long-held idea that selfers face high extinction
Divergence time (mya)
Rangesize(km2)
55400300022000160000
0.14 0.4 1 3 8 20 55
Selfer
Outcrosser
Figure 3 Range size as a function of divergence time for selﬁng and
outcrossing sister species. The size of open circles represents the
proportion of the posterior sample of trees in which the focal pair were
each other’s closest relatives. The line segments represent the linear
regression results for selfers and outcrossers (black and gray lines
respectively). Range size and divergence time axes are natural logarithmic
scale (back-transformed). See Table 2 for statistical results.
Table 2 Results of four separate linear mixed models analysing the effect
of divergence time, mating system and their interaction on species’ range
size and latitude
Response
Fixed effects
Divergence
time
Mating
system
Time*Mating
system
LR P LR P LR P
Natural log range size 3.86 0.05 4.67 0.03 54.37 0.001
Minimum latitude 2.44 0.12 0.53 0.47 2.71 0.10
Midpoint latitude 0.00 0.99 0.00 0.99 2.02 0.16
Maximum latitude 0.00 0.99 3.07 0.08 2.90 0.09
Signiﬁcance of ﬁxed effects was assessed by likelihood ratio tests (LR)
using single term deletions. Range size was estimated in c. 100 km2
grid
cells; results are consistent across all spatial scales (not presented).
© 2015 John Wiley & Sons Ltd/CNRS
Letter Range size is predicted by mating system 711
rates relative to outcrossers (reviewed in Igic & Busch 2013).
One possible resolution is that selfers do go extinct more frequently,
but the ones we observe are those that happen to
attain large range size (as in the example above). Another
potential resolution is that extinction in selﬁng plants occurs
by rapid extirpations across the entire range, rather than a
gradual elimination of populations until the range dwindles
and disappears. This could be due to rapid ﬂuctuation of the
geographic distribution of selfers, or to stronger autocorrelation
of extinction risk across the range. For example, if selfers
have less genetic variation (Stebbins 1957; Hamrick & Godt
1996; Crawford & Whitney 2010; Sheth & Angert 2014) and
steadily accumulate deleterious mutations (Wright et al.
2013), they may be vulnerable to sudden changes in the biotic
or abiotic environments, leading to rapid, range-wide extirpation
and extinction. A ﬁnal potential resolution could be the
combination of the phylogenetic scale over which selﬁng species
go extinct (e.g. selfers may give rise to other selfers prior
to going extinct) and ascertainment bias on the scale of our
study: there are many species, both selﬁng and outcrossing,
whose sister species is not of the opposite mating system and
which are consequently not included in our data set. All these
potential resolutions warrant future research.
Other contributors to geographic range size
The search for reasons underlying the massive variation in species’
geographic range sizes has a long history but reveals few
universal explanations (Brown et al. 1996; Gaston 2003). The
results here suggest that mating system is a strong predictor of
range size, explaining up to 20% of the variation in species’
geographic ranges. To put this in perspective, body size is
among the best-studied correlates of range size and only
explains about 6% of the overall variation in the range size of
birds and mammals (averaged across studies; Agosta et al.
2013), which is typical of other predictors of range size (Brown
et al. 1996). We note that the proportion of variation in range
size that we explain is likely an overestimate: we chose clades in
which mating system was sufﬁciently variable so as to potentially
explain a reasonable portion of the variation. It is not
clear how this ascertainment scheme compares to previous
investigations of other traits’ inﬂuences on range size.
In addition to mating system, recent shared ancestry may
shape species’ geographic ranges (Jablonski 1987; BohningGaese
et al. 2006; Martin & Husband 2009). For instance,
species generally arise in the same region of the world as their
close relatives, which in turn may inﬂuence range size, for
example, Rapoport’s Rule (Stevens 1989). In addition, life history
traits that are shared due to recent common ancestry
(e.g., dispersal, size, environmental tolerance) may inﬂuence
species’ ranges (reviewed in Brown et al. 1996; Gaston 2003).
In our study, shared ancestry (at the level of genus and sister
pair) and mating system together explained 90–95% of the
variation in range size.
CONCLUSION
Our observation that selfers have larger ranges than outcrossers
is consistent with the idea that mate availability at the
colonisation stage may limit species’ range size. This implies
that the ability to ﬁnd a mate and establish in a novel habitat
may have as great an inﬂuence on the species’ range as
environmental tolerance or interspeciﬁc competition. Furthermore,
this suggests that traits that increase the odds of ﬁnding
a mate during colonisation (e.g., sperm storage in
females), may result in increased geographic range size.
Unfortunately, despite the critical role of mate limitation in
slowing range expansions (Shaw & Kokko 2015), studies
assessing the impact of mate availability on range size in
animals or other taxa are lacking, as much of the last century
of research has instead focused on traits inﬂuencing dispersal
and environmental tolerance (reviewed in Brown et al.
1996; Gaston 2003). We suggest that in many taxa, focusing
on traits that encourage mate-ﬁnding during colonisation
may be central to understanding the puzzle of geographic
range size.
ACKNOWLEDGEMENTS
Boris Igic, April Randle, Sue Kalisz and the Baker’s Law
working group provided stimulating discussions. Anne
Worley, Barbara Neuffer, Andress Franzke, Jeremiah Busch
and Justen Whittall provided expert advice on mating systems
and phylogeny construction.
AUTHORSHIP
All authors contributed to framing the study questions. DG
constructed phylogenies, analysed data and wrote the ﬁrst
draft of the manuscript. DG, RBR and YB compiled mating
system and geographic data. RBR, EEG and YB contributed
to analysis and revision.
REFERENCES
Agosta, S.J., Bernardo, J., Ceballos, G. & Steele, M.A. (2013). A
macrophysiological analysis of energetic constraints on geographic
range size in mammals. PLoS ONE, 8, e72731.
Baker, H.G. (1955). Self compatibility and establishment after long
distance dispersal. Evolution, 9, 347–349.
Baker, H.G. (1966). The evolution, functioning and breakdown of
heteromorphic incompatibility systems. I. The Plumbaginaceae.
Evolution, 20, 349–368.
Baker, H.G. (1974). The evolution of weeds. Annu. Rev. Ecol. Syst., 5, 1–24.
Barrett, S.C.H., Harder, L.D. & Worley, A.C. (1996). The comparative
biology of pollination and mating in ﬂowering plants. Philos. Trans. R.
Soc. B Biol. Sci., 351, 1271–1280.
Barringer, B. (2007). Polyploidy and self-fertilization in ﬂowering plants.
Am. J. Bot., 94, 1527–1533.
Bell, G. (1982). The Masterpiece of Nature: The Evolution and Genetics of
Sexuality. Berkeley: University of California Press.
Bierzychudek, P. (1985). Patterns in plant parthenogenesis. Experientia,
41, 1255–1264.
B€ohning-Gaese, K., Caprano, T., van Ewijk, K. & Veith, M. (2006).
Range size: disentangling current traits and phylogenetic and
biogeographic factors. Am. Nat., 167, 555–567.
Brown, J.H., Stevens, G.C. & Kaufman, D.M. (1996). The geographic
range: size, shape, boundaries, and internal structure. Annu. Rev. Ecol.
Syst., 27, 597–623.
Cheptou, P. & Massol, F. (2009). Pollination ﬂuctuations drive
evolutionary syndromes linking dispersal and mating system. Am. Nat.,
174, 46–55.
© 2015 John Wiley & Sons Ltd/CNRS
712 D. Grossenbacher et al. Letter
Crawford, K.M. & Whitney, K.D. (2010). Population genetic diversity
inﬂuences colonization success. Mol. Ecol., 19, 1253–1263.
Crawford, D.J., Archibald, J.K., Stoermer, D., Mort, M.E., Kelly, J.K. &
Santos-Guerra, A. (2008). A test of Baker’s law: breeding systems and
the radiation of Tolpis (Asteraceae) in the Canary Islands. Int. J. Plant
Sci., 169, 782–791.
Darwin, C. (1859). On the Origin of Species by Means of Natural
Selection. Murray, London.
Gaston, K.J. (2003). The Structure and Dynamics of Geographic Ranges.
Oxford University Press, Oxford, UK.
Glesener, R.R. & Tilman, D. (1978). Sexuality and the components of
environmental uncertainty: clues from geographic parthenogenesis in
terrestrial animals. Am. Nat., 112, 659–673.
Goodwillie, C., Kalisz, S. & Eckert, C.G. (2005). The evolutionary enigma
of mixed mating systems in plants: occurrence, theoretical explanations,
and empirical evidence. Annu. Rev. Ecol. Evol. Syst., 36, 47–79.
Grifﬁn, P.C. & Willi, Y. (2014). Evolutionary shifts to self-fertilisation
restricted to geographic range margins in North American Arabidopsis
lyrata. Ecol. Lett., 17, 484–490.
Haag, C.R. & Ebert, D. (2004). A new hypothesis to explain geographic
parthenogenesis. Ann. Zool. Fennici, 41, 539–544.
Hamrick, J.L. & Godt, M.J.W. (1996). Effects of life history traits on
genetic diversity in plant species. Philos. Trans. R. Soc. London. Ser. B
Biol. Sci., 351, 1291–1298.
Hargreaves, A.L. & Eckert, C.G. (2014). Evolution of dispersal and
mating systems along geographic gradients: implications for shifting
ranges. Funct. Ecol., 28, 5–21.
Heller, R. & Smith, J.M. (1978). Does Muller’s ratchet work with selﬁng?
Genet. Res., 32, 289–293.
Henslow, G. (1879). On the self-fertilization of plants. Trans. Linn. Soc.
London, 1, 317–398.
H€orandl, E. (2010). The evolution of self-fertility in apomictic plants. Sex.
Plant Reprod., 23, 73–86.
Igic, B. & Busch, J.W. (2013). Is self-fertilization an evolutionary dead
end? New Phytol., 198, 386–397.
Jablonski, D. (1987). Heritability at the species level: analysis of
geographic ranges of cretaceous mollusks. Science, New Series, 238,
360–363.
Johnson, M.T.J., Smith, S.D. & Rausher, M.D. (2009). Effects of plant
sex on range distributions and allocation to reproduction. New Phytol.,
186, 769–779.
Johnson, A.L., Govindarajulu, R. & Ashman, T. (2014). Bioclimatic
evaluation of geographical range in Fragaria (Rosaceae): consequences
of variation in breeding system, ploidy and species age. Bot. J. Linn.
Soc., 176, 99–114.
Jordaens, K., Geenen, S., Reise, H., Van Riel, P., Verhagen, R. &
Backeljau, T. (2000). Is there a geographical pattern in the breeding
system of a complex of hermaphroditic slugs (Mollusca: Gastropoda:
Carinarion)? Heredity (Edinb), 85, 571–579.
Kay, K.M., Whittall, J.B. & Hodges, S.A. (2006). A survey of nuclear
ribosomal internal transcribed spacer substitution rates across
angiosperms: an approximate molecular clock with life history effects.
BMC Evol. Biol., 6, 36.
Lloyd, D.G. (1980). Demographic factors and mating patterns in
angiosperms. In: Demogr. Evol. plantPopul. (ed. Solbrig, O.T.).
Blackwell, Oxford, pp. 67–88.
Lowry, E. & Lester, S.E. (2006). The biogeography of plant reproduction:
potential determinants of species’ range sizes. J. Biogeogr., 33, 1975–
1982.
Lynch, M. (1984). Destabilizing hybridization, general-purpose genotypes
and geographic parthenogenesis. Q. Rev. Biol., 59, 257–290.
Martin, S.L. & Husband, B.C. (2009). Inﬂuence of phylogeny and ploidy
on species ranges of North American angiosperms. J. Ecol., 97, 913–
922.
Mittelbach, G.G., Schemske, D.W., Cornell, H.V., Allen, A.P., Brown, J.M.,
Bush, M.B. et al. (2007). Evolution and the latitudinal diversity gradient:
speciation, extinction and biogeography. Ecol. Lett., 10, 315–331.
Pannell, J.R. & Barrett, S.C.H. (1998). Baker’s law revisited: reproductive
assurance in a metapopulation. Evolution, 52, 657–668.
Pigot, A.L., Owens, I.P.F. & Orme, C.D.L. (2012). Speciation and
extinction drive the appearance of directional range size evolution in
phylogenies and the fossil record. PLoS Biol., 10, e1001260.
Pujol, B., Zhou, S.-R., Vilas, J.S. & Pannell, J.R. (2009). Reduced
inbreeding depression after species range expansion. Proc. Natl Acad.
Sci., 106, 15379–15383.
Randle, A., Slyder, J. & Kalisz, S. (2009). Can differences in autonomous
selﬁng ability explain differences in range size among sister-taxa pairs
of Collinsia (Plantaginaceae)? An extension of Baker’s. New Phytol.,
183, 618–629.
Robertson, K., Goldberg, E.E. & Igic, B. (2011). Comparative evidence
for the correlated evolution of polyploidy and self-compatibility in
Solanaceae. Evolution, 65, 139–155.
Sexton, J.P., McIntyre, P.J., Angert, A.L. & Rice, K.J. (2009). Evolution
and ecology of species range limits. Annu. Rev. Ecol. Evol. Syst., 40,
415–436.
Shaw, A.K. & Kokko, H. (2015). Dispersal evolution in the presence of allee
effects can speed up or slow down invasions. Am. Nat., 185, 631–639.
Sheth, S.N. & Angert, A.L. (2014). The evolution of environmental
tolerance and range size: a comparison of geographically restricted and
widespread Mimulus. Evolution, 68, 2917–2931.
Stebbins, G.L. (1950). Variation and Evolution in Plants. Columbia
University Press, New York, London.
Stebbins, G.L. (1957). Self fertilization and population variability in the
higher plants. Am. Nat., 91, 337–354.
Stevens, G.C. (1989). The latitudinal gradient in geographical range: how
so many species coexist in the tropics. Am. Nat., 133, 240–256.
Wright, S.I., Kalisz, S. & Slotte, T. (2013). Evolutionary consequences of
self-fertilization in plants. Proc. R. Soc. B Biol. Sci., 280, 20130133.
SUPPORTING INFORMATION
Additional Supporting Information may be downloaded via
the online version of this article at Wiley Online Library
(www.ecologyletters.com).
Editor, John Pannell
Manuscript received 5 January 2015
First decision made 13 February 2015
Manuscript accepted 9 April 2015
© 2015 John Wiley & Sons Ltd/CNRS
Letter Range size is predicted by mating system 713