ArticlePDF Available

Genome divergence and diversification within a geographic mosaic of coevolution

Authors:
Article

Genome divergence and diversification within a geographic mosaic of coevolution

Abstract and Figures

Despite substantial interest in coevolution's role in diversification, examples of coevolution contributing to speciation have been elusive. Here, we build upon past studies that have shown both coevolution between South Hills crossbills and lodgepole pine (Pinus contorta), and high levels of reproductive isolation between South Hills crossbills and other ecotypes in the North American red crossbill (Loxia curvirostra) complex. We used genotyping by sequencing to generate population genomic data and applied phylogenetic and population genetic analyses to characterize the genetic structure within and among nine of the ecotypes. Although genome-wide divergence was slight between ecotypes (FST = 0.011–0.035), we found evidence of relative genetic differentiation (as measured by FST) between and genetic cohesiveness within many of them. As expected for nomadic and opportunistic breeders, we detected no evidence of isolation by distance. The one sedentary ecotype, the South Hills crossbill, was genetically most distinct because of elevated divergence at a small number of loci rather than pronounced overall genome-wide divergence. These findings suggest that mechanisms related to recent local coevolution between South Hills crossbills and lodgepole pine (e.g. strong resource-based density dependence limiting gene flow) have been associated with genome divergence in the face of gene flow. Our results further characterize a striking example of coevolution driving speciation within perhaps as little as 6000 years.
Content may be subject to copyright.
Genome divergence and diversification within a
geographic mosaic of coevolution
THOMAS L. PARCHMAN,* C. ALEX BUERKLE,V
ICTOR SORIA-CARRASCOand
CRAIG W. BENKMAN§
*Department of Biology, University of Nevada Reno, Reno, NV 89557, USA, Department of Botany, University of Wyoming,
Laramie, WY 82071, USA, Department of Animal and Plant Sciences, University of Sheffield, Sheffield S10 2TN, UK,
§Department of Zoology and Physiology, University of Wyoming, Laramie, WY 82071, USA
Abstract
Despite substantial interest in coevolution’s role in diversification, examples of coevo-
lution contributing to speciation have been elusive. Here, we build upon past studies
that have shown both coevolution between South Hills crossbills and lodgepole pine
(Pinus contorta), and high levels of reproductive isolation between South Hills cross-
bills and other ecotypes in the North American red crossbill (Loxia curvirostra) com-
plex. We used genotyping by sequencing to generate population genomic data and
applied phylogenetic and population genetic analyses to characterize the genetic struc-
ture within and among nine of the ecotypes. Although genome-wide divergence was
slight between ecotypes (F
ST
=0.0110.035), we found evidence of relative genetic dif-
ferentiation (as measured by F
ST
) between and genetic cohesiveness within many of
them. As expected for nomadic and opportunistic breeders, we detected no evidence of
isolation by distance. The one sedentary ecotype, the South Hills crossbill, was geneti-
cally most distinct because of elevated divergence at a small number of loci rather than
pronounced overall genome-wide divergence. These findings suggest that mechanisms
related to recent local coevolution between South Hills crossbills and lodgepole pine
(e.g. strong resource-based density dependence limiting gene flow) have been associ-
ated with genome divergence in the face of gene flow. Our results further characterize
a striking example of coevolution driving speciation within perhaps as little as
6000 years.
Keywords: coevolution, divergent selection, ecological speciation, genetic differentiation, Loxia,
population genomics
Received 31 May 2016; revision received 10 August 2016; accepted 18 August 2016
Introduction
Coevolution, the process of reciprocal adaptation by
two or more species in response to reciprocal selection,
is thought to be a major driver of biological diversifica-
tion (Ehrlich & Raven 1964; Thompson 1994, 2005).
However, demonstrating coevolution has been challeng-
ing (Gomulkiewicz et al. 2007). Moreover, few studies
link coevolution directly to speciation and diversifica-
tion (Althoff et al. 2014; Hembry et al. 2014). Coevolu-
tion between the South Hills crossbill (Loxia curvirostra
complex) and Rocky Mountain lodgepole pine (Pinus
contorta latifolia) is one of the best-documented exam-
ples of coevolution (Thompson 2005; Gomulkiewicz
et al. 2007), and of coevolution generating reproductive
isolation (Althoff et al. 2014; Hembry et al. 2014). In the
absence of the red squirrel (Tamiasciurus hudsonicus;a
predispersal seed predator), crossbills in the South
Hills, Idaho, USA, are resident and much more abun-
dant and exert stronger selection on lodgepole pine
cones, causing the evolution of enhanced seed defences
directed at crossbills. Where red squirrels occur, they
are superior competitors for the lodgepole pine seeds
and crossbills are much less abundant. Under these
conditions, cones evolve mostly in response to selection
Correspondence: Thomas L. Parchman, Fax: 775 784 2851;
E-mail: tparchman@unr.edu
©2016 John Wiley & Sons Ltd
Molecular Ecology (2016) doi: 10.1111/mec.13825
exerted by red squirrels rather than crossbills and have
less crossbill-directed defence, which favours smaller
beaked crossbills (Benkman 1999; Benkman et al. 2001,
2003). The result is a geographic mosaic of coevolution,
where crossbills coevolve in an arms race with lodge-
pole pine in the absence of red squirrels but not in their
presence (Benkman 1999; Benkman et al. 2001, 2003,
2013).
The South Hills crossbill is one of 10 morphologically
and vocally differentiated ecotypes (‘call types’) of the
North American red crossbill complex (Groth 1993;
Benkman et al. 2009; Irwin 2010) that have evolved in
response to selection for specialization on seeds in the
cones of different conifer species (Benkman 1993; Benk-
man et al. 2003). This hypothesis was tested in past
studies that quantified feeding performance for five eco-
types, including the South Hills crossbill (Type 9) and
the ecotype specialized on lodgepole pine where red
squirrels occur (Type 5; Benkman 1993, 2003). Feeding
performance varied in relation to beak depth (influences
efficiency of seed extraction from conifer cones) and
groove width in the horny palate (influences seed husk-
ing ability), with each ecotype having beak traits that
approximate the predicted optima for foraging on seeds
of their ‘key’ conifer (i.e., conifers that reliably produce
and hold seeds in cones; Benkman 1993, 2003). The
close fit between trait means and both their predicted
optima and survival selection strongly implicates
resource-based divergent selection in driving this
adaptive radiation (Benkman 1993, 2003).
Because divergent selection can reduce gene flow,
divergent selection could lead to genetic differentiation,
even in the absence of geographic isolation (Endler
1973; Nosil et al. 2008; Shafer & Wolf 2013). In cross-
bills, reproductive isolation is related to divergent selec-
tion among ecotypes, because of expected lower fitness
of potentially intermediate (hybrid) phenotypes, habitat
isolation, low immigrant reproduction and several
forms of behavioural isolation (Smith & Benkman 2007;
Snowberg & Benkman 2007, 2009; Smith et al. 2012). In
the South Hills, the combination of well-defended
lodgepole pine cones (Benkman 1999; Benkman et al.
2001, 2003, 2013), local adaptation by South Hills cross-
bills and strong density-dependent food limitation pre-
vents all but a few individuals of the nonlocally
adapted ecotypes from persisting prior to and during
pairing by South Hills crossbills (Smith & Benkman
2007; see Bolnick 2011). Strong density dependence
arises because of very stable seed renewal in the South
Hills, unlike the episodic abundance of resources that
other ecotypes experience (Benkman et al. 2012). This
contributes to the high levels of premating reproductive
isolation between the South Hills crossbill and the two
other ecotypes that breed in the South Hills [0.999 on a
scale from 0 (no isolation) to 1 (complete reproductive
isolation); Smith & Benkman 2007; Benkman et al. 2009].
However, whether local adaptation and strong contem-
porary premating reproductive isolation have been of
sufficient duration to cause genome divergence is
unknown.
The ecology and evolutionary history of the red
crossbill complex have resulted in limited genetic differ-
entiation among the ecotypes. Ecotype diversification
probably occurred in the Holocene following the retreat
of glaciers and expansion of conifers (Benkman 1993).
The distributions of some key conifers relied upon by
the ecotypes were so restricted in the late Pleistocene
(e.g. coastal Douglas fir Pseudotsuga m. menziesii and
Rocky Mountain ponderosa pine P. ponderosa scopulo-
rum; Gugger et al. 2010; Potter et al. 2013) that frequent
regional cone crop failures likely prevented earlier spe-
cialization by crossbills (Benkman 1993). Such recent
divergence is indicated by coalescent analyses of Euro-
pean common (red) crossbill mtDNA haplotypes sug-
gesting that ecotype diversification occurred rapidly
over the last 11 000 years (Bjorklund et al. 2013). Pat-
terns of mtDNA divergence and diversity are similar
among North American ecotypes (Questiau et al. 1999).
In addition, due to regular localized cone crop failures
most ecotypes are nomadic and move long distances
between natal and breeding locations, and between
breeding locations (up to ~3000 km; Newton 2006),
often breeding opportunistically and sympatrically in
areas with abundant conifer seeds (Groth 1993; Sum-
mers et al. 2007). In the South Hills, multiple ecotypes
breed (Smith & Benkman 2007). Such conditions and
behaviour likely allowed extensive gene flow through-
out the young radiation. Indeed, prior analyses using
mtDNA and AFLPs found little evidence for genetic
differentiation among ecotypes (Questiau et al. 1999;
Parchman et al. 2006). Nonetheless, the widely sym-
patric occurrence of distinct ecotypes suggests that
adaptation and reproductive isolation have evolved
despite the large potential for homogenizing gene flow.
Recent innovations in DNA sequencing have dramati-
cally increased our ability to address how geography,
ecology and history shape genome divergence during
the early phases of divergence (Alcaide et al. 2014;
Lamichhaney et al. 2015; Martin et al. 2015). Here, we
use genotyping by sequencing (GBS) to evaluate the
pattern and extent of genome divergence across the
North American red crossbill complex. We sampled
multiple, geographically dispersed populations within
each ecotype (Table S1, Fig. S1, Supporting information)
to test the hypothesis that ecotypes are genetically cohe-
sive and differentiated from one another. We tested for
isolation by ecology, in the form of the evidence that
genetic divergence was associated with divergent
©2016 John Wiley & Sons Ltd
2T. L. PARCHMAN ET AL.
selection (Nosil et al. 2008; Shafer & Wolf 2013), but
anticipate little geographic structure (isolation by dis-
tance) based on the aforementioned biology of cross-
bills. Although our focus is on the South Hills crossbill,
we include nine ecotypes to provide a broader context
of divergence among closely related forms, and because
the form of divergent selection and opportunities for
geographic isolation are not uniquely different because
of coevolution. Finally, the fossil record indicates a
large reduction in the amount of lodgepole pine in the
South Hills region (Mehringer 1985; Davis et al. 1986)
during a several thousand-year warm period centred
around 6000 BP (Bartlein et al. 2014). Because this could
have prevented the persistence of a local crossbill popu-
lation (See Siepielski & Benkman 2005), we estimate the
distribution of lodgepole pine 6000 BP to characterize
the time period over which this coevolutionary interac-
tion could have persisted.
Materials and methods
DNA sequencing, assembly and variant calling
We sequenced DNA from 219 red crossbills represent-
ing nine morphologically and vocally differentiated eco-
types, as well as 12 white-winged crossbills (L. l.
leucoptera; Table S1, Fig. S1, Supporting information).
We utilized a GBS protocol that we have used in previ-
ous studies (Gompert et al. 2012; Nosil et al. 2012;
Parchman et al. 2012, 2013), and generated three lanes
of single-end 100-base sequencing on an Illumina HiSeq
2000 at the National Center for Genome Resources
(Santa Fe, NM, USA). We used Perl scripts to remove
contaminant DNA, trim barcodes and match barcodes
to individual sample information. We first used SEQMAN
NGEN 3.0.4 (DNASTAR) to perform a de novo assembly
for a subset of 30 million reads sampled randomly from
the sequencing data for all individuals. The purpose of
this step was to produce a consensus GBS reference of
the genomic regions represented in our libraries.
Because our library preparation method produces reads
identical in length from genomic regions beginning
with EcoRI cut sites, reads typically align neatly into
rectangular contigs, the consensus sequences of which
represent a reference of genomic regions sampled by
GBS. After removing low-quality or overassembled con-
tigs, we generated a reference of 349 865 contig consen-
sus sequences. We then aligned all reads for each
individual onto the GBS reference using BWA 0.7.8 (Li &
Durbin 2009). We used SAMTOOLS 1.19 and BCFTOOLS 1.19
(Li et al. 2009) to identify bi-allelic single nucleotide
polymorphisms (SNPs), and only called variants when
98% of the individuals had at least one read. We used
this high threshold here to obtain SNPs with higher
coverage and genotypes having relatively high levels of
statistical certainty. While we could have called more
SNPs with a lower threshold, we were interested mostly
in genome-wide parameter estimates for this study and
hope to generate more comprehensive resequencing
data for locus-specific analyses in the future. We ran-
domly selected a single variant from each contig to
increase the independence of loci, and limited analyses
to loci with minor allele frequencies >0.03 for popula-
tion genetic analyses. Further details on assembly and
variant calling are in the Supporting information.
Phylogenetic analyses
Phylogenetic analyses included the 219 red crossbills,
and 12 white-winged crossbills for use as an outgroup.
We first generated a second data set of SNPs present in
alignments of both crossbill species using BWA,SAMTOOLS
and BCFTOOLS. While calling variants, we disregarded
insertions and deletions, and only considered SNPs
when 95% of the individuals had at least one read at
that locus. We used a custom Perl script to produce a
multiple alignment by defining the DNA state of each
individual and variant as the genotype with the highest
likelihood. Heterozygotes were coded using IUPAC
ambiguities (i.e. M for A/C, R for A/G, W for A/T, S
for C/G, Y for C/T and K for G/T), and loci with too
much uncertainty (i.e. equal likelihoods for the three
genotypes) were encoded as missing data. This resulted
in a multiple alignment of 238 615 positions and 231
individuals. We inferred maximum-likelihood (ML)
trees using EXAML 2.0.4 (Stamatakis & Aberer 2013), and
executed 25 independent ML searches using as starting
points 25 parsimony trees inferred using PARSIMONATOR
1.0.3 (Stamatakis 2014). We conducted ML inferences
using a GTR +Γsubstitution model and performed 500
bootstrap replicates using EXAML with a GTR substitu-
tion model using the CAT approximation. We produced
the bootstrapped ML analysis with RAXML 8.0.20 (Sta-
matakis 2014) and used parsimonator as before to
obtain starting trees from every alignment. We summa-
rized the ML analyses using RAXML in two different
ways: (i) drawing bootstrap support values onto the
best-supported ML tree and (ii) computing a bootstrap
consensus tree using the majority rule extended
criterion.
Population genetic analyses
We used analyses based on allele frequencies and geno-
type probabilities to quantify patterns of genetic struc-
ture within and among the red crossbill ecotypes using
the 18 385 high-coverage SNPs described above. We
estimated population allele frequencies and genotype
©2016 John Wiley & Sons Ltd
GENOME DIVERGENCE IN CROSSBILLS 3
probabilities based on genotype likelihoods estimated
with BCFTOOLS using a hierarchical Bayesian model
(Gompert et al. 2012). This model treats population
allele frequencies and individual genotypes as unknown
model parameters and utilizes Markov Chain Monte
Carlo (MCMC). We used this model to estimate allele
frequencies and genotype probabilities for each geo-
graphically separate sample within each ecotype
(Table S1, Supporting information). These 22 population
samples included multiple samples from within five of
the ecotypes (types 2, 3, 4, 5 and 7) that were from geo-
graphically distant regions (Fig. S1, Supporting informa-
tion). We ran MCMC chains for 20 000 steps, discarded
5000 as burn-in and recorded every fifth step. We first
summarized genotypic variation across all individuals
of the red crossbill complex using principal components
analysis (PCA). We generated a genetic covariance
matrix based on the genotype point estimates for each
bird and performed the PCA on this genetic covariance
matrix using the PRCOMP function in R(R Core Team,
2013). We tested for significant differentiation between
ecotypes and for significant differentiation among pop-
ulations within ecotypes using permutational multivari-
ate analysis of variance (PERMANOVA; Anderson 2001)
based on Euclidian distances of the first two principal
components using the vegan package in R(Oksanen
et al. 2013).
We used allele frequency estimates to calculate Nei’s
genetic distance (Nei’s D; Nei 1972) among ecotypes
and among all samples (populations) within ecotypes.
We calculated pairwise Hudson’s F
ST
(Hudson et al.
1992) based on estimated allele frequencies at all loci
for each ecotype and each population, using code writ-
ten in R. We generated ML estimates of the folded-site
allele frequency spectrum, nucleotide diversity (p) and
expected heterozygosity as indicators of genetic varia-
tion within each ecotype using the expectation maxi-
mization algorithm of Li (2011) as implemented in
SAMTOOLS and ran the algorithm for 20 iterations for each
population.
We further investigated hierarchical patterns of genetic
structure across the ecotypes and populations within eco-
types using a hierarchical Bayesian model that is similar
to the correlated allele frequency model of STRUCTURE
(Pritchard et al. 2000; Falush et al. 2003). We used this
model (hereafter ENTROPY, described in Gompert et al.
2014) to characterize population structure and estimate
admixture proportions for individuals in the absence of
information on sample origin. Importantly, ENTROPY
allows for stochastic variation in sequence coverage
across individuals and loci and estimates allele frequency
and genotype probability parameters along with admix-
ture proportions. Similar to the admixture model in
STRUCTURE,ENTROPY assumes that the genome of each
individual consists of loci with ancestry from one of k
ancestral populations and makes no a priori assumptions
about the population or cluster origin of individual sam-
ples. Admixture proportions, which represent the frac-
tion of an individual’s genome inherited from each of the
kclusters, are estimated for each individual. In addition,
ENTROPY generates estimates of deviance information cri-
terion (DIC) as a metric for model choice and compar-
ison; models with lower DIC values are those that fit the
data better (Gompert et al. 2014).
To facilitate the convergence and stabilization of
MCMC chains, we initialized individual admixture pro-
portions in the chains using probabilities of cluster
membership based on k-means clustering of the princi-
pal component scores (equivalent to a no-admixture
model; Falush et al. 2003). Specifically, we used k-means
clustering (KMEANS package in R) based on the principal
components estimated from genotypes in a linear dis-
criminant analysis (LDA package in R; Jombart et al.
2010). This provided reasonable starting values of qto
initialize MCMC and ensured proper mixing and con-
vergence of MCMC chains. Importantly, this approach
uses genotypic data without reference to sample origin
and does not constrain posterior sampling. We ran
ENTROPY separately for predefined values of k=19 and
ran five independent chains for each k. Each chain used
the probability of cluster membership as mean expecta-
tion for the admixture proportion q, but random devi-
ates with a precision scalar of 20 were drawn from a
Dirichlet distribution to initialize qfor each chain. We
used an upper value of 9 for k, representing the number
of ecotypes included in our analysis. We ran each
MCMC chain for 80 000 steps following 60 000 steps
that were discarded as burn-in and saved every 10th
step. We estimated posterior medians, and 95% credible
intervals for parameters of interest. We checked for
mixing and convergence of posterior parameter esti-
mates by plotting MCMC steps for different parameter
sets and inspected mixing during the burn-in period
and convergence among chains.
The localities sampled are geographically separated
and could have allele frequencies that differ due to
genetic drift, with population homogenization due to
migration declining with distance, leading to isolation
by distance. Likewise, divergent selection could reduce
gene flow and potentially lead to differences in popu-
lation allele frequencies (Endler 1973; Nosil et al. 2008;
Shafer & Wolf 2013). To investigate the extent to
which allele frequency differences can be attributed to
geographic and phenotypic distances between popula-
tions, we modelled pairwise genetic distances (Nei’s
D) between populations as a function of geographic
and phenotypic distances (difference in mean beak
depth) between populations. Beak depth data (sample
©2016 John Wiley & Sons Ltd
4T. L. PARCHMAN ET AL.
sizes for males and females in parentheses) were from
Groth (1993) for ecotypes 1 (39 and 33) and 7 (5 and
1), unpublished measurements of live birds and
museum specimens by CWB for ecotypes 2 (226 and
149), 3 (47 and 34), 4 (25 and 15), 5 (61 and 31) and 6
(150 and 71), Benkman et al. (2013) for ecotype 9 (471
and 335) and Irwin (2010) for ecotype 10 (54 and 35).
Phenotypic distances were calculated based on differ-
ences between ecotype means. The use of mean trait
values should be conservative, because it provides less
power to detect patterns than analyses based on mea-
surements from each individual. Geographic distances
were calculated based on Haversine distances, as
implemented in the Rpackage FOSSIL (Vavrek 2011).
Geographic and phenotypic distances were normalized
(transformed to Z-scores) so that their coefficients
would be on the same scale. Genetic distances were
logit-transformed and centred on the mean so as to
not be bounded by zero and one. We used a Bayesian
linear model that did not require all observations in
the response variable to be independent, but instead
modelled random effects for all population pairs
(Clarke et al. 2002; Gompert et al. 2014) and over all
coefficients for geographic and genetic distances, and
separately for phenotypic and genetic distances. The
model was specified in JAGS (version 3.4; Plummer
2003), and samples were gathered from the R interface
to JAGS (RJAGS; R core Team 2013). After discarding
2000 steps as burn-in, we obtained 2000 samples of
the posterior distributions from each of three chains,
by retaining every fifth iteration of 10 000 MCMC
steps. All chains were inspected graphically for
adequate convergence and mixing.
Past forest distribution estimation
Random Forests (Breiman 2001) is a nonparametric clas-
sification and regression tree approach that we used to
model the distribution of lodgepole pine, because of its
past success when true absence data are available, and
its ability to identify nonlinear relationships and inter-
action terms (Cutler et al. 2007). We used the USFS For-
est Inventory and Analysis data (O’Connell et al. 2014)
for 3406 presences and 10 855 absences of lodgepole
pine in Idaho, Montana, Wyoming, Colorado and Utah
to estimate the distribution of Rocky Mountain lodge-
pole pine. This was augmented with 50 000 ‘likely-
absence’ points randomly placed within cells classified
as ‘unforested’ in the LANDFIRE Forest Canopy Cover
data set; this likely-absence data set was reduced by 59
points by eliminating all points within 1 km of lodge-
pole pine presence points to allow for error or lack of
precision in the canopy cover layer, resulting in a total
of 60 796 absences.
Covariates used for modelling were the BIOCLIM set
(Hijmans et al. 2005) and elevation (Gesch et al. 2002).
BIOCLIM values are similar to climate predictors used
to model the distribution of tree species in other stud-
ies, including lodgepole pine, by other researchers
(BoucherLalonde et al. 2012; Bell et al. 2014), but have
the advantage of having been modelled for the mid-
Holocene (6000 BP). Covariates were resampled to 1 km
spatial resolution for modelling. We iteratively gener-
ated 100 models, each of which used a subsample of
the absence data so that there were three times the
number of absence points as presence points. This was
done because Random Forests performs poorly when
classes are highly imbalanced (Chen et al. 2004). The
100 models were combined into a single classification
model. Summary statistics and graphs of covariance
convergence for the subsampled absence data were
used to evaluate the stability of the model. The out-of-
bag (Breiman 2001) error rate in predicting known pres-
ences and absences was 9%. This model was applied to
the historical distribution of climate variables in the
BIOCLIM data sets to predict the past distribution of
lodgepole pine. We estimated the relative amount of
lodgepole pine forest 6000 BP compared with the current
amount based on the combined area and relative
probabilities of occurrence during the two time periods.
Results
After removing barcodes from the raw reads, and dis-
carding contaminant reads, we retained 321 627 388
reads representing all 231 individuals. Initial de novo
assembly placed 24 352 918 reads into 403 678 contigs;
the 349 865 highest quality contigs from this assembly
were used as a GBS reference. We subsequently aligned
reads from all individuals to this reference using BWA.
After using SAMTOOLS and BCFTOOLS to call variant sites,
discarding loci with minor allele frequency <0.03, and
randomly sampling a single SNP per contig, we
retained a final set of 18 385 SNPs (mean coverage per
individual per locus of 7.29) for population genetic
analyses across the red crossbill complex. Phylogenetic
analyses were based on a set of 238 615 SNPs that were
called in the alignments of red crossbills and white-
winged crossbills, as described above and in the Sup-
porting information.
Phylogenetic analyses revealed topologies with white-
winged crossbills and red crossbills each forming
strongly supported monophyletic groups (Fig. 1), con-
sistent with previous studies (Questiau et al. 1999;
Parchman et al. 2006). In contrast to past studies (Parch-
man et al. 2006), South Hills crossbills formed a strongly
supported monophyletic group and were the only
monophyletic red crossbill lineage (Fig. 1). Individuals
©2016 John Wiley & Sons Ltd
GENOME DIVERGENCE IN CROSSBILLS 5
of the remaining ecotypes were dispersed throughout
the tree and showed no evidence of clustering, with the
exception of Type 6, for which bootstrap support was
weak (Fig. 1). Type 6 is the only ecotype that could be
considered allopatric to the other ecotypes, as it is con-
fined mainly to Mexico south of the other ecotypes
(Groth 1993). Consistent with phylogenetic analyses,
population genetic analyses based on 18 385 SNPs
revealed low levels of genome-wide genetic differentia-
tion between the different ecotypes, as indicated by
small Nei’s Dand F
ST
estimates (mean F
ST
=0.021,
range: 0.0110.035; Table S2, Supporting information).
Despite low levels of divergence, genetic differentia-
tion among the ecotypes and similarity among geo-
graphically dispersed samples within individual
ecotypes was evident in the PCA (Fig. 2), where the
first two principal components differed significantly
among ecotypes (PERMANOVA,F
8, 210
=1220; R
2
=0.98;
P<0.001). South Hills crossbills were the most distinct
in these analyses and were separated from other eco-
types along PC1, while the remaining ecotypes were
separated mostly along PC2 (Fig. 2). The four smallest
ecotypes (types 1, 3, 4 and 10) have the highest PC2
scores (Fig. 2), three of which are found mostly in the
Pacific Northwest (types 3, 4 and 10), whereas Type 1 is
the one ecotype found mostly in eastern North America
(Fig. S1, Supporting information; Groth 1993). Interme-
diate-sized ecotypes are found in the middle cluster,
including the two most abundant ecotypes in the Rocky
Mountain region (types 2 and 5) and Type 7, which is
uncommon but found within the geographic ranges of
types 2 and 5 in the northern Rocky Mountains and
west to the Cascades (C. W. Benkman, personal obser-
vations; Groth 1993). The largest ecotype is Type 6,
which has the smallest PC2 scores (Fig. 2) and occurs
mostly in Mexico, allopatric to the other ecotypes. Geo-
graphically dispersed samples from within the same
ecotype overlapped extensively in PC space, indicating
genetic cohesiveness within ecotypes.
Support for the distinctiveness of the South Hills
crossbill and genetic similarity of geographically dis-
persed samples within each of the ecotypes was also
found in Bayesian clustering analyses (ENTROPY; Gom-
pert et al. 2014). DIC values were similar from k=2
through k=6 (Table S3, Supporting information), and
all five models led to conclusions consistent with PCAs
above. Inspection of MCMC chains indicated sufficient
mixing and convergence only for k6 models. Subtle
allele frequency differences among some of the clusters
likely caused problems with the mixing of the MCMC
chains for k>6 models. We highlight results from
k=2, 3 and 5. In the k=2 model, South Hills crossbills
were assigned to one cluster, whereas all other individ-
uals were assigned to the other (Fig. 3A). The k=3
model also assigned South Hills crossbills to a single
cluster and assigned types 1, 3, 4 and 10 to a second
cluster, and types 2, 5, 6 and 7 to a third (Fig. 3B). The
k=5 model assigned individuals to clusters that largely
reflect the four nonoverlapping groups of ecotypes in
the PCA (Fig. 2), with types 5, 6 and 9 each assigned to
their own clusters, types 1, 3, 4 and 10 assigned to one
cluster, and types 2 and 7 assigned to the fifth cluster
(Fig. 3C).
We detected no evidence for isolation by distance.
Pairwise genetic distances were unrelated to geographic
distances (Fig. 4A), both for pairs of samples within the
same ecotype and between all 22 geographically sepa-
rate samples [the credible interval for the slope for the
relationship between genetic and geographic distance
included zero; slope for full analysis: 1.5 910
6
; 95%
credible or equal-tail probability interval (ETPI):
1.8 910
5
to 4.2 910
5
]. In contrast to the lack of
isolation by distance, ecotype beak depth tended to
decrease with increasing PC2 values (Fig. 2), and these
groupings of ecotypes were detected using ENTROPY
(Fig. 3). In addition, genetic distance tended to increase
with increasing beak depth divergence among ecotypes
(Fig. 4B), but this pattern was not statistically significant
(slope: 0.212; 95% ETPI: 0.008 to 0.439).
Although South Hills crossbills stand out in the phy-
logenetic and population genetic analyses (Figs 13),
point estimates of genome-wide differentiation were not
greater than in pairwise comparisons among all of the
other ecotypes (Fig. 5A). Instead, the upper tails of the
F
ST
distributions for the South Hills crossbill had higher
Fig. 1 A maximum-likelihood tree for the 219 red crossbills
(Loxia curvirostra) and 12 white-winged crossbills (L. l. leu-
coptera) based on 238 615 SNPs. Bootstrap support values on
the nodes are based on 500 bootstrap replicates and are only
shown for major nodes having >75% support; bootstrap
support for monophyly of Type 6 was 10.
©2016 John Wiley & Sons Ltd
6T. L. PARCHMAN ET AL.
densities (more loci with especially high F
ST
) (Fig 5B),
and these loci differentiated South Hills crossbills in
PCA and Bayesian clustering analyses. For example, in
pairwise comparisons between South Hills crossbills
and other ecotypes, there was a strong relationship
between locus-specific F
ST
and the strength of PC1 load-
ing (Fig. 5C), a pattern that does not exist in compar-
isons among the other ecotypes (Fig. 5D). Similarly, the
0.8 quantiles of the genome-wide F
ST
distributions were
higher for pairwise analyses involving South Hills
crossbills (Fig. 5B). Thus, elevated divergence in a
restricted number of genomic regions, rather than mean
genome-wide genetic divergence, distinguished South
Hills crossbills in PCA and ENTROPY analyses.
Estimates of nucleotide diversity indicate that South
Hills crossbills harbour lower levels of genetic diversity
Type 3, 8.15 mm
Type 4, 8.68 mm
Type 1, 8.78 mm
Type 7, 9.53 mm
Type 5, 9.33 mm
Type 2, 9.57 mm
Type 6, 10.99 mm
Type 10, 8.46 mm
PC 1 (11.4%)
PC 2 (2.4%)
Type 9, 9.79 mm
–0.05 0.00 0.05 0.10 0.15
–0.06
–0.04
–0.02
0.00
0.02
0.04
0.06
5
5
5
44
9
1
1
1
7
7
6
10
3
3
2
2
2
2
2
22
Fig. 2 Genotypic variation (based on 18 385 SNPs) among individuals summarized by the first two principal components from a
PCA of the matrix of genotype covariances between individuals. Lines connect individual PC values to the mean for each sampled
population, with the mean represented by circles. Numbers and colours correspond to ecotypes (call types), and different geographi-
cally separated samples from a given ecotype have the same number and colour. All geographically separate samples within an eco-
type, with the exception of Type 7, overlap in PC1-PC2 space. To the left, are representative study skins and the corresponding mean
beak depth of seven of the ecotypes (photograph from Groth 1993). Dotted lines connect the specimen images to their ecotype’s mean
PC values.
k
k
k
Admixture proportion (q)
0
1
0
(A)
(B)
(C)
1
0
1
Ecotype (call type) and sample location
A
L
KMNBC DE FGHJ
I
OPQS
R
310 4157 26
310 4157 26
310 4157 269
9
9
South Hills crossbill
DIC:
8.6 x 106
DIC:
8.7 x 106
DIC:
8.5 x 106
Fig. 3 Admixture proportion estimates
(q) from the hierarchical Bayesian model
implemented in ENTROPY. Each vertical
bar represents a bird, and bars are
coloured to reflect the posterior medians
of each individual’s admixture propor-
tions for each of kclusters. Results with k
equal to 2, 3 and 5 are shown. Numbers
along the abscissa represent ecotype, and
letters for geographically separate popu-
lations correspond to those given in
Table S1 (Supporting information). The
grey and black bars indicate boundaries
between population samples.
©2016 John Wiley & Sons Ltd
GENOME DIVERGENCE IN CROSSBILLS 7
than the other ecotypes (Fig. 7), which could reflect the
fact that South Hills crossbills reside in only ~70 km
2
of
lodgepole pine forest (Fig. 6) and likely have a much
smaller effective population size than other ecotypes.
Our historical reconstruction of lodgepole pine distribu-
tion in the region where South Hills crossbills occur
suggests that there was little pine forest available only
6000 years ago during a period of warming (Fig. 6B, D).
Unless the South Hills crossbill and pine began coevolv-
ing elsewhere and subsequently codispersed to the
South Hills, these results suggest that coevolution and
genome divergence occurred within the last 6000 years.
Discussion
Although levels of genetic differentiation were low,
many ecotypes correspond to genetically cohesive
groups that are differentiated from other such groups
(Fig. 2). The low levels of genetic differentiation in our
results and those of previous studies (Parchman et al.
2006; Bjorklund et al. 2013) are consistent with the eco-
type diversification occurring recently, in the face of
gene flow, or both. Glacial advances during the Pleis-
tocene caused severe reductions in habitat that likely
eroded ecotype diversity before glacial retreats allowed
a vast expansion of conifers and an ensuing diversifica-
tion of ecotypes (Benkman 1993; Dynesius & Jansson
2000). The absence of isolation by distance (Fig. 4A),
consistent with ecotype nomadism and opportunistic
breeding, indicates that genetic differentiation was not
dependent on geographic distance or isolation. Instead,
our results highlight the importance of adaptation to
alternative conifer species (Benkman 1993, 2003) in con-
tributing to reproductive isolation and genetic differen-
tiation. These results contrast with the evidence that
divergence without geographic isolation appears
uncommon in birds (Price 2008). Although this differ-
ence might be attributable to the use of much smaller
sets of genetic markers in past studies (but see Poelstra
et al. 2014; Mason & Taylor 2015), strong reproductive
isolation as a by-product of adaptation to alternative
resources (Smith et al. 1999, 2012; Smith & Benkman
2007) distinguishes crossbills from most bird species
(Price 2008).
Population genomic analyses indicate that the South
Hills crossbill (Type 9) was the most genetically distinct
ecotype (Figs 2 and 3A), and it was the only mono-
phyletic ecotype in phylogenetic analyses (Fig. 1). This
pattern of genetic differentiation indicates that previ-
ously documented patterns of divergent selection, adap-
tation (Benkman 1999; Benkman et al. 2003), and
reproductive isolation (Smith & Benkman 2007; Benk-
man et al. 2009) associated with a local coevolutionary
arms race have contributed to genome divergence. Our
results show that, rather than overall genome-wide
divergence, elevated genetic differentiation in a small
number of genomic regions characterizes divergence in
the South Hills crossbill (Fig. 5). This pattern is
expected when adaptive divergence occurs in the face
of gene flow (Peccoud et al. 2009; Feder et al. 2012),
although such a pattern could also arise from selective
sweeps in the absence of reproductive isolation (Cruick-
shank & Hahn 2014). Our findings are consistent with
recent studies of other vertebrate taxa with phenotypic
differentiation and reproductive isolation (Poelstra et al.
2014; Malinsky et al. 2015; Mason & Taylor 2015), where
differentiation is restricted to few genomic regions
across a background of genomic homogeneity. While
GBS data offer a coarse assessment of patterns of eco-
type genome divergence, whole genome resequencing
Geographic distance (km)
Genetic distance (Nei's D)
0.010
0.015
0.020
0.025
0.030
0.035
0.040
0 1000 2000 3000 4000 5000
Genetic distance (Nei's D)
Beak depth difference (mm)
0.010
0.015
0.020
0.0 0.5 1.0 1.5 2.0 2.5
(A) (B)
Fig. 4 (A) Genetic distances were unrelated to geographic distances between geographically separate samples within ecotypes (black
symbols) and between all 22 samples, and (B) genetic distances between ecotypes did not increase significantly with increasing beak
depth divergence among ecotypes (B includes only between-ecotype comparisons). Estimates of the effect of geographic distance or
beak depth on genetic distance are for appropriately transformed variables (see ‘Materials and methods’), rather than the
untransformed values in the plots.
©2016 John Wiley & Sons Ltd
8T. L. PARCHMAN ET AL.
could eventually resolve the size and organization of
genomic regions involved in divergence and speciation.
Genome divergence in the South Hills crossbill could
be influenced by geographic isolation, strength of repro-
ductive isolation and effective population size. South
Hills crossbills are not geographically isolated, as other
nomadic ecotypes regularly move through and breed in
the South Hills annually (Smith & Benkman 2007; Benk-
man et al. 2009). Moreover, the large scale over which
we were unable to detect isolation by distance (Fig. 4A),
suggests that 150 km of forestless area separating the
South Hills from the vast forests to the north is unlikely
to affect the opportunity for gene flow. Alternatively,
reproductive isolation is potentially stronger in the
South Hills than elsewhere, because strong density-
dependent food limitation (Benkman et al. 2012) and
cones with elevated defences against crossbills make it
more difficult for nonlocally adapted ecotypes to persist
and breed (Smith & Benkman 2007). Our test of isola-
tion by ecology was not statistically significant (Fig. 4B),
indicating that increasing divergent selection alone did
not result in an increase in genetic differentiation. How-
ever, as implied above, our measure of divergent selec-
tion (beak depth divergence) does not capture certain
0.035
0.030
0.025
0.020
0.015
0.010
1234571069 All other ecotypes Ecotype 6 Ecotype 9
30
25
20
15
10
5
0
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.00
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.00
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
FST
Ecotype Ecotype comparison
Locus-specific FST Locus-specific FST
Number of loci FST > 0.8
Absolute value of locus-specific PC1 loading
Absolute value of locus-specific PC1 loading
R2 = 0.48
R2 = 0.02
(A) (B)
(C) (D)
Fig. 5 Ecotypes 6 and 9, shown on the right in A and B, do not have particularly large pairwise F
ST
estimates relative to those includ-
ing only the other ecotypes (A), but they do have more numerous locus-specific F
ST
estimates >0.8 (of 18 385 loci) (B) in comparison
with pairwise estimates between the other ecotypes (ecotype 6 vs. all others: Wilcoxon pairwise test, Z=2.48, P=0.013; ecotype 9
vs. all others: Z=3.57, P=0.0004; ecotype 6 vs. 9: Z=0.16, P=0.16; similar patterns were found for F
ST
>0.9, but are not shown).
Analyses of PC1 loadings in relation to locus-specific F
ST
estimates for pairs of ecotypes revealed much stronger relationships when
ecotype 9 was included (C: ecotypes 2 and 9 shown; P<2.2 910
16
) than when it was excluded (D: ecotypes 2 and 10 shown;
P<2.2 910
16
). Box plots (A, B) show minimum, first quartile, median, third quartile and maximum. Dashed lines (C, D) represent
least-squares linear regressions.
©2016 John Wiley & Sons Ltd
GENOME DIVERGENCE IN CROSSBILLS 9
elements of ecology such as strength of density depen-
dence that likely affect reproductive isolation (Bolnick
2011). Furthermore, given the heterogeneity of differen-
tiation across the genome (Nosil et al. 2009), average
genome-wide divergence is a poor measure for such
analyses when divergence occurs with gene flow. An
additional factor that could contribute to the greater
genetic distinctiveness of the South Hills crossbill is its
small effective population size, as genetic drift could
increase relative genetic differentiation (F
ST
). Indeed,
ecotype-level estimates of heterozygosity (p) are lowest
for the South Hills crossbill (Fig. 7), consistent with a
small effective population size [our current (October
November 2015) total population estimate is N
c
~4000
birds; N. Behl and C. W. Benkman, unpublished data].
Small population size, in addition to geographic isola-
tion, has likewise been suggested to contribute to the
relatively elevated levels of genetic divergence of the
crossbill endemic to the Aleppo pine forests (P. halepen-
sis) on the island of Mallorca (L. c. balearica; Bjorklund
et al. 2013).
Reconstructions of historical lodgepole pine distribu-
tion using classification models (Random Forests) based
on BioClim data were consistent with palaeobotanical
(A) (B)
(C) (D)
Fig. 6 The distribution of Rocky Mountain lodgepole pine based on Random Forests models that infer the probability of occurrence.
A and C (inset in A enlarged) show the current distribution predicted by the models, which matches the actual distribution well
except in Nevada, and within the area of Utah in C, where lodgepole pine does not occur. B and D (inset in B enlarged) represent
the predicted lodgepole pine distribution 6000 BP. The amount of lodgepole pine forest in the South Hills and Albion Mountains,
Idaho, 6000 BP is estimated to have been 86% less than its current abundance in these two mountain ranges where South Hills
crossbills currently reside. Lodgepole pine did not occur in northwest Utah 6000 BP (Mehringer 1985) contra D.
©2016 John Wiley & Sons Ltd
10 T. L. PARCHMAN ET AL.
studies (Mehringer 1985; Davis et al. 1986) indicating
that pine forest was sparse during several thousand
years of warming centred on 6000 BP in the region
where South Hills crossbills currently reside (Fig. 6).
Given the current South Hills crossbill population size,
such a large (~86%) reduction in the amount of lodge-
pole pine would likely have prevented a distinct cross-
bill population from persisting in this region (Siepielski
& Benkman 2005). This conclusion is supported by the
evidence for exceptionally high temperatures 50007000
BP (Bartlein et al. 2014) conducive to lodgepole pine
experiencing frequent catastrophic fires (Westerling
et al. 2011), which would further reduce habitat for
crossbills.
Thus, the South Hills crossbill either diverged as it
coevolved with lodgepole pine in the South Hills dur-
ing the last 6000 years, or it diverged elsewhere then
subsequently colonized the expanded lodgepole pine
forests of the South Hills. The latter is unlikely, as our
reconstructions of historical lodgepole pine distributions
for 22 000 BP (not shown) and palaeobotantical studies
(Mehringer 1985) provide no indication that large
forested areas occurred in this region at an earlier time.
Furthermore, coevolving crossbills and pines were unli-
kely to have moved to the South Hills from elsewhere,
because to the east and north red squirrels are wide-
spread and lodgepole pine cones there reflect strong
selection exerted by red squirrels suggesting a history
of interaction (Benkman 1999; see Arbogast et al. 2001).
To the west and south, lodgepole pine does not and has
not occurred (Wells 1983). Finally, the high level of
reproductive isolation in the South Hills crossbill is
related to both the local resource characteristics that
have evolved in response to the absence of red squirrels
(Benkman & Siepielski 2004; Benkman et al. 2012) and
crossbill-pine coevolution resulting in strong density
dependence and local adaptation (Smith & Benkman
2007).
Conclusions
Coevolution has often been invoked to explain patterns
of macroevolutionary diversification (Ehrlich & Raven
1964; Thompson 2005; Jablonski 2008), and some com-
ponents of coevolution have been documented in
numerous natural populations (Thompson 1994, 2005).
However, clear examples of reciprocal selection and
adaptation driving speciation (the link between coevo-
lution as a micro- and macroevolutionary process) are
largely lacking (Althoff et al. 2014; Hembry et al. 2014).
Past studies on the South Hills crossbill have provided
strong evidence for the role of coevolution in driving
morphological divergence and reproductive isolation
(Benkman 1999, 2003; Benkman et al. 2001, 2013; Smith
& Benkman 2007). Our results indicate that the high
contemporary measures of premating reproductive iso-
lation (Smith & Benkman 2007; Benkman et al. 2009)
reflect a longer term barrier to gene flow. Moreover, the
nomadic behaviour of crossbills, their common sym-
patric occurrence and the absence of isolation by dis-
tance across all ecotypes (Fig. 4A) suggest that local
coevolution rather than geographic isolation per se is
responsible for the high levels of reproductive isolation
for the South Hills crossbill. Model-based reconstruc-
tions of the past lodgepole pine distribution in the
South Hills region (Fig. 6) and other lines of the evi-
dence indicate that the ancestors of South Hills cross-
bills became resident and began coevolving with
lodgepole pine more recently than 6000 BP. If the South
Hills crossbill evolved so recently, it could represent
one of the fastest examples of speciation in birds (Price
2008). Unfortunately, it is likely that this most geneti-
cally differentiated New World red crossbill lineage will
go extinct within this century due to climate change
and loss of suitable habitat (Santisteban et al. 2012;
Benkman in press).
The evidence presented here for genetic differentia-
tion associated with resource specialization in the
absence of clear geographic isolation is rare, but similar
to that seen in host races of insects (Mallet 2008; Pec-
coud et al. 2009; Nosil 2012; Powell et al. 2013). More
generally, there is growing evidence that reproductive
isolation can arise in part as a by-product of adaptation
to alternative resources (i.e. ecological speciation; Sch-
luter 2000; Nosil 2012). Crossbills might be unusual
among birds, which diverge primarily in allopatry
Ecotype
1234567910
2.90
2.95
3.00
3.05
Heterozygosity (x 10−3)
Fig. 7 Ecotype 9 (South Hills crossbills) exhibits much lower
genetic diversity (likelihood estimates of heterozygosity) than
the other ecotypes. The dashed line represents the overall
mean.
©2016 John Wiley & Sons Ltd
GENOME DIVERGENCE IN CROSSBILLS 11
(Price 2008), but the mechanisms underlying their
diversification could prove general considering the
tremendous diversity of host-specific insects (Mallet
2008) and that coevolution is thought to be a major dri-
ver of diversification (Thompson 2005; Althoff et al.
2014; Hembry et al. 2014). In particular, geographic vari-
ation in the coevolutionary process has been docu-
mented as an important source of divergent selection
for many interactions (Thompson 2005, 2009). In cases
where such divergent selection generates reproductive
isolation, the geographic mosaic of coevolution could
contribute prominently to ecological speciation.
Acknowledgements
This research was funded by NSF grants to C.A.B., the Robert
B. Berry Chair to C.W.B. and UNR Start-Up Funds to T.L.P.
We thank Carla Cicero, Museum of Vertebrate Zoology at The
University of California at Berkeley, for providing tissue sam-
ples that Jeff Groth had collected, Pim Edelaar for additional
samples and Mark Andersen for conducting the forest distribu-
tion modelling. We thank Pim Edelaar, Darren Irwin, Joshua
Jahner, Patrik Nosil, Cody Porter and Katie Wagner for
comments on drafts of the manuscript.
References
Alcaide M, Scordato ES, Price TD, Irwin DE (2014) Genomic
divergence in a ring species complex. Nature,511,8385.
Althoff DM, Segraves KA, Johnson MTJ (2014) Testing for
coevolutionary diversification: linking pattern with process.
Trends in Ecology & Evolution,29,8289.
Anderson M (2001) A new method for non-parametric multi-
variate analysis of variance. Austral Ecology,26,3246.
Arbogast BS, Browne RA, Weigl PD (2001) Evolutionary genet-
ics and Pleistocene biogeography of North American tree
squirrels (Tamiasciurus). Journal of Mammalogy,82, 302319.
Bartlein PJ, Hostetler SW, Alder JR (2014) Paleoclimate. In: Cli-
mate Change in North America (ed. Ohring G), pp. 151.
Springer International Publishing, Cham, Switzerland.
Bell DM, Bradford JB, Lauenroth WK (2014) Early indicators of
change: divergent climate envelopes between tree life stages
imply range shifts in the western United States. Global Ecol-
ogy and Biogeography,23, 168180.
Benkman CW (1993) Adaptation to single resources and the
evolution of crossbill (Loxia) diversity. Ecological Monographs,
63, 305325.
Benkman CW (1999) The selection mosaic and diversifying
coevolution between crossbills and lodgepole pine. The Amer-
ican Naturalist,154, S75S91.
Benkman CW (2003) Divergent selection drives the adaptive
radiation of crossbills. Evolution,57, 11761181.
Benkman CW (in press) The natural history of the South Hills
crossbill in relation to its impending extinction. The American
Naturalist,188.
Benkman CW, Siepielski AM (2004) A keystone selective
agent? Pine squirrels and the frequency of serotiny in lodge-
pole pine. Ecology,85, 20822087.
Benkman CW, Holimon WC, Smith JW (2001) The influence of
a competitor on the geographic mosaic of coevolution
between crossbills and lodgepole pine. Evolution,55, 282
294.
Benkman CW, Parchman TL, Favis A, Siepielski AM (2003)
Reciprocal selection causes a coevolutionary arms race
between crossbills and lodgepole pine. The American Natural-
ist,162, 182194.
Benkman CW, Smith JW, Keenan PC, Parchman TL, Santiste-
ban L (2009) A new species of the red crossbill (Fringillidae:
Loxia) from Idaho. Condor,111, 169176.
Benkman CW, Fetz T, Talluto MV (2012) Variable resource
availability when resource replenishment is constant: the
coupling of predators and prey. The Auk,129, 115123.
Benkman CW, Smith JW, Maier M, Hansen L, Talluto MV (2013)
Consistency and variation in phenotypic selection exerted by a
community of seed predators. Evolution,67, 157169.
Bjorklund M, Alonso D, Edelaar P (2013) The genetic structure
of crossbills suggests rapid diversification with little niche
conservatism. Biological Journal of the Linnean Society,109,
908922.
Bolnick DI (2011) Sympatric speciation in threespine stickle-
back: why not? International Journal of Ecology,2011,
942847. doi:10.1155/2011/942847.
BoucherLalonde V, Morin A, Currie DJ (2012) How are tree
species distributed in climatic space? A simple and general
pattern. Global Ecology and Biogeography,21, 11571166.
Breiman L (2001) Random forests. Machine Learning,45,532.
Chen LJ, Lee DS, Song ZP, Suh HS, Lu BR (2004) Gene flow
from cultivated rice (Oryza sativa) to its weedy and wild rela-
tives. Annals of Botany,93,17.
Clarke R, Rothery P, Raybould A (2002) Confidence limits for
regression relationships between distance matrices: estimat-
ing gene flow with distance. Journal of Agricultural, Biological,
and Environmental Statistics,7, 361372.
Cruickshank TE, Hahn MW (2014) Reanalysis suggests that
genomic islands of speciation are due to reduced diversity,
not reduced gene flow. Molecular Ecology,23, 31333157.
Cutler DR, Edwards Jr TC, Beard KH, Cutler A, Hess KT
(2007) Random forests for classification in ecology. Ecology,
88, 27832792.
Davis OK, Sheppard JC, Robertson S (1986) Contrasting cli-
matic histories for the Snake River Plain, Idaho, resulting
from multiple thermal maxima. Quaternary Research,26, 321
339.
Dynesius M, Jansson R (2000) Evolutionary consequences of
changes in species’ geographical distributions driven by
Milankovitch climate oscillations. Proceedings of the National
Academy of Sciences USA,97, 91159120.
Ehrlich PR, Raven PH (1964) Butterflies and plants: a study in
coevolution. Evolution,18, 586608.
Endler J (1973) Gene flow and population differentiation.
Science,179, 243250.
Falush D, Stephens M, Pritchard JK (2003) Inference of popula-
tion structure using multilocus genotype data: linked loci
and correlated allele frequencies. Genetics,164, 15671587.
Feder JL, Egan SP, Nosil P (2012) The genome of speciation-
with-gene-flow. Trends in Ecology & Evolution,28, 342350.
Gesch D, Oimoen M, Greenlee S, Nelson C, Steuck M, Tyler D
(2002) The national elevation dataset. Photogrammetric Engi-
neering and Remote Sensing,68,532.
©2016 John Wiley & Sons Ltd
12 T. L. PARCHMAN ET AL.
Gompert Z, Lucas LK, Nice CC, Fordyce JA, Forister ML, Buer-
kle CA (2012) Genomic regions with a history of divergent
selection affect fitness of hybrids between two butterfly spe-
cies. Evolution,66, 21672181.
Gompert Z, Lucas LK, Buerkle CA, Forister ML, Fordyce JA,
Nice CC (2014) Admixture and the organization of genetic
diversity in a butterfly species complex revealed through
common and rare genetic variants. Molecular Ecology,23,
45554573.
Gomulkiewicz R, Drown DM, Dybdahl MF et al. (2007) Dos
and don’ts of testing the geographic mosaic theory of coevo-
lution. Heredity,98, 249258.
Groth JG (1993) Evolutionary Differentiation in Morphology, Vocal-
izations, and Allozymes Among Nomadic Sibling Species in the
North American Red Crossbill (Loxia curvirostra) Complex
Vol. 127. University of California Press, Berkeley, CA, USA.
Gugger PF, Sugita S, Cavender Bares J (2010) Phylogeography
of Douglas-fir based on mitochondrial and chloroplast DNA
sequences: testing hypotheses from the fossil record. Molecu-
lar Ecology,19, 18771897.
Hembry DH, Yoder JB, Goodman KR (2014) Coevolution and
the diversification of life. The American Naturalist,184, 425
438.
Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A (2005)
Very high resolution interpolated climate surfaces for global
land areas. International Journal of Climatology,25, 19651978.
Hudson RR, Slatkin M, Maddison W (1992) Estimation of
levels of gene flow from DNA sequence data. Genetics,132,
583589.
Irwin K (2010) A new and cryptic call type of the red crossbill.
Western Birds,41,1025.
Jablonski D (2008) Biotic interactions and macroevolution:
extensions and mismatches across scales and levels. Evolu-
tion,62, 715739.
Jombart T, Devillard S, Balloux F (2010) Discriminant analy-
sis of principal components: a new method for the analy-
sis of genetically structured populations. BMC Genetics,11,
94.
Lamichhaney S, Berglund J, Alm
en MS et al. (2015) Evolution
of Darwin’s finches and their beaks revealed by genome
sequencing. Nature,518, 371375.
Li H (2011) A statistical framework for SNP calling, mutation
discovery, association mapping and population genetical
parameter estimation from sequencing data. Bioinformatics,
27, 29872993.
Li H, Durbin R (2009) Fast and accurate short read alignment
with Burrows-Wheeler transform. Bioinformatics,25, 1754
1760.
Li H, Handsaker B, Wysoker A et al. (2009) The sequence
alignment/map format and SAMtools. Bioinformatics,25,
20782079.
Malinsky M, Challis RJ, Tyers AM et al. (2015) Genomic islands
of speciation separate cichlid ecomorphs in an East African
crater lake. Science,350, 14931498.
Mallet J (2008) Hybridization, ecological races and the nature
of species: empirical evidence for the ease of speciation.
Philosophical Transactions of the Royal Society of London B: Bio-
logical Sciences,363, 29712986.
Martin CH, Cutler JS, Friel JP, Dening CT, Coop G, Wain-
wright PC (2015) Complex histories of repeated gene flow in
Cameroon crater lake cichlids cast doubt on one of the
clearest examples of sympatric speciation. Evolution,69,
14061422.
Mason NA, Taylor SA (2015) Differentially expressed genes
match bill morphology and plumage despite largely undif-
ferentiated genomes in a Holarctic songbird. Molecular Ecol-
ogy,24, 30093025.
Mehringer Jr PJ (1985) Late-Quaternary pollen record from the
interior Pacific Northwest and northern Great Basin of the
United States. In: Pollen Records of Late-Quaternary North
America Sediments (eds Bryant Jr WM, Holloway RG), pp.
167189. American Association of Stratigraphy and Palynol-
ogy Foundation, Dallas, Texas.
Nei M (1972) Genetic distance between populations. The Ameri-
can Naturalist,106, 283392.
Newton I (2006) Movement patterns of common crossbills Loxia
curvirostra in Europe. Ibis,148, 782788.
Nosil P (2012) Ecological Speciation. Oxford University Press,
Oxford, UK.
Nosil P, Egan SP, Funk DJ (2008) Heterogeneous genomic dif-
ferentiation between walking-stick ecotypes: “isolation by
adaptation” and multiple roles for divergent selection. Evolu-
tion,62, 316336.
Nosil P, Funk DJ, Ortiz-Barrientos D (2009) Divergent selection
and heterogeneous genomic divergence. Molecular Ecology,
18, 375402.
Nosil P, Gompert Z, Farkas TE et al. (2012) Genomic conse-
quences of multiple speciation processes in a stick insect.
Proceedings of the Royal Society B: Biological Sciences,279,
50585065.
O’Connell BM, LaPoint EB, Turner JA, Ridley T, Pugh SA, Wil-
son AM, Waddell KL, Conkling BL (2014) The Forest Inven-
tory and Analysis database. In: Database description and user
guide version 6.0 for phase 2. pp. 182 US Department of Agri-
culture Forest Service, Raleigh, NC.
Oksanen J, Blanchet F, Kindt R et al. (2013) Vegan: Community
Ecology Package. R package version 2.0-9.
Parchman TL, Benkman CW, Britch SC (2006) Patterns of
genetic variation in the adaptive radiation of New World
crossbills (Aves: Loxia). Molecular Ecology,15, 18731887.
Parchman TL, Gompert Z, Mudge J, Schilkey F, Benkman CW,
Buerkle CA (2012) Genome-wide association genetics of an
adaptive trait in lodgepole pine. Molecular Ecology,21, 2991
3005.
Parchman TL, Gompert Z, Braun MJ et al. (2013) The genomic
consequences of adaptive divergence and reproductive isola-
tion between species of manakins. Molecular Ecology,22,
33043317.
Peccoud J, Ollivier A, Plantegenest M, Simon JC (2009) A con-
tinuum of genetic divergence from sympatric host races to
species in the pea aphid complex. Proceedings of the National
Academy of Sciences USA,106, 74957500.
Plummer M (2003) JAGS: a program for analysis of Bayesian
graphical models using Gibbs sampling. In: Proceedings of the
3rd international workshop on distributed statistical computing.
Vol. 124.
Poelstra JW, Vijay N, Bossu CM et al. (2014) The genomic land-
scape underlying phenotypic integrity in the face of gene
flow in crows. Science,344, 14101414.
Potter KM, Hipkins VD, Mahalovich MF, Means RE (2013)
Mitochondrial DNA haplotype distribution patterns in Pinus
ponderosa (Pinaceae): range-wide evolutionary history and
©2016 John Wiley & Sons Ltd
GENOME DIVERGENCE IN CROSSBILLS 13
implications for conservation. American Journal of Botany,100,
15621579.
Powell THQ, Hood GR, Murphy MO et al. (2013) Genetic
divergence along the speciation continuum: the transition
from host race to species in Rhagoletis (Diptera: Tephritidae).
Evolution,67, 25612576.
Price T (2008) Speciation in Birds. Roberts and Company Pub-
lishers, Greenwood Village, Colorado.
Pritchard JK, Stephens M, Donnelly P (2000) Inference of popu-
lation structure using multilocus genotype data. Genetics,
155, 945959.
Questiau S, Gielly L, Clouet M, Taberlet P (1999) Phylogeo-
graphical evidence of gene flow among common crossbill
(Loxia curvirostra, Aves, Fringillidae) populations at the conti-
nental level. Heredity,83, 196205.
R Core Team (2013) R: A Language and Environment for Statisti-
cal Computing. R Foundation for Statistical Computing,
Vienna, Austria.
Santisteban L, Benkman CW, Fetz T, Smith JW (2012) Survival
and population size of a resident bird species are declining
as temperature increases. Journal of Animal Ecology,81, 352
363.
Schluter D (2000) The Ecology of Adaptive Radiation. Oxford
University Press, Oxford, UK.
Shafer A, Wolf JB (2013) Widespread evidence for incipient
ecological speciation: a meta-analysis of isolation-by-ecology.
Ecology Letters,16, 940950.
Siepielski AM, Benkman CW (2005) A role for habitat area in
the geographic mosaic of coevolution between red crossbills
and lodgepole pine. Journal of Evolutionary Biology,18, 1042
1049.
Smith JW, Benkman CW (2007) A coevolutionary arms race
causes ecological speciation in crossbills. The American Natu-
ralist,169, 455465.
Smith JW, Benkman CW, Coffey K (1999) The use and misuse
of public information by foraging red crossbills. Behavioral
Ecology,10,5462.
Smith JW, Sjoberg SM, Mueller MC, Benkman CW (2012)
Assortative flocking in crossbills and implications for ecolog-
ical speciation. Proceeding of the Royal Society B: Biological
Sciences,279, 42234229.
Snowberg LK, Benkman CW (2007) The role of marker traits in
the assortative mating within red crossbills, Loxia curvirostra
complex. Journal of Evolutionary Biology,20, 19241932.
Snowberg LK, Benkman CW (2009) Mate choice based on a
key ecological performance trait. Journal of Evolutionary Biol-
ogy,22, 762769.
Stamatakis A (2014) RAxML version 8: a tool for phylogenetic
analysis and post-analysis of large phylogenies. Bioinformat-
ics,30, 13121313.
Stamatakis A, Aberer AJ (2013) Novel parallelization schemes
for large-scale likelihood-based phylogenetic inference. In:
Parallel & Distributed Processing (IPDPS), 2013 IEEE 27th Inter-
national Symposium, pp. 11951204. IEEE Computer Society,
New York.
Summers RW, Dawson RJ, Phillips RE (2007) Assortative mat-
ing and patterns of inheritance indicate that the three cross-
bill taxa in Scotland are species. Journal of Avian Biology,38,
153162.
Thompson JN (1994) The Coevolutionary Process. University of
Chicago Press, Chicago, Illinois.
Thompson JN (2005) The Geographic Mosaic of Coevolution.
University of Chicago Press, Chicago, Illinois.
Thompson JN (2009) The coevolving web of life. The American
Naturalist,173, 125140.
Vavrek MJ (2011) Fossil: palaeoecological and palaeogeographi-
cal analysis tools. Palaeontologia Electronica,14, 16.
Wells PV (1983) Paleobiogeography of montane islands in the
Great Basin since the last glaciopluvial. Ecological Mono-
graphs,53, 341382.
Westerling AL, Turner MG, Smithwick EAH, Romme WH,
Ryan MG (2011) Continued warming could transform
greater Yellowstone fire regimes by mid-21st century. Pro-
ceedings of the National Academy of Sciences USA,108, 13165
13170.
T.L.P., C.A.B. and C.W.B. designed research; T.L.P. gen-
erated and assembled data; T.L.P., C.A.B. and V.S.C.
analysed data; and T.L.P. and C.W.B. wrote paper with
contributions from C.A.B. and V.S.C.
Data accessibility
A detailed description of our genotyping by sequencing
protocol, a matrix of genotype probabilities and. bam
files from BWA assemblies are available at the Dryad
digital repository (http://dx.doi.org/10.5061/dryad.
65d97).
Supporting information
Additional supporting information may be found in the online ver-
sion of this article.
Appendix S1 Methods.
Table S1 The number of individuals sampled of Loxia curviros-
tra ecotypes, including geographic locations of separate sam-
ples within each ecotype, as well as white-winged crossbills
Loxia leucoptera.
Table S2 Pairwise estimates of Nei’s D(upper triangle) and
F
ST
(lower triangle) among Loxia curvirostra ecotypes (call
types).
Table S3 Deviance information critserion (DIC) estimates for
entropy models run for k=2 through k=9. Lower estimates
of DIC reflect better model fit.
Fig. S1 The map illustrates sampling localities for red crossbill
(Loxia curvirostra complex). Individual points refer to geogra-
phically separate collection locations and correspond with
Table S1. Circles of the same number and color represent dif-
ferent geographical samples from a given ecotype.
©2016 John Wiley & Sons Ltd
14 T. L. PARCHMAN ET AL.
... S eed dispersal by birds is an essential ecological process that contributes to the regeneration and viability of plant communities [1][2][3] . The resulting mutualistic relationship between seed-dispersing birds and their food plants potentially leads to resource specialization and can subsequently act as a driving force in speciation 2,[4][5][6][7] . Nutcrackers (genus Nucifraga) exemplify a case of mutualism and are known for their preference for caching pine seeds. ...
... Despite previous reports of phenotypic variation in beak morphology between nutcracker subspecies, it is still unclear whether the mutualism between nutcrackers and pines resulted in co-evolutionary processes in which local adaptation to pine cone morphology was the driving factor of beak variation. For example, crossbills (Loxia) illustrate a typical co-evolving arms-race with lodgepole pines (Pinus contorta latifolia) where divergent selection on beak morphology contributes to reproductive isolation of populations resulting in different crossbill species 6,68,69 . Moreover, only limited genome-wide differentiation was found between crossbill ecotypes, present only at a small number of loci that arose over as little as 6000 years 6,7 . ...
... For example, crossbills (Loxia) illustrate a typical co-evolving arms-race with lodgepole pines (Pinus contorta latifolia) where divergent selection on beak morphology contributes to reproductive isolation of populations resulting in different crossbill species 6,68,69 . Moreover, only limited genome-wide differentiation was found between crossbill ecotypes, present only at a small number of loci that arose over as little as 6000 years 6,7 . If nutcrackers represent a similar case of co-evolution, we would expect that phenotypic differentiation between subspecies was primarily driven by variation in beak morphology. ...
Article
Full-text available
Bird-mediated seed dispersal is crucial for the regeneration and viability of ecosystems, often resulting in complex mutualistic species networks. Yet, how this mutualism drives the evolution of seed dispersing birds is still poorly understood. In the present study we combine whole genome re-sequencing analyses and morphometric data to assess the evolutionary processes that shaped the diversification of the Eurasian nutcracker (Nucifraga), a seed disperser known for its mutualism with pines (Pinus). Our results show that the divergence and phylogeographic patterns of nutcrackers resemble those of other non-mutualistic passerine birds and suggest that their early diversification was shaped by similar biogeographic and climatic processes. The limited variation in foraging traits indicates that local adaptation to pines likely played a minor role. Our study shows that close mutualistic relationships between bird and plant species might not necessarily act as a primary driver of evolution and diversification in resource-specialized birds. Genomic and phylogeographic analyses indicate that resource-specialization did not play a major role in the diversification and speciation of seed dispersing nutcrackers
... The red crossbill complex includes the three populations elevated to species status by checklist committees -parrot crossbill, Scottish crossbill and Cassia crossbill -as well as several distinctive populations in southern Europe and nine call types in North America (Smith and Benkman 2007). Based on climate models and projections of how long a redsquirrel-free South Hills lodgepole pine ecosystem likely existed in Idaho, Parchman et al. (2016) concluded the Cassia crossbill lineage is about 6000 years old. The ages of the parrot crossbill and Scottish crossbill lineages are estimated with less certainty, but they cannot be older than 11 000 years. ...
... As a follow-up to earlier genetic analyses, Parchman et al. (2016) compared almost 240 000 single-nucleotide polymorphisms (SNPs) isolated from 230 individual crossbills drawn from all North American call types including Cassia crossbills and found evidence of genetic structure among North American crossbill populations (Parchman et al. 2016). A principal components analysis (PCA) of genetic data that included over 18 000 SNPs indicated that individuals from the Cassia crossbill population fell in an isolated cluster along the primary axis of genetic variation. ...
... However, in this analysis, the call type 6 population also showed monophyly, albeit with much lower support. Call type 6 individuals have the largest bill size of North American populations surveyed (akin to the relatively larger size of the parrot crossbill in Europe) and are considered nearly allopatric to other crossbill populations (Parchman et al. 2016); yet call type 6 crossbills are not recognized as a distinct species. ...
Article
Full-text available
The process of speciation remains poorly understood and hence definitions of species continue to be a focus of sharp debate. A key unresolved issue in speciation debates is whether local populations that are somewhat distinct from a parent population in morphology or genotype but that remain connected to the parent population via gene flow should be recognized as species. The species status of the Cassia crossbill Loxia sinesciuris, as well as other distinctive populations in the red crossbill Loxia curvirostra clade including the parrot crossbill L. pytyopsittacus and the Scottish crossbill L. scotica, are test cases for whether the hundreds of other avian ecomorphs should be taxonomically elevated to full species. We argue that these regional populations of red crossbills do not meet stated criteria for species designation under the Biological Species Concept and, regardless of species concept, fail as species simply following logic, parsimony, and consistency. There are no diagnostic morphological characteristics that definitively separate birds in the Cassia crossbill, parrot crossbill, or Scottish crossbill populations from birds in other sympatric crossbill populations. Call type, the behavioral characteristic that is proposed to distinguish Cassia crossbills, parrot crossbills and Scottish crossbills from other populations of crossbills, is learned and can change within the life of an individual. There is evidence of substantial on-going gene flow between Cassia crossbills, parrot crossbills, and Scottish crossbill and other populations of red crossbills, and there is no divergence in mitochondrial genotype. Instead, there exist few fixed genetic differences in nuclear genotype between individuals from these local crossbill populations and individuals from other red crossbill populations. We propose that the recognition of the Cassia crossbill, parrot crossbill and Scottish crossbill as species sets a poor precedent in avian taxonomy and muddles understanding of the process of speciation.
... The development of many new methods in phylogeography and systematics was inspired in part by the challenge of understanding speciation during this dynamic historical period, including statistical phylogeography (Knowles and Maddison 2002), and phylogenetic methods that account for ILS Knowles and Carstens 2007;Kubatko and Degnan 2007) and gene flow (Long and Kubatko 2018). With the increasing availability of genomic data, these methods at the intersection of population genetics and phylogenetics are beginning to unravel cases of complex speciation in North America (e.g., Parchman et al. 2016;Morales et al. 2017;Colella et al. 2018Colella et al. , 2021Irwin et al. 2018;Linck et al. 2019;Battey 2020;Manthey et al. 2021;Mikkelsen and Irwin 2021) and across the globe (e.g., Chan et al. 2017Chan et al. , 2021Georges et al. 2018;Hosegood et al. 2020;Martin et al. 2020;Newton et al. 2020;Erickson et al. 2021;Schield et al. 2021). ...
Article
Full-text available
Complex speciation, involving rapid divergence and multiple bouts of post-divergence gene flow, can obfuscate phylogenetic relationships and species limits. In North America, cases of complex speciation are common, due at least in part to the cyclical Pleistocene glacial history of the continent. Scrub-jays in the genus Aphelocoma provide a useful case study in complex speciation because their range throughout North America is structured by phylogeographic barriers with multiple cases of secondary contact between divergent lineages. Here, we show that a comprehensive approach to genomic reconstruction of evolutionary history, i.e., synthesizing results from species delimitation, species tree reconstruction, demographic model testing, and tests for gene flow, is capable of clarifying evolutionary history despite complex speciation. We find concordant evidence across all statistical approaches for the distinctiveness of an endemic southern Mexico lineage (A. w. sumichrasti), culminating in support for the species status of this lineage under any commonly applied species concept. We also find novel genomic evidence for the species status of a Texas endemic lineage A. w. texana, for which equivocal species delimitation results were clarified by demographic modeling and spatially explicit models of gene flow. Finally, we find that complex signatures of both ancient and modern gene flow between the non-sister California Scrub-Jay (A. californica) and Woodhouse's Scrub-Jay (A. woodhouseii), result in discordant gene trees throughout the species' genomes despite clear support for their overall isolation and species status. In sum, we find that a multi-faceted approach to genomic analysis can increase our understanding of complex speciation histories, even in well-studied groups. Given the emerging recognition that complex speciation is relatively commonplace, the comprehensive framework that we demonstrate for interrogation of species limits and evolutionary history using genomic data can provide a necessary roadmap for disentangling the impacts of gene flow and incomplete lineage sorting to better understand the systematics of other groups with similarly complex evolutionary histories.
... For example climate, habitat and diet have been found to influence genetic variation for a wide variety of aquatic and terrestrial species (Pilot et al., 2006;Selkoe et al., 2010;MacDonald et al., 2020). When gene flow is restricted between habitats or ecosystems, it can lead to the formation of genetically distinct ecotypes (Kess et al., 2018;Parchman et al., 2016). ...
Article
Full-text available
Ecological, environmental, and geographic factors all influence genetic structure. Species with broad distributions are ideal systems because they cover a range of ecological and environmental conditions allowing us to test which components predict genetic structure. This study presents a novel, broad geographic approach using molecular markers, morphology, and habitat modeling to investigate rangewide and local barriers causing contemporary genetic differentiation within the geographical range of three white-crowned sparrow (Zonotrichia leucophrys) subspecies: Z. l. gambelii, Z. l. oriantha, and Z. l. pugetensis. Three types of genetic markers showed geographic distance between sampling sites, elevation, and ecosystem type are key factors contributing to population genetic structure. Microsatellite markers revealed white-crowned sparrows do not group by subspecies, but instead indicated four groupings at a rangewide scale and two groupings based on coniferous and deciduous ecosystems at a local scale. Our analyses of morphological variation also revealed habitat differences; sparrows from deciduous ecosystems are larger than individuals from coniferous ecosystems based on principal component analyses. Habitat modeling showed isolation by distance was prevalent in describing genetic structure, but isolation by resistance also had a small but significant influence. Not only do these findings have implications concerning the accuracy of subspecies delineations, they also highlight the critical role of local factors such as habitat in shaping contemporary population genetic structure of species with high dispersal ability.
... The North American Red crossbill complex is comprised of 10 "call types" or ecotypes (Types 1-8, 10, 11) that differ in vocalizations and morphology including bill size (Benkman, 1999;Groth, 1993;Young, 2011) and palate structure (Benkman, 1993(Benkman, , 2003Benkman and Young, 2020;Irwin, 2010) with most of the call types specialized on a single conifer (Benkman and Young, 2020). Call type 9 was recently given species status, now Cassia Crossbill (Loxia sinesciuris), because of its genetic distinctiveness (Parchman et al., 2016) and strong and constant reproductive isolation (Benkman et al., 2009;Smith and Benkman, 2007). During the Red crossbill irruption in Arkansas in 2012-13, Type 2 (Ponderosa Pine crossbill), Type 3 (Western Hemlock crossbill), and Type 5 (Lodgepole Pine crossbill) were reported (Smith et al., 2015). ...
Article
Unpredictable movement patterns of irruptive migrants make them difficult to study through structured survey methods. We used citizen science data to assess the distribution and habitat usage of an irruptive migrant outside its core area of occurrence. We curated 50 years of citizen science data on Red crossbills (Loxia curvirostra) in Arkansas and field observations to 1) assess their distribution and habitat use outside their core area of occurrence, 2) determine if Red crossbills breed in Arkansas, 3) characterize the occurrence and abundance of Red crossbills in Arkansas, and 4) identify “call types” or ecotypes of Red crossbills present in Arkansas during 2017–18 irruption. We constructed a sample bias-corrected species distribution model using elevation, land cover type, and monthly temperature and precipitation as predictors for MaxEnt algorithm. The most significant predictors of Red crossbill distribution were land cover type, elevation, and precipitation of October. The probability of Red Crossbill occurrence was high in developed areas, areas with large water bodies, open areas, and evergreen forests. Red crossbills were found to breed in Arkansas based on field observations and historical records. The time-series captures the major irruption years within last 50 years in Arkansas. It also depicts months when the probability of finding Crossbills in Arkansas is highest. In addition, we documented four “call types” or ecotypes of Red crossbills in Arkansas, including type 1 and 4 that were previously unreported. Our study illustrates utility of citizen science observations for understanding the occurrence, distribution, and habitat use of difficult to study nomadic species.
... sistant to the homogenizing effects of gene flow and stand out in contrast to the rest of the genome(Nosil & Feder, 2012), as has been documented in other systems (e.g.,Marques et al., 2016;Tavares et al., 2018). Indeed, Vermivora are just one of a growing number of systems where strong reproductive isolating barriers are present, yet genome-wide divergence is weak (e.g.,Karrenberg et al., 2019;Parchman et al., 2016Parchman et al., , 2018Sambatti et al., 2012;Turbek et al., 2021). ...
Article
Full-text available
Toews et al. assert that strong reproductive isolation in Vermivora is inconsistent with other lines of evidence. Here, we discuss how strong yet incomplete reproductive isolation is consistent with other results from this system.
Article
Studying the relationship between diversification and functional trait evolution among broadly co‐occurring clades can shed light on interactions between ecology and evolutionary history. However, evidence from many studies is compromised because of their focus on overly broad geographic or narrow phylogenetic scales. We addressed these limitations by studying 46 independent, biogeographically delimited clades of songbirds that dispersed from the Eastern Hemisphere into the Americas and assessed (1) whether diversification has varied through time and/or among clades within this assemblage, (2) the extent of heterogeneity in clade‐specific morphological trait disparity and (3) whether morphological disparity among these clades is consistent with a uniform diversification model. We found equivalent support for constant rates birth–death and density‐dependent speciation processes, with notable outliers having significantly fewer or more species than expected given their age. We also found substantial variation in morphological disparity among these clades, but that variation was broadly consistent with uniform evolutionary rates, despite the existence of diversification outliers. These findings indicate relatively continuous, ongoing morphological diversification, arguing against conceptual models of adaptive radiation in these continental clades. Additionally, they suggest surprisingly consistent diversification among the majority of these clades, despite tremendous variance in colonization history, habitat valences and trophic specializations that exist among continental clades of birds. Diversification trends vary between songbird clades in the Americas, but morphological evolution appears surprisingly uniform.
Article
Species delineation has a long and contentious history, yet most agree that sympatric populations exhibiting high levels of reproductive isolation and evolving independently are species. In an opinion piece, Hill and Powers (2021; hereafter H&P) claim that several recognized species of crossbills (Loxia spp.) do not represent species because by no measure are they discrete, the vocalizations used to categorize crossbills are learned, modified and can switch to that of a different species, and reproductive isolation is incomplete and weak. We argue that the behavioral and genetic evidence indicate that Cassia crossbills L. sinesciuris, which we focus on because the data relevant to species status are more diverse and extensive, are genetically discrete; call modification rarely leads to crossbill misclassification and overwhelmingly results in call divergence and enhanced discrimination; and are nearly completely reproductively isolated with little evidence of introgression from sympatric red crossbills. The differences in our conclusions result in part from H&P mischaracterizing and misconstruing the ecology of Cassia crossbills, geographic context of their divergence, and evidence for reproductive isolation. H&P seemingly require that crossbills must adhere to the typical model of bird speciation–protracted divergence in allopatry, followed by a gradual increase in sympatry if reproductive isolation and ecological divergence allow–and require evidence such as initial long periods of allopatry, FST values > 0.2, divergent mtDNA and intrinsic postzygotic isolation. Although such evidence commonly distinguishes bird species, an increasing number of studies show that such criteria are not necessary to indicate sympatric, evolutionarily independent lineages.
Article
Theoretical models indicate that speciation, especially when the scope for gene flow is great (e.g., sympatric speciation), is most likely when strong performance trade-offs coincide with reproduction. We tested this classic hypothesis using measures of the strength of three prezygotic reproductive isolating barriers (habitat isolation, reduced immigrant fecundity, and behavioral isolation) between two young (~2,000 years) and sympatric red crossbill (Loxia curvirostra) ecotypes. All three isolating barriers increased with increases in performance trade-offs, with total reproductive isolation varying between 0.72 and 1 (0 represents random mating, and 1 represents complete reproductive isolation). Strong trade-offs led to strong habitat isolation, an inability to breed in the "wrong" habitat, and more assortative flocks, with the latter leading to stronger behavioral isolation. Reproductive isolation decreased as resource availability increased relative to the demands of breeding, with higher resource availabilities eliminating the positive relationship between reproductive isolation and performance trade-offs. This latter result is consistent with previous work suggesting that increasing resource abundance dampens the effect of strong performance trade-offs on evolutionary divergence. Because many organisms, with the notable exception of host-specific phytophagous insects, rely on abundant food resources with weak performance trade-offs while breeding, our results may explain why sympatric speciation is uncommon.
Article
The study of reciprocal adaptation in interacting species has been an active and inspiring area of evolutionary research for nearly 60 years. Perhaps owing to its great natural history and potential consequences spanning population divergence to species diversification, coevolution continues to capture the imagination of biologists. Here we trace developments following Ehrlich and Raven's classic paper, with a particular focus on the modern influence of two studies by Dr. May Berenbaum in the 1980s. This series of classic work presented a compelling example exhibiting the macroevolutionary patterns predicted by Ehrlich and Raven and also formalized a microevolutionary approach to measuring selection, functional traits, and understanding reciprocal adaptation between plants and their herbivores. Following this breakthrough was a wave of research focusing on diversifying macroevolutionary patterns, mechanistic chemical ecology, and natural selection on populations within and across community types. Accordingly, we breakdown coevolutionary theory into specific hypotheses at different scales: reciprocal adaptation between populations within a community, differential coevolution among communities, lineage divergence, and phylogenetic patterns. We highlight progress as well as persistent gaps, especially the link between reciprocal adaptation and diversification. This article is protected by copyright. All rights reserved
Article
We compare the utility of two methods for estimating the average levels of gene flow from DNA sequence data. One method is based on estimating FST from frequencies at polymorphic sites, treating each site as a separate locus. The other method is based on computing the minimum number of migration events consistent with the gene tree inferred from their sequences. We compared the performance of these two methods on data that were generated by a computer simulation program that assumed the infinite sites model of mutation and that assumed an island model of migration. We found that in general when there is no recombination, the cladistic method performed better than FST while the reverse was true for rates of recombination similar to those found in eukaryotic nuclear genes, although FST performed better for all recombination rates for very low levels of migration (Nm = 0.1).
Book
In Speciation in Birds, Trevor Price, a University of Chicago professor and leading expert in the field, has written the most authoritative and modern synthesis on the subject to date. In clear and engaging prose and through beautiful illustrations, Price shows us why the field is as exciting and vibrant as ever. He evaluates the roles of natural selection and sexual selection. He asks how speciation contributes to some of the great patterns in species diversity such as the large number of species in the tropics, and the many endemic species on isolated islands. Throughout the book, Price emphasizes the integration of behavior, ecology, and genetics.
Article
We describe a model-based clustering method for using multilocus genotype data to infer population structure and assign individuals to populations. We assume a model in which there are K populations (where K may be unknown), each of which is characterized by a set of allele frequencies at each locus. Individuals in the sample are assigned (probabilistically) to populations, or jointly to two or more populations if their genotypes indicate that they are admixed. Our model does not assume a particular mutation process, and it can be applied to most of the commonly used genetic markers, provided that they are not closely linked. Applications of our method include demonstrating the presence of population structure, assigning individuals to populations, studying hybrid zones, and identifying migrants and admixed individuals. We show that the method can produce highly accurate assignments using modest numbers of loci—e.g., seven microsatellite loci in an example using genotype data from an endangered bird species. The software used for this article is available from http://www.stats.ox.ac.uk/~pritch/home.html.