Skip to main content
NCBI home page
Proceedings of the Royal Society B: Biological Sciences logoLink to Proceedings of the Royal Society B: Biological Sciences
. 2016 Jul 13;283(1834):20160731. doi: 10.1098/rspb.2016.0731

Mutations in different pigmentation genes are associated with parallel melanism in island flycatchers

J Albert C Uy 1,✉, Elizabeth A Cooper 1,2, Stephen Cutie 1, Moira R Concannon 3,4, Jelmer W Poelstra 3,5, Robert G Moyle 6,7, Christopher E Filardi 8
PMCID: PMC4947890  PMID: 27412275

Abstract

The independent evolution of similar traits across multiple taxa provides some of the most compelling evidence of natural selection. Little is known, however, about the genetic basis of these convergent or parallel traits: are they mediated by identical or different mutations in the same genes, or unique mutations in different genes? Using a combination of candidate gene and reduced representation genomic sequencing approaches, we explore the genetic basis of and the evolutionary processes that mediate similar plumage colour shared by isolated populations of the Monarcha castaneiventris flycatcher of the Solomon Islands. A genome-wide association study (GWAS) that explicitly controlled for population structure revealed that mutations in known pigmentation genes are the best predictors of parallel plumage colour. That is, entirely black or melanic birds from one small island share an amino acid substitution in the melanocortin-1 receptor (MC1R), whereas similarly melanic birds from another small island over 100 km away share an amino acid substitution in a predicted binding site of agouti signalling protein (ASIP). A third larger island, which separates the two melanic populations, is inhabited by birds with chestnut bellies that lack the melanic MC1R and ASIP allelic variants. Formal FST outlier tests corroborated the results of the GWAS and suggested that strong, directional selection drives the near fixation of the MC1R and ASIP variants across islands. Our results, therefore, suggest that selection acting on different mutations with large phenotypic effects can drive the evolution of parallel melanism, despite the relatively small population size on islands.

Keywords: melanism, MC1R, ASIP, Agouti, parallel evolution, Monarcha

1. Introduction

Convergent evolution of phenotypic traits across taxa, such as limb length in Anolis lizards [1] and pelage colour in Peromyscus mice [2], provides some of the strongest evidence of natural selection in the wild [3]. Although there is clear evidence that these convergent traits evolve in response to similar ecological pressures, surprisingly little is known about their underlying genetic basis [4,5]. Uncovering the genetic basis of convergent or parallel traits is essential, as it provides key insights into the genetic and developmental constraints that underlie diverse phenotypes [5–10], and the operation of selection as it acts on existing or novel mutations to drive convergence [11–13]. For instance, in threespine sticklebacks Gasterosteus aculeatus, multiple populations evolve similar plate armour morphology as they colonize fresh water lakes from marine habitats [14]. This parallel adaptation is mediated by repeated selection on allelic variants of the Ectodysplasin (EDA) gene [7]. These EDA variants are present in low frequency in marine populations, suggesting that standing genetic variation facilitates the repeated and rapid colonization of freshwater lakes by sticklebacks [7]. Knowledge of the genetic basis of parallel traits, combined with understanding the evolutionary processes that favour convergence, can therefore provide a comprehensive understanding of how natural selection operates [5].

In island birds, one of the most variable traits among closely related populations is plumage colour, with geographically disjunct populations sometimes converging on similar colour patterns [15,16]. Unlike, for example, bill variation, which has been linked to feeding ecology, the genetic basis of convergent plumage colour and the evolutionary processes that mediate its occurrence across islands remain little understood [15,17]. Because the efficacy of selection should be reduced in islands with small effective populations (Ne), the evolution of plumage traits on islands has been attributed to alternative mechanisms, such as relaxed selection or pleiotropy (reviewed by Grant [15]). This is because reduced Ne should affect the efficacy of selection either directly (i.e. Ne × s) or indirectly through the Hill–Robertson effect, which enhances interference by background, deleterious mutations through reduced recombination rates in smaller populations [18]. However, observations of multiple taxa converging on similar colour patterns implicate selection in driving the predictable evolution of plumage colour [19]. Information on the genetic basis of and the evolutionary processes that mediate parallel plumage colour could, therefore, provide unique insights into the evolutionary processes that operate in small populations.

In his seminal work on speciation, Ernst Mayr [20] presented the variable chestnut-bellied flycatcher Monarcha castaneiventris complex as a prime example of incipient speciation and parallel evolution of plumage colour. The M. castaneiventris clade comprises at least six named taxa that differ in plumage colour across narrow geographical gradients throughout the Solomon Islands [16,20]. A notable pattern is that subspecies in the southeastern part of the archipelago are similar in body size and bill morphology, but strikingly different in plumage colour [20]. A population on the large island of Makira has a chestnut belly and iridescent blue-black upper-parts (Monarcha castaneiventris megarhynchus; ‘chestnut-bellied form’), whereas nearby populations, approximately 8 km away, on the small satellite islands to the north (Ugi & Three Sisters) and southeast (Santa Ana and Santa Catalina) are entirely iridescent blue-black (Monarcha castaneiventris ugiensis, ‘melanic’; figure 1). The northern and southeastern melanic populations are over 100 km apart, yet are considered a single subspecies because of their similarity in plumage colour [20]. Phylogenetic analyses of the M. castaneiventris complex indicate that M. c. megarhynchus and M. c. ugiensis form a separate, well-supported clade that is sister to the other M. castaneiventris subspecies [21]. Neither taxon, however, is reciprocally monophyletic owing to gene flow between islands and the retention of ancestral polymorphisms, reflecting their recent divergence [22,23]. More recent phylogenomic analyses, however, suggest that the melanic population of Santa Ana is sister to the chestnut-bellied population of Makira, with the melanic population of Ugi as the outgroup to the two taxa [23]. Given that chestnut belly is the most likely ancestral state for the M. castaneiventris clade [21] and the entire Monarcha genus, in general [24], the phylogenomic pattern suggests that melanism evolved convergently between the southeastern Santa Ana and northern Ugi populations of M. c. ugiensis [23]. Alternatively, hybridization could explain similar plumage colour between melanic populations, given evidence of gene flow among islands. Information on the genetic basis of plumage colour could, therefore, help resolve the origins of melanism in this clade.

Figure 1.

Figure 1.

Open in a new tab

Plumage colour phenotype, and MC1R and ASIP genotype frequencies for melanic and chestnut-bellied subspecies of Monarcha castaneiventris, which consists of the chestnut-bellied Monarcha castaneiventris megarhynchus on Makira (orange island) and melanic Monarcha castaneiventris ugiensis subspecies on Ugi and Three Sisters to the north, and Santa Ana to the southeast (black islands). The top row of pie diagrams shows proportion of individuals, by locality, that were chestnut-bellied (orange), melanic (black) and partially melanic (grey) in plumage colour. The middle and bottom rows of pie diagrams show proportions of individuals that were homozygous for the ancestral alleles (orange), homozygous for the derived alleles (black) and heterozygous for the derived and ancestral alleles (grey) for MC1R Asp119Asn and ASIP Iln55Thr, respectively. Asterisks indicate study sites where samples were collected. See figure 3 for the phenotypes of the MC1R and ASIP heterozygotes (grey).

In the past century, over 100 genes have been identified that influence colour in vertebrates [25–27]. Most notably, mutations in the melanocortin-1 receptor (MC1R) and its antagonist, agouti signalling protein (ASIP), have been shown experimentally to have large effects on the expression of melanin-based coloration [28–31]. MC1R is a seven-transmembrane receptor that regulates melanogenesis in vertebrates. When bound to alpha melanocyte stimulating hormone (α-MSH), MC1R signals the production of eumelanin in melanosomes, which creates black to grey coloration. By contrast, when bound to ASIP, MC1R signals the production of phaeomelanin in melanosomes, which creates chestnut to brown coloration [27]. Association studies, complemented by direct mutagenesis, biochemical and pharmacological experiments, show that simple, point mutations in either of these genes can result in large changes in colour [25,26]. Taking advantage of this well-known genetic pathway and complemented by reduced representation genomic sequencing data [32], we explore the genetic basis of and the evolutionary processes that mediate parallel melanism between the two geographically isolated populations of M. c. ugiensis inhabiting small islands to the north and to the southeast of Makira Island (figure 1). We discuss our results in light of the expectation that a smaller effective population size should reduce the efficacy of natural selection on small islands.

2. Material and methods

(a). Collection

From 2006 through to 2012, tissue or blood samples were collected from 45 chestnut-bellied birds in Makira, 22 melanic birds in Santa Ana, 29 melanic birds in Ugi and 10 melanic birds in Three Sisters (electronic supplementary material, table S1). In addition, blood samples were taken from six birds with intermediate belly colours (e.g. partial melanic): three individuals from Makira and three individuals from Santa Ana. Birds were caught with mist nets then measured, tagged, blood-sampled and released as part of a long-term study [21,22]. Genomic DNA was extracted from blood samples using standard protocols (DNEasy DNA Extraction Kit, Qiagen, Valencia, CA, USA).

(b). Candidate gene and next generation sequencing approaches

For our candidate gene study, we sequenced the following genes/loci: MC1R (813 bp of the first and only exon) and ASIP (411 bp of all three exons, 704 bp of intron 1, 532 bp of intron 2 548 bp of the upstream region (approx. 400 bp from start of exon 1)). Ancestral state for the MC1R and ASIP allelic variants was inferred by sequencing the sister species Monarcha cinerascens (electronic supplementary material, table S1). Primers for sequencing MC1R and ASIP were developed using the zebra finch genome [33] and are detailed in the electronic supplementary material, table S2. Unincorporated primers and dNTPs from PCR products were removed using Exo-SAP-IT following the manufacturer's protocol (USB, Cleveland, OH, USA). Sequencing reactions using both forward and reverse primers were carried out at Cornell University's Life Sciences Core Laboratories Center and at the University of Miami's Molecular Core Facilities in the Department of Biology using an ABI 3730xl capillary DNA sequencer. Sequences were aligned and annotated using Sequencher 4.9 (Gene Codes Corporation, Ann Arbor, MI, USA). Population genetic parameters, such as FST (Wright's estimator) and diversity (π), were calculated using the program DnaSP v. 5.10 [34].

We complemented the candidate gene approach with genome-wide single nucleotide polymorphism (SNP) data recovered from restriction site associated DNA sequencing (RAD-seq) [32] of 22 chestnut-bellied and 2 partially melanic birds from Makira, 1 partially melanic and 19 melanic birds from Santa Ana, and 24 melanic birds from Ugi (a subset of birds used in the candidate gene approach). Multiplexed RAD tag libraries were constructed for each population following a modified version of the protocol described by Parchman et al. [35]. Briefly, the genomic DNA was double digested with EcoRI and MseI, and then ligated to barcoded Illumina sequencing adapters, which had been customized to recognize the restriction sites. Adapter-ligated DNA was then purified using Beckman-Coulter's AMPure purification kit (cat. no A63880) to remove adapter dimers, amplified via PCR and size-selected using gel electrophoresis. Samples were organized into libraries according to population, resulting in three libraries with a maximum of 24 barcoded individuals each. Libraries were sequenced at the Hussman Institute of Human Genetics, University of Miami in one-half of a lane on an Illumina HiSeq 2000, which produced 100 bp paired-end reads. Raw Illumina reads were sorted by barcode and filtered for quality (any read with a phred quality <20 was removed) using process_radtags from the Stacks package [36]. Filtered RAD tags were de novo assembled using the Stacks package, and individual reads were re-aligned to the consensus using BWA [37], with up to 10 mismatches allowed (note, however, that 99% of the aligned reads had five or fewer mismatches).

SNPs were called using vcfutils in the SAMtools package [37], with a minimum quality threshold of 20, and a minimum and maximum individual read depth threshold of 10 and 50, respectively. The read depth cut-offs were determined by examining the empirical distribution of coverage in our aligned data, which indicated that a depth above 50 represented an outlier. Using custom Perl scripts, we removed any sites with more than two alleles or fewer than 10 individuals per population, and then verified the heterozygous and homozygous variants called by SAMtools. We removed polymorphisms with less than 5% minor allele frequency in order to eliminate potential sequencing errors. We also required at least 20% of reads within an individual to contain the minor allele in order for that individual to be called heterozygous; individuals with between 0 and 20% minor alleles at a particular site were treated as missing data at that locus. These procedures were similar to the methods and cut-offs used by Rheindt et al. [38], although we were more conservative because we had lower overall coverage in our data. More details about our pipeline and the scripts can be found online at: https://github.com/eacooper400/RAD_Pipeline_Info.

Our study system comprises geographically isolated populations with nearly fixed phenotypes but limited population genetic structure [22] (see the electronic supplementary material, figure S1). We, therefore, used two complementary approaches to recover potential candidate genes for plumage colour and to infer the evolutionary processes that mediate the differences in plumage colour across islands: (i) a genome-wide association study (GWAS), which is more appropriate for studies of groups with little background, genetic structuring; and (ii) an FST outlier test, which is more appropriate for studies of distinct populations.

Combining the genome-wide SNP data from RAD-seq, and the MC1R and ASIP SNPs from the candidate gene approaches, we conducted a GWAS to scan for genetic variation associated with plumage colour using the mixed linear model implemented in the software package GEMMA [39]. Because plumage colour is correlated with locality (i.e. islands are fixed or nearly fixed for divergent plumage colour), we controlled for intrinsic, population genetic structure by running a principal component analysis (PCA) to calculate the major axis of genetic variation between populations (implemented in PLINK v. 1.7; [40]), then used the major PC axis as a covariate in the GWAS. Given that our previous results from a candidate gene approach suggested that the underlying genetic basis of melanism between the Santa Ana and Ugi populations are different [22], we ran separate GWAS for the Makira and Ugi (8907 SNPs), the Makira and Santa Ana (14 305 SNPs), and Santa Ana and Ugi (13 030 SNPs) comparisons. Chestnut-bellied birds were assigned a phenotypic value of 0, melanic birds were assigned a phenotypic value of 1, and partially melanic birds were assigned an intermediate phenotypic value of 0.5. For each SNP, significant association with plumage colour was determined using Wald's statistic, which was Bonferroni corrected for multiple comparisons (i.e. cut-offs for calling candidate SNPs were p = 5.6 × 10−6, 3.5 × 10−6 and 3.8 × 10−6 for the Makira and Ugi (8907 SNPs), Makira and Santa Ana (14 305 SNPs) and Santa Ana and Ugi (13 030 SNPs) comparisons, respectively).

Given that our comparisons involve distinct islands, we complemented the GWA approach by implementing FST outlier tests for the Makira and Ugi (8907 SNPs), Makira and Santa Ana (14 305 SNPs) and Santa Ana and Ugi (13 030 SNPs) comparisons to uncover SNPs that may be subject to directional selection and thus likely linked to candidate genes for divergent plumage colour [41]. We executed FST outlier tests in Bayescan v. 2.1 [42], which implements a Bayesian method to estimate the posterior probability that a particular locus or SNP is subject to selection. That is, Bayescan calculates a posterior probability (α) for a model in which selection better explains the data than the null model of neutrality, with positive values suggesting directional selection and negative values suggesting stabilizing selection. We used the default parameters for our runs; however, the false discovery rate (FDR) was set at 0.05 and the prior odds for the neutral model was set at 100 (i.e. the neutral model was set at 100 times more likely than a model with selection) to reduce the likelihood of false positives [41]. We checked for adequate mixing and convergence of our Markov chain Monte Carlo runs by evaluating trace plots of FST and Geweke's diagnostic tests [43], executed in CODA [44].

3. Results and discussion

(a). A point mutation in MC1R predicts melanism in the southeastern Santa Ana population

Consistent with our previous work [22], a comparison of most of the coding region of MC1R, here with a substantially larger sample size, indicates that a derived amino acid substitution at position 119 of MC1R (aspartic acid → asparagine; Asp119Asn) is strongly associated with melanism in the southeast Santa Ana population (figure 1; electronic supplementary material, table S3). All melanic birds from Santa Ana (n = 22) were homozygous for Asn119, while all chestnut-bellied birds from Makira (n = 45) were homozygous for Asp119. A mixed linear model GWAS using 14 305 SNPs distributed throughout the genome of 20 birds from Santa Ana and 24 birds from Makira recovered MC1R Asn119 as the best predictor of melanism, with only two other SNPs approaching MC1R's strong association (electronic supplementary material, table S3). That is, when the MC1R SNP is combined with the RAD-seq dataset, the GWAS that incorporated intrinsic population structure as a covariate recovered MC1R as the best predictor of plumage colour between melanic Santa Ana and chestnut-bellied Makira birds, with Asp119Asn having a β-value of 0.50 ± 0.00 (p = 0.0). Only two other SNPs were significant, with β-values of 0.31 ± 0.06 (p = 2.6 × 10−6) and 0.29 ± 0.06 (p = 3.3 × 10−6). The two additional, significant SNPs did not map onto or near known genes in the zebra finch genome. Likewise, the Bayescan FST outlier test using the same dataset recovered MC1R Asp119Asn as the only significant outlier (figure 2a).

Figure 2.

Figure 2.

Open in a new tab

Results of the Bayescan FST outlier analyses for the comparisons between (a) Makira (chestnut-bellied) and Santa Ana (southeast melanic; 14 303 SNPs), (b) Makira and Ugi (north melanic; 8905 SNPs) and (c) Santa Ana and Ugi (13 208 SNPs) populations. The horizontal axis shows log10 of the q-value, which is the false discovery rate (FDR) analogue of the p-value. The vertical axis shows the FST for each SNP (open circles), with the diamonds indicating outliers detected by Bayescan. The vertical line indicates the log10 of FDR = 0.05. (Online version in colour.)

Additional lines of evidence strongly suggest a causal role for Asn119 in melanism for Santa Ana birds. First, we sequenced MC1R for six unusual birds with intermediate belly colour from Makira (n = 3 of 48 birds) and Santa Ana (n = 3 of 25 birds). All six were heterozygous for Asp119Asn, suggesting that this MC1R allele is partially dominant. Note also that the genome-wide level of heterozygosity based on approximately 14 000 genome-wide SNPs did not differ among the three partially melanic, 22 chestnut-bellied and 19 melanic birds included in the RAD-seq study (electronic supplementary material, figure S2), suggesting that heterozygosity in Asp119Asn for partially melanic birds is not simply due to them being F1 hybrids of matings between Makira and Santa Ana birds (e.g. overall high heterozygosity in the genomic background). Second, experimental and association studies in other taxa indicate that point mutations in this region of MC1R can lead to melanism [25], and, in fact, Asn119 is the same mutation in the homologous position of MC1R responsible for melanism in some breeds of sheep and pigs [45,46]. Finally, in vitro functional experiments indicate that the position corresponding to Asp119Asn is necessary for high affinity ligand binding [45], and that the Asn119 mutation may activate MC1R by mimicking ligand binding [47]. In summary, the co-segregation of genotype and phenotype, including six birds with intermediate belly colour, and the results from previous functional experiments and comparative studies strongly implicate Asn119 as the cause of melanism in the southeastern Santa Ana birds.

(b). A point mutation in ASIP is the best predictor of melanism in the northern Ugi and Three Sisters populations

Because birds from the southeastern and northern satellite islands are both melanic in plumage colour (figure 1), we tested if MC1R Asn119 likewise predicts melanism in the northern satellite islands of Ugi and Three Sisters. We found a much weaker association between MC1R Asn119 and melanism (figure 1), suggesting that northern melanic birds may have a different or additional genetic basis for melanism. That is, the derived MC1R Asn119 allele is present in high frequency in the melanic populations of Ugi and Three Sisters, but the association between colour and genotype is weak (figure 1). In fact, several melanic birds were homozygous for the ancestral, chestnut Asp119 allele (figure 1). This pattern suggests the possibility that a different or additional mutation mediates melanism in the northern M. c. ugiensis populations of Ugi and Three Sisters.

We, therefore, sequenced the entire coding region of ASIP (three exons), the antagonist of MC1R, and found a polymorphic site in exon 1 that results in an amino acid substitution at position 55 of ASIP (isoleucine → threonine; Ile55Thr). Exons 2 and 3 were not variable. All melanic birds from the northern satellite islands (n = 29 on Ugi Island; n = 10 on Three Sisters) are homozygous for threonine in this position (Thr55; figure 1). The Thr55 allele, however, is nearly absent in the chestnut-bellied Makira (n = 48) and melanic Santa Ana (n = 25) populations, with both melanic and chestnut-bellied birds carrying copies of isoleucine instead of threonine (Ile55; figure 1). In the chestnut-bellied population of Makira, the Thr55 allele is present as a single copy in a heterozygote, and this bird was intermediate in belly colour. In the melanic Santa Ana population, we likewise caught a single bird heterozygous for ASIP position 55 but this individual was entirely melanic. A GWAS that explicitly controlled for population structure, and used 8907 SNPs from the combined RAD-seq and candidate gene approaches recovered ASIP Ile55Thr as the best predictor of plumage colour for the melanic birds of Ugi and chestnut-bellied birds of Makira. ASIP Ile55Thr had the highest β-value of 0.49 ± 0.02 (p = 1.18 × 10−25), and only one other candidate marker was uncovered by the GWAS, which had a β-value of 0.29 ± 0.05 (p = 1.8 × 10−7; see the electronic supplementary material, table S4). This SNP did not map onto a known gene in the zebra finch genome. MC1R Asp119Asn was not recovered by the GWAS as a predictor of plumage colour. Similar patterns were found in the Ugi and Santa Ana comparison, with ASIP Ile55Thr having the highest β-value of 0.46 ± 0.03 (p = 3.7 × 10−21). Seven other candidate SNPs were uncovered by the GWAS, but none mapped onto or near annotated genes in the zebra finch genome (see the electronic supplementary material, table S5). The Bayescan FST outlier analyses corroborated the results of the GWAS, recovering ASIP Ile55Thr as the only significant outlier for the Makira versus Ugi, and Ugi versus Santa Ana comparisons (figure 2b,c).

Based on sequences from an outgroup species, M. cinerascens, and other more distantly related avian species, isoleucine is the ancestral state for ASIP's 55th amino acid residue [48]. Alignment with other vertebrate ASIP protein sequences indicates that Ile55Thr is in the N-terminal chain of ASIP [48]. The N-terminal chain is predicted to bind to attractin, an accessory receptor adjacent to MC1R thought to mediate ASIP's antagonistic activity [48]. In fact, because of its important role in binding, the N-terminal chain is highly conserved across distantly related mammalian and avian taxa [48]. Further, the Ile55Thr mutation results in a change from a non-polar to a polar residue (hydrophobic to hydrophilic, respectively), which can influence the affinity of the N-terminal chain to attractin [49] and could thus limit ASIP's antagonistic activity on MC1R. In the Japanese quail Coturnix japonica, dark plumage (recessive black) results from a frameshift mutation in the N-terminal chain that causes the dysfunction of ASIP as an antagonist of α-MSH [50], while light plumage (yellow) results from a large > 90 kb deletion in RALY, which regulates ASIP [51]. Similarly, in two Peromyscus mice populations, a premature stop codon eliminating the N-terminus of ASIP is associated with melanism [29], while a deletion in the N-terminal chain is associated with light pelage colour [12]. Although the point mutation in the final Peromyscus example results in light and not melanic coloration, the result from this study suggests that relatively simple mutations in the N-terminal domain can have large effects on ASIP's interaction with α-MSH and MC1R. Overall, therefore, a causal role for ASIP Thr55 in mediating melanism in the northern satellite islands of Ugi and Three Sisters is suggested by the clear genotype-phenotype association from the candidate gene and GWAS approaches, coupled with the experimental and comparative data indicating the N-terminal chain's critical function in receptor binding and highly conserved nature across vertebrates [48,49].

Without functional experiments to directly quantify the effects of ASIP Thr55, we cannot definitively exclude the possibility that this point mutation is simply in linkage disequilibrium with other mutation(s) in ASIP. We explored this hypothesis by surveying the scale of linkage disequilibrium in ASIP, sequencing upstream and downstream of ASIP's first exon where Thr55 resides.

Between the northern melanic M. c. ugiensis and chestnut-bellied Makira M. c. megarhynchus populations, the FST for ASIP Ile55Thr is 0.99 when partially melanic birds are included or 1.0 when only chestnut-bellied and melanic birds are included (electronic supplementary material, figure S3). By contrast, the FST values downstream of the first exon (i.e. portions of introns 1 and 2) are much lower, with a sharp decline within 1500 bp downstream (electronic supplementary material, figure S3). The significant reduction in population structure for the intronic sequences downstream of exon 1 (and lack of variation in exons 2 and 3) suggests that the causal mutation is not downstream of ASIP's first exon.

In the region adjacent and upstream of exon 1, we found six segregating sites in 516 bp sequenced. Five of the six SNPs showed only minor variation across islands. However, a single point mutation is fixed in the northern melanic populations (428 bp upstream of exon 1; g.428A>T), resulting in relatively high FST values between islands for this SNP (electronic supplementary material, figure S3). The strong linkage disequilibrium between g.428T and Thr55 makes it difficult to definitively exclude g.428T as the causal mutation for melanism in the northern satellite islands. Genotypes of individuals outside the northern satellite islands, however, provide insights on its possible role in melanism.

The derived g.428T allele is present in three heterozygotes in the chestnut-bellied Makira population. All three birds, which were homozygous for ancestral ASIP Ile55 and MC1R Asp119 alleles, had entirely chestnut bellies despite carrying a copy of the derived g.428T allele. Further, the single bird that was heterozygous for ASIP Ile55Thr was homozygous for the ancestral g.428A allele, yet it was partially melanic in plumage colour (electronic supplementary material, figure S3). Finally, in Monarcha castaneiventris obscurior, a subspecies that is polymorphic in plumage colour (i.e. melanic and chestnut-bellied birds) and found in the Russell Islands over 300 km from Makira, four birds were homozygous for the derived g.428T allele but homozygous for the ancestral ASIP Ile55 and the ancestral MC1R Asp119 allelic variants. If g.428T mediates melanism, these four birds should be melanic. Only two of the four birds, however, were melanic, with the remaining two having pure chestnut bellies (electronic supplementary material, figure S4). The imperfect association between the g.428A > T allelic variants and plumage colour, especially the cases where chestnut-bellied birds were homozygous for the derived g.428T allele and a partially melanic bird was homozygous for the ancestral g.428A allele, suggests that g.428T does not cause melanism and is, instead, probably in strong linkage disequilibrium with Thr55 in the northern M. c. ugiensis birds. Overall, however, the strong linkage disequilibrium between g.428A > T and Ile55Thr prevents us from completely ruling out additional mutations upstream of g.428A > T as the causal mutation(s) for melanism in the northern populations. Regardless of which of the two point mutations is causal or if there are additional mutations upstream of Ile55Thr, the clear association between these ASIP variants and melanism suggests that mutation(s) in ASIP mediate(s) melanism in the northern populations of M. c. ugiensis.

Breeding and functional experiments will provide more direct tests of the causal role of ASIP and MC1R mutations in melanism. However, our results showing the clear association between mutations in genes known to mediate pigmentation, after controlling for population structure, implicate mutations in MC1R and ASIP in causing parallel melanism in the two, geographically separated populations of M. c. ugiensis.

(c). Interaction between ASIP and MC1R allelic variants

With each M. c. ugiensis and M. c. megarhynchus individual genotyped for the coding sequences of ASIP and MC1R and having sampled six birds of intermediate belly colour in Makira and Santa Ana, we can infer how alleles from the two loci interact to create specific plumage colours. Figure 3 summarizes the genotype combinations and corresponding belly colour in a 3 × 3 genotype interaction table. Individuals homozygous for the ancestral alleles for both MC1R and ASIP had chestnut bellies (top left in figure 3). Individuals homozygous for the derived ASIP allele were melanic regardless of their MC1R genotype (right column in figure 3). Similarly, individuals homozygous for the derived MC1R allele were melanic regardless of their ASIP genotype (bottom row in figure 3). Individuals heterozygous for MC1R or ASIP were intermediate in belly colour as long as they were not homozygous for either the derived MC1R or ASIP alleles. These patterns suggest that: (i) the derived MC1R and ASIP alleles are partially dominant; (ii) being homozygous for either the derived ASIP or MC1R alleles is, by itself, sufficient to cause melanism; and (iii) ASIP and MC1R interact epistatically. Overall, these patterns suggest a simple genetic mechanism for melanism, similar to other avian and mammalian species (reviewed in [25,26]).

Figure 3.

Figure 3.

Open in a new tab

Association between MC1R and ASIP genotypes, and belly colour. The islands where each genotype combination was sampled (and corresponding sample size) are listed below each representative photograph of an individual's chest and belly.

(d). Directional selection drives the differentiation of MC1R and ASIP between islands

Results from our FST outlier analyses are consistent with the hypothesis that MC1R and ASIP are subject to strong directional selection (figure 2). In the comparison between the chestnut-bellied Makira and the melanic Santa Ana populations, MC1R Asp119Asn was the only significant outlier detected, with an α score of 2.02 (figure 2a). Similarly, in the comparison between the chestnut-bellied Makira and the melanic Ugi population to the north, ASIP Ile55Thr was the only outlier and had an α score of 1.76 (figure 2b). Finally, in the comparison between the two melanic populations of Santa Ana and Ugi, ASIP Ile55Thr was the only outlier detected, with an α score of 1.59 (figure 2c). Overall, the significant and positive alpha scores, as estimated by Bayescan, indicate that MC1R and ASIP are under strong directional selection, which drives the near fixation of alternative MC1R or ASIP allelic variants between the chestnut-bellied Makira population and the melanic populations of Santa Ana and Ugi, respectively.

Because islands typically have a smaller effective population size (Ne) than their mainland counterparts (or larger islands in a chain), the efficacy of natural selection should be reduced on small islands. Based on field observations of territory size in this species [22] and the size of each island, we estimate the effective population size at 2342 and 6746 for the Santa Ana and Ugi populations, respectively (see the electronic supplementary material for our methods of estimating Ne). By contrast, assuming that birds only breed in comparable habitats (lowland, secondary growth rainforest, which is approx. 1/3 of Makira) in the large island of Makira, a very conservative estimate of Ne for this population is 149 942, which is about 20 to 60 times greater than the melanic populations of the satellite islands. As such, despite Ugi and Santa Ana's relatively small effective population size, our results suggest that strong directional selection can drive the evolution of parallel melanism on very small islands.

4. Conclusion

In mice (Peromyscus spp.; Chaetodipus intermedius) and lizards (Sceloporus undulates, Aspidoscelis inornata and Holbrookia maculata), natural selection for crypsis across a variable visual background maintains a stable colour polymorphism that is similarly mediated by mutations in MC1R or ASIP [10–12,52]. Conversely, loss of pigmentation across independent populations of the cave-dwelling fish Astyanax mexicanus is mediated by unique mutations in MC1R [9]. These studies suggest that parallel ecological selection can favour the evolution and maintenance of convergent phenotypes between independent populations, and that these convergent traits can be mediated by the same [53,54] or different [9,10,29,52] genetic mechanisms. Likewise, our results suggest that parallel, diversifying selection may favour similar phenotypes that are associated with different pigmentation genes across two geographically separated populations of a single subspecies. One possible mechanism of selection between islands is assortative mating or conspecific recognition based on plumage colour. Our previous field experiments indicate that divergent plumage colour is used by melanic birds from Santa Ana and chestnut-bellied birds from Makira in recognizing conspecifics, suggesting divergent plumage colour as the basis for premating reproductive isolation or discrimination against hybrid phenotypes [22,55]. Mate choice experiments should provide more direct insights into the role of divergent plumage colour in assortative mating as a mechanism for disruptive selection. Regardless of the specific mechanism, however, our results highlight the diversity of simple molecular pathways that can give rise to similar morphological traits between populations of a single species, as well as the effectiveness of natural selection even in small island populations.

Supplementary Material

Electornic Supplementary Materials
rspb20160731supp1.pdf (2.5MB, pdf)

Acknowledgements

For assistance in the Solomon Islands, we thank the Murray family and Jerry Tuari of Kirakira, and the staff and students of Pawa Secondary School (Ugi Island). For excellent assistance in the field, we thank Lonsdale Taka, James Suafuria and George Wabeasi of Star Harbour, and Henry Pirigua of Santa Ana. We thank the American Museum of Natural History, University of Kansas Natural History Museum and the Burke Museum, University of Washington for access to tissue samples. For helpful advice on the implementation of GEMMA, we thank A. Comeault. We thank R. C. Albertson, H. E. Hoekstra, D. C. Presgraves, R. J. Safran, W. T. Starmer, F. Mora-Kepfer Uy, members of the Uy & W. E. Searcy laboratories, and anonymous reviewers and Associate Editor P. Nosil for thoughtful advice and comments on various versions of this manuscript.

Ethics

This research adhered to the Institutional Animal Care and Use Committee guidelines for the use of vertebrate animals in research (protocol no. 11–116) and the legal requirements of the country in which the work was carried out. Permission to work in the Solomon Islands was granted by the Ministry of Education, and the Ministry of Environment, Conservation and Meteorology of the Solomon Islands.

Data accessibility

RAD-seq SNP data for GWAS: Dryad ID 877n3 (http://dx.doi.org/10.5061/dryad.877n3). DNA sequence: GenBank submission numbers: 1842234, 1842238–40, 1842246–47, 1842249, 1842252, 1842255, 1842259, 1842266.

Authors' contributions

J.A.C.U., lead and corresponding author: research design, fieldwork, laboratory procedures, funding, data analysis and primary writing; EAC, co-author: laboratory procedures, bioinformatics and writing; S.C., co-author: laboratory procedures and writing; M.R.C., co-author: laboratory procedures and writing; J.W.P., co-author: laboratory procedures and writing; R.G.M., co-author: laboratory procedures, fieldwork and writing. C.E.F., co-author: fieldwork and writing.

Competing interests

The authors declare no conflict of interest.

Funding

This work was supported by a National Science Foundation CAREER award (IOS 1137624/0643606), a National Geographic Society CRE award (9023-11), the College of Arts and Sciences of the University of Miami and the Aresty Chair in Tropical Ecology to J.A.C.U.

References

  • 1.Losos JB, Jackman TR, Larson A, de Queiroz K, Rodriguez-Schettino L. 1998. Historical contingency and determinism in replicated adaptive radiations of island lizards. Science 279, 2115–2118. ( 10.1126/science.279.5359.2115) [DOI] [PubMed] [Google Scholar]
  • 2.Steiner CC, Römpler H, Boettger LM, Schöneberg T, Hoekstra HE. 2009. The genetic basis of phenotypic convergence in beach mice: similar pigment patterns but different genes. Mol. Biol. Evol. 26, 35–45. ( 10.1093/molbev/msn218) [DOI] [PubMed] [Google Scholar]
  • 3.Schluter D, Clifford EA, Nemethy M, McKinnon JS. 2004. Parallel evolution and inheritance of quantitative traits. Am. Nat. 163, 809–822. ( 10.1086/383621) [DOI] [PubMed] [Google Scholar]
  • 4.Manceau M, Domingues VS, Linnen CR, Rosenblum EB, Hoekstra HE. 2010. Convergence in pigmentation at multiple levels: mutations, genes and function. Phil. Trans. R. Soc. B 365, 2439–2450. ( 10.1098/rstb.2010.0104) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Stern DL. 2013. The genetic causes of convergent evolution. Nat. Rev. Genet. 14, 751–764. ( 10.1038/nrg3483) [DOI] [PubMed] [Google Scholar]
  • 6.Sucena E, Delon I, Jones I, Payre F, Stern DL. 2003. Regulatory evolution of shavenbaby/ovo underlies multiple cases of morphological parallelism. Nature 424, 935–938. ( 10.1038/nature01768) [DOI] [PubMed] [Google Scholar]
  • 7.Colosimo PF, et al. 2005. Widespread parallel evolution in sticklebacks by repeated fixation of Ectodysplasin alleles. Science 307, 1928–1933. ( 10.1126/science.1107239) [DOI] [PubMed] [Google Scholar]
  • 8.Protas ME, Hersey C, Kochanek D, Zhou Y, Wilkens H, Jeffrey WR, Zon LI, Borowsky R, Tabin CJ. 2006. Genetic analysis of cavefish reveals molecular convergence in the evolution of albinism. Nat. Gen. 38, 107–111. ( 10.1038/ng1700) [DOI] [PubMed] [Google Scholar]
  • 9.Gross JB, Borowsky R, Tabin CJ. 2009. A novel role for Mc1r in the parallel evolution of depigmentation in independent populations of the cavefish Astyanax mexicanus. PLoS Genet. 5, e1000326 ( 10.1371/journal.pgen.1000326) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Rosenblum EB, Römpler H, Schöneberg T, Hoekstra HE. 2010. Molecular and functional basis of phenotypic convergence in white lizards at White Sands. Proc. Natl Acad. Sci. USA 107, 2113–2117. ( 10.1073/pnas.0911042107) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Nachman MW, Hoekstra HE, D'Agostino SL. 2003. The genetic basis of adaptive melanism in pocket mice. Proc. Natl Acad. Sci. USA 100, 5268–5273. ( 10.1073/pnas.0431157100) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Linnen CR, Kingsley EP, Jenson JD, Hoekstra HE. 2009. On the origin and spread of an adaptive allele in deer mice. Science 325, 1095–1098. ( 10.1126/science.1175826) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Soria-Carrasco V, et al. 2014. Stick insect genomes reveal natural selection's role in parallel speciation. Science 344, 738–742. ( 10.1126/science.1252136) [DOI] [PubMed] [Google Scholar]
  • 14.Marchinko KB. 2009. Predation's role in repeated phenotypic and genetic divergence of armor in threespine stickleback. Evolution 63, 127–138. ( 10.1111/j.1558-5646.2008.00529.x) [DOI] [PubMed] [Google Scholar]
  • 15.Grant PR. 2001. Reconstructing the evolution of birds on islands: 100 years of research. Oikos 92, 385–403. ( 10.1034/j.1600-0706.2001.920301.x) [DOI] [Google Scholar]
  • 16.Mayr E, Diamond JM. 2001. The birds of Northern Melanesia. New York, NY: Oxford University Press. [Google Scholar]
  • 17.Grant PR. 1965. Plumage and the evolution of birds on islands. Syst. Zool. 14, 47–52. ( 10.2307/2411902) [DOI] [Google Scholar]
  • 18.Keightley PD, Otto SP. 2006. Interference among deleterious mutations favours sex and recombination in finite populations. Nature 443, 89–92. ( 10.1038/nature05049) [DOI] [PubMed] [Google Scholar]
  • 19.Uy JAC, Vargas-Castro LE. 2015. Island size predicts the frequency of melanism in a polymorphic flycatcher. Auk 132, 787–794. ( 10.1642/AUK-14-284.1) [DOI] [Google Scholar]
  • 20.Mayr E. 1942. Systematics and the origin of species. New York, NY: Columbia University Press. [Google Scholar]
  • 21.Uy JAC, Moyle RG, Filardi CE. 2009. Plumage color and song differences mediate species recognition between incipient flycatcher species of the Solomon Islands. Evolution 63, 153–164. ( 10.1111/j.1558-5646.2008.00530.x) [DOI] [PubMed] [Google Scholar]
  • 22.Uy JAC, Moyle RG, Filardi CE, Cheviron ZA. 2009. Difference in plumage color used in species recognition between incipient species is linked to a single amino acid substitution in the melanocortin-1 receptor. Am. Nat. 174, 244–254. ( 10.1086/600084) [DOI] [PubMed] [Google Scholar]
  • 23.Cooper EA, Uy JAC. Submitted. The genomics of speciation-with-gene-flow in an island bird.
  • 24.Andersen MJ, Hosner PA, Filardi CE, Moyle RG. 2015. Phylogeny of the monarch flycatchers reveals extensive paraphyly and novel relationships within a major Australo-Pacific radiation. Mol. Phylogenet. Evol. 83, 118–136. ( 10.1016/j.ympev.2014.11.010) [DOI] [PubMed] [Google Scholar]
  • 25.Hoekstra HE. 2006. Genetics, development and evolution of adaptive pigmentation in vertebrates. Heredity 97, 222–234. ( 10.1038/sj.hdy.6800861) [DOI] [PubMed] [Google Scholar]
  • 26.Hubbard JK, Uy JAC, Hauber ME, Hoekstra HE, Safran RJ. 2010. Vertebrate pigmentation: from underlying genes to adaptive function. Trends Genet. 26, 231–240. ( 10.1016/j.tig.2010.02.002) [DOI] [PubMed] [Google Scholar]
  • 27.Barsh GS. 1996. The genetics of pigmentation: from fancy genes to complex traits. Trends Genet. 12, 299–305. ( 10.1016/0168-9525(96)10031-7) [DOI] [PubMed] [Google Scholar]
  • 28.Mundy NI, Badcock NS, Hart T, Scribner K, Janssen K, Nadeau NJ. 2004. Conserved genetic basis of a quantitative plumage trait involved in mate choice. Science 303, 1870–1873. ( 10.1126/science.1093834) [DOI] [PubMed] [Google Scholar]
  • 29.Kingsley EP, Manceau M, Wiley CD, Hoekstra HE. 2009. Melanism in Peromyscus is caused by independent mutations in agouti. PLoS ONE 4, e6435 ( 10.1371/journal.pone.0006435) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Theron E, Hawkins K, Bermingham E, Ricklefs RE, Mundy NI. 2001. The molecular basis of an avian plumage polymorphism in the wild: a melanocortin-1-receptor point mutation is perfectly associated with the melanic plumage morph of the bananaquit, Coereba flaveola. Curr. Biol. 11, 550–557. ( 10.1016/S0960-9822(01)00158-0) [DOI] [PubMed] [Google Scholar]
  • 31.Rosenblum EB, Hoekstra HE, Nachman MW. 2004. Adaptive reptile color variation and the evolution of the Mc1r gene. Evolution 58, 1794–1808. ( 10.1554/03-741) [DOI] [PubMed] [Google Scholar]
  • 32.Etter PD, Bassham S, Hohenlohe PA, Johnson EA, Cresko WA. 2011. SNP discovery and genotyping for evolutionary genetics using RAD sequencing. Methods Mol. Biol. 772, 157–178. ( 10.1007/978-1-61779-228-1_9) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Warren W, et al. 2010. The genome of a songbird. Nature 464, 757–762. ( 10.1038/nature08819) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Librado P, Rozas J. 2009. DnaSP v. 5: a software for comprehensive analysis of DNA polymorphism data. Bioinformatics 25, 1451–1452. ( 10.1093/bioinformatics/btp187) [DOI] [PubMed] [Google Scholar]
  • 35.Parchman TL, et al. 2013. The genomic consequences of adaptive divergence and reproductive isolation between species of manakins. Mol. Ecol. 22, 3304–3317. ( 10.1111/mec.12201) [DOI] [PubMed] [Google Scholar]
  • 36.Catchen J, Hohenlohe PA, Bassham S, Amores A, Cresko WA. 2013. Stacks: an analysis tool set for population genomics. Mol. Ecol. 22, 3124–3140. ( 10.1111/mec.12354) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, Marth G, Abecasis G, Durbin R. 2009. The sequence alignment/map format and SAMtools. Bioinformatics 25, 2078–2079. ( 10.1093/bioinformatics/btp352) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Rheindt FE, Cuervo AM, Brumfield RT. 2013. Rampant polyphyly indicates cryptic diversity in a clade of Neotropical flycatchers (Aves: Tyrannidae). Biol. J. Linn. Soc. 108, 889–900. ( 10.1111/j.1095-8312.2012.02036.x) [DOI] [Google Scholar]
  • 39.Zhou X, Carbonetto P, Stephens M. 2013. Polygenic modeling with Bayesian sparse linear mixed models. PLoS Genet. 9, e1003264 ( 10.1371/journal.pgen.1003264) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Purcell S, et al. 2007. PLINK: a toolset for whole-genome association and population-based linkage analysis. Am. J. Hum. Genet. 81, 559–575. ( 10.1086/519795) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Lotterhos KE, Whitlock MC. 2014. Evaluation of demographic history and neutral parameterization on the performance of FST outlier tests. Mol. Ecol. 23, 2178–2192. ( 10.1111/mec.12725) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Foll M, Gaggiotti O. 2008. A genome-scan method to identify selected loci appropriate for both dominant and codominant markers: a Bayesian perspective. Genetics 180, 977–993. ( 10.1534/genetics.108.092221) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Geweke J. 1991. Evaluating the accuracy of sampling-based approaches to the calculation of posterior moments. Minneapolis, MN: Federal Reserve Bank of Minneapolis, Research Department Minneapolis. [Google Scholar]
  • 44.Plummer M, Best N, Cowles K, Vines K. 2006. CODA: convergence diagnosis and output analysis for MCMC. R News 6, 7–11. [Google Scholar]
  • 45.Våge DI, Klungland H, Lu D, Cone RD. 1999. Molecular and pharmacological characterization of dominant black coat color in sheep. Mamm. Genome 10, 39–43. ( 10.1007/s003359900939) [DOI] [PubMed] [Google Scholar]
  • 46.Kijas JMH, Wales R, Törnsten A, Chardon P, Moller M, Andersson L. 1998. Melanocortin receptor 1 (MC1R) mutations and coat color in pigs. Genetics 150, 1177–1186. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Lu DS, Vage DI, Cone RD. 1998. A ligand-mimetic model for constitutive activation of the melanocortin-1 receptor. Mol. Endocrinol. 12, 592–604. ( 10.1210/mend.12.4.0091) [DOI] [PubMed] [Google Scholar]
  • 48.Jackson PJ, Douglas NR, Chai B, Binkley J, Sidow A, Barsh GS, Millhauser GL. 2006. Structural and molecular evolutionary analysis of agouti and agouti-related proteins. Chem. Biol. 13, 1297–1305. ( 10.1016/j.chembiol.2006.10.006) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.He L, Gunn TM, Bouley DM, Lu XY, Watson SJ, Schlossman SF, Duke-Cohan JS, Barsh GS. 2001. A biochemical function for attractin in agouti-induced pigmentation and obesity. Nat. Genet. 27, 40–47. ( 10.1038/83741) [DOI] [PubMed] [Google Scholar]
  • 50.Hiragaki T, Inoue-Murayama H, Miwa M, Fujiwara A, Mizutani M, Minvielle F, Ito S. 2008. Recessive black is allelic to the yellow plumage locus in Japanese quail and associated with a frameshift deletion in the ASIP gene. Genetics 178, 771–775. ( 10.1534/genetics.107.077040) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Nadeau NJ, Minvielle F, Ito S, Inoue-Murayama H, Gourichon D, Follet SA, Burke T, Mundy NI. 2008. Characterization of Japanese quail yellow as a genomic deletion upstream of the avian homolog of the mammalian ASIP (agouti) gene. Genetics 178, 777–786. ( 10.1534/genetics.107.077073) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Steiner CC, Weber JN, Hoekstra HE. 2007. Adaptive variation in beach mice produced by two interacting pigmentation genes. PLoS Biol. 5, 1880–1889. ( 10.1371/journal.pbio.0050219) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Cresko WA, Amores A, Wilson C, Murphy J, Currey M, Phillips P, Bell MA, Kimmel CB, Postlethwait JH. 2004. Parallel genetic basis for repeated evolution of armor loss in Alaskan threespine stickleback populations. Proc. Natl Acad. Sci. USA 101, 6050–6055. ( 10.1073/pnas.0308479101) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Kronforst MR, Kapan DD, Gilbert LE. 2006. Parallel genetic architecture of parallel adaptive radiations in mimetic Heliconius butterflies. Genetics 174, 535–539. ( 10.1534/genetics.106.059527) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Uy JAC, Safran RJ. 2013. Variation in the temporal and spatial use of signals and its implications for multimodal communication. Behav. Ecol. Sociobiol. 67, 1499–1511. ( 10.1007/s00265-013-1492-y) [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Electornic Supplementary Materials
rspb20160731supp1.pdf (2.5MB, pdf)

Data Availability Statement

RAD-seq SNP data for GWAS: Dryad ID 877n3 (http://dx.doi.org/10.5061/dryad.877n3). DNA sequence: GenBank submission numbers: 1842234, 1842238–40, 1842246–47, 1842249, 1842252, 1842255, 1842259, 1842266.

Articles from Proceedings of the Royal Society B: Biological Sciences are provided here courtesy of The Royal Society

RESOURCES