Biodiversity Genomics of North American Dryobates Woodpeckers

applyparastyle "fig//caption/p[1]" parastyle "FigCapt"

AmericanOrnithology.org

Volume XX, 2019, pp. 1–12 DOI: 10.1093/auk/ukz015 RESEARCH ARTICLE

Biodiversity genomics of North American Dryobates woodpeckers reveals Downloaded from https://academic.oup.com/auk/advance-article-abstract/doi/10.1093/auk/ukz015/5454745 by New York University user on 28 April 2019 HeadA=HeadB=HeadA=HeadB/HeadA little gene flow across the D. nuttallii x D. scalaris contact zone HeadB=HeadC=HeadB=HeadC/HeadB Extract2=HeadB=Extract=HeadB Joseph D. Manthey,1,2,3,*, Stéphane Boissinot,2 and Robert G. Moyle3 Extract2=HeadA=Extract=HeadA 1 Extract3=HeadA=Extract1=HeadA Department of Biological Sciences, Texas Tech University, Lubbock, Texas, USA 2New York University Abu Dhabi, Abu Dhabi, UAE Extract3=HeadB=Extract1=HeadB 3Biodiversity Institute and Department of Ecology and Evolutionary Biology, University of Kansas, Lawrence, Kansas, USA HeadA=HeadB=HeadA=HeadB/HeadA *Corresponding author: [email protected] HeadB=HeadC=HeadB=HeadC/HeadB Submission Date: September 7, 2018; Editorial Acceptance Date: February 1, 2019; Published April 13, 2019 HeadC=HeadD=HeadC=HeadD/HeadC Extract3=HeadA=Extract1=HeadA REV_HeadA=REV_HeadB=REV_HeadA=REV_HeadB/HeadA ABSTRACT Evolutionary biologists have long used behavioral, ecological, and genetic data from contact zones between closely REV_HeadB=REV_HeadC=REV_HeadB=REV_HeadC/HeadB related species to study various phases of the speciation continuum. North America has several concentrations of REV_HeadC=REV_HeadD=REV_HeadC=REV_HeadD/HeadC avian contact zones, where multiple pairs of sister lineages meet, with or without hybridization. In a southern California REV_Extract3=REV_HeadA=REV_Extract1=REV_HeadA contact zone, 2 species of woodpeckers, Nuttall’s Woodpecker (Dryobates nuttallii) and the Ladder-backed Woodpecker EDI_HeadA=EDI_HeadB=EDI_HeadA=EDI_HeadB/HeadA (D. scalaris), occasionally hybridize. We sampled these 2 species in a transect across this contact zone and included EDI_HeadB=EDI_HeadC=EDI_HeadB=EDI_HeadC/HeadB samples of their closest relative, the Downy Woodpecker (D. pubescens), to obtain large single nucleotide polymorphism EDI_HeadC=EDI_HeadD=EDI_HeadC=EDI_HeadD/HeadC panels using restriction-site associated DNA sequencing (RAD-seq). Furthermore, we used whole-genome resequencing EDI_Extract3=EDI_HeadA=EDI_Extract1=EDI_HeadA data for 2 individuals per species to identify whether patterns of diversity inferred from RAD-seq were representative of whole-genome diversity. We found that these 3 woodpecker species are genomically distinct. Although low levels CORI_HeadA=CORI_HeadB=CORI_HeadA=CORI_HeadB/HeadA of gene flow occur between D. nuttallii and D. scalaris across the contact zone, there was no evidence for widespread CORI_HeadB=CORI_HeadC=CORI_HeadB=CORI_HeadC/HeadB genomic introgression between these 2 species. Overall patterns of genomic diversity from the RAD-seq and whole- CORI_HeadC=CORI_HeadD=CORI_HeadC=CORI_HeadD/HeadC genome datasets appear to be related to distributional range size and, by extension, are likely related to effective CORI_Extract3=CORI_HeadA=CORI_Extract1=CORI_HeadA population sizes for each species. ERR_HeadA=ERR_HeadB=ERR_HeadA=ERR_HeadB/HeadA Keywords: contact zone, Dryobates, genomics, hybrid zone, Picoides, RAD-seq, woodpeckers ERR_HeadB=ERR_HeadC=ERR_HeadB=ERR_HeadC/HeadB ERR_HeadC=ERR_HeadD=ERR_HeadC=ERR_HeadD/HeadC La biodiversidad genómica de los carpinteros Dryobates de América del Norte revela poco flujo génico a Keywords=HeadA=Keywords=HeadA_First través de la zona de contacto de D. nuttallii x D. scalaris CORI_Text_First=CORI_Text=CORI_Text_First=CORI_TextInd RESUMEN Box_Head=Box_AHead=Box_Head=Box_AHead/Head Los biólogos evolutivos han usado durante mucho tiempo datos comportamentales, ecológicos y genéticos de las zonas de contacto entre especies cercanamente emparentadas para estudiar varias fases del proceso continuo de especiación. América del Norte tiene varias concentraciones de zonas de contacto de aves, dónde múltiples pares de linajes hermanos se encuentran, con o sin hibridación. En una zona de contacto del sur de California, dos especies de carpinteros, Dryobates nuttallii y D. scalaris, hibridan ocasionalmente. Muestreamos estas dos especies en un transecto a través de esta zona de contacto e incluimos muestras de su pariente más cercano, D. pubescens, para obtener grandes paneles de polimorfismo de un solo nucleótido (SNP) usando secuenciación de sitios de restricción asociados al ADN (RAD-seq por sus siglas en inglés). Además, usamos datos de re-secuenciación de todo el genoma para dos individuos por especie, para identificar si los patrones de diversidad inferidos a partir de RAD-seq fueron representativos de toda la diversidad genómica. Encontramos que estas tres especies de carpinteros son genómicamente distintas. Aunque existen bajos niveles de flujo génico entre D. nuttallii y D. scalaris a través de la zona de contacto, no hubo evidencia de una amplia introgresión genómica entre estas dos especies. Los patrones generales de diversidad genética a partir de RAD-seq y los datos del genoma completo parecen estar relacionadas con el tamaño del rango de distribución, y por extensión, están probablemente relacionadas con los tamaños poblacionales efectivos para cada especie. Palabras clave: carpinteros, Dryobates, genómica, Picoides, RAD-seq, zona de contacto, zona híbrida

INTRODUCTION ecosystems in continental North America, as well as tran- sitions between ecosystems, has resulted in a plethora of Contact zones, where species may or may not hybridize, are closely related species complexes with partially overlap- ideal locations to study the effects of barriers to gene flow ping distributions (i.e. contact zones; Remington 1968). and introgression on the process of speciation between With geographic contact between closely related species closely related species (Hewitt 1988). The multitude of comes the opportunity for gene flow and introgression

across species boundaries. In North American birds, these Nuttall’s Woodpecker (D. nuttallii) (Figure 1A, B). D. sca- contact zones are concentrated in the Great Plains, the laris and D. nuttallii are known to occasionally hybridize transition between the Rocky Mountains and Cascade in southern California, USA, and Northern Baja California, Mountains, and in the sky islands of southwestern USA Mexico (Short 1971) (Figure 1), although they do not have and northwestern Mexico (Remington 1968, Rising 1983, a broad or well-documented hybrid zone. Downloaded from https://academic.oup.com/auk/advance-article-abstract/doi/10.1093/auk/ukz015/5454745 by New York University user on 28 April 2019 Swenson and Howard 2005). D. nuttallii and D. scalaris are distinguishable by In some North American regions, hybridization is fre- slight differences in morphology and plumage, including quent between closely related but morphologically di- bill length and patterning of black and white feathers on vergent species, resulting in gene flow across species head, back, and rectrices (Short 1971). Additionally, hab- boundaries, as documented in buntings (Passerina), and itat preferences and vocalizations differ between these grosbeaks (Pheucticus) (Carling and Brumfield 2008, sister species; D. nuttallii and D. scalaris differ in their Mettler and Spellman 2009). In contrast, hybridization calls and drumming (Winkler and Short 1978, Stark et al. in other North American birds may be infrequent—indi- 1998), and D. nuttallii inhabits oak and riparian forests in cated by the detection of little gene flow or introgression California (Figure 1A), whereas D. scalaris inhabits dry between lineages— as is seen in treecreepers (Certhia) and stunted forest, scrub, and desert across southwestern USA wood-pewees (Contopus), despite general morphological and northwestern Mexico (Figure 1A). The distinct habitat similarity between sister taxa (Manthey and Robbins 2016, preferences and song divergence of D. nuttallii and D. sca- Manthey et al. 2016). In cases of limited gene flow between laris likely limit their opportunities for hybridization and closely related phenotypically similar taxa, behavioral rec- subsequent introgression between the species; they come ognition (e.g., song) and ecological preferences or adapta- into contact in areas where the lower elevation oak forest tions are potential pre-mating isolation mechanisms rather or riparian forest meets drier habitat, such as Joshua tree than visual recognition (Slabbekoorn and Smith 2002). In (Yucca brevifolia) forest abutting riverine areas. Because of these systems, hybridization is physically possible but in- these distinct habitat differences, the contact zone between frequent, and leads to the question: Does infrequent hy- these 2 Dryobates woodpeckers occurs in a geographic re- bridization influence genomic distinctiveness of sister gion lacking other North American avian contact zones species? (Remington 1968, Swenson and Howard 2005). Several North American woodpeckers (Picidae, genera: Regardless of their differences, D. nuttallii x D. scalaris Colaptes, Dryobates, Melanerpes, and Sphyrapicus) have hybrids have been identified in multiple contexts. Short contact zones with varying levels of hybridization: from a (1971) studied the morphology of hundreds of D. nuttallii broad contact zone with frequent hybridization in Colaptes and D. scalaris specimens, identifying a handful of hybrids. auratus color morphs to nearly none in Dryobates sister Images of these hybrid specimens and typical pure speci- species, an ideal system to assess the effects of rare hy- mens and their data are presented in figure 10 and table 21 bridization (Short 1971, Grudzien et al. 1987, Smith 1987, in Short (1971), respectively. Furthermore, birders occa- McCarthy 2006, Seneviratne et al. 2016). The woodpecker sionally post documentation of intermediate-plumaged genus Dryobates contains 5 species, of which 3 are found individuals on eBird (Sullivan et al. 2009); eBird sightings only in North America: Downy Woodpecker (D. pube- of potential D. nuttallii x D. scalaris individuals with pho- scens), Ladder-backed Woodpecker (D. scalaris), and tographs can be searched for directly on the ebird.org site

FIGURE 1. Distribution and sampling maps for North American Dryobates woodpeckers. (A) Distribution of D. nuttallii and D. scalaris, and the contact zone of the 2 species. (B) Distribution of D. pubescens (note that the distribution extends into Alaska). (C) Sampling localities for specimens used in this study, with the colors indicating species matching those of the distribution maps. In panel (C), stars indicate collecting localities of specimens that Short (1971) identified as hybrids (n = 8).

by exploring species maps of the putative cross of these any individuals that were a priori identified as hybrids. 2 species and filtering for observations labeled with “rich For all new samples, we extracted total genomic DNA media.” using DNeasy blood and tissue kits (QIAGEN, Hilden, Although Short (1971) examined the morphology of Germany) following the manufacturer’s protocol. We purported D. nuttallii x D. scalaris hybrids, the extent of quantified DNA in each extract using a Qubit Fluorometer Downloaded from https://academic.oup.com/auk/advance-article-abstract/doi/10.1093/auk/ukz015/5454745 by New York University user on 28 April 2019 hybridization and potential introgression between these (Life Technologies, Carlsbad, California, USA) high sensi- species has not been studied. This provides the opportu- tivity assay and then standardized the concentration of all nity to investigate whether rare hybridization influences extracts. the distinctiveness of sister species, across both genomic and geographic landscapes. Using mitochondrial DNA Mitochondrial DNA Sequencing and Analyses and a reduced representation approach to sequence subge- We amplified a portion of the mtDNA gene NADH de- nomic-scale data for many individuals, and whole-genome hydrogenase subunit 2 (ND2) for each sample using pol- scale data for 2 individuals per species, we investigated the ymerase chain reaction (PCR) and primers L5215 and following question and hypotheses: H6313 (Hackett 1996, Johnson and Sorenson 1998). We In cases of rare hybridization between sister taxa, what used 10 μL PCRs with the following thermal cycling regime: are the effects of gene flow on the genomic distinctiveness initial denaturation at 95°C for 10 min, 35 cycles of 95°C for of the 2 species? 30 s, 54°C for 45 s, and 72°C for 1 min, and a final extension at 72°C for 2 min. We used a mixture of exonuclease H0: Rare hybrids are evolutionary dead-ends and result and shrimp alkaline phosphatase to purify PCR products, in no significant gene flow between sister species. and then cycle-sequenced the products using 7 μL ABI HA1: Rare hybridization results in gene flow between BigDye 3.1 (Applied Biosystems, Foster City, California, sister species, but that gene flow is restricted to indi- USA) reactions, purified the final reactions with an eth- viduals in the contact zone and does not permeate anol precipitation, and sequenced the forward strand on across the 2 species’ geographic distributions. an ABI 3130 Genetic Analyzer. We used Geneious 11.1.2 HA2: Rare hybridization results in widespread gene flow (BioMatters, Aukland, New Zealand) to align, check, and between sister species, but certain genomic regions curate the mtDNA ND2 sequences. With this alignment, show no evidence of gene flow due to selective we used PopArt (Leigh and Bryant 2015) to create a min- pressures. imum spanning network of the ND2 sequences.

METHODS Restriction-site Associated DNA Sequencing and Bioinformatics Sampling We used a modified RAD-seq protocol (Miller et al. We analyzed a total of 44 Dryobates tissue samples for this 2007) with dual barcodes (Andolfatto et al. 2011) iden- study, including newly collected material and specimen tical to the protocol of Manthey and Moyle (2015) to se- loans from museums (Table 1, Supplementary Material quence reduced-representation genomic libraries for all Table S1). We also obtained mitogenome data from NCBI’s individuals and obtain single nucleotide polymorphisms Genbank for 4 individuals and whole-genome data for 6 (SNPs) for population genomic inferences. Briefly, we individuals from a previously published study (Manthey digested all samples with the restriction enzyme NdeI, et al. 2018) (Table 1, Supplementary Material Table S1). We and then multiplexed all individuals with unique barcodes sampled the D. nuttallii and D. scalaris individuals along (Andolfatto et al. 2011). We then further reduced librar- a transect across their region of sympatry (Figure 1C) to ies by size-selecting fragments between 500 and 600 base assess whether individuals in sympatry would contain pairs (bp) with a Pippin Prep electrophoresis cassette (Sage more admixture than those in allopatry. We did not sample Science, Beverly, Massachusetts, USA). We dual-indexed

TABLE 1. Sampling of Dryobates for this study. Detailed sampling information is available in Supplementary Material Table S1 and Figure 1C. Species mtDNAa RAD-seqb Allopatricc Sympatricc Genomesd D. nuttallii 23 23 17 6 2 D. pubescens 8 3 — — 2 D. scalaris 17 16 13 3 2 a Samples with mitochondrial data. b Samples with restriction-site associated DNA sequencing (RAD-seq) data. c Number of allopatric and sympatric D. nuttallii and D. scalaris individuals with RAD-seq data (also see Figure 1c). d Number of samples with whole-genome data.

samples with a limited-cycle PCR, cleaned samples with D. nuttallii and D. scalaris, we created 2 additional SNP ma- a magnetic bead clean-up, and quantified the pool using trices that required SNP sites to be present and genotyped in a Qubit Fluorometer. The pooled library was tested for 50% or 75% of both species (Table 2). quality and quantity using quantitative PCR and an Agilent We used the program STRUCTURE (Pritchard et al. Tapestation, and then multiplexed and sequenced with 2000, Falush et al. 2003) to investigate differentiation and Downloaded from https://academic.oup.com/auk/advance-article-abstract/doi/10.1093/auk/ukz015/5454745 by New York University user on 28 April 2019 other libraries using several lanes of an Illumina HiSeq2500 potential admixture in the dataset without a priori input. with 100 bp single-end reads at the University of Kansas For the input SNP matrix, we filtered our dataset to in- Genome Sequencing Core Facility. clude SNPs only >5,000 bp apart from each other to re- We used the STACKS 1.48 (Catchen et al. 2011) Python duce effects of linkage. We initially ran STRUCTURE with script process_RADtags to filter our sequences and demul- one assumed population (k = 1) to estimate a value for the tiplex. During filtering, we removed sequences with pos- parameter λ. We subsequently used a fixed value of λ, the sible adapter contamination, lacking the restriction cut site, admixture model, and correlated allele frequencies to run or containing any windows of ≥15 bp with an average Phred STRUCTURE. We ran the program for 100,000 burn-in score ≤10. We trimmed the restriction sites using the FASTX generations followed by an additional 100,000 MCMC toolkit (Gordon and Hannon 2010). Next, for each indi- generations for several assumed numbers of populations vidual, we used the Burrows-Wheeler transform algorithm (k = 1–4), with 5 replicates for each value of k. We used the implemented by the function BWA-MEM in the program log probability of the model and the ΔK statistic of Evanno BWA (Li and Durbin 2009) to align all filtered sequences et al. (2005) to choose the appropriate number of genetic to the published Downy Woodpecker (D. pubescens) ge- clusters. As an additional view of genetic structure, we nome (April 28, 2014 version, dx.doi.org/10.5524/101012). created a phylogeny using maximum likelihood with the We then used Samtools 1.7 (Li et al. 2009) to convert the program RAxML 8 (Stamatakis 2014). With RAxML, we SAM files produced by BWA into BAM files. Next, we used used the GTRGAMMA model of sequence evolution and the pstacks, cstacks, and sstacks modules of the STACKS 100 quick bootstrap replicates to assess support for the pipeline to identify polymorphisms, to create a catalogue tree with the highest likelihood. Heterozygous sites were of all loci in the dataset, and to map all individual informa- coded with IUPAC ambiguity codes. tion to the complete catalogue of loci, respectively. We then We used the ∂a∂i software (Gutenkunst et al. 2009) to ran the rxstacks module of STACKS to correct genotype explore multiple divergence scenarios between D. nuttal- and haplotype calls by using information from the whole lii and D. scalaris. ∂a∂i uses a log-likelihood-based multi- population sample, and then reran the cstacks and sstacks nomial approach (Gutenkunst et al. 2009) with input site modules of STACKS using these corrected calls. frequency spectra to test different demographic scenar- We required SNPs to have sequencing coverage of at least ios. For these analyses, we used the SNP dataset where 5 reads in at least half of the individuals and for the loci to we required SNPs to be present and genotyped in at least have coverage in at least 1 individual of each of the 3 spe- 50% of individuals in both D. nuttallii and D. scalaris to cies. With the final SNP dataset, we estimated patterns of be included (Table 2). From this SNP matrix, we derived diversity using R (R Development Core Team 2014). When site frequency spectra for all bi-allelic SNPs using de- calculating observed heterozygosity, we did not include indi- rived and ancestral states (unfolded terminology) based viduals with >75% missing data to prevent potential biases on calls relative to the D. pubescens reference genome. in diversity distributions. Furthermore, to investigate fixed, Using ∂a∂i, we performed 3 replicate analyses for each of shared, and private polymorphisms within and between several D. nuttallii and D. scalaris divergence scenarios (Figure 2): (1) no divergence between species, (2) divergence and no subsequent gene flow, (3) divergence with TABLE 2. RAD-seq SNP datasets. equal bidirectional gene flow after the split, and (4) di- Dataset No. of Loci No. of SNPs % missing vergence with asymmetric migration after the split. We NPS 1PP a 5,780 14,764 29.25 also used ∂a∂i to fit 3 demographic history scenarios of NS 50CM b,c 4,506 8,784 28.24 D. nuttallii and D. scalaris (Figure 2): (1) constant popu- b NS 75CM 1,193 2,202 16.28 lation size through time; (2) two epoch model, with an in- a Dataset with all 3 species (D. nuttallii, D. pubescens, and D. scalaris). stantaneous population size change (positive or negative) Restricted to SNPs present and genotyped in ≥50% of individuals at some time in the past; and (3) expansion, with a popu- and at least 1 individual per species. Used for analyses of genetic lation size expansion starting at a given time in the past. structure (Figure 3) and observed heterozygosity (Figure 4). b Datasets for inference of polymorphisms (Figure 5a) between We obtained the output variant call format (VCF) D. nuttallii and D. scalaris and inclusive of at least 65% or 80% of files from low coverage genomic sequencing of 2 indi- each species’ sampled individuals. viduals of each of the 3 species of Dryobates from a c Dataset for divergence scenario inferences between D. nuttallii previous study (Manthey et al. 2018). The VCF files and D. scalaris using ∂a∂i. included SNP polymorphism data, small insertions

The Auk: Ornithological Advances XX:1–12, © 2019 American Ornithological Society J. D. Manthey, S. Boissinot, and R. G. Moyle Dryobates biodiversity genomics 5 Downloaded from https://academic.oup.com/auk/advance-article-abstract/doi/10.1093/auk/ukz015/5454745 by New York University user on 28 April 2019

FIGURE 2. Demographic and divergence scenarios tested using ∂a∂i. In demographic scenarios, NA and NC represent ancestral and contemporary effect population sizes, respectively, and t indicates the time of population size changes. In divergence scenarios, NA, NN, and NS represent effective population sizes of ancestral, D. nuttallii, and D. scalaris populations, respectively. tdiv is the estimated divergence time between species, and 2Nm is the effective migration rate between lineages since lineage divergence. and deletions (indels) that were genotyped using the RESULTS genome analysis toolkit (GATK) (McKenna et al. 2010), and retrotransposon polymorphisms identi- Mitochondrial DNA Patterns fied using the mobile element locator tool (MELT) We sequenced 500 bp of the mitochondrial ND2 gene (Gardner et al. 2017). The assayed retrotransposons and identified 16 unique haplotypes among the 3 species. (Superfamily: CR1; Family: J3_Pass) are representative D. scalaris had the most haplotype diversity (Figure 3), of a large woodpecker-specific genomic expansion of with only a few haplotypes present for both D. nuttallii and transposable elements (TEs) (Manthey et al. 2018). We D. pubescens. This is not surprising for D. pubescens, given used these VCF files to assess genomic-wide hetero- the small number of individuals sampled from a relatively zygosity for different types of polymorphisms. With limited geographic area. As expected, based on previous both the RAD-seq and whole-genome VCF files, we phylogenetic work (Shakya et al. 2017), D. nuttallii and used VCFtools 0.1.15 (Danecek et al. 2011) to estimate D. scalaris were closely related in the haplotype network genetic differentiation (FST) for each SNP and mean FST (Figure 3), with ~1–2% uncorrected sequence divergence per reference genome scaffold. between individuals of the 2 species. D. pubescens was more

distantly related to the other 2 species of Dryobates, with RAD-seq and whole-genome datasets (Figure 5C). We may >6% uncorrected sequence divergence among mitochon- consider the RAD-seq dataset as a better estimate of ge- drial haplotypes (Figure 3). Overall, the mtDNA haplotypes nomic differentiation due to the small sample sizes (n = 2 are in perfect concordance with the 3 morphospecies. per species) in the whole-genome datasets. Downloaded from https://academic.oup.com/auk/advance-article-abstract/doi/10.1093/auk/ukz015/5454745 by New York University user on 28 April 2019 Genetic diversity, as measured by observed hetero- Genomic Patterns and Processes Based on RAD-seq zygosity in the RAD-seq data, was generally highest in and Whole-Genome Data D. pubescens and lowest in D. nuttallii individuals (Figure We sequenced a total of 69.4 M reads for 42 individu- 6A), consistent with patterns from hundreds of thousands als from the RAD-seq libraries, with a median and SD of SNPs and tens of thousands of indels found across of 1.65 M and 1.1 M reads per individual, respectively whole-genome samples (Figure 6B). By contrast, the diver- (Supplementary Material Table S1). With this coverage, we sity of TE insertions did not correlate with polymorphism recovered a mean of 46,320 RAD-seq loci per individual of SNPs and indels; in D. nuttallii and D. scalaris, there was (Supplementary Material Table S1). We used the NPS 1PP substantially higher heterozygosity for TEs relative to SNPs dataset (a minimum of 1 individual for each of the 3 spe- and indels, whereas the 3 marker classes showed about cies for a SNP to be included), with 14,764 SNPs (Table 2), the same level of diversity for D. pubescens (Figure 6B). to assess phylogenetic relationships among samples, and Using 2 datasets specific to D. nuttallii and D. scalaris (NS to estimate genetic structure within and between groups 50CM and NS 75CM; Table 2), a large proportion of poly- (Figure 4). We identified strong structuring of all individu- morphisms between the 2 taxa were private (species-spe- als into 3 clades, representative of the 3 a priori species cific) or fixed between them, with ~6% of polymorphisms identifications (Figure 4). STRUCTURE analyses recov- shared between species (Figure 7A,B). Furthermore, we ered a best number of populations equal to the number of wanted to identify whether patterns of fixed differences species investigated (k = 3), with strong genetic structure between D. scalaris and D. nuttallii were similar in the and little evidence for admixture across all individuals, ir- contact zone relative to regions of allopatry (Figure 7C). respective of geographic proximity with the contact zone To do this, we looked for fixed differences between our (Figure 4). eastern-most (D. scalaris) and western-most (D. nuttallii) Genetic differentiation (F ) showed no clear trends ST populations, and assessed whether these polymorphisms across the D. pubescens genomic scaffolds (Figure 5A), remained fixed differences between species in the contact with no long stretches of low or high F estimates. In ST zone. Here, we used the populations sampled from New general, genetic differentiation was moderate between Mexico (D. scalaris) and Los Padres National Forest near D. nuttallii and D. scalaris across most scaffolds (mean F ST San Luis Obispo, California (D. nuttallii) (Supplementary ~0.4; Figure 5B), and was generally consistent between the Material Table S1) as hypothesized “pure” parentals. We required putatively fixed SNPs to be found with at least 6 copies for each parental population, as well as at least 4 copies for each species in sympatry (contact zone) and allopatry, resulting in 227 assayed SNPs. We found >90% of SNPs fixed between parental populations were also fixed between species in the contact zone (Figure 7C). Using ∂a∂i to infer divergence scenarios between D. nuttallii and D. scalaris, we found that a model including a split and then isolation with low migration was the best fit to the RAD-seq data (Table 3). Regardless of which divergence scenario was utilized, little gene flow has occurred per generation (2Nm <0.1) since the split of the 2 lineages. We used a substitution rate estimate of 2.42e-9 mutations per site per year, estimated from non-coding regions of the D. pubescens genome (Jarvis et al. 2014, Zhang et al. 2014) and a conservative one year generation-time estimate of Dryobates species to associate times with the divergence, which ranged between ~0.37–1.2 mya, depending on FIGURE 3. Median-joining mitochondrial haplotype network based on ND2 gene sequences from Dryobates woodpeckers. whether gene flow was allowed in the model (Table 3). The Dashes between circles indicate mutational steps connecting generation time estimate we used is conservative because haplotypes, and the size of the circles indicates sample size of it is the age of sexual maturity for the Downy Woodpecker each haplotype. reported in The Animal Aging and Longevity Database

The Auk: Ornithological Advances XX:1–12, © 2019 American Ornithological Society J. D. Manthey, S. Boissinot, and R. G. Moyle Dryobates biodiversity genomics 7 Downloaded from https://academic.oup.com/auk/advance-article-abstract/doi/10.1093/auk/ukz015/5454745 by New York University user on 28 April 2019

FIGURE 4. Phylogeny and genetic structure analysis based on analysis of RAD-seq data. The phylogeny at left includes all individuals in this study and was estimated using RAxML. Asterisks indicate 100% bootstrap support based on bootstrap analysis. STRUCTURE plots at the tip of the phylogeny indicate the inferred probability of ancestry for each individual sample to 1 of 3 genetic clusters. Individuals with white dots next to the STRUCTURE plots are individuals sampled from the contact zone.

(Tacutu et al. 2018), and average generation time will likely flow between species (Table 3), and polymorphism pat- be longer. Doubling the generation time would effectively terns are generally similar between allopatric and sym- double divergence timing, as divergence estimates are patric parts of these 2 species’ distributions (Figure 7). scaled with generation time and mutation rate. All of these patterns suggest that although D. nuttallii Contemporary effective population sizes in both D. nut- and D. scalaris can hybridize, little of the genomic con- tallii and D. scalaris have likely increased from lower pop- tent moves between species. In North American birds, it ulation sizes during the Pleistocene (Table 4). Although is clear that different processes shape genomic diversity the current effective population sizes of both species are and differentiation between sister species or lineages, approximately of the same magnitude, D. nuttallii has ex- and this manifests in many cases where genetic differen- panded from a much smaller ancestral population size tiation does not match phenotypic differentiation. The than D. scalaris (Table 4), which is consistent with smaller spectrum of patterns ranges from morphologically sim- refugial population sizes in D. nuttallii during periods of ilar lineages with strong genetic differentiation and little the Pleistocene. to no hybridization, such as treecreepers (Manthey et al. 2016) and scrub jays (Gowen et al. 2014), to morpholog- DISCUSSION ically variable taxa that differ in little to none of their genomes, such as in redpolls (Mason and Taylor 2015) Distinctiveness of North American Dryobates Species and warblers (Toews et al. 2016). With such a wide range With reduced-representation genomic sequencing, we of morphological and genetic variation, a multitude of showed that the 3 species of Dryobates in North America processes clearly has shaped diversification histories in are genetically distinct (Figures 3, 4, and 5). This is con- North American birds. Here, we show that Dryobates sistent with the 3 species’ distinct morphological and woodpeckers, specifically D. nuttallii and D. scalaris, fall ecological characteristics (Short 1971). Two of these in the middle of the spectrum, where moderate genomic species, D. nuttallii and D. scalaris, infrequently hy- differentiation is associated with moderate phenotypic bridize (Short 1971); however, this results in little gene differentiation.

The Auk: Ornithological Advances XX:1–12, © 2019 American Ornithological Society 8 Dryobates biodiversity genomics J. D. Manthey, S. Boissinot, and R. G. Moyle Downloaded from https://academic.oup.com/auk/advance-article-abstract/doi/10.1093/auk/ukz015/5454745 by New York University user on 28 April 2019

FIGURE 5. Genetic differentiation between D. nuttallii and D. scalaris. (A) FST estimate for each SNP sampled in the RAD-seq dataset in position along the D. pubescens genome assembly (scaffolds). (B) Mean FST per scaffold for the RAD-seq and whole-genome datasets, including only scaffolds with a minimum of 20 SNPs. Blue lines indicate mean of entire sample. (C) Relationship of FST estimated in RAD- seq or whole-genome datasets for each scaffold that has a minimum of 20 SNPs in both datasets.

TABLE 3. Divergence scenarios between D. nuttallii and D. scalaris as estimated with ∂a∂i. Three replicates were performed for each scenario, with their mean values reported here. Models and parameter estimates are described in Figure 2 and the text.

Model Likelihood ΔAICc 2Nm 2Nmn->s 2Nms->n tdiv (k yr) No divergence –39,171 78,344 Split, no gene flow –2,472 3,243 370 Split, bidirectional migration –1,742 251.3 0.026 1,122 Split, isolation with migration –1,615 1.3 <0.001 0.033 1,175

Processes shaping Genomic Distinctiveness between Behavioral differences are potential pre-mating isolation Dryobates Sister Lineages mechanisms that would limit gene flow between different Two main processes may contribute to the lack of gene flow species. Woodpeckers likely use call and drumming behavior and genome homogenization between D. nuttallii and D. sca- to recognize conspecifics. Multiple studies of vocalizations laris: (1) behavioral or ecological differences that mostly en- in D. nuttallii and D. scalaris suggest that both their calls sure pre-mating isolation except in rare circumstances, and and drumming are significantly different (Winkler and Short (2) reduced hybrid fitness (i.e. postzygotic isolation). 1978, Stark et al. 1998), behaviors which may limit contact

TABLE 4. Demographic scenarios for D. nuttallii and D. scalaris as estimated with ∂a∂i. Three replicates were performed for each scenario, with their mean values reported here. Models and parameter estimates are described in Figure 2 and the text.

Species Model Likelihood ΔAICc NC / NA NC (k individuals) T (k yr) D. nuttallii Constant –234.6 297.3 111 Two epoch –85.7 3.4 0.19 384 36.8 Downloaded from https://academic.oup.com/auk/advance-article-abstract/doi/10.1093/auk/ukz015/5454745 by New York University user on 28 April 2019 Expansion –84.0 0 0.15 471 65.5 D. scalaris Constant –130.4 16.6 226 Two epoch –120.1 0 0.70 294 77.6 Expansion –120.1 0 0.63 328 118.8

FIGURE 6. Genomic diversity of Dryobates species inferred from (A) RAD-seq and (B) whole-genome resequencing data. between these different species and consequently limit in- between D. nuttallii and D. scalaris have reduced fitness, terspecific gene flow. Furthermore, different ecological pref- hybridization would be less likely to cause meaningful erences or adaptations in sister species would promote the amounts of genomic introgression between species. Of maintenance of species boundaries even in the face of little course, a combination of both prezygotic and postzy- genome-wide differentiation. In North America, differences gotic barriers could exist between these 2 species, and in habitat preferences among Contopus pewees (Manthey this would manifest in a sort of tension zone, as has been and Robbins 2016) and differences in ecological adaptations shown in grosbeaks (Mettler and Spellman 2009). In these among Ammodramus sparrows (Walsh et al. 2016) result in circumstances, hybridization would be rare except in situations where hybridization is either rare or occurs only in the hybrid zone, and hybrid individuals would be less fit a narrow contact zone between species, despite limited ge- than either parental form. The tension zone scenario is nomic differentiation. Similarly, D. nuttallii and D. scalaris unlikely for D. nuttallii and D. scalaris because hybridiza- have recently diversified and are capable of hybridizing but tion between the 2 species appears to be rare, and there are likely kept isolated by different ecological preferences. are no known regions where there are many hybrid indi- Another potential reason for lack of gene flow between viduals. Nonetheless, this investigation into the contact closely related sympatric species could be reduced hybrid zone between sister woodpecker species provides another fitness, as has been hypothesized in chickadees (McQuillan example of processes shaping diversification at the time et al. 2018) and different lineages of the Yellow-rumped point in speciation where hybridization is possible, but Warbler (Dendroica coronata) complex (Brelsford and rare, and leads to little gene flow, thereby preserving the Irwin 2009). If the offspring of the heterospecific pairings genomic identity of these 2 species.

The Auk: Ornithological Advances XX:1–12, © 2019 American Ornithological Society 10 Dryobates biodiversity genomics J. D. Manthey, S. Boissinot, and R. G. Moyle Downloaded from https://academic.oup.com/auk/advance-article-abstract/doi/10.1093/auk/ukz015/5454745 by New York University user on 28 April 2019

FIGURE 7. SNP polymorphism patterns between D. nuttallii and D. scalaris. (A–B) Proportion of polymorphisms shared or fixed between species, and private within species. (C) Geographic patterns of SNPs that are fixed between the assumed “genetically pure parental” populations (stars). Pie charts indicate proportion of fixed differences from pure populations found in allopatric and sympatric distributional areas for the 2 species.

Genomic, morphological, ecological, and behavioral dif- diversity and population size for each of the species; the ferences between D. nuttallii and D. scalaris are consistent species with the largest range has the highest genomic with defining these 2 taxa as species under the biological, diversity (D. pubescens) and the species with the smallest evolutionary, and phylogenetic species concepts. This pat- range has the lowest diversity (D. nuttallii) (Figure 1). Of tern of broadly distinctive yet closely related species in course, effective population sizes and geographic range contact contrasts with other hybridizing North American sizes are dynamic through time (e.g., varying through woodpeckers (Picidae) such as Colaptes and Sphyrapicus, Pleistocene climatic cycles), and this pattern may simply which hybridize and have been studied with genomic-scale be a signature of the current climatic epoch. For example, data. The red-shafted and yellow-shafted morphological it is likely that D. pubescens had a much smaller range dur- groups of the Northern Flicker (Colaptes auratus) complex ing the Last Glacial Maximum and has since undergone have a wide hybrid zone, frequent hybridization, and little a large demographic expansion (Pulgarín-R and Burg genetic differentiation between taxa (Grudzien et al. 1987, 2012). Similarly, although both D. nuttallii and D. sca- Aguillon et al. 2018). Similarly, 3 species of Sphyrapicus laris have a similar magnitude of contemporary effective sapsuckers hybridize relatively frequently in western population sizes as estimated by modeling of site fre- North American contact zones, with small to moderate quency spectra (Table 4), D. nuttallii has expanded from genetic distinctiveness (Seneviratne et al. 2016). In both a much smaller population size during the Pleistocene, the Colaptes and Sphyrapicus woodpecker cases, frequent whereas populations of D. scalaris have either remained observation of intermediate-plumage individuals has pre- relatively constant or expanded only minimally (Table 4). cluded their definition as species under a strict biological This may suggest that current genomic diversity in North species concept. However, generally assortative mating in American Dryobates woodpeckers could be a combina- both the Colaptes and Sphyrapicus hybrid zones has pre- tion of contemporary range size and population size in vented the collapse of species limits and these taxa remain Pleistocene refugia during glacial maxima. Overall, these distinct species under the evolutionary species concept patterns and processes suggest that largely neutral pat- (Grudzien et al. 1987, Seneviratne et al. 2016, Aguillon terns created by the dynamics between population size— et al. 2018). both current and changes through time—and genetic drift have shaped the overall patterns of genomic diver- Processes shaping Genomic Diversity in North sity in these 3 species. American Dryobates In contrast to the genomic diversity patterns observed We identified different patterns of genomic diver- in putatively neutral SNPs and indels, we found that TE sity in the 3 Dryobates species investigated here using polymorphisms (of the family CR1 J3_Pass) show a dif- both reduced representation genomic data and whole- ferent pattern that is inversely correlated with neutral genome resequencing data (Figure 6). Here, we found markers (albeit with a small sample size; Figure 6). This that D. pubescens has the highest genomic diversity, and suggests that TE polymorphism frequency is influenced D. nuttallii has the lowest diversity (Figure 6), a pattern by non-neutral processes. This pattern is consistent consistent across both SNPs and indels. This pattern with selection acting against TE activity, and subse- is likely due to a simple relationship between genomic quent maintenance of TE diversity, in these woodpecker

species (Manthey et al. 2018). Since selection is more sequence data are deposited in NCBI’s sequence read archive effective in eliminating deleterious alleles in large popu- (SRA) under BioProject PRJNA515065. lations (such as in D. pubescens), the higher TE diversity in D. nuttallii and D. scalaris could be explained by the LITERATURE CITED effect of genetic drift counter-acting purifying selection Downloaded from https://academic.oup.com/auk/advance-article-abstract/doi/10.1093/auk/ukz015/5454745 by New York University user on 28 April 2019 against deleterious TEs in these species. As such, a com- Aguillon, S. M., L. Campagna, R. G. Harrison, and I. J. Lovette (2018). A flicker of hope: Genomic data distinguish Northern Flicker bination of the influences of demography and selection taxa despite low levels of divergence. The Auk: Ornithological likely influences the genomic diversity of TE polymor- Advances 135:748–766. phisms in these woodpeckers. Future work investigat- Andolfatto, P., D. Davison, D. Erezyilmaz, T. T. Hu, J. Mast, T. ing genomic variation across chromosomes and genetic Sunayama-Morita, and D. L. Stern (2011). Multiplexed shotgun variant types will require larger sample sizes of whole- genotyping for rapid and efficient genetic mapping. Genome genome resequencing data and will also be useful for Research 21:610–617. identifying whether any regions of Dryobates species Brelsford, A., and D. E. Irwin (2009). Incipient speciation despite genomes introgress widely across species boundaries. little assortative mating: The Yellow-Rumped Warbler hybrid zone. Evolution 63:3050–3060. Carling, M. D., and R. T. Brumfield (2008). Haldane’s rule in an avian system: Using cline theory and divergence population ACKNOWLEDGMENTS genetics to test for differential introgression of mitochondrial, autosomal, and sex-linked loci across the Passerina bunting We thank the following individuals for help with obtain- hybrid zone. Evolution 62:2600–2615. ing tissue loans for this work: Chris Witt, Donna Dittmann, Catchen, J. M., A. Amores, P. Hohenlohe, W. Cresko, and Kevin Burns, Phil Unitt, Kimball Garrett, and Mark J. H. Postlethwait (2011). Stacks: Building and genotyping loci Robbins. We are thankful for the use of vouchered speci- de novo from short-read sequences. G3 1:171–182. mens from the following biodiversity collections: Museum Danecek, P., A. Auton, G. Abecasis, C. A. Albers, E. Banks, M. A. of Southwestern Biology Division of Birds, Louisiana State DePristo, R. E. Handsaker, G. Lunter, G. T. Marth, S. T. Sherry, G. University Museum of Natural Science, San Diego State McVean, and R. Durbin; 1000 Genomes Project Analysis Group University Museum of Biodiversity and San Diego Natural (2011). The variant call format and VCFtools. Bioinformatics History Museum, Natural History Museum of Los Angeles 27:2156–2158. County, and University of Kansas Biodiversity Institute. We Evanno, G., S. Regnaut, and J. Goudet (2005). Detecting the number thank John Kelly and Patrick Monnahan for help during of clusters of individuals using the software STRUCTURE: early stages of learning RAD-seq methods. We thank Mark A simulation study. Molecular Ecology 14:2611–2620. Geiger for assistance with portions of the fieldwork, and the Falush, D., M. Stephens, and J. K. Pritchard (2003). Inference state, provincial and federal agencies and numerous wildlife of population structure using multilocus genotype data: officers for their cooperation and help obtaining permits to Linked loci and correlated allele frequencies. Genetics conduct this research. 164:1567–1587. Funding statement: This work was funded by an American Gardner, E. J., V. K. Lam, D. N. Harris, N. T. Chuang, E. C. Scott, W. S. Museum of Natural History Frank M. Chapman Memorial Pittard, R. E. Mills, and S. E. Devine; 1000 Genomes Project Fund Grant (J.D.M.), an American Ornithologists’ Union Consortium (2017). The mobile element locator tool (MELT): student research grant (J.D.M.), a University of Kansas Population-scale mobile element discovery and biology. Biodiversity Institute Panorama award (J.D.M.), and multiple Genome Research 27:1916–1929. National Science Foundation (NSF) grants (DEB-1406989 Gordon, A., and G. Hannon (2010). Fastx-toolkit: FASTQ/A short-reads to J.D.M. and R.G.M., DEB-1557053 and DEB-1241181 to preprocessing tools. http://hannonlab.cshl.edu/fastx_toolkit. R.G.M.). This research was supported by New York University Gowen, F. C., J. M. Maley, C. Cicero, A. T. Peterson, B. C. Faircloth, T. C. Abu Dhabi Research Funds AD180 (to S.B.). The NYUAD Warr, and J. E. McCormack (2014). Speciation in Western Scrub- Sequencing Core is supported by NYUAD Research Institute Jays, Haldane’s rule, and genetic clines in secondary contact. grant G1205-1205A to the NYUAD Center for Genomics and BMC Evolutionary Biology 14:135. Systems Biology. Grudzien, T. A., W. S. Moore, J. R. Cook, and D. Tagle (1987). Genic Ethics statement: We followed the approved IACUC pro- population structure and gene flow in the Northern Flicker tocol (R.G.M.): 174-01. (Colaptes auratus) hybrid zone. The Auk 4:654–664. Author contributions: J.D.M. and R.G.M. conceived the idea, Gutenkunst, R. N., R. D. Hernandez, S. H. Williamson, and design, and experiment. J.D.M. collected the data and con- C. D. Bustamante (2009). Inferring the joint demographic ducted the research. J.D.M., S.B., and R.G.M. wrote the paper history of multiple populations from multidimensional SNP and/or substantially edited the paper. J.D.M. analyzed the frequency data. Plos Genetics 5:e1000695. data. J.D.M., S.B., and R.G.M. contributed substantial materi- Hackett, S. J. (1996). Molecular phylogenetics and biogeography als, resources, or funding. of tanagers in the genus Ramphocelus (Aves). Molecular Date deposits: All mitochondrial sequence data are acces- Phylogenetics and Evolution 5:368–382. sioned in NCBI’s Genbank and all accession numbers are re- Hewitt, G. M. (1988). Hybrid zones-natural laboratories for ported in Supplementary Material Table S1. All RAD-seq raw evolutionary studies. Trends in Ecology & Evolution 3:158–167.

Jarvis, E. D., S. Mirarab, A. J. Aberer, B. Li, P. Houde, C. Li, S. Y. Ho, B. C. Pulgarín-R, P. C., and T. M. Burg (2012). Genetic signals of Faircloth, B. Nabholz, J. T. Howard, et al. (2014). Whole-genome demographic expansion in Downy Woodpecker (Picoides analyses resolve early branches in the tree of life of modern pubescens) after the last North American glacial maximum. birds. Science 346:1320–1331. PLoS One 7:e40412.

Johnson, K. P., and M. D. Sorenson (1998). Comparing molecular R Development Core Team (2014). R: A language and environment Downloaded from https://academic.oup.com/auk/advance-article-abstract/doi/10.1093/auk/ukz015/5454745 by New York University user on 28 April 2019 evolution in two mitochondrial protein coding genes for statistical computing. R Foundation for Statistical Computing, (cytochrome b and ND2) in the dabbling ducks (Tribe: Anatini). Vienna, Austria. http://www.R-project.org/. Molecular Phylogenetics and Evolution 10:82–94. Remington, C. L. (1968). Suture-zones of hybrid interaction Leigh, J. W., and D. Bryant (2015). Popart: Full-feature software between recently joined biotas. In Evolutionary Biology for haplotype network construction. Methods in Ecology and (T. Dobzhansky, M. K. Hecht, and W. C. Steere, Editors). Springer, Evolution 6:1110–1116. Boston, Massachusetts, USA. pp. 321–428. Li, H., and R. Durbin (2009). Fast and accurate short read Rising, J. (1983). The Great Plains hybrid zones. In Current alignment with Burrows-Wheeler transform. Bioinformatics Ornithology (R. F. Johnston, Editor). Springer, Boston, 25:1754–1760. Massachusetts, USA. pp. 131–157. Li, H., B. Handsaker, A. Wysoker, T. Fennell, J. Ruan, N. Homer, G. Seneviratne, S. S., P. Davidson, K. Martin, and D. E. Irwin (2016). Low Marth, G. Abecasis, and R. Durbin; 1000 Genome Project Data levels of hybridization across two contact zones among three Processing Subgroup (2009). The sequence alignment/map species of woodpeckers (Sphyrapicus sapsuckers). Journal of format and SAMtools. Bioinformatics 25:2078–2079. Avian Biology 47:887–898. Manthey, J. D., and R. G. Moyle (2015). Isolation by environment in Shakya, S. B., J. Fuchs, J. M. Pons, and F. H. Sheldon (2017). Tapping white-breasted nuthatches (Sitta carolinensis) of the madrean the woodpecker tree for evolutionary insight. Molecular archipelago sky islands: A landscape genomics approach. Phylogenetics and Evolution 116:182–191. Molecular Ecology 24:3628–3638. Short, L. L. (1971). Systematics and behavior of some North Manthey, J. D., R. G. Moyle, and S. Boissinot (2018). Multiple and American woodpeckers, genus Picoides (Aves). Bulletin of the independent phases of transposable element amplification in American Museum of Natural History, New York 145:1–118. the genomes of Piciformes (woodpeckers and allies). Genome Slabbekoorn, H., and T. B. Smith (2002). Bird song, ecology and Biology and Evolution 10:1445–1456. speciation. Philosophical Transactions of the Royal Society of Manthey, J. D., and M. B. Robbins (2016). Genomic insights into London. Series B, Biological Sciences 357:493–503. hybridization in a localized region of sympatry between Smith, J. I. (1987). Evidence of hybridization between red-bellied pewee sister species (Contopus sordidulus x C. virens) and their and golden-fronted woodpeckers. The Condor 89:377–386. chromosomal patterns of differentiation. Avian Research 7:6. Stamatakis, A. (2014). Raxml version 8: A tool for phylogenetic Manthey, J. D., M. B. Robbins, and R. G. Moyle (2016). A genomic analysis and post-analysis of large phylogenies. Bioinformatics investigation of the putative contact zone between divergent 30:1312–1313. brown creeper (Certhia americana) lineages: Chromosomal Stark, R. D., D. J. Dodenhoff, and E. V. Johnson (1998). A quantitative patterns of genetic differentiation. Genome 59:115–125. analysis of woodpecker drumming. The Condor 100:350–356. Mason, N. A., and S. A. Taylor (2015). Differentially expressed Sullivan, B. L., C. L. Wood, M. J. Iliff, R. E. Bonney, D. Fink, and S. Kelling genes match bill morphology and plumage despite largely (2009). eBird: A citizen-based bird observation network in the undifferentiated genomes in a Holarctic songbird. Molecular biological sciences. Biological Conservation 142:2282–2292. Ecology 24:3009–3025. Swenson, N. G., and D. J. Howard (2005). Clustering of contact McCarthy, E. M. (2006). Handbook of Avian Hybrids of the World. zones, hybrid zones, and phylogeographic breaks in North Oxford University Press, New York, New York, USA. America. The American Naturalist 166:581–591. McKenna, A., M. Hanna, E. Banks, A. Sivachenko, K. Cibulskis, A. Tacutu, R., D. Thornton, E. Johnson, A. Budovsky, D. Barardo, T. Kernytsky, K. Garimella, D. Altshuler, S. Gabriel, M. Daly, Craig, E. Diana, G. Lehmann, D. Toren, J. Wang, V. E. Fraifeld, and and M. A. DePristo (2010). The genome analysis toolkit: J. P. de Magalhães (2018). Human ageing genomic resources: new A MapReduce framework for analyzing next-generation DNA and updated databases. Nucleic Acids Research 46:D1083–D1090. sequencing data. Genome Research 20:1297–1303. Toews, D. P., S. A. Taylor, R. Vallender, A. Brelsford, B. G. Butcher, P. W. McQuillan, M. A., T. C. Roth II, A. V. Huynh, and A. M. Rice (2018). Messer, and I. J. Lovette (2016). Plumage genes and little else Hybrid chickadees are deficient in learning and memory. distinguish the genomes of hybridizing warblers. Current Evolution 72:1155–1164. Biology 26:2313–2318. Mettler, R. D., and G. M. Spellman (2009). A hybrid zone revisited: Walsh, J., W. G. Shriver, B. J. Olsen, and A. I. Kovach (2016). Molecular and morphological analysis of the maintenance, Differential introgression and the maintenance of species movement, and evolution of a Great Plains avian (Cardinalidae: boundaries in an advanced generation avian hybrid zone. Pheucticus) hybrid zone. Molecular Ecology 18:3256–3267. BMC Evolutionary Biology 16:65. Miller, M. R., J. P. Dunham, A. Amores, W. A. Cresko, and Winkler, H., and L. L. Short (1978). A comparative analysis of E. A. Johnson (2007). Rapid and cost-effective polymorphism acoustical signals in pied woodpeckers (Aves, Picoides). identification and genotyping using restriction site associated Bulletin of the American Museum of Natural History, New York DNA (RAD) markers. Genome Research 17:240–248. 160:1–109. Pritchard, J. K., M. Stephens, and P. Donnelly (2000). Inference Zhang, G., B. Li, C. Li, M. T. Gilbert, E. D. Jarvis, and J. Wang; Avian of population structure using multilocus genotype data. Genome Consortium (2014). Comparative genomic data of Genetics 155:945–959. the avian phylogenomics project. Gigascience 3:26.