Molecular Phylogenetics and Evolution 113 (2017) 67–75 Contents lists available at ScienceDirect Molecular Phylogenetics and Evolution journal homepage: www.elsevier.com/locate/ympev Cryptic speciation and gene flow in a migratory songbird Species Complex: Insights from the Red-Eyed Vireo (Vireo olivaceus) ⇑ C.J. Battey a, , John Klicka b a University of Washington, Department of Biology, Kincaid Hall, Box 351800, Seattle, WA 98195, USA b Burke Museum of Natural History and Culture, University of Washington, Box 353010, Seattle, WA 98195, USA article info abstract Article history: Migratory species that alternate between sympatry and allopatry over the course of an annual cycle are Received 25 January 2017 promising subjects for studies seeking to understand the process of speciation in the absence of strict Revised 7 April 2017 geographic isolation. Here we sought to identify cryptic species and assess rates of gene flow in a clade Accepted 8 May 2017 of neotropical migrant songbirds in which geography and taxonomy are currently out of sync: the Red- Available online 12 May 2017 Eyed Vireo (V. olivaceus) Species Complex. Phylogenetic, clustering, and statistical species delimitation analyses found that V. olivaceus includes two non-sister lineages migrating in opposite directions across Keywords: the equator. Analyses of gene flow identified low levels of introgression between two species pairs, but Speciation none between northern and southern olivaceus. We also identified substantial well-supported conflicts Hybridization Phylogeography between nuclear and mitochondrial topologies. Although the geographic distribution of mito-nuclear dis- Systematics cordance is suggestive of hybridization and mitochondrial capture, we found no evidence of introgression Birds in the nuclear genome of populations with discordant mitochondrial gene trees. Our study finds that spe- Migration cies boundaries match breeding range and migratory phenology rather than the existing taxonomy in this Genomics group, and demonstrates the utility of genomic data in inferring species boundaries in recently diverged Phylogenetics clades. Ó 2017 Elsevier Inc. All rights reserved. 1. Introduction The extensive spatial mixing associated with seasonal migra- tion suggests that migratory species are promising subjects for Geographic isolation is thought to be the dominant factor in ini- studies seeking to better understand the role of gene flow in speci- tiating speciation in birds (Mayr, 1942; Mayr and O’Hara, 1986; ation. Recent studies in songbirds such as Redpolls (Acanthis; Coyne and Price, 2000; Phillimore et al., 2008), and comparative Mason and Taylor, 2015), Greenish Warblers (Phylloscopus trochi- studies often focus on the role of range gaps and geographic barri- loides; Irwin et al., 2005), and Darwin’s Finches (Geospiza; ers in generating species diversity (Smith et al., 2014; Barber and Lamichhaney et al., 2015) have found evidence of extensive intro- Klicka, 2010; Zink et al., 2001). For migratory species, the role of gression among morphologically divergent populations. In some allopatry in lineage divergence is unusual. Most individuals are cases (e.g. Geospiza, Acanthis) phenotypic plasticity and divergent physiologically capable of moving between disjunct ranges even selection on a small number of genes is thought to drive morpho- when separated by large distances, but populations are isolated logical variation despite a largely panmictic nuclear genome. One by heritable variation in the direction and timing of migration recent whole-genome sequencing study of hybridizing Golden- (Helbig, 1991; Pulido et al., 2001; Delmore and Irwin, 2014). In and Blue-winged warblers (Vermivora chrysoptera & V. cyanoptera, some cases sister species are allopatric during the breeding season respectively), found that just 6 genomic regions appear to control and sym- or parapatric during the nonbreeding season, as occurs in phenotypic variation between species despite extensive introgres- many neotropical migrants breeding in northern temperate sion across the genome (Toews et al., 2016). These studies are part regions and wintering in Central and South America. Rates of gene of an increasing focus on the extent and impacts of hybridization flow across the ranges of these species are dependent on levels of on speciation (and, in corollary, the biological meaning of taxo- migratory connectivity and distance of natal dispersal, rather than nomic designations). simple inertia in natal location. In this study we assess the consequences of divergent migratory behavior on speciation and gene flow in a systematically recalcitrant ⇑ Corresponding author. clade of neotropical songbirds: the Red-Eyed Vireo (V. olivaceus) E-mail address: [email protected] (C.J. Battey). Species Complex. This group includes three widespread migratory http://dx.doi.org/10.1016/j.ympev.2017.05.006 1055-7903/Ó 2017 Elsevier Inc. All rights reserved. 68 C.J. Battey, J. Klicka / Molecular Phylogenetics and Evolution 113 (2017) 67–75 species that are sympatric in northern South America during the putatively orthologous loci are then aligned with Muscle (Edgar, nonbreeding season and para- or allopatric during the breeding sea- 2004) and output in a variety of sequence alignment formats for son. Unusually, the most widespread species in the group – V. oli- downstream analysis. We set a minimum read depth of 6 for call- vaceus – includes populations migrating both north and south of ing consensus sequences within samples and used a clustering the equator, as well as sedentary populations breeding in northern threshold of 0.90 for all Usearch runs. We filtered out all loci shar- South America. Due to similar habits, vocalizations, and morpholo- ing heterozygotic sites across more than 8 individuals to avoid gies, earlier taxonomists typically recognized three (Mayr and Short, including paralogs in our sequencing alignment. 1970) or four (Paynter, 1968; American Ornithologists’ Union, 1983) Because sample coverage – the proportion of sampled individu- species in the group. The first genetic study of the group (Johnson als sequenced for any given locus – is known to have a large effect and Zink, 1985) found the North- and South American olivaceus on the total number of loci recovered (Leaché et al., 2014; Huang monophyletic relative to the Central American flavoviridis, resulting and Knowles, 2014; DaCosta and Sorenson, 2016), we prepared in the current five-species taxonomy (Johnson and Zink, 1985; two sequence assemblies for downstream analyses by varying this American Ornithologists’ Union, 1998). Surprisingly, a recent study parameter in pyRAD. The ‘‘MD10” alignment includes all loci of mitochondrial DNA suggested the presence of at least six geo- sequenced for at least 90% of samples, while the ‘‘MD50” alignment graphically and genetically distinct lineages (Slager et al., 2014), includes loci sequenced for at least 50% of samples. To explore the though basal nodes in the group were poorly supported. source of missing data in our alignments we also used a Mantel Here we sequenced a genome-wide sampling of SNP’s from Test (Mantel, 1967) to test for a correlation between pairwise museum specimens caught across the western hemisphere in genetic distance and the number of shared loci among samples. order to assess the strength and distribution of species boundaries Sequence alignments for concatenated phylogenetic analyses across the group. Specifically, we ask: (1) What is the optimal spe- and structure input files were prepared with PyRAD’s standard cies delimitation scheme for partitioning diversity in the Red-Eyed output options. STRUCTURE and Adegenet analyses used the Vireo Complex? (2) Do migrant species breeding in parapatry show MD50 alignment. SNAPP and RAxML were run on both MD50 evidence of gene flow after lineage divergence? And (3) Does intro- and MD10 alignments, though only MD10 alignments were used gression lead to erosion of neutral genetic differentiation among for Bayes Factor Delimitation (BFD⁄). Both STRUCTURE and SNAPP putative species? input files include one random SNP per locus. For SNAPP, pyRAD’s ‘‘unlinked_snps” output was converted to a three-state numeric input file (two alleles plus heterozygotes) suitable for analysis in 2. Methods SNAPP using custom R scripts written by Dr. Barbara Banbury and available online on the shinyphry webserver (https://rstudio. 2.1. Sampling and library preparation stat.washington.edu/shiny/phrynomics/). We developed additional custom R scripts for visualizing patterns of missing data, writing We obtained frozen tissue samples from vouchered natural his- D-test input files, summarizing D-test output, and plotting STRUC- tory museum specimens representing 40 individuals and 6 species TURE results with consistent color matching (github.com/cjbattey/ of Vireo. These included four members of the Red-eyed vireo com- radplots/). All analyses were conducted in the R programming plex (V. olivaceus, V. flavoviridis, V. altiloquus, V. magister) and two language (R Development Core Team, 2014). outgroup taxa (V. gilvus, V. plumbeus)(Fig. 1, Supplementary Table 1). No tissue samples of V. gracilirostris (endemic to Noronha Island off the coast of Brazil) could be located, so this taxon was not 2.3. Phylogenetics included in our study. Southern and northern V. olivaceus were dis- tinguished by subspecific identity in museum records, or by loca- We ran two phylogenetic analyses on both the MD50 and MD10 tion and time of capture for any specimens lacking subspecific alignments.