Open Collections

Walsh, Jennifer; Aguillon, Stepfanie; Chan, Yvonne; Arcese, Peter; Benham, Phred; Lovette, Irby; Mikles, Chloe

<b>Abstract</b><br/>

Elucidating forces capable of driving species diversification in the face of gene flow remains a key goal in evolutionary biology. Song sparrows, Melospiza melodia, occur as 25 subspecies in diverse habitats across North America, are among the continent’s most widespread vertebrate species, and are exemplary of many highly variable species for which the conservation of locally adapted populations may be critical to their range-wide persistence. We focus here on six morphologically distinct subspecies resident in the San Francisco Bay region, including three salt-marsh endemics and three residents in upland and riparian habitats adjacent to the Bay. We used reduced-representation sequencing to generate 2,773 SNPs to explore genetic differentiation, spatial population structure, and demographic history. Clustering separated individuals from each of the six subspecies, indicating subtle differentiation at microgeographic scales. Evidence of limited gene flow and low nucleotide diversity across all six subspecies further supports a hypothesis of isolation among locally adapted populations. We suggest that natural selection for genotypes adapted to salt marsh environments and changes in demography over the past century have acted in concert to drive the patterns of diversification reported here. Our results offer evidence of microgeographic specialization in a highly polytypic bird species long discussed as a model of sympatric speciation and rapid adaptation, and they support the hypothesis that conserving locally adapted populations may be critical to the range-wide persistence of similarly highly variable species.</p>; <b>Methods</b><br />

<span><span style="font-style:normal;"><span><span style="font-weight:normal;"><span style="letter-spacing:normal;"><span><span><span style="white-space:normal;"><span><span><span>Sequence quality was assessed using <span><span><span><span><span><span>FastQC</span></span></span></span></span></span> version 0.11.8. (www.bioinformatics.babraham.ac.uk/projects/fastqc). Individuals were filtered for quality using the <span><span><span><span><span><span>Fast-X</span></span></span></span></span></span> Toolkit (http://hannonlab.cshl.edu/fastx_toolkit), removing sequences with Phred quality scores below 10 and sequences with more than 5% of bases with Phred quality scores below 20. We demultiplexed sequences using the command “process_radtags” in <span><span><span><span><span><span>Stacks</span></span></span></span></span></span> version 1.48 (Catchen et al., 2011) and additionally filtered samples to only retain reads that passed the Illumina chastity filter, contained an intact <i>Sbf</i>I RAD site, contained a unique sample barcode, and did not contain Illumina indexing adapters. To account for differences in length, the remaining filtered and demultiplexed reads were trimmed to 94 base pairs at the 3’ end using <span><span><span><span><span><span>Fast-X Trimmer</span></span></span></span></span></span> (FAST-X Toolkit). </span></span></span></span></span></span></span></span></span></span></span></p>

<span><span style="font-style:normal;"><span><span style="font-weight:normal;"><span style="letter-spacing:normal;"><span><span><span style="white-space:normal;"><span><span><span>Sequences were aligned to a song sparrow (<i>Melospiza melodia</i>) reference genome (Feng et al., in review) using <span><span><span><span><span><span>Bowtie2 </span></span></span></span></span></span>version 2.3. Mapped reads were then analyzed using the ref_map.pl pipeline in <span><span><span><span><span><span>Stacks</span></span></span></span></span></span>. We allowed five mismatches between sample loci and required a minimum of ten identical raw reads to make a stack. We ran the Populations module in <span><span><span><span><span><span>Stacks</span></span></span></span></span></span> for one population (-p) and required that a locus be present in a minimum of 80% of individuals to be processed (-r). In addition to obtaining all SNPs per locus (10,270), we created a subset of SNPs that included only the first SNP per stack (2,773). To avoid bias associated with physical linkage (O’Leary et al. 2018), we used the reduced, unlinked dataset with 2,773 SNPs for all analyses, unless otherwise stated.</span></span></span></span></span></span></span></span></span></span></span></p>; <b>Usage notes</b><br />

All sample information for individuals included in this VCF can be found in Supporting Information Table S1. This is the filtered VCF used.</p>

Dataset

2020-09-01

This dataset is made available under a Creative Commons CC0 license with the following additional/modified terms and conditions: CC0 Waiver

https://doi.org/10.5683/SP2/VJH4C8

https://creativecommons.org/publicdomain/zero/1.0/