Western burrowing owl genomics

Description

Notes

Methods

Details regarding sample collection, genome sequencing, and sequence processing may be found in Methods S1; but notably, we sequenced a reference genome to high coverage and 202 burrowing owl samples collected across their migratory and resident breeding range to low coverage. Because our resequencing dataset was low coverage, we used variant detection and analytical methods that largely did not require called genotypes. This included both genotype likelihoods as estimated in the program ANGSD (Korneliussen et al., 2014) and a single‐read‐sampling (SRS) method that randomly selects one read per variant to temper the bias of high variation in locus‐to‐locus depths. Using these methods, files were prepared for analyses as described below using the following four filtering and genotyping frameworks and conditions: (1) Using ANGSD to produce genotype likelihood files for all individuals in the BEAGLE format (‐doGlf 3) and a minor allele frequency file (‐domaf 1) with restrictive filtering that uses a conservative minimum minor allele frequency (‐minmaf 0.05), a low maximum likelihood of being polymorphic (‐SNP_pval 1e‐6), adjusting mapQ scores for excessive mismatches from the reference genome (‐C 50), and confirming variants using a base alignment quality estimation (‐baq 1). (2) For SRS analyses, we used the 'HaplotypeCaller' module in GATK (McKenna et al., 2010) to call genotypes for all individuals sequenced, filtered by removing insert/deletion variants, and kept only biallelic variants found in 50% of the individuals. (3) We used ANGSD to create population‐specific site frequency spectra (SFSs) from site allele frequency files using the reference genome to polarize allele calls (‐anc), adjusting frequencies using individual F_IS (‐indF), and with strict filtering conditions including discarding reads without unique mapping (‐uniqueOnly 1), removing bad reads (‐remove_bads 1), using only reads for which mates are mapped (‐only_proper_pairs 1), discarding reads with low mapping quality (‐minMapQ 1), keeping reads with high base quality (‐minQ 20), dropping reads with low or high depth across samples (‐setMinDepth 10 ‐setMaxDepth 500), keeping only biallelic sites (‐skipTriallelic 1), and also previously described conditions (‐minMaf 0.05 ‐C 50 ‐baq 1). (4) Minor allele frequency files (MAFs) were also generated for each sample site (‐doMaf 1), sampling all the sites identified in the overall MAF file, and only generating minor allele frequencies for variants found in a minimum of four individuals in each population.

Additional details