Selective Sweep Detection
Selective sweep detection was performed to identify genomic regions undergoing selection. Three separate techniques were used: extended haplotype homozygosity mapping, a composite likelihood ratio test (Huber, DeGiorgio et al. 2016), and a Tajima’s D approach (see methods). The Extended Haplotype Homozygosity test identifies genomic regions with high local haplotype homozygosity, considered to be a strong indication of signatures of positive selection (Sabeti, Reich et al. 2002). The Composite Likelihood Ratio test calculates the likelihood ratio of the null hypothesis, calculated from the neutral (genome-wide) frequency spectrum, whilst the alternative hypothesis is calculated using a model where neutral selection has been altered by recent selection. This technique can separate out footprints of positive selection from background selection (Charlesworth 2012). Bermuda (n=21) and Hawaii (n=25) samples were analysed using each sweep detection technique separately.
Extended Haplotype Homozygosity (EHH) Mapping
For the Bermuda dataset, a total of 386 putative selective sweep regions were identified within Bermuda G. gallus genomes using EHH mapping (Supplementary Table 1). A threshold of 1% of the genome was selected. The mean selective sweep length was 24kb bases (median 20kb, sd 12.2kb). For the Kauai population, 351 selective sweeps were detected, with a mean selective sweep length of 26kb (median 20kb, sd 14.6kb), see Supplementary Table 1. Seventeen of the selective sweeps detected in the two feral populations overlapped with one another (permutation test, p < 0.001, see methods and Supplementary Table 2). To ascertain how many of the selective sweep regions overlapped with domestication selective sweeps, the selective sweeps identified in prior studies that focused on G. gallusdomestication were compared with the Kauai and Bermuda selective sweeps. Qanbari et al (Qanbari, Rubin et al. 2019) identified 304 domestication/ improvement selective sweeps in their analysis, with five of these overlapping Bermudian EHH sweeps (permutation test, p > 0.05), and four of these overlapping Kauai EHH selective sweeps (permutation test p > 0.05) (see Supplementary Table 3).
Composite Likelihood Ratio (CLR) Mapping
In the Bermudian population we detected a total of 455 selective sweep regions using CLR mapping (mean 32kb, median 20kb, SD 24.8kb; Supplementary Table 4), with a significance threshold of 1% of the genome used. In the Kauai population a total of 456 selective sweep regions were identified with CLR (mean 32kb, median 20kb, SD 29.6kb). Of the selective sweep regions identified using CLR in the feral populations, 44 of these overlap (permutation test, p < 0.001) (see Supplementary Table 5). When comparing with the domesticated selective sweeps identified by Qanbari et al, ten of the selective sweeps detected in Bermuda overlap (permutation test, p > 0.05), and 21 of the Kauai selective sweeps overlap (permutation test, p < 0.001; see Supplementary Table 3).
Tajimas’s D Mapping
In the Bermuda population, 168 selective sweep regions were identified (mean 57kb, median 40kb, SD 51.9kb), whilst in the Kauai population 165 selective sweep regions were identified (mean 58kb, median 40kb, SD 38.1 kb), see Supplementary Table 6. Between these two feral populations, 39 selective sweep regions were shared (permutation test, p < 0.001) (Supplementary Table 7). In contrast, Bermuda had 18 overlaps (permutation test, p < 0.001) with the domestication selective sweeps from Qanbari et al, and Kauai had 17 overlaps (permutation test, p < 0.001) (see Supplementary Table 3).