Global selection at candidate loci
To assess whether candidate loci were under divergent selection across the six populations, we compared global estimates of Hedrick’sG′ST (Hedrick, 2005), calculated using the ‘diveRsity’ package (Keenan et al., 2013) in R, between neutral microsatellite and functional SNP markers. Hedrick’sG′ST is suitable for comparing genetic divergence measures among different marker types since it standardizes for heterozygosity (Hedrick, 2005). To assess an overall signature of selection at SNP loci relative to microsatellite markers across all populations, we first developed a ‘neutral range’ meanG′ST with 99% confidence intervals (CI) for the microsatellite marker G′ST values using the ‘diffCalc’ function of the R package ‘diveRsity’. Specifically, we used bias-corrected bootstrapping across microsatellite loci to estimate the neutral CI range, representing neutrality expectation (presumably due to genetic drift) based on the G′ST distribution of microsatellite markers. Next, we determined whether theG′ST values for individual SNP loci fell outside of the neutral ranges, as such loci are likely to be under selection. Since the calculated neutral range for G’ST did not include zero, we were able to identify SNP genes under stabilizing (lower than neutral expected) and divergent (higher than neutral expected) selection.