Pairwise selection at SNP loci
The overall pairwiseG’ST values ranged from 0.007 to 0.4508 for the microsatellite markers, and from 0.0076 to 0.0655 for the SNP locus markers among the fifteen comparisons, depending on the population compared (Supplementary Material S8).
Controlling for the neutral processes (99% CI using microsatellite genotype data) for all fifteen pairwise population comparisons of the 101 SNP loci (1286 possible G’ST values), we mostly observed signatures of stabilizing selection (51%) and genetic drift (38%), followed by undetermined (7%) and divergent selection (4%). We observed roughly equivalent patterns of divergence among the gene function categories (Supplementary Material S9). The distribution of the types of selection did not differ significantly among different functional categories (χ2 = 20.33, p = 0.32).
We detected signatures of stabilizing selection across all SNP loci in all but six population comparisons (Figure 4), where the neutralG’ST ranges included zero (Alert/Utqiagvik, Alert/Mitivik Island, Alert/Svalbard, Utqiagvik/Mitivik Island, Utqiagvik/Svalbard, and Mitivik Island/Svalbard). For these six comparisons, the SNP loci that had G’ST values less than the neutral expected range (i.e., negativeG’ST values) were identified as “undetermined”. Therefore, we are likely underestimating overall stabilizing selection effects. Overall, we observed that 67.5% - 96.3% of the SNP loci are under stabilizing selection among the nine pairwise population comparisons where we were able to test for signatures of stabilizing selection (Figure 4, Supplementary Material S8).
We did not observe any signatures of divergent selection in six (Aleutian Islands/Alert, Aleutian Islands/Utqiagvik, Aleutian Islands/Mitivik Island, Aleutian Islands/Pribilof Islands, Aleutian Islands/Svalbard, and Alert/Utqiagvik) out of fifteen pairwise population comparisons (Figure 4). For the remaining nine population pairs, we observed that 1.09%-23.1% of the SNP loci exhibited divergent selection (Figure 4, Supplementary Material S8). The Utqiagvik/Svalbard population comparison showed the most divergence (23.1%) based on our selected functional locus SNPs, followed by Pribilof Islands/Svalbard and Alert/Mitivik Island population comparisons which exhibited directional divergence at 7.23%, and 6.60% of tested SNP loci, respectively (Supplementary Material S8).
To compare selection signatures across all pairwise population comparisons, SNP marker data would have to be 100% available (i.e., not classified as “undetermined” in any comparison). Thus, only a minority (11/101) of SNP marker loci could be broadly compared across all pairwise comparisons (Figure 5; Table 5). Based on those 11 SNP marker loci, high levels of stabilizing selection signatures were generally observed when the non-migratory populations (Aleutian & Pribilof islands) were compared to each other or with migratory populations (Alert, Utqiagvik, Svalbard, and Mitivik Island), whereas comparisons among the migratory populations showed mixed signatures of neutral processes and divergent selection depending on the SNP locus (Table 5). Of these 11 SNP loci, 7 showed evidence of divergent selection in at least one population comparison, while the Utqiagvik/Svalbard population pair comparison showed 6 of these 11 loci under divergent selection (Table 5). Broadly, the divergent genes from pairwise comparisons in this subset were associated with cellular housekeeping, lipid metabolism, nervous system development, reproduction, and stress (Table 5, Supplementary Material S10).