Pairwise selection at SNP loci
While it is possible for individual functional SNP markers to show a
global selection signature, others may only show signatures of divergent
or stabilizing selection at the population level due to specific
differences in local conditions among individual populations. To assess
genetic divergence patterns among pairs of snow bunting populations, we
calculated pairwise estimates of G’ST using both
microsatellite and SNP genotype data and compared the SNP loci pairwiseG’ST values with the presumed-neutral
microsatellite loci range (created using ‘diffCalc’ function’s
bias-corrected bootstrapping loci approach as explained above) at the
99.9% CI to detect signatures of divergent and stabilizing selection.
We used higher CI (99.9% versus 99% neutral CI used in global
comparison) to avoid detection of false positives for pairwise
comparisons since we are assessing 101 SNPs (useable SNPs
post-filtration step; see “Bioinformatics” section above for details)
and fifteen pairwise population comparisons. Corrections for multiple
comparisons were not necessary as neutral range was individually
developed for each comparison. We first combined all the results from
the pairwise comparisons to investigate overall levels of genetic drift
and selection, and also conducted a Chi-squared test to assess whether
the pattern of selection signatures differed across the seven gene
function categories. However, for some population pairs it was not
possible to identify SNPs under stabilizing selection since the neutralG’ST range for that pairwise comparison included
zero. As such, we have reported the SNP loci showing likely signals of
divergent selection for all fifteen pairwise comparisons, but
stabilizing selection for only nine of fifteen comparisons. For the six
comparisons which had neutral ranges that included zero, the SNP loci
with G’ST values less than expected neutral range
(i.e., negative G’ST values) were identified as
“undetermined”.
To gain further insight into specific genes that showed evidence for
divergent selection, we explored the function of selected SNP loci withG’ST values that had no “undetermined”
classifications across any of the fifteen pairwise comparisons.
Therefore, each SNP locus in this subset was identified as either under
neutral processes, stabilizing selection or divergent selection for all
fifteen pairwise comparisons. This approach allowed us to assess the
selection status of divergent SNP loci across all other population pairs
– this allows the comparison of the role of these functional markers
across all other population comparison(s) to highlight specific
differences, allowing us to identify specific genes contributing to
population divergence and local adaptation.