Supplementary Figure 10: Genome-wide patterns of population differentiation.
(a) The mean fixation index (FST) of coding and noncoding SNPs detected in three comparisons (top, A versus MC; middle, C versus AM; bottom, M versus AC; A, Africa; M, northern/western Europe; C, southern/eastern Europe) as estimated for genes of different caste expression and CpGO/E categories. The FST of SNPs in coding regions (CDS + UTRs) is not significantly higher than the level of differentiation detected in noncoding (intron + intergenic) regions. There are no signs of genome-wide signals of positive selection as detected by FST on coding regions across all genes in the genome or among genes with caste-biased overexpression. The coding regions of low-GC/CpG genes have a small (∼12%) increase in mean FST compared to high-GC/CpG genes, a signal that is less clear in noncoding regions (NS, genes with unbiased expression; Q, queen-biased genes; W, worker-biased genes; D, drone-biased genes; LCpG, low-CpG genes; HCpG, high-CpG genes; LGC, low-GC genes; HGC, high-GC genes; 95% confidence intervals were estimated by bootstrap). (b) GC content in a 100-bp window around each SNP (y axes) is weakly correlated with FST (x axes; R2 < 2%) and mainly detected as a 10–20% drop in the extreme tail along the FST distribution for SNPs in coding regions. (c) Substitutions in coding regions were analyzed for codon changes (proportion of nonsynonymous SNPs; y axes) across the FST spectrum (x axes). High-FST substitutions (FST ≥ 0.9; fnon = 0.42) were found to be significantly enriched for nonsynonymous changes between African and European bees (A versus MC) compared to low-FST substitutions (FST < 0.9; fnon = 0.28) and in the C versus AM comparison (FST ≥ 0.9; fnon = 0.31 versus FST < 0.9; fnon = 0.26) but not the M versus AC comparison (FST ≥ 0.9; fnon = 0.28 versus FST < 0.9; fnon = 0.28) by bootstrapping the SNPs and assessing the variation around the mean at the 5% level. The large enrichment of highly differentiated nonsynonymous SNPs suggests that protein sequences are under divergent positive selection in African and European bees.
