Figure 3. PCA plot of first two principal components based on (a) 88 SNPs: offshore (n = 3), resident (n = 11), Bigg's (n = 30) from data in Morin et al. [15]; (b) 26 microsatellites: offshore (n = 5), resident (n = 250), Bigg's (n = 116) (samples genotyped at ≥20 loci) [56]; unpublished); (c) 3678 RADseq SNPs: offshore (n = 7), resident (n = 52) and Bigg's (n = 37) populations [57,62]; (d) 1 00 000 (subset from 6 371 282) SNPs from 147 high-coverage genomes of offshore (n = 7), Bigg's (n = 14) and resident (n = 126) samples from multiple geographically and behaviourally defined subpopulations (Alaska, northern and southern resident populations) (based on subset of SNP genotype data from [113]. See Supplementary Materials for methods and data set information.