Genomic divergence
Using 1 205 604 SNP markers across 186 individuals, principal component analysis (PCA), pairwise FST and admixture analyses all revealed little genome-wide differentiation. First, the first two principal components (PC1 and PC2) explained only 0.009 and 0.008% of variance, respectively (see supplement). There was some clustering of individuals along PC1 (Fig S4), which could be indicative of clustering based on genotype (associated with, for example, spatial structure, sex or phenotype), on areas of low recombination or, alternatively, arise as a statistical artefact. Our PCA did not cluster individuals according to any a priori hypotheses for which we had data (i.e., according to basin, habitat type, site, sex, phenotype or tapeworm presence). Second, pairwise Nei’s Fstwere very low at both habitat and site level (Table S3). Third, admixture analyses using snmf revealed that the model with K = 1 had the lowest cross-entropy score, suggesting that K = 1 best predicted the population genetic structure in our dataset (Figure 3). This suggests that Mývatn stickleback form one panmictic population. The effective populations size (Ne) for the population was estimated as 1976 (± 640).