Figure S1minor Allele Frequency Variation for Fstoutlier Snps Across the Three Mainland

Figure S1Minor allele frequency variation for FSToutlier SNPs across the three mainland populations. Filled symbols indicate the population average for Cambria (), Monterey (), Año Nuevo ()with standard deviation. Unfilled symbols indicate frequency for stands within populations.

Figure S2Box and whisker plots indicating the distribution of FST values for putatively adaptive outlier SNPs, putatively neutral non-outlier SNPs and putatively neutral SSR markers. SSR FST value were estimated using the data set from Karhu et al (2006). Boxes indicates the median, 25th and 75th quantiles, vertical lines indicate minima and maxima of the sample range. Distributions were similar for non-outlier and SSR marker sets (ANOVA: P < 0.9) and were both significantly different to the outlier SNP set (ANOVA: P < 0.001).

Figure S3 Comparison of allele frequency at SNP 57 (top), specific surface area (middle) and environment (bottom) across the three mainland populations: Cambria (), Monterey () and Año Nuevo (▲). Populations are arranged by deg. latitude along the x-axis.

Figure S4Bar plots of individual proportional memberships in K hypothetical ancestral clusters averaged over 3 replicates for the three mainland populations:Año Nuevo, Monterey and Cambria, for (top) 139 putatively neutral SNPs, (middle) 8 putatively adaptive outlier SNPs loading to SNPPCA1, and (bottom)2putatively adaptive outlier SNPs loading to SNPPCA2.

Figure S5 a) Averagelog probability (Ln(K)) and standard deviation for varying number of ancestral populations (K) estimated in Structure for neutral (non-outlier) SNPs (solid line: y-axis left), outlier SNPs loading to SNPPCA1 or group 1 (dotted line: y-axis right) and outlier SNPs loading to SNPPCA2 or group 2 (dashed line: y-axis left). Probabilities for both groups of outlier SNPs plateaued quickly after K=2, whereas values continue to rise gradually for the non-outlier SNP set. Log probabilities for the group 2 outliers were scaled by a factor of four to allow comparison. b) The mode ofK (K = m(L”K)/ s(L(K))) (Evanno et al. 2005) calculated for the outlier group 1 (dotted line: y-axis left), and outlier group 2 (dashed line: y-axis right) SNP setsinferred two populations in both cases. In the case of neutral SNP markers K initially implied 2 populations (solid line: y axis left), however when Monterey and Cambria were analysedin isolation a third population could be resolved (dotted dashed line: y axis left).