Figure legends
Figure 1: Geographic map of the 16 localities (15 islands) under study, sampled in the archipelagos of Mallorca, Cabrera, and Menorca.
Figure 2: Chromosome distribution of raw and filtered SNP data for each of the three datasets (COMBINED, GBS and RADSeq).
Figure 3: Genetic diversity estimates per population (Pi: nucleotide diversity, Ho: Observed heterozygosity, and Fis: inbreeding coefficient) with corresponding error bars, based on 5523 SNPs (COMBINED dataset).
Figure 4: Heatmap of pairwise genetic differentiation (Fst) between the 16 populations of P. lilfordi (COMBINED dataset, 2876 SNPs). All Fst estimates were significant (p< 0.01 after permutations).
Figure 5: Population genetic structure based on 2876 SNPs (COMBINED dataset, after LD). A) DAPC retaining the first 64 PCs (65% of the explained variance). B) Unsupervised clustering by ADMIXTURE depicting K =2 to K=9 (best value) clusters.
Figure 6: Midpoint-rooted ML tree based on consensus sequences per population after SNP concatenation (2876 SNPs, only variable sites, model GTR+ASC, 10,000 bootstrap replicates). Node numbers indicate bootstrap support. The asterisk indicates the most probable ancestral population according to IQ-TREE 2 root analysis.
Figure 7: Candidate SNP outliers identified for GBS (n= 222), RADSeq (n=190), and COMBINED (n=18). A) Outliers chromosome distribution (barplot) and proportion per gene category (coding, CDS and non-coding regions) (pie chart). B) Proteins with outliers detected by multiple datasets and corresponding number of outliers. The asterisk indicates proteins with shared SNP outliers among datasets. C) Distance biplot of a Principal Component Analysis (PCA) based on the outlier genotype distance matrix for the COMBINED dataset (18 SNPs). Circles are specimens plot to optimally reproduce their Euclidean distances; arrows are unscaled SNPs scores showing their contribution to each principal component.