RDA analysis
Uncorrelated variables (r<0.7) retained for the RDA analysis were BIO4 (temperature seasonality –standard deviation ✕ 100–), BIO6 (minimum temperature of the coldest month), BIO16 (precipitation of wettest quarter) and BIO17 (precipitation of driest quarter). Permutation tests showed statistical significance for the first two axes and all four environmental variables (p-value<0.01). Bioclimatic variables included in the RDA explained 17.2% of the genetic variation (inertia of constrained axes), and the remaining 82.8% of the variation is unexplained and not accounted for by the included bioclimatic variables (inertia of unconstrained axes). Eigenvalues for the two significant constrained axes were 84.65 (proportion of variance explained (PVE): 0.70) for RDA1 and 19.41 (PVE: 0.16) for RDA2. RDA analysis (Fig. 3a) showed a clear association between C. paulo-vargasii genotypic information and BIO4, and between Iberian C. binervis populations and BIO17. On the other hand, Eurosiberian C. binervis populations, in congruence with the phylogenetic relationships, formed a distinct cluster that overlapped with C. camposii . Finally, C. laevigatapopulations showed a wider spread across RDA axes, although two groups of populations mostly fitted BIO16 and BIO17 values, whereas another group placed more along BIO6.
We obtained a set of 72 outlier loci highly putatively associated with the set of retained environmental predictors (Fig. 3b). Thus, BIO6, BIO17, BIO16 and BIO4 showed the strongest correlation with 43, 14, 8 and 7 outlier SNPs, respectively. Remarkably, the chromosome number of individuals used as a response variable was retrieved as an “outlier locus” in the RDA analysis and showed the strongest correlation with BIO6 (Fig. 3b).
There are four loci that are significantly associated with BIO6 and highly correlated with chromosome number (Table S2). One of the four loci is linked to genome regions with high amounts of repetitive DNA. The other three loci are linked to important metabolic genes. One locus was found within an urease subunit (related to urea metabolism; although four other important genes were very close: lipoxygenase protein, aspartyl aminopeptidase, adenylate kinase and homologous to rice gene LOC). A second locus was found within a diacylglycerol kinase (DGK, related to diacylglycerol metabolism, a lipid from cell membrane; a second protein kinase domain was found very close). Finally, the last locus was quite close to a formyl transferase.