FIGURE 6.

SNP turnover in allele frequencies (panel a) of Primula elatior as predicted by the gradient forest model with the climatic partial RDA outliers as response. Principal component analysis (PCA) was used to reduce the genetic turnover of each predicted variable into three principal components that were each assigned a RGB colour. The colour code does not have a scale but similarity between colours indicates the relation of predicted adaptive genetic frequencies between populations. The relation of the cumulative importance (R ²) of neutral and adaptive SNPs to explanatory climate variables can be found in panel b.