Fig. 2. Gene richness (total number of unique seed orthologues/strain) significantly decreases as environmental stress increases across four gradients (A acidity, B aridity, C heat, D salinity).

Fitted lines (with 95% CI) show model predictions after accounting for environmental collinearity in a multiple regression model (see Method D). P values from Table S2 model results are shown, along with partial R² measures (see Supplementary Methods S10). Each data point is a strain (n = 374). Raw gene richness values are plotted. Singleton accessory genes and core genes are removed from the gene richness count, n = 74,089 genes.