Figure legends
Figure 1. Glucosinolate (GSLs) chemical distance versus phylogenetic distance across 14 species of Cardamine . Shown is the non-metric multidimensional scaling (NMDS) plot of the GSLs of all species grouped based on 95% confidence interval ellipses. Colours represent the different groups (1: C. alpina, C. resedifolia, C. rivularis , C. amara , 2: C. kitaibelii, C. pentaphyllos, C. heptaphylla , 3: C. trifolia and C. pratensis , 4: C. hirsuta, C. impatiens, C. flexuosa, C. matthioli and C. bulbifera ). Finally, each species in the NMDS plot is assigned to its corresponding phylogenetic position in the pruned cladogram depicted on top.
Figure 2. Coinertia analysis figure based on correlated structure between the matrix of plant functional traits and the GSL matrix across species. Species are color-coded based on the assigned group. Individual GSLs are not shown to avoid confusion on the figure but are discriminated according to the groups in Fig. 3.
Figure 3. (a) Linear discriminant (LD) analysis of the differences in the distribution of leaf GSLs profiles among the four groups of Cardamine species. Histograms show the distribution of discriminant scores of leaf GSL profiles produced by plant species across different groups. The first LD1 explains 90.8% of the between-group variance. n = the number of species in each group.
Figure 4. Boxplots representing the average growth of different herbivores across the different groups of Cardamine species. Significant differences among groups were tested with a linear model followed by post-hoc analysis with Tukey HSD test.