Principal component analysis
A PCA, using MAP and leaf-level structural and anatomical traits associated with gm and Kleaf, was performed to complement the trait-to-trait comparisons. The first two major axes (PC1 and PC2) along with the average position of 18 C4 grasses in PC1-PC2 space are shown in Fig. 6. The, first three axes with eigenvalues and scores are shown in Table S4. PC1 explained about 50.56 % of the total variation in the C4 grasses. PC1 scaled positively with gm, SDada, Smes, SR, Narea, IVD, leaf thickness and average VED but negatively with total VLA, BSias and MAP. Thus, PC1 delineated the C4 grasses into those which occupy relatively drier habitats and show traits associated with greater gm and hence photosynthetic C-gain (higher score on PC1)) from those which occupy relatively wetter habitats and show traits associated with greater Kleaf and water-loss (lower score on PC1). PC2 explained about 16 % of the total variation and scaled positively with MAP and BS area ratio but negatively with total VLA and BSCW. PC3 explained 10.25% of total variation and scaled positively only with MAT. Thus, MAT was unrelated to any of the leaf-level structural and anatomical traits measured in the current study. Together, the first three axes explained about 77% of the total variation. To test if in addition to greater gm species adapted to drier habitats also exhibit greater Anet and leaf-level WUE, we looked at the relationship of the PC1 with Anet (Fig. 7a) and two measures of leaf-level WUE, that is, instantaneous water-use efficiency (Anet/E, Fig. 7b) and intrinsic of water-use efficiency (Anet/gsw, Fig. 7c). A higher score on PC1 indicates presence of a ‘suite’ of traits associated with greater gm and photosynthetic C-gain and lower Kleaf and transpirational water-loss. There was a significant positive relationship between Anet and PC1 (R 2 = 0.56, P = 0.002, Fig.7a). Furthermore, Anet/E showed a significant positive relationship with PC1 (R 2 = 0.25, P = 0.035, Fig.7b). However, Anet/gs did not relate with PC1. In summary, C4 grasses with greater score on PC1 also showed greater Anet and Anet/E.