Comparing morphological vs eco-physiological traits
To compare the importance of morphological vs eco-physiological traits in explaining species’ demographic responses, we performed two additional PCAs: one with only morphological/leaf structural traits (height, width, Rootd, Rootw, CSI, LMA, LA, LT, LDMC, C, N and C:N), and another with only eco-physiological traits (defined as describing internal chemical or physical processes and their responses to the environment: A, E, GS, WUEintr, WUEinst, PNUE, R, GSnight, Enight, δ13C , δ18O, depth, Topt, Ω and Q10). We then built LMMs as described above with the two first axes of both PCA’s to evaluate their ability to predict demographic change.
To investigate if morphological traits could be used as proxies of physiology, we first calculated correlation coefficients for each physiological-morphological trait pair. We then tested if the whole morphological trait space correlated with the eco-physiological trait space by performing a co-inertia analysis between the two PCAs (Drayet al. 2003). The significance was tested with a permutation test (n = 99). Finally, we evaluated how predictions of species’ vulnerability to climate warming would differ if we based the predictions on either morphological or eco-physiological traits only. To this end, we ranked species from least to most vulnerable to warming based on their locations on the trait axes.