Leaf toughness is a better predictor of herbivory and plant performance than leaf mass per area (LMA) in temperate evergreens

The mechanical strengthening of leaves protects seedlings from herbivore damage, particularly in shade-tolerant evergreens. Interspecific studies have shown that leaf mass per area (LMA) and leaf toughness (force-to-punch) can play this role. Here we compared the influence of LMA and leaf toughness on herbivory and plant performance in a temperate rainforest. In seedlings of 14 evergreen species, we addressed the across-species relationship between LMA and force-to-punch, and compared the strength of their associations with herbivory and with species’ light requirements. Moreover, in four understory species we performed a multivariate analysis within-species, analogue to phenotypic selection analysis, evaluating the correlation between seedling performance, estimated as chlorophyll fluorescence (F_v/F_m), and force-to-punch, LMA, lamina density and lamina thickness. LMA and force-to-punch were positively associated across species. Herbivory was negatively correlated with both force-to-punch and LMA, but a stepwise multiple regression showed that force-to-punch was a better predictor of herbivory. Neither leaf lamina density nor thickness were associated with herbivore damage. Those species that were more shade-tolerant had leaves with higher force-to-punch and higher LMA, and less slender seedlings. In the within-species analyses in four shade-tolerant species, seedling performance was generally positively associated with force-to-punch, but not with LMA, lamina thickness, or lamina density. Both interspecific and within-species analyses showed that force-to-punch is more strongly related to herbivore damage and plant performance than LMA. This consistency between interspecific patterns of trait covariation and within-species trait-performance associations suggests that natural selection could have shaped the relationships between mechanical traits and ecological roles observed across species.