Community-wide herbivory
Consistent with our hypothesis, we found that community-wide (CWM) herbivory decreased as latitude increased from N 26°59′ to N 38°34′ (Slope = -0.203, R2 = 0.079, P = 0.012) (Figure 2); this latitudinal pattern remained even after accounting for the effects of longitude and altitude (Slope = -0.155, P = 0.069) (Table S1). Model selection based on AICcsuggested that both latitude (AICc = 222.557,w AIC = 0.525) and altitude (AICc = 224.140, w AIC = 0.238) better predicted community-wide herbivory than did the null (intercept-only) model (AICc = 224.391, w AIC = 0.210) (Table 1). Among the climatic factors, temperature seasonality decreased CWM-herbivory (Slope = -0.005, R2 = 0.107, P = 0.003), while precipitation in the wettest month (Slope = 0.023, R2 = 0.101, P= 0.004) and mean annual precipitation (Slope = 0.004, R2 = 0.098, P = 0.004) increased CWM-herbivory (Table 1; Figure S2). As for the edaphic factors, soil nitrogen (Slope = 0.185, R2 = 0.078, P = 0.003) and soil water content (Slope = 0.037, R2 = 0.076, P = 0.005) increased CWM-herbivory, while soil pH showed the opposite effect (Slope = -0.639, R2 = 0.093, P = 0.003) (Table 1; Figure S2). Community-wide herbivory was not associated with any plant community factors in the linear mixed-effect models (Table 1).
Intraspecific variability and species’ turnover represent additive components of community-wide herbivory; intraspecific variability increased with latitude (Slope = -0.131, R2 = 0.054,P = 0.018), while turnover was not related to altitude, latitude or longitude (Figure 2). Among the predicted factors, turnover was not well explained by any climatic, plant community or soil factors (Table S2). Meanwhile, soil electrical conductivity increased intraspecific variability (Slope = 0.007, R2 = 0.053, P = 0.013), but temperature seasonality (Slope = -0.003, R2 = 0.051, P = 0.022) and soil pH (Slope = -0.405, R2 = 0.059, P = 0.010) decreased intraspecific variability (Table S3; Figure S3).