Generalized linear models built to predict herbivore, natural enemy, and predation services.

Here we present the best models for each response variable (models <2 AIC in difference included in the table for each response variable). Abundance and richness for herbivores are from visual surveys. Abundance and richness for natural enemies are combined values from visual surveys and pitfall traps. Intercepts are scaled values. Significance is denoted as (*) for P values, as * between 0.01–0.05, as ** for values <0.01, as *** for values <0.001. (AIC = Akaike Information Criterion; df = degrees of freedom; LRR = Log Response Ratio; VCI = vegetation complexity index).