Quantitative synthesis of context dependency in ant-plant protection mutualisms.

Context dependency, variation in the outcome of species interactions with biotic and abiotic conditions, is increasingly considered ubiquitous among mutualisms. Despite several qualitative reviews of many individual empirical studies, there has been little quantitative synthesis examining the generality of context dependency, or conditions that may promote it. We conducted a meta-analysis of ant-plant protection mutualisms to examine the generality of context-dependent effects of ants on herbivory and plant performance (growth, reproduction). Our results show that ant effects on plants are not generally context dependent, but instead are routinely positive and rarely neutral, as overall effect sizes of ants in reducing herbivory and increasing plant performance were positive and significantly greater than 0. The magnitude of these positive effects did vary, however. Variation in plant performance was not explained by the type of biotic or abiotic factor examined, including plant rewards (extrafloral nectar, food bodies, domatia), ant species richness, plant growth form, or latitude. With the exception of plant growth form, these factors did contribute to the effects of ants in reducing herbivory. Reductions in herbivory were greater for plants with than without domatia, and greatest for plants with both domatia and food bodies. Effect sizes of ants in reducing herbivory decreased, but remained positive, with latitude and ant species richness. Effect sizes in reducing herbivory were greater in tropical vs. temperate systems. Although ant-plant interactions have been pivotal in the study of context dependency of mutualisms, our results, along with other recent meta-analyses, indicate that context dependency may not be a general feature of mutualistic interactions. Rather, ant-plant protection mutualisms appear to be routinely positive for plants, and only occasionally neutral.