Background/Question/Methods Mutualistic interactions lie at the core of ecological communities. Although many mutualisms were initially studied as two-partner systems, most mutualisms are comprised of complex communities of interacting species. Symbiotic mutualisms between ants and plants, in which plants provide ant colony space and ants defend against leaf-eating herbivores, are frequently associated with hemipteran phloem-sucking insects. Hemipterans are antagonists of the plant, but symbiotic ants tend hemipterans, consuming their honeydew waste. If ants utilize honeydew as a primary food source, do hemipterans have indirect, positive effects on plants, mediated by symbiotic ants? We studied the interactions among a common neotropical tree, Cordia alliodora, its symbiotic ants, Azteca pittieri, and several species of hemipterans at two sites in seasonally dry tropical forest in Middle America. We asked whether ants defend tree leaves from herbivory by excluding ants from leaves. We then asked whether densities of ants and hemipterans in plant nodes are correlated with levels of herbivory. Finally, we asked whether ant defense against leaf-eating herbivores is improved via interactions with hemipterans because: (1) well fed ants are more active or more aggressive defenders; and/or (2) ant colony size is limited by food availability, not by space, and larger colonies defend better. We tested (1) by keeping ants in the lab on diets of high or low sugar and quantifying activity and aggression against caterpillars at the end of three weeks. We tested (2) by collecting trees, counting the total number of ants and hemipterans, measuring the total amount of colony space, and quantifying herbivory. Results/Conclusions We found that A. pittieri ants do defend C. alliodora leaves from herbivory, although the strength of this effect depends on both site and season. Ants are better defenders late in the rainy season when herbivory pressure is lower. Densities of both ants and hemipterans were negatively correlated with levels of herbivory in multiple regression models, but herbivory was more strongly correlated with hemipteran densities than with ant densities. Ants on high sugar diets were significantly more aggressive against caterpillars but were not more active than ants on low sugar diets. In multiple regression models, the number of hemipterans (food availability) was significantly positively correlated with the number of ants, whereas the amount of colony space was not. Larger ant colonies appear to defend better, although this effect depends on site. We conclude that hemipterans may provide important, indirect benefits to plants that rely on ant defenders.