Background/Question/Methods   Tropical communities are characterized by complex biological interactions, involving multiple species, and many trophic levels. In coffee farms in Chiapas, Mexico, complex insect food webs involve non-consumptive effects of a Pseudacteon phorid parasite on Azteca instabilis (a keystone ant species), trait-mediated effects of the phorid on Azteca competitors and prey, and interactions among multiple species of twig-nesting ants. Several ant species (including Azteca, Pseudomyrmex simplex and Procryptocerus hylaeus) prey on the coffee berry borer, limiting its ability to attack coffee fruits. Furthermore, Azteca attacks on borers are limited by the phorids. However, the multi-predator effects of ants on borers in this system, and the potential influence of phorid presence on multi-predator interactions are not known. We examined the following questions: 1) What are the combined (multi-predator) effects of ants on the coffee berry borer, and 2) How do the non-consumptive effects of the Pseudacteon phorid on Azteca ants alter multi-predator effects? We conducted experiments in insect arenas with coffee branches with 10 fruits, 20 borers, and different ant-phorid combinations. We used all combinations of zero, one, two, and three predator treatments, and added 2 phorids to half of cages with Azteca alone or with other ant species. After 24 hours, we counted number of attacked fruits.

Results/Conclusions   In behavioral trials, phorids had no effect on non-Azteca ant species, so phorids were not added to treatments where Azteca ants were absent. Consistent with previous research, we found that borers, when alone, entered 3.83 fruits on average, between 25-43% more than when with one ant species (Azteca = 1.807, P. simplex = 2.8, and P. hylaeus = 1.66). Additionally, phorids limited Azteca abilities to prevent borers from entering fruits (3.46 fruits attacked). Two-ant-species together performed as well as or better (e.g. prevented more borers from entering fruits) than single-species treatments (1.28 – 1.5 fruits attacked), but three ant species were not better than two (1.6 fruits attacked). Thus, there is weak evidence for non-additive effects of ant predator species on the coffee berry borer. Importantly, however, in the presence of phorid flies, other ant species compensated for the reduced effects of Azteca; only between 1.0-1.8 fruits were attacked where Azteca, phorids, and 1-2 other ant species were present. Thus, non-consumptive effects of specialist predators may affect multi-predator impacts. This may be especially important for maintenance of biological control in agricultural systems.