The indirect consequences of a mutualism: comparing positive and negative components of the net interaction between honeydew-tending ants and host plants.

1. In ecological webs, net indirect interactions between species are composed of interactions that vary in sign and magnitude. Most studies have focused on negative component interactions (e.g. predation, herbivory) without considering the relative importance of positive interactions (e.g. mutualism, facilitation) for determining net indirect effects. 2. In plant/arthropod communities, ants have multiple top-down effects via mutualisms with honeydew-producing herbivores and harassment of and predation on other herbivores; these ant effects provide opportunities for testing the relative importance of positive and negative interspecific interactions. We manipulated the presence of ants, honeydew-producing membracids and leaf-chewing beetles on perennial host plants in field experiments in Colorado to quantify the relative strength of these different types of interactions and their impact on the ant's net indirect effect on plants. 3. In 2007, we demonstrated that ants simultaneously had a positive effect on membracids and a negative effect on beetles, resulting in less beetle damage on plants hosting the mutualism. 4. In 2008, we used structural equation modelling to describe interaction strengths through the entire insect herbivore community on plants with and without ants. The ant's mutualism with membracids was the sole strong interaction contributing to the net indirect effect of ants on plants. Predation, herbivory and facilitation were weak, and the net effect of ants reduced plant reproduction. This net indirect effect was also partially because of behavioural changes of herbivores in the presence of ants. An additional membracid manipulation showed that the membracid's effect on ant activity was largely responsible for the ant's net effect on plants; ant workers were nearly ten times as abundant on plants with mutualists, and effects on other herbivores were similar to those in the ant manipulation experiment. 5. These results demonstrate that mutualisms can be strong relative to negative direct interspecific interactions and that positive interactions deserve attention as important components of ecological webs.