Ant-plant mutualisms, in which plants provide food and shelter for ants and in turn are protected from herbivores, are a common feature in many ecosystems. The effect of ants on their plant hosts is frequently variable, and this variation can play an important role in governing the productivity and stability of the ecological communities to which these species belong. This dissertation explores sources of variation in the effect of ants on a common mangrove, Conocarpus erectus (Combretaceae), which attracts ants using extrafloral nectar.

The first chapter examines the effect of leaf pubescence on the ant- C. erectus mutualism. Observational studies showed that C. erectus plants with pubescent leaves produced less extrafloral nectar than plants without pubescent leaves. An ant-exclusion experiment demonstrated that pubescent plants have less herbivory and receive less of a benefit from ant mutualists than non-pubescent plants. I suggest that redundancy between pubescence and ant mutualists has led to the evolution of trade-offs between these anti-herbivore defenses.

The second chapter describes a field experiment investigating the effects of disturbance on the ant-C. erectus mutualism using pruning and ant-exclusion treatments. Pruned C. erectus plants produced more extrafloral nectar, less pubescence, less tough foliage, and more tannins. In addition, pruned plants had higher densities of both ants and herbivores. These changes were expected to increase the beneficial effects of ants on pruned plants, but this did not occur. Hunting spiders in the family Clubionidae were most abundant on pruned, ant-exclusion plants, and I suggest that compensatory predation by these alternate predators diminished the effect of ant-exclusion on pruned plants.

The third chapter describes a field experiment examining the effect of a vertebrate predator, the lizard Anolis sagrei, and resource subsidies, in the form of seaweed, on the ant-C. erectus mutualism. The experiment utilized ant exclusions and seaweed manipulations on islands with and without lizards. Ants and lizards had a synergistic effect on herbivory in the absence of resource subsidies. This synergy probably arises from temporal partitioning of the habitat (the dominant ant species is nocturnal, while the lizards are diurnal), which may eliminate temporal refuges for herbivores. In the presence of resource subsidies the synergy disappears, likely due to changes in lizard and ant foraging behavior when seaweed-derived prey are available.