One approach to understanding how mutualisms function in community settings is to model well-studied pairwise interactions in the presence of the few species with which they interact most strongly. In nature, such species are often specialized antagonists of one or both mutualists. Hence, these models can also shed light on the problem of when and how mutualisms are able to persist in the face of exploitation. We used spatial stochastic simulations to model the ecological dynamics of obligate, species-specific mutualisms between plants and pollinating seed parasite insects (e.g., yuccas and yucca moths) in the presence of one of two obligate antagonist species: flower-feeding insects (florivores) or insects that parasitize seeds but fail to pollinate (exploiters). Our results suggest that mutualisms can persist surprisingly well in the presence of highly specialized antagonists but that they exhibit distinctly different temporal and spatial dynamics when antagonists are present. In our models, antagonists tend to induce oscillations in the mutualist populations. As the number of per capita visits by antagonists increase, the system's oscillatory dynamics become more extreme, finally leading to the extinction of one or more of the three species. When the antagonists exhibit high per capita visitation frequencies and long dispersal distances, significant spatial patchiness emerges within these tripartite interactions. We found surprisingly little difference between the ecological effects of florivores and exploiters, although in general florivores tended to drive themselves (and sometimes the mutualists) to extinction at parameter values at which the exploiters were able to persist. These theoretical results suggest several testable hypotheses regarding the ecological and evolutionary persistence of mutualisms. More broadly, they point to the critical importance of studying the dynamics of pairwise interactions in community contexts.