The influence of habitat autocorrelation on plants and their seed-eating pollinators

Model systems for studying mutualism costs and benefits include the many species of plants that rely on seed-eating pollinators for their reproduction. Empirical studies of these interactions show that mutual- ism costs and benefits can vary greatly within populations. Here we investigate the role of plant habitat autocorrelation on mutualism properties when mutualist dispersal is limited. We build a spatially explicit individual-based model of an obligate mutualism that includes a plant and its obligate seed-eating polli- nator. We also model exploiters of this mutualism, which do not pollinate, but compete with pollinators for pollinated plant ovules in which to develop. We test how the autocorrelation of habitable plant envi- ronment affects pollinator production, seed production, pollinator visitation to plants, and the persistence of exploiters at different dispersal distances and rates of exploitation. We find that positive habitat auto- correlation increases the mean number of pollinator visits to plants. More frequent pollinator visitation to plants increases the probability that a random plant will be pollinated, but also the probability of pollinator oviposition into plant ovules at the cost of a developing seed. This process leads to spatial variation in the production of pollinators versus seeds. For a given scale of habitat autocorrelation, the turnover of this variation decreases when pollinator dispersal distance is high. Exploiters of the mutual- ism dramatically lower the number of pollinator visits per flower, which decreases pollinator production, seed production, and mutualist densities. Exploiters persist with mutualists when the mean number of pollinator visits per plant is neither too low, nor too high. When the mean number of pollinator visits a plant receives is too low, overexploitation and the extinction of both mutualists and exploiters follows; a high mean number of pollinator visits results in the competitive exclusion of exploiters by pollinators. Because the autocorrelation of habitat strongly affects the number of pollinator visits per flower, our results show that habitat autocorrelation can influence key mutualism properties and the susceptibility of mutualisms to exploitation