ABSTRACT
The study of pollinator-plant interactions has traditionally taken an individual-based approach. The patterns of movement of individual pollinators, patterns of floral choice, and the evolution of floral traits in individual plants have been major foci for field and laboratory investigations. While considerable research has addressed the evolutionary implications of individual behavior in pollinator and plant traits, little attention has been directed at elucidating the ecological implications of pollinator behavior on plant populations. For example, we still do not possess a clear picture of the relationships between pollinator visitation frequencies, seed production, and recruitment of new individuals into the plant population. In many cases, pollinator behavior can act as a strong selective force on floral traits with no appreciable influence on overall plant population dynamics. In some cases, the effects of pollinator behavior on plant population dynamics can be quite direct, as in the transmission of pollinator-borne diseases.
As a model system, we have examined the interaction between the anther smut fungus, Ustilago violacea, and its host plant, Silene alba. Plants infected with the anther smut are rendered sterile and produce flowers with anther sacs filled with spores. Insect pollinators which land on infected flowers act as vectors of the fungus. We have demonstrated experimentally that pollinators exhibit behavioral preferences with respect to disease status of the flowers. In particular, bumble bees preferentially visit healthy rather than diseased Silene plants. We have used an individual-based modeling approach to evaluate some of the intricacies of this complex plant-vector-pathogen interaction. We ultimately hope to understand how patterns of vector movement and transfer of spores may account for the spatial distribution and temporal spread of another smut disease within the plant population.
Thus far, we have constructed simple computer simulation models of individual vector movement among a patch of healthy and diseased plants with a spatial structure identical to a particular natural population. Our analysis of spatial structure in this natural population, using Morisita’s index of dispersion and the parameter k of the negative binomial, demonstrates that clumping is greater in the diseased subset of the population than in the total population of healthy and diseased plants. In our simulations, each plant has just one flower. During each time interval in the simulation there is a probability α that a bee will enter the patch. Each simulation begins with a disease-carrying bumble bee entering the patch of flowers, at which point the bee has equal probability of landing on any particular flower. Further movements of the bee differ between simulations of nondiscriminating vectors, which move randomly with respect to disease status of flowers encountered, and those of discriminating vectors, which visit diseased flowers at a lower probability than healthy flowers. Results indicate that a discriminating vector can spread disease as rapidly as the nondiscriminating vector given enough disease initially present in the population. However, where disease was absent prior to the introduction of the first smutted bee, the nondiscriminating vector spread the disease at a higher rate than the discriminating vector.
