Habitat Heterogeneity, Dispersal, and Local Risk of Exposure to Lyme Disease

Spatial heterogeneity presents a fundamental challenge to conventional ecological theory. Although ecological systems are usually heterogeneous, it is not clear how often heterogeneity fundamentally alters their behavior. We addressed this issue with a study of the infection of ticks (Ixodes scapularis) by the causative agent of Lyme disease (the spirochete Borrelia burgdorferi) in multiple habitats within a semirural landscape, combining both field and modeling approaches. We sampled the densities and infection prevalences of ticks in five habitats over two years in southeastern New York. There were consistent differences among habitats in adult infection prevalence, which was unrelated to tick density, suggesting that local habitat features exert some control over local risk of exposure to infected ticks. Other results underscored the importance of processes taking place on larger scales. We observed a positive relationship between the change in tick density within a cohort from the nymphal to adult stages, and changes in prevalence over the same period. Habitats with many adults relative to the number of nymphs several months earlier showed increasing prevalence of B. burgdorferi. Presumably these habitats were receiving immigrating ticks that became infected on their dispersing hosts. We designed a computer model, patterned after the life cycle of I. scapularis, to determine whether patterns observed in the field could be explained by dispersal among habitats differing in host species composition. The model showed that habitat-related variation in tick density and spirochete prevalence was maintained even with moderate dispersal, as long as the different habitats supported distinct assemblages of hosts. Dispersal produced nonlinear or threshold responses under many conditions, due to positive and negative feedbacks. Such feedback is a general feature of many ecological systems, which implies that the behavior of heterogeneous systems will very often be unpredictable from an understanding of isolated components.