COS 16-7
Landscape predictors of pathogen prevalence in bumble bees

Monday, August 10, 2015: 3:40 PM
339, Baltimore Convention Center
Scott H. McArt, Entomology, Cornell University, Ithaca, NY
Christine Urbanowicz, Biology, Dartmouth College
Rebecca Irwin, Biology, Dartmouth College
Lynn Adler, Biology, UMass-Amherst
Background/Question/Methods

Bumble bees are important pollinators in both wild and managed systems. In recent years, it has become evident that widespread declines of several Bombus species are occurring. Furthermore, these declines are associated with infection by pathogens, including the microsporidian Nosema bombi. The mechanisms responsible for associations between pathogen infection and bee declines are not clear, however. Specifically, we know little regarding how ecological variables are associated with disease prevalence within and among bee species. Here, we utilize the broadest survey to date on Bombus infection patterns in North America (Cameron et al. 2011 PNAS, Cordes et al. 2012 J Invert Path) to assess how landscape variables are related to the prevalence of Crithidia bombi and Nosema bombi within and among 34 Bombusspecies. The dataset is comprised of 16,788 individual bees, sampled from 382 locations between 2007-2009.

Results/Conclusions

We quantified the area of cultivated crops, percent agricultural area, percent forest area, area of developed land, area of natural land, forest fragmentation, natural area fragmentation, and human population density at 0.5, 1.0, 2.5, and 5.0 km radii surrounding each of 382 sampling locations using GIS. In addition, we used the USGS national pesticide synthesis database to predict levels of several pesticides applied within each radius at each sampling location. In general, we found that intermediate radii provided the best fit between landscape variables and infection with both pathogens. Several landscape variables and predicted pesticide inputs were associated with disease prevalence in bees, including percent agricultural area, percent forest area, natural area fragmentation, and human population density. In addition, several interactions between bee species and landscape variables exist, suggesting different bee species may respond differently to changing land use patterns in terms of their infection dynamics.  Overall, our results show that landscape factors can be an important determinant of pathogen infection in bumble bees. Given recent pollinator declines caused in part by pathogens, human-driven land use change is an important consideration with regards to pollinator health.