The Influence of Pesticide Use on Amphibian Chytrid Fungal Infections Varies with Host Life Stage across Broad Spatial Scales

Aim: Widespread pesticides might influence pathogen distributions across landscapes via effects on host–pathogen interactions. Empirical research supports two hypotheses regarding effects of pesticides on amphibians and the aquatic fungal pathogen Batrachochytrium dendrobatidis (Bd): (a) pesticides can be toxic to Bd, reducing infection risk of aquatic larval amphibians; and, (b) exposure to pesticides early in amphibian development can increase disease risk after metamorphosis. The aim of this research is to evaluate whether these patterns from laboratory experiments are consistent across host species and occur in the field at broad spatial scales. Location: Contiguous U.S.A. Time Period: 1998–2009. Major Taxa Studied: Amphibian hosts and Bd. Methods: Our data included 3,946 individuals evaluated for Bd infection across 49 amphibian species, which resulted in 155 estimates of Bd prevalence in populations. We used multimodel inference to examine associations between Bd infection prevalence in amphibian populations and (a) total pesticide use, (b) pesticide use by type (herbicide, insecticide or fungicide), and (c) the most commonly used pesticide compounds across life stages, controlling for several factors documented to affect the distribution of Bd. Results: Consistent with laboratory findings, our results indicate that use of multiple herbicide compounds is most closely associated with low infection risk in the aquatic larval stage but high risk in the terrestrial post‐metamorphic stage when controlling for insecticide and fungicide use. We caution against assuming that insecticides and fungicides do not influence Bd distributions, because use of herbicides, insecticides and fungicides are all correlated positively. Main Conclusions: The effects that pesticides can have on disease distributions are complex and should be considered strongly at broad scales and across host species, especially in environments in which use and exposure are widespread. Accurate predictions of disease distributions may lead to more effective management strategies to limit disease spread