AMPHIBIAN PARASITES EXHIBIT IDIOSYNCRATIC RELATIONSHIPS WITH SPATIOTEMPORAL ENVIRONMENTAL AND HOST-COMMUNITY VARIATION.

Why parasites occur in certain hosts in certain locations has been a long-standing question among ecological and evolutionary parasitologists. Encounter and compatibility filters summarize the likelihood that a host and parasite will physically interact and establish an infection upon contact. Encounter and compatibility filters are not fixed and, among multiple locations, the abiotic environmental characteristics and biotic community composition that contribute to the filters often vary spatially and temporally. Abiotic variation may directly affect hosts or parasites-particularly parasites with 1 or more free-living stages-whereas the local biotic community may dilute or amplify parasite transmission. Unlike directly transmitted parasites, complex-life cycle parasites use multiple hosts, thus having life cycles that, we hypothesize, are highly susceptible to the effects of spatiotemporal environmental variation. We modeled infection probability relationships of endohelminths from post-metamorphic wood frogs (Rana [Lithobates] sylvatica) and northern leopard frogs (Rana pipiens) with wetland characteristics, landscape composition, and the anuran species within the local community. Parasites included complex-life cycle trematodes that use amphibians as definitive hosts (Haematoloechus spp., Glypthelmins quieta) or as intermediate hosts (Alaria sp., Neodiplostomum sp., echinostomatids, and Lechriorchis) and nematodes with direct or indirect life cycles (Cosmocercoides and Oswaldocruzia). Although our results demonstrate that distributions of parasites with complex and direct life cycles are correlated with some abiotic and biotic characteristics of the environment, there were few general trends. Each parasite's distribution had its own unique relationship with wetland, landscape, and amphibian-community variables, and there was overall low predictability for most species. One landscape feature-the number of wetlands within the vicinity of the site of amphibian capture-was commonly included in top models for leopard frogs and could be associated with how definitive hosts (e.g., amphibians, mammals, and birds) and intermediate hosts (e.g., snails and odonates) use the landscape. The amphibian community at any given site also commonly affected infection probabilities, such that the local presence of other species tended to reduce infection probabilities in sampled frogs, lending support to the dilution effect at the landscape level. Our research highlights the need to consider spatiotemporal sampling, environmental variation, and host-community variation when studying parasite prevalence in any given component community.