Landscape influence on parasite distribution in wood frogs (Rana sylvatica) in the prairie pothole region of North Dakota

Frog populations are spatially structured due to their dependence on water.  Parasites using frogs as a resource may have similar spatial structure to the extent that their dispersal and survival depend on the frog host versus factors that affect other hosts used in their life cycles.  In this study we sought to understand how landscape and habitat characteristics influence the spatial distribution of helminth parasites in wood frogs.  We sampled up to five male frogs per pond each year for a total of 444 frogs from 59 ponds during the springs of 2004 - 2006.  Using aerial photographs and GIS we estimated landscape characteristics (wetlands, agriculture, grass, trees, roads, highways, and building sites) at several spatial scales (50, 100, 200, 300, 400 and 500 meters) with the wetland as the focal point.  We used logistic regression and model selection to determine which variables provided the best prediction of parasite occurrence. We also considered the problem of understanding spatial distribution when limited sampling might fail to detect species presence.  We identified 11 species of helminth parasites, including one new species of lung nematode (Rhabdias bakeri).  The majority of the parasites were trematodes, which use the wood frog as an intermediate host.  Nine of the parasites occurred at ≥10% of locations.  Two parasites, including the directly-transmitted Rh. bakeri, were nearly ubiquitous, indicating that their distribution was not limited on this landscape.  For other species, wetlands, proximity of trees, grass, and roads were the most common predictors of parasite occurrence on the landscape, although scale and direction of impact varied with the parasite.  Analyses that simultaneously estimated occupancy and detectability resulted in different models than analyses not including detectability.  Results suggest that multiple spatial scales are required to understand the relationship between parasite distribution and landscape features, and accounting for detectability is crucial.