Predicting amphibian and reptile distributions and species richness across urban parks in the eastern United States

Urban parks and greenspaces protect biodiversity in cities and provide important opportunities for city residents to interact with wildlife. We used a large community-science database to examine the predictors of amphibian and reptile species richness across urban areas in the eastern United States. Overall, we found that park area, wetland habitat diversity, and park connectivity were related to amphibian and reptile species richness, reducing prediction error by around 50% compared to a null model. Additionally, forest cover within parks was a predictor of amphibian richness, but not a predictor for reptiles. The richness of uncommon species (fewer than 50 observations in our sample) was related only to park area and connectivity. Our model was effective at predicting herp species richness at parks in "new" urban areas (i.e. those not included in the original model fitting) for some locations (e.g., Philadelphia, Indianapolis, Manhattan) but also tended to under-estimate the most species-rich parks in some places (e.g., Washington DC and Pittsburgh) and over-estimate richness in others (e.g., Staten Island). For individual species, presence in parks tended to follow expected habitat associations, with salamanders mostly responding to forest cover, and frogs and turtles primarily responding to park area and wetland diversity. Our results demonstrate that large urban parks with can harbor high species richness, particularly those with a high diversity of aquatic and terrestrial habitats. However, these patterns are only moderately predictable, and individual parks may have higher or lower conservation value value than might be expected from simple models.