Stream and Riparian Habitat Use by Anurans along a Forested Gradient in Western Georgia, USA
Urban development is known to alter the structure, chemistry, and biota associated with stream systems; however, little is known about the dynamics of anurans that breed in and around streams. We used automated devices to record breeding anurans for one year across a forested gradient in an effort to identify species-specific sensitivities to urbanization. Six of 13 total species found during the study were present in surveyed streams that were representative of a gradient ranging from heavily urbanized to largely forested, and maximum species richness occurred in watersheds with rapid urban development but low values of impervious surfaces such as pavement and roof tops. The best landscape-scale predictor of assemblage metrics such as species richness, total species activity, and species diversity was generally percent forest cover at the watershed scale (or in very large buffers around the sample point). We used Bayesian inference to estimate detection probability and occupancy for 11 species. While detection probability varied across sampling occasions, there was no evidence that occupancy was a function of forest loss for any species. Urban streams and riparian areas are often severely altered when compared to similar habitats in forested areas. Anuran richness and diversity from urban areas was not altered to the same extent as caudates, which have been sampled from these same streams in previous studies. Increased vagility of anurans, coupled with different breeding strategies, may help to explain this discrepancy.
Location of study sites used to monitor anuran calling activity. The white counties on the inset map of Georgia represent the region depicted in the main map. Local stream names, accompanying the stream codes provided here, are given in Table 1.
Rarefaction curves and associated standard errors for (left) species richness and (right) Shannon's index as calculated from frog call sampling across three land use categories (n = 3 streams for each category) in western Georgia, USA.
Estimated mean occupancy and detection probabilities (with associated 95% confidence intervals) for the 11 species used in multi-species occupancy models. Species are ordered by increasing width of 95% CI around occupancy probability. Four species, Bird-voiced Treefrogs (Hyla avivoca), Upland Chorus Frogs (Pseudacris feriarum), Barking Treefrogs (Hyla gratiosa), and Cricket Frogs (Acris sp.), did not have sufficient data to estimate occupancy probability. Species codes are defined in Table 2.
Contributor Notes
Associate Editor: J. W. Snodgrass.