Habitat Associations and Predictive Presence Modeling for Plethodontid Eurycea hillisi in West Georgia

Abstract

The purpose of this study was to identify habitat associations of Eurycea hillisi and use those characteristics to develop a predictive model. In Chapter 1, I gave an overview of the taxonomic history of the species and what general habitat usage has been observed. During preliminary data collection in stream floodplains, I consistently observed salamanders along one side of a given stream at several sites, rather than both sides. These “detected” and “undetected” stream sides served as a basis of comparison of species presence throughout the thesis. In Chapter 2, I sampled sites during winter and summer seasons using a transect-based protocol within detected/undetected stream sides and used linear mixed models to determine the difference in mean values between the two sides. I found that habitat characteristics associated with the detected side tended to have association with moisture related vegetation. Soil moisture content exhibited a shift in associations between the seasons, likely due to flood potential and soil composition. I took the significant variables and ran a Non-metric Multidimensional Scaling (NMDS) analysis to demonstrate the similarities and dissimilarities of the habitat variables and the strength of the significance of each variable as they related to detected/undetected. In Chapter 3, I used the significant variables to develop a predictive model using generalized linear mixed models for both winter and summer seasons. I discovered that the top winter model contained top of bank height, leaf litter depth, percent woody debris, percent sphagnum cover, and soil moisture content. The top two summer models were not significantly different from one another and both contained bank angle, percent woody debris cover, percent fern cover, soil moisture content, and soil pH. Percent sphagnum cover was also present in the second top model.