Authors: April Bledsoe*, , T. Andrew Joyner, East Tennessee State University, Ingrid Luffman, East Tennessee State University, Jim Mead, The Mammoth Site
Topics: Biogeography, Spatial Analysis & Modeling, Animal Geographies
Keywords: ecological niche modeling, prairie dogs, bioclimatic variables
Session Type: Poster
Presentation File: No File Uploaded
Ecological niche models (ENMs) were constructed from commonly used bioclimatic variables and non-traditional variables (e.g., ecoregions and other ecological delineations) for two species of prairie dog, C. leucurus and C. ludovicianus. Commonly used bioclimatic variables generated training an Area Under the Curve of 0.942 for C. ludovicianus, and of 0.983 for C. leucurus. Non-traditional environmental variables generated training AUCs of 0.944 for C. ludovicianus, and of 0.984 for C. leucurus. Results included smaller differences between testing and training AUCs for non-traditional environmental variables which may indicate that those variables may explain species presence better than the commonly used bioclimatic variables. Perhaps, ENM Relative Occurrence Rates constructed from a combination of the best performing non-traditional and bioclimatic variables, as indicated by preliminary model jackknife results, would be highly informative. Comparison of jackknife results on model AUC values indicated that the best preforming variable in the common-variable models outperforms the best performing variable in the non-traditional variable models for generating higher model AUC values. Results may justify the inclusion of Environmental Protection Agency and The Nature Conservancy Ecoregions in future ENMs as these variables contributed the most to model AUC values. As environmental variables, ecoregions are constructed from decades of ecological research. Inclusion of a combination of commonly used and non-traditional environmental variables that are highly indicative of the given species’ ecosystems and are biologically informed, may provide researchers with more informative current ENMs.