Identifying habitat variables important to the rare Columbia spotted frog in utah (USA): An information-theoretic approach

We identified habitat associations of the rare Columbia spotted frog ( Rana luteiventris). We used an information-theoretic approach that encouraged careful consideration of previous studies, demanded a priori formulation of hypotheses and models, and provided interpretable results while avoiding some criticisms of traditional statistical analyses. We formulated hypotheses about habitat associations based on conductivity, emergent vegetation, littoral zone depth, pond hydrology, and water temperature. We modeled hypothesized associations with logistic regression and used Akaike's information criterion to quantify evidence for models, weigh the relative importance of each habitat variable, and select predictive models. Although variable, results suggested that spotted frogs are more likely to occur in ponds that do not shrink in size seasonally, maintain relatively constant seasonal water temperature, and have high emergent vegetation cover. Associations we identified will assist actions in Utah that may be critical to local persistence of spotted frogs. Moreover, this approach for identifying associations has great potential for other rare species.