Evaluating Habitat-Association Models for the Saltmarsh Sparrow

Abstract

Habitat models that associate organisms with features of their environment can help identify areas for planning conservation strategies. These models, however, should be tested with new data before their conclusions are widely accepted. The Saltmarsh Sparrow (Ammodramus caudacutus) is a species of growing conservation concern along the Atlantic coast of North America. In a previous study, we developed models for Saltmarsh Sparrow presence and nesting habitat. Sparrow occupancy was best predicted from raw reflective properties of marshes derived from remote sensing, whereas sparrow nesting was best predicted from a classification of marsh plant communities. To test these models, we surveyed a stratified random sample of sites for which the probability of sparrow presence had been predicted and compared the observations to the predicted probability of presence and nesting generated from the models. The models' performance, assessed by the area under a receiver-operating characteristic curve and the deviance of observations from the models' predictions, was significantly better than expected by chance alone. Because sparrows are predicted to occur in many places where they are unlikely to nest, monitoring sparrow presence alone will not identify areas important for their nesting.