The use of statistical models to predict the likely occurrence or distribution of species is becoming an increasingly important tool in conservation planning and wildlife management. Evaluating the predictive performance of models using independent data is a vital step in model development. Such evaluation assists in determining the suitability of a model for specific applications, facilitates comparative assessment of competing models and modelling techniques, and identities aspects of a model most in need of improvement. The predictive performance of habitat models developed using logistic regression needs to be evaluated in terms of two components: reliability or calibration (the agreement between predicted probabilities of occurrence and observed proportions of sites occupied), and discrimination capacity (the ability of a model to correctly distinguish between occupied and unoccupied sites). Lack of reliability can be attributed to two systematic sources, calibration bias and spread. Techniques are described for evaluating both of these sources of error. The discrimination capacity of logistic regression models is often measured by cross-classifying observations and predictions in a two-by-two table, and calculating indices of classification performance. However, this approach relies on the essentially arbitrary choice of a threshold probability to determine whether or not a site is predicted to be occupied. An alternative approach is described which measures discrimination capacity in terms of the area under a relative operating characteristic (ROC) curve relating relative proportions of correctly and incorrectly classified predictions over a wide and continuous range of threshold levels. Wider application of the techniques promoted in this paper could greatly improve understanding of the usefulness, and potential limitations, of habitat models developed for use in conservation planning and wildlife management. (C) 2000 Elsevier Science B.V. All rights reserved.
