Exposure analysis of species distribution modeling results

A species distribution model uses a species' observed distribution and biological characteristics to predict its actual (or potential) distribution. The availability of a large amount of environmental data and powerful computing techniques has made the modeling feasible and affordable. A natural tool to use in this context is data mining. The primary goal of data mining is to explore and discover efficient and accurate models. Even though the accuracy is the primary factor in evaluating these modeling tools, generating better predictions is only a stepping stone on the path to identifying better models. The primary goal for scientists is to explore the model space so that they can better understand the phenomena behind the data. In other words, the evaluation and the interpretation of the model are at least as important as building the fitted model itself. In this paper, we provide explore analysis to the species distribution modeling results generated by the Maxent model. Through the process of correlation analysis, Bayesian classification, decision tree induction, and rule induction, the exposure analysis helps an ecologist gain better insights into the relationships between the raw environmental data and the potential species geographic distributions.