Ecological niche modeling in Maxent: the importance of model complexity and the performance of model selection criteria.

摘要: Maxent, one of the most commonly used methods for inferring species distributions and environmental tolerances from occurrence data, allows users to fit models arbitrary complexity. Model complexity is typically constrained via a process known as L1 regularization, but at present little guidance available setting appropriate level effects inappropriately complex or simple are largely unknown. In this study, we demonstrate use information criterion approaches regularization in compare selected using criteria other that common literature. We evaluate model performance data generated "true" initial Maxent model, several different metrics quality transferability. show reduced ability infer habitat quality, relative importance variables constraining species' distributions, transferability time periods. also may offer significant advantages over