A retrospective approach for evaluating ecological niche modeling transferences over time: The case of Mexican endemic rodents
Abstract
Ecological niche modeling (ENM) is an approach to infer the suitable conditions for species’ persistence and their potential geographic distributions. ENM is extensively used to assess the potential effects of climate change on species’ distributions, although modeling algorithms are acknowledged as an important source of uncertainty in climate change projections. A problem is that these models cannot be properly assessed in the context of future projections. In this study, we evaluated the performance of seven popular modeling algorithms (Bioclim, Generalized Additive Models (GAM), Generalized Linear Models (GLM), Boosted Regression Trees (BRT), Maxent, Random Forest (RF), and Support Vector Machine (SVM)) for transferring ENM over time for Mexican endemic rodents. To do so, we followed a retrospective approach by transferring models from the near past (1950-1979) to the present (1980-2009) and vice versa. We found three important results: (1) The quality of input data and the algorithm had a significant effect on model output and performance; (2) algorithm performance was different for transferring models to the future than to the past; and (3) the most robust algorithms were RF, BRT and Maxent, whereas Bioclim was the least consistent. In conclusion, algorithm choice is critical for transferring ENM over time. Since no algorithm performed consistently better than the rest, we recommend to test different algorithms prior transferring models to future scenarios under a retrospective approach.