A comparison of several statistical techniques common in species distribution modeling was developed during this study to evaluate and obtain the statistical model most accurate to predict the distribution of different forest tree species (in our case presence/absence data) according environmental variables. During the process we have developed maximum entropy (MaxEnt), classification and regression trees (CART), multivariate adaptive regression splines (MARS), showing the statistical basis of each model and, at the same time, we have developed a specific additive model to compare and validate their capability. To compare different results, the area under the receiver operating characteristic (ROC) function (AUC) was used. Every AUC value obtained with those models is significant and all of the models could be useful to represent the distribution of each species. Moreover, the additive model with thin plate splines gave the best results. The worst capability was obtained with MARS. This model’s performance was below average for several species. The additive model developed obtained better results because it allowed for changes and calibrations. In this case we were aware of all of the processes that occurred during the modeling. By contrast, models obtained using specific software, in general, perform like “hermetic machines”, because it could sometimes be impossible to understand the stages that led to the final results.
This is an open access article distributed under the Creative Commons Attribution License
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited