Performance of Maxent models in predicting climatic suitability in modern (m) or historical (h) native ranges.

Note: Models were tested using random pseudo-presence data that was generated separately from training data. The AUC values were averaged over 10 runs for each species/time period. Kappa statistics were calculated from cumulative MTP and MTSS thresholded model outputs and a set of separately generated random pseudo-presence and pseudo-absence points. *Thesholded Maxent predictions generated using modern range training data were evaluated using test files that corresponded with historical ranges.