Table 2. Model evaluation of the nine species distribution methods (see the methods section for abbreviations) and their ensemble prediction (EP).

Model	AUC	TSS	BI
ANN	0.917 ± 0.006*	0.807 ± 0.071*	0.802 ± 0.032*
BRT	0.972 ± 0.027*	0.891 ± 0.043*	0.873 ± 0.041*
CTA	0.915 ± 0.023*	0.849 ± 0.057*	0.811 ± 0.022*
FDA	0.905 ± 0.025*	0.803 ± 0.002*	0.918 ± 0.014*
GAM	0.947 ± 0.021*	0.864 ± 0.077*	0.873 ± 0.041*
GLM	0.911 ± 0.086*	0.805 ± 0.011*	0.982 ± 0.017*
MARS	0.904 ± 0.066*	0.801 ± 0.021*	0.909 ± 0.057*
MAXENT	0.942 ± 0.056*	0.865 ± 0.082*	0.964 ± 0.035*
RF	0.998 ± 0.002*	0.989 ± 0.011*	0.804 ± 0.088*
EP	0.951 ± 0.048*	0.902 ± 0.022*	0.981 ± 0.019*

Area Under the Curve (AUC) ranges between 0 and 1 (worse than a random model and best discriminating model, respectively). True Skill Statistic (TSS) and Boyce’s Index (BI) ranges between −1 and 1 (higher values indicate a good predictive accuracy, while 0 indicates random prediction). Average values ± standard deviations are shown (*: P < 0.001).