Table 1. Results of the Linear Mixed Effect models for the AUC, Kappa, IFK, FGM and DFAC (deviance from average variable contribution).
| Algorithms | AUC | Kappa | IFK | FGM | DFAC |
| Max vs ANN | (+) *** | (+) *** | (+) *** | (+) *** | (−) *** |
| Max vs GAM | ns | ns | ns | ns | (−) *** |
| Max vs GBM | (+) *** | (+) *** | ns | (−) *** | (+) *** |
| Max vs GLM | (+) *** | (+) *** | ns | ns | (−) *** |
| Max vs RF | (+) *** | ns | ns | (−) *** | (+) *** |
| Max vs Con | (−) * | (+) *** | (+) ** | ns | Na |
| ANN vs GAM | (−) *** | (−) *** | (−) *** | (−) *** | (+) *** |
| ANN vs GBM | (−) ** | (−) *** | (−) *** | (−) *** | (+) *** |
| ANN vs GLM | (−) *** | (−) *** | (−) *** | (−) *** | ns |
| ANN vs RF | (−) * | (−) *** | (−) *** | (−) *** | (+) *** |
| ANN vs Con | (−) *** | (−) *** | (−) *** | (−) *** | na |
| GAM vs GBM | (+) *** | ns | ns | (−) *** | (+) *** |
| GAM vs GLM | (+) * | ns | ns | ns | (−) *** |
| GAM vs RF | (+) *** | ns | ns | (−) *** | (+) *** |
| GAM vs Con | (−) *** | (+) * | ns | ns | na |
| GBM vs GLM | ns | ns | ns | (+) *** | (−) *** |
| GBM vs RF | ns | ns | ns | (−) ** | (+) *** |
| GBM vs Con | (−) *** | ns | ns | (+) *** | na |
| GLM vs RF | ns | ns | ns | (−) *** | (+) *** |
| GLM vs Con | (−) *** | ns | ns | ns | na |
| RF vs Con | (−) *** | (+) * | ns | (+) *** | na |
| Max vs Records | (−) *** | ns | ns | (−) ** | na |
| ANN vs Records | ns | (+) * | (+) * | ns | na |
| GAM vs Records | (−) *** | ns | ns | (−) ** | na |
| GBM vs Records | (−) * | ns | ns | (−) *** | na |
| GLM vs Records | (−) ** | ns | ns | (−) ** | na |
| RF vs Records | (−) * | ns | ns | (−) *** | na |
| Con vs Records | (−) *** | ns | ns | (−) *** | na |
| Max vs Distribution | (−) * | ns | ns | (−) * | na |
| ANN vs Distribution | ns | (+) *** | (+) ** | (+) * | na |
| GAM vs Distribution | ns | ns | ns | (−) ** | na |
| GBM vs Distribution | ns | ns | ns | ns | na |
| GLM vs Distribution | ns | ns | ns | (−) * | na |
| RF vs Distribution | ns | ns | ns | ns | na |
| Con vs Distribution | ns | ns | ns | ns | na |
The significance of the pairwise algorithms comparisons, their interaction with the number of records and spatial distribution is presented. The positive and negative signs apply for the first algorithm being compared against the second. For the first four measures the positive sign points to algorithms that render higher values -better fits and maps similarities. In the DFAC, the negative signs point to a more consistent algorithm as it renders lower deviances than the second. Max = Maxent, Con = Consensus approach; ns = no significant; na = not applicable. Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05. Corrected Tukey’s P values reported.