Skip to main content
. 2015 Apr 9;10(4):e0122721. doi: 10.1371/journal.pone.0122721

Table 1. Model runs used to assess the potential impact of climate and land cover change on species and genera richness in Madagascar.

Threshold technique Scenario Taxonomic level Mean Model performance (SD) Pixel level loss/gain in richness
AUC COR Kappa Minimum Mean Maximum
T1 Land cover only Species 0.876 (0.05) 0.684 (0.08) 0.646 (0.12) -699 -38.82 213
T3 Land cover only Genera 0.859 (0.04) 0.638 (0.06) 0.599 (0.09) -368 1.10 151
T3 Climate only Species 0.923 (0.04) 0.765 (0.06) 0.731 (0.10) -705 -39.50 199
T2 Climate only Genera 0.907 (0.03) 0.721 (0.05) 0.681 (0.08) -260 19.82 187
T1 Combined Species 0.944 (0.03) 0.803 (0.06) 0.789 (0.10) -791 -27.69 212
T2 Combined Genera 0.927 (0.03) 0.758 (0.05) 0.735 (0.07) -383 18.80 218

Only those thresholds selected for further analyses are shown (See S1 Table for complete table and discussion of model sensitivity to threshold selection). T1 runs were thresholded using the “sensitivity = specificity” rule; T2 runs were thresholded using Youden’s index; and T3 runs were thresholded using the “predicted prevalence = observed prevalence” rule. The mean values (and their associated standard deviations) for AUC, COR and Kappa are shown. Values for the model performance metrics are based on the average of 828 genera and 2186 species. The mean, minimum and maximum pixel level loss/gain in richness is also given.