Predicting plant species distributions using climate?based model ensembles with corresponding measures of congruence and uncertainty

Aim The increasing availability of regional and global climate data presents an opportunity to build better ecological models; however, it is not always clear which dataset most appropriate. aim this study was understand the impacts that alternative datasets have on modelled distribution plant species, develop systematic approaches enhancing their use in species models (SDMs). Location Victoria, southeast Australia Himalayan Kingdom Bhutan. Methods We compared statistical performance SDMs for 38 Victoria 12 Bhutan with multiple algorithms using globally regionally calibrated datasets. Individual were against one another as SDM ensembles explore potential predictions improve performance. two new spatially continuous metrics support interpretation ensemble by characterizing per-pixel congruence variability contributing models. Results There no consensus performed best across all either region. On average, multi-model (across same different data) increased AUC/TSS/Kappa/OA up 0.02/0.03/0.03/0.02 0.06/0.11/0.11/0.05 Ensembles than single both (AUC = 85%; TSS 68%) 86%; 69%). fitted and/or each provided a significant improvement over model runs. Main conclusions Our results demonstrate ensembles, built variables, can quantify identify regions highest uncertainty while mitigating risk erroneous predictions. Algorithm selection known be large source error SDMs, our comparably uncertainty.