Overprediction of species distribution models in conservation planning: A still neglected issue with strong effects

Abstract

Species distribution models (SDM) are increasingly used in conservation planning to identify priority areas for the establishment of protected areas. Nevertheless, the quality of SDM varies widely and may compromise the effectiveness of protected areas. Here we reviewed whether SDM overprediction is considered in spatial conservation prioritization exercises and evaluated how model overprediction influences the effectiveness and the spatial arrangement of priority areas. To do so, we carried out a systematic review to analyze how researchers have handled SDM overprediction when identifying priority areas for conservation. To show how spatial conservation prioritization outcomes are affected by SDM overprediction, we used SDM of native palm at three geographic scales (Neotropics, Amazon ecoregion, and Ecuadorian Amazon). We found that only 10% of the evaluated manuscripts accounted for model overprediction. Our spatial conservation prioritization based on SDM with overprediction conferred high priority rank values in a region where species do not occur, underestimated the efficiency of selected priority areas, and over or underestimated the efficiency of current protected areas. Such effects were lower at smaller geographic extents. Our findings highlight the importance of improving future spatial conservation prioritization studies through the correction of SDM overprediction, resulting in the detection of more adequate areas for species conservation, especially at broader extents.