Evaluating the ecological realism of plant species distribution models with ecological indicator values

摘要: Species distribution models (SDMs) are routinely applied to assess current as well future species distributions, for example impacts of environmental change on biodiversity or underpin conservation planning. It has been repeatedly emphasized that SDMs should be evaluated based not only their goodness fit the data, but also realism modeled ecological responses. However, possibilities latter hampered by limited knowledge true responses a lack quantitative evaluation methods. Here we compared niche optima obtained from European-scale 1476 terrestrial vascular plant with empirical indicator values indicating preferences key conditions. For each first fitted an ensemble SDM including three modeling techniques (GLM, GAM and BRT) extracted climate, soil, land use nitrogen deposition variables large explanatory power occurrence species. We then these SDM-derived means bivariate correlation analysis. found weak moderate correlations in expected direction between values. The strongest occurred growing degree days temperature. Correlations were weaker more distal relationship (notably precipitation soil moisture). Further, consistently highest BRT, followed GLM GAM. Our method gives insight into projected core habitats can used improve making informed selection techniques.