Evaluating predictive performance of statistical models explaining wild bee abundance in a mass-flowering crop

摘要: Wild bee populations are threatened by current agricultural practices in many parts of the world, which may put pollination services and crop yields at risk. Loss can potentially be predicted models that link abundances with landscape-scale land-use, but there is little knowledge on degree to these statistical transferable across time space. This study assesses transferability for wild abundance a mass-flowering space (from one region another) year another). The used existing data bumblebee solitary winter oilseed rape fields, together high-resolution land-use crop-cover semi-natural habitats data, from studies conducted five different regions located four countries (Sweden, Germany, Netherlands UK), three years (2011, 2012, 2013). We developed hierarchical model combining all evaluated using cross-validation. found both cover crops permanent habitats, including grasslands forests, important drivers regions. However, while negative effect increasing density pollinators consistent between studies, direction habitat variable studies. limited, especially regions, also time. Our demonstrates limits conjunction widely available classes extrapolating pollinator likely part because input variables such as poorly capture variability resources years. (Less)