Effects of wildflower strips, landscape structure and agricultural practices on wild bee assemblages - A matter of data resolution and spatial scale?

Agricultural systems in Central Europe were redesigned during the last century to attain maximum yields. The results often lead to homogeneous landscapes with only few structures of ecological value and have concurrently exacerbated habitat fragmentation. Perennial wildflower strips have become a significant agri-environmental measure (AEM) to counteract the ecological consequences for wild bees and other pollinators in agricultural landscapes. The effectiveness of AEMs depends on the landscape context, but information about geodata sources and spatial scales relevant for the analysis of landscape effects on wild bees is lacking. This study uses data from various sources on land cover and agricultural practices to assess their applicability in an evaluation of perennial wildflower strips as AEM for wild bees in Saxony-Anhalt, Germany. We investigated the relationships of wild bee species diversity and abundance to the landscape context at spatial scales from 200 m to 10 km considering several factors: land cover/land use, protected areas, crop types, agri-environment schemes/greening, intensity of agriculture, and intensity of grassland farming. In general, our results revealed that landscape effects were more relevant for solitary than social wild bees on flower strips, pointing to a higher limitation of solitary wild bees in nesting resources as compared to social wild bees. Numbers of wild bee species and individuals benefitted from bare soil and ecological focus areas in the surroundings up to 3 km distance, whereas the share of Red List solitary bee species was positively influenced by a variety of factors (e.g., wood structures and grasslands) especially at large scales up to 10 km. The comparison of models based on different land cover data sources showed that the lack of geodata resolution can mask landscape effects on wild bees. Altogether, our results suggest a high potential of data from the Basic Digital Landscape Model (DLM), together with the Integrated Administration and Control System (IACS), to indicate effects of landscape structures and agricultural practices on the species composition and distribution of wild bee assemblages in Germany.