Historical changes in the importance of climate and land use as determinants of Dutch pollinator distributions

Aim: Species distribution models are often used to project species distributions to different environmental conditions. However, most models do not consider whether the importance of abiotic factors may change over time. If they change, this has implications for the assessment of how abiotic changes affect species distributions. Here, we use spatially explicit historical data on species occurrences, climate and land use to test whether the importance of different climatic and land-use drivers as determinants of species distributions has remained constant over a period of > 60 years (1951–2014). Location: The Netherlands. Methods: Using species distribution models and a comprehensive country-wide dataset at 5 × 5 km resolution, we modelled the distribution of a total of 398 pollinator species (bees, butterflies and hoverflies) for three periods (1951–1970, 1971–1990 and 1998–2014). We then evaluated whether the importance of variables related to climate (precipitation, temperature) and land use (landscape composition and habitat fragmentation) in determining pollinator distributions has changed over time. Results: Variables related to landscape composition were highly important in determining pollinator distributions in all periods. Precipitation was generally less important than temperature, and habitat fragmentation less than landscape composition. Land-use variables remained equally important across time for all pollinator groups, except for bees where the importance of habitat fragmentation decreased significantly over time. Among climate variables, the importance of precipitation did not change across time for any pollinator group. However, the importance of temperature increased significantly in recent times for bees and hoverflies. Main conclusions: Determinants of species distributions can change in their importance over time when changes in the magnitude and range of environmental conditions occur. Given future temperature rises, our results imply that species distribution models calibrated with current climatic conditions may not adequately predict the future importance of environmental factors in driving species distributions