Accounting for spatial interactions in the upscaling of ecosystem services

Abstract Maps of ecosystem service (ES) supply are frequently used to guide spatial planning, policymaking and ecosystem management. However, these are typically based upon coarse land‐cover proxies. This approach lacks a strong mechanistic basis and neglects spatial biodiversity dynamics and interactions among landscape properties that can modify ES provision. We present an analytical framework for ES upscaling that incorporates spatial interactions between landscape properties, which determine ES supply. The resulting models can be viewed as a spatially informed ES production function. Key aspects of our synthetic framework include (i) the systematic assessment of multiple drivers across many levels of abiotic and biotic organization to formulate the statistical ES production function, (ii) the inclusion of spatial interactions with the surrounding environment into the ES production function and (iii) the use of expert input to inform ES production functions. We demonstrate the approach using two example ES from German grasslands: biodiversity conservation and water supply. We show that the inclusion of spatial interactions in the upscaling model improved model predictions by 15%–33%, depending on the ES evaluated. In addition, inclusion of spatial interactions led to reduced error associated with the upscaled estimates. By overcoming several shortcomings of existing upscaling approaches, we generate maps of ES supply that can more reliably inform spatial planning. Further, the underlying models allow for simulation of changes in the drivers of ES supply and estimation of respective outcomes. These advantages have the potential to better link detailed local‐scale ecological understanding and land management with large‐scale ES supply mapping, and thus better inform decision making.