Accounting for spatial interactions in the upscaling of ecosystem services

O_LIMaps of ecosystem service (ES) supply are frequently used to guide spatial planning, policy making, 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. C_LIO_LIWe present an analytical framework for ES upscaling that incorporates spatial interactions between landscape properties to determine ES supply. The resulting models can be viewed as a spatially informed ES production function. The approach comprises seven steps that include several elements absent from most existing approaches, notably a procedure for identifying geodata variables that represent the true mechanistic drivers, the inclusion of spatial interactions in the upscaling model, and modification following expert feedback on the selected model. C_LIO_LIWe 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 from 15% to 33% depending on the ES evaluated. In addition, inclusion of spatial interactions led to reduced error associated with the upscaled estimates. C_LIO_LIBy overcoming several shortcomings of existing, upscaling approaches we generate resulting 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 and spatial planning. C_LI