ReviewPathways to bridge the biophysical realism gap in ecosystem services mapping approaches

•Ecosystem service models focussing on biotic contribution.•Data sources for five types of ecosystem service models.•Reducing uncertainty in ecosystem service maps by approaches fitted to user resources 3.

The mapping of ecosystem service supply has become quite common in ecosystem service assessment practice for terrestrial ecosystems, but land cover remains the most common indicator for ecosystems ability to deliver ecosystem services. For marine ecosystems, practice is even less advanced, with a clear deficit in spatially-explicit assessments of ecosystem service supply. This situation, which generates considerable uncertainty in the assessment of ecosystems' ability to support current and future human well-being, contrasts with increasing understanding of the role of terrestrial and marine biodiversity for ecosystem functioning and thereby for ecosystem services. This paper provides a synthesis of available approaches, models and tools, and data sources, that are able to better link ecosystem service mapping to current understanding of the role of ecosystem service providing organisms and land/seascape structure in ecosystem functioning. Based on a review of literature, models and associated geo-referenced metrics are classified according to the way in which land or marine use, ecological processes and especially biodiversity effects are represented. We distinguish five types of models: proxy-based, phenomenological, niche-based, trait-based and full-process. Examples from each model type are presented and data requirements considered. Our synthesis demonstrates that the current understanding of the role of biota in ecosystem services can effectively be incorporated into mapping approaches and opens avenues for further model development using hybrid approaches tailored to available resources. We end by discussing ways to resolve sources of uncertainty associated with model representation of biotic processes and with data availability.