Adrienne Grêt-Regamey, Bettina Weibel, Kenneth J. Bagstad, Marika Ferrari, Davide Geneletti, Hermann Klug, Uta Schirpke, Ulrike Tappeiner

Ecosystems provide life-sustaining services upon which human civilization depends, but their degradation largely continues unabated. Spatially explicit information on ecosystem services (ES) provision is required to better guide decision making, particularly for mountain systems, which are characterized by vertical gradients and isolation with high topographic complexity, making them particularly sensitive to global change. But while spatially explicit ES quantification and valuation allows the identification of areas of abundant or limited supply of and demand for ES, the accuracy and usefulness of the information varies considerably depending on the scale and methods used. Using four case studies from mountainous regions in Europe and the U.S., we quantify information gains and losses when mapping five ES - carbon sequestration, flood regulation, agricultural production, timber harvest, and scenic beauty - at coarse and fine resolution (250 m vs. 25 m in Europe and 300 m vs. 30 m in the U.S.). We analyze the effects of scale on ES estimates and their spatial pattern and show how these effects are related to different ES, terrain structure and model properties. ES estimates differ substantially between the fine and coarse resolution analyses in all case studies and across all services. This scale effect is not equally strong for all ES. We show that spatially explicit information about non-clustered, isolated ES tends to be lost at coarse resolution and against expectation, mainly in less rugged terrain, which calls for finer resolution assessments in such contexts. The effect of terrain ruggedness is also related to model properties such as dependency on land use-land cover data. We close with recommendations for mapping ES to make the resulting maps more comparable, and suggest a four-step approach to address the issue of scale when mapping ES that can deliver information to support ES-based decision making with greater accuracy and reliability.