The spatial level of analysis affects the patterns of forest ecosystem services supply and their relationships.

The implementation of the Ecosystem Services (ES) framework (including supply and demand) should be based on accurate spatial assessments to make it useful for land planning or environmental management. Despite the inherent dependence of ES assessments on the spatial resolution at which they are conducted, the studies analyzing these effects on ES supply and their relationships are still scarce. To study the influence of the spatial level of analysis on ES patterns and on the relationships among different ES, we selected seven indicators representing ES supply and three variables that describe forest cover and biodiversity for Catalonia (NE Iberian Peninsula). These indicators were estimated at three different scales: local, municipality and county. Our results showed differences in the ES patterns among the levels of analysis. The higher levels (municipality/county) removed part of the local heterogeneity of the patterns observed at the local scale, particularly for ES indicators characterized by a finely grained, scattered distribution. The relationships between ES indicators were generally similar at the three levels. However, some negative relationships (potential trade-offs) that were detected at the local level changed to positive (and significant) relationships at municipality and county. Spatial autocorrelation showed similarities between patterns at local and municipality levels, but differences with county level. We conclude that the use of high-resolution spatial data is preferable whenever available, in particular when identifying hotspots or trade-offs/synergies is of primary interest. When the main objective is describing broad patterns of ES, intermediate levels (e.g., municipality) are also adequate, as they conserve many of the properties of assessments conducted at finer scales, allowing the integration of data sources and, usually, being more directly relevant for policy-making. In conclusion, our results warn against the uncritical use of coarse (aggregated) spatial ES data and indicators in strategies for land use planning and forest conservation.