Evaluation of the accuracy of land-use based ecosystem service assessments for different thematic resolutions

The demand for pragmatic tools for mapping ecosystem services (ES) has led to the widespread application of land-use based proxy methods, mostly using coarse thematic resolution classification systems. Although various studies have demonstrated the limited reliability of land use as an indicator of service delivery, this does not prevent the method from being frequently applied on different institutional levels. It has recently been argued that a more detailed land use classification system may increase the accuracy of this approach. This research statistically compares maps of predicted ES delivery based on land use scoring for three different thematic resolutions (number of classes) with maps of ES delivery produced by biophysical models. Our results demonstrate that using a more detailed land use classification system does not significantly increase the accuracy of land-use based ES assessments for the majority of the considered ES. Correlations between land-use based assessments and biophysical model outcomes are relatively strong for provisioning services, independent of the classification system. However, large discrepancies occur frequently between the score and the model-based estimate. We conclude that land use, as a simple indicator, is not effective enough to be used in environmental management as it cannot capture differences in abiotic conditions and ecological processes that explain differences in service delivery. Using land use as a simple indicator will therefore result in inappropriate management decisions, even if a highly detailed land use classification system is used.