Quantification of ecosystem services (ES) - the benefits that people obtain from the environment - can be valuable for managing multifunctional landscapes. Proxy measures and models are often used to estimate ES in the absence of direct measurements, but projections based on composition of different landcover types have been shown to be poor representations of the true distribution of service provision. The use of estimates derived from composition neglects the critical role of landscape configuration (spatial arrangement of landcover types) in the flow and quality of benefits. A set of landscape metrics characterizing aspects of both landscape configuration (describing landscape shape, position, and connectivity) and composition were calculated for 136 municipalities in two watersheds in Southern Quebec. Through comparison with measurements of 10 ES in the same municipalities, structural equation models were used to combine statistical relationships with hypothesized causal pathways and determine which mechanisms and interactions were responsible for observed relationships.
Results/Conclusions
Nine models included a causal path from the latent variable shape to measured service provision, eight models included paths from each of composition and position to measured service provision, and five models included a path from connectivity to measured service provision. Models for five services included paths from composition to one or more of the three latent configuration variables, which generated additional indirect paths from composition to service provision. For all categories of ES, the aggregate effects of landscape configuration were found to exceed that of composition. The proportion of all model-explained variation that was attributable to configuration was 0.60±0.39 (mean ± standard deviation) for regulating services, 0.55±0.31 for cultural services, and 0.72±0.27 for provisioning services. Composition emerged as the dominant characteristic for only three of ten individual services (carbon sequestration, tourism, and maple syrup production). Many multi-service models of ecosystem service provision rely heavily on information about the amount of each landcover type present in the landscape. The higher relative influence of configuration characteristics of landscape structure on measured ES provision suggests that benefits will not be well represented by models based on landcover composition.