As governments, business, and lending institutions are increasingly considering investments in natural capital as one strategy to meet their operational goals and society’s demands for sustainable development, the importance of accurate, accessible information on ecosystem services for use in decision-making has never been greater. However, most ecosystem services models and decision support tools are based on categorical representation of land-use and land-cover (LULC), with the assumption that all habitat within each LULC type is identical, which poses challenges for both the accuracy and accessibility of the information. Linking biodiversity and ecosystem services is also an important step forward for support of decisions on conservation and sustainable development, as currently even diversity-driven services like pollination and wildlife-based tourism use land-use as a surrogate instead of species richness or ecosystem functional diversity. Here we present the integration of Essential Biodiversity Variables (EBVs) derived from satellite information to improve the accuracy, accessibility, and relevance of ecosystem service modeling. We model EBVs for forest, wildlife, and pollinators, and use these as inputs for ecosystem service models for carbon storage, sediment retention, wildlife-based tourism and pollination in Costa Rica, where national accounting for ecosystem services is being piloted by the Central Bank.
Results/Conclusions
For carbon storage, we compare EBV-derived biomass based on a spatial regression with climatic and soil variables to an LULC-based inventory assigning biomass to each land cover class. For sediment retention, we estimate the cover factor (C-factor) of the InVEST sediment model based on NDVI and compare this to an LULC-based approach for setting this parameter. For wildlife-based tourism, we regress proxies for visitation (photo-user days from the Flickr database) against a set of biophysical predictors with and without bird biodiversity. For pollination we set relative pollinator abundance using species distribution models and nesting and floral resources using ecosystem functional diversity, and compare to static relative abundances and LULC-based nesting and floral parameters. Model results derived from EBV approaches are closer to observations than LULC-based approaches, and reflect finer-scale variations impacted by landscape heterogeneity. These findings highlight that the application of EBVs in ecosystem service models is a simple and promising way to improve the accuracy and readiness of this information at a country or global scale.