Are productivity and biodiversity adequate predictors for rapid assessment of forest ecosystem services values?

Ecosystem service valuation (ESV) has become a widely recognized means of orienting decision-making about environmental management. However, the economic valuation methods with primary data are complicated and time-consuming. Motivated by widely used land cover-based (using ecosystem services per unit area as a proxy) or productivity-based valuation approaches, this study determines the feasibility and reliability of an approach for rapid ESV modeling using a limited number of indicators. The ecosystem service cascade shows that indicators representing ecosystem structure and function are critical for ESV. We selected productivity (net primary productivity [NPP]) and biodiversity (with Shannon–Wiener index class [SIC] as a proxy) as the key predictors. Taking China’s forest ecosystems as an example, we conducted a literature search to collect primary valuation data following a standardized procedure, and a total of 847 value estimates for 11 regulating services were used for predictive modeling. We set up the initial regression models incorporating four additional qualitative context variables. A strong hierarchical lasso was used to identify the interaction items within them, and 10-fold cross-validation was performed to select the optimal model. The results show that at least one variable has a significant effect on the final model for each ecosystem service. Five models yielded an R2 value of > 70 %, and four others yielded an R2 value of approximately 50 %. Nine models included both NPP and SIC, but these two indicators did not necessarily have a dominant effect. Our approach was also found to greatly improve the performance of models using single-indicator-based proxies. We conclude that this valuation approach based on productivity and biodiversity achieves a new balance of rapidity and accuracy for ESV and can be used to redesign the application of unit value transfer to improve the validity of ESV under certain budget or data constraints.