Inferring scalable productivity-related grassland functional diversity in combination with in-situ leaf spectra and Sentinel-2 data.

The positive relationships between biodiversity and ecosystem productivity have been broadly recognized in aquatic and terrestrial ecosystems, such as grasslands. However, remotely sensed assessment of functional diversity (FD) and its relationships with productivity across large regions are less studied in grasslands. In this study, we first examined the potential of spectral retrieval of 13 leaf functional traits from a species spectra-trait library to complement field measurements across three types of grassland communities in the Xinlingol grassland located in northern China. We then pre-selected the key traits out of 13 functional traits from 1664 plant individuals of 112 species to calculate in-situ productivity-related FD, and explored the multi-scale relationships of single-trait community weighed mean (CWM) and FD index (Rao's quadratic entropy, RaoQ) with ecosystem productivity at plot level (1 m × 1 m) and site level (30 m × 30 m), respectively. Finally, we applied Sentinel-2 satellite data to infer regional FD through statistical relationships with direct spectral association using partial least squares regression (PLSR). With the leaf spectral prediction (cross-validation