Community structure and functional traits mediate hyperspectral prediction of biomass in restored grasslands

Abstract Introduction Hyperspectral techniques enable rapid non‐destructive prediction of vegetation characteristics, offering an effective method for monitoring grassland restoration status. There are still gaps in the utilization of field hyperspectral techniques to assess the specific grassland characteristics on the Loess Plateau. Objectives This study aimed to develop a predictive model for spectral parameters and community characteristics of restored grasslands using hyperspectral techniques, providing a feasible method for monitoring grassland restoration status. Methods A field spectrometer was applied to collect spectral parameters of visible, red‐edge, and near‐infrared (NIR) wavebands, and 16 vegetation indices (VIs) were calculated. A field investigation was conducted on community structural characteristics, functional traits, and aboveground biomass (AGB). Results Grassland community diversity and height were primarily correlated with the blue‐violet and NIR wavebands. The red and blue‐violet wavebands exhibited the highest sensitivity to canopy pigment content. The red‐edge and visible wavebands were the most critical for estimating grassland community AGB and canopy water content. The dominance of soil‐adjusted indices (e.g. Optimized Soil Adjusted Vegetation Index, Modified Soil Adjusted Vegetation Index) could effectively estimate canopy coverage, water content, and AGB. Significant indirect relationships were observed between VIs and AGB. Conclusions The optimized prediction model was developed using field hyperspectral techniques, highlighting the mediating roles of community structure and functional traits in predicting biomass from vegetation indices.