Long-Term Biodiversity Monitoring Using Remote Sensing

Abstract

Remote sensing -- the acquisition of information about Earth's surface through satellite-borne or airborne sensors - provides a transformative capability for biodiversity monitoring at spatial and temporal scales impossible with field surveys alone, enabling continuous tracking of habitat extent, condition, and structural properties that are mechanistically linked to biodiversity. This study developed and validated a Multi-Sensor Biodiversity Monitoring Framework (MSBMF) combining Sentinel-2 optical (10 m; NDVI, EVI, NDWI, red-edge indices), Sentinel-1 SAR (10 m; backscatter coefficient for vegetation structure), and GEDI (Global Ecosystem Dynamics Investigation) LiDAR (25 m footprint; canopy height, biomass) across 42 European terrestrial ecosystem sites (2,840 km2 total; 2017-2022) to predict field-surveyed species richness, NDVI-derived productivity, and habitat quality indices. MSBMF-predicted species richness correlated with field-surveyed richness at r = +0.84 for all habitats combined (p < 0.001) and r = +0.88-0.92 for structurally complex habitats (woodland, wetland). Vegetation spectral diversity (coefficient of variation of NDVI across 100 m grid cells) was the strongest single remote sensing predictor of species richness (r = +0.78; p < 0.001), consistent with the spectral variation hypothesis. Long-term MSBMF time-series (2017-2022) detected biodiversity-relevant habitat change at 84.4% accuracy versus aerial photographic interpretation as ground truth, enabling annual updates of species richness predictions without field surveys. A Remote Sensing Biodiversity Index (RSBI) combining all sensor inputs explained 72.4% of variance in field-surveyed species richness (R2 = 0.724) and was significantly negatively correlated with habitat degradation indicators (NDVI decline trend r = -0.74; fragmentation index r = -0.68). MSBMF reduced per-site biodiversity monitoring cost by 74.4% relative to annual field surveys at equivalent coverage.