Soil biodiversity plays a central role in regulating ecosystem functioning and maintaining soil health through its influence on nutrient cycling, carbon dynamics, soil structure formation and water regulation. At the same time, environmental change and land-use intensification alter soil physicochemical conditions and vegetation dynamics, with cascading consequences for belowground communities and ecosystem functioning. Despite increasing recognition of the importance of soil biodiversity within European soil health initiatives, identifying scalable and operational indicators capable of capturing soil ecological responses across contrasting pedoclimatic regions remains a major challenge.
Using harmonized datasets generated within the SOB4ES project, we integrated soil biodiversity indicators, soil physicochemical properties, land-use information and Earth Observation-derived climatic, vegetation and topographic variables across European sites. The analyses employed machine-learning approaches and explainable artificial intelligence methods to investigate large-scale relationships among soil biodiversity, environmental gradients and ecosystem service-related functions under different climatic and land-use contexts.
Across multiple organism groups, diversity-based metrics generally showed stronger and more consistent relationships with environmental gradients than abundance- or density-based indicators, supporting their relevance for large-scale soil health assessment frameworks. Soil pH, soil organic carbon, moisture conditions and soil texture parameters emerged as dominant cross-taxa predictors, while climatic and vegetation-related variables derived from Earth Observation data explained a substantial proportion of spatial variation in belowground biodiversity patterns. However, the influence of environmental predictors frequently differed among land-use types, indicating that biodiversity responses are strongly context-dependent and shaped by interactions among soil conditions, vegetation structure and management intensity.

Overall, the integration of harmonized biodiversity observations, remote-sensing data and machine-learning approaches demonstrated strong potential for supporting spatially explicit soil health monitoring and the identification of robust biological indicators under global change and land-use pressures across Europe.

Acknowledgments: The work and all the authors were supported by the Horizon Europe project SOB4ES under Grant Agreement No. 101112831. We acknowledge all participating investigators from the SOB4ES consortium who contributed to the existing sample collection and the field sampling for the generation of the spatial database used in the current analysis. Partners from KNAW, UVIGO, NUID UCD, UNICT, KU Leuven, CU, ARO, IBB, UL, UoC, SLU, EFWSL, Airfield, MFO, and INRAe provided these contributions.