Soil arthropod communities are better predictors of soil quality than nematodes in agricultural soils

Soil ecosystems are shaped by complex interactions between their biological communities, environmental conditions and physicochemical properties. However, higher-trophic soil fauna is often neglected in soil risk assessments. This study utilized environmental DNA (eDNA) barcoding to describe arthropod and nematode communities from 445 agricultural sites in north-east China. We analyzed these community profiles in combination with key soil physicochemical variables to predict agricultural soil quality, degradation and health indices using machine learning models. Arthropod community composition consistently outperformed that of nematodes in predicting individual soil variables (e.g., pH, total carbon, soil organic carbon, total nitrogen), generating stronger correlations. Predictions of soil degradation and health indices also showed higher accuracy when based on arthropod communities. Notably, keystone arthropod taxa (e.g., Diptera, Araneae and Coleoptera) explained over two-thirds of the predictive power achieved by the full community models, indicating their potential to underpin accurate soil quality assessments. Our study demonstrates the potential value of keystone arthropods as bioindicators and provides a foundation for developing biologically-informed soil health frameworks for sustainable land management.