Assessing soil biological properties as quality indicators across a fertility gradient in agricultural areas of Shaanxi Province, China

Background and aims Tradional assessments frequently fail to capture the dynamics of soil quality by exclusively measuring physicochemical properties. A comprehensive soil quality assessment system necessitates the consideration of biological properties. This study identified the biological indicators of changes in soil fertility in agricultural areas of Shaanxi Province, China. Methods Five major crop types were selected across four distinct ecological regions—the Northern Sandy Area, the Weibei Dry Plateau, the Guanzhong Plain, and the Southern Qinling Mountains. Fertility levels were classified using fuzzy membership functions based on soil physicochemical properties and crop yields. Soil biological properties were assessed at different fertility levels. The key determinants of soil quality were identified as the most discriminant biological indicators by clustering, correlation, and random forest analyses. Results Microbial physiological activity and enzyme activity were enhanced in high-fertility plots than in low-fertility plots (e.g., by 52.2–156.4% for soil nitrification intensity) and were positively correlated with soil nutrient contents. High-fertility plots showed greater microbial alpha-diversity (Shannon-Wiener index), with increased proportions of 12 genus-level biomarkers (e.g., Saccharomonospora , Massilia , Mesorhizobium ). Two biomarker clusters were significantly correlated with soil nutrient contents, microbial physiological activity, and enzyme activity. The biomarker network in high-fertility plots exhibited higher complexity and stability (positive edges: 82.2%) compared to that of low-fertility plots (70.6%). Conclusions Twenty key biological indicators effectively distinguished between high- and low-fertility plots. This study provides a calibrated set of metrics for assessing soil quality in (semi)arid agricultural areas of Shaanxi Province and, with local benchmarking, in ecologically comparable regions.