Objectives Soils in the Loess Plateau are characterized by nutrient deficiency and pronounced spatial heterogeneity, limiting sustainable agricultural development. We studied the spatial patterns of soil multifunctionality and the key driving factors at the county scale.

Methods This study was conducted in Wuzhai County, a representative region in the northern Loess Plateau. One hundred topsoil samples (0-20 cm) were collected from farmland to measure soil nutrient content, microbial biomass, enzyme activities, and community diversity. A multifunctionality evaluation system was constructed based on four functional dimensions—nutrient supply, nutrient cycling, nutrient storage, and carbon storage, and a soil multifunctionality (SMF) index was calculated. Correlation analysis, random forest modeling, and partial least squares path modeling were employed to identify the main drivers of SMF and their regulatory pathways.

Results The average soil chemical properties of the 100 farmland samples were: organic matter 7.09 g/kg, total N 0.80 g/kg, available P 12.64 mg/kg, and available K 131 mg/kg, with coefficients of variation of 38.62%, 47.42%, 67.72%, and 54.10%, indicating that the nutrient content in the farmlands were generally low and highly spatial heterogeneity. SMF was positively correlated with individual functional indicators and crop yield (r = 0.45–0.75, P<0.05). Spatially, higher SMF values occurred in the southern and southeastern areas, whereas lower values were found in the northern and northwestern regions, reflecting the spatial variation of key driving factors. Random forest analysis identified β-glucosidase, microbial biomass carbon, alkaline phosphatase, microbial biomass nitrogen, and total N as the main predictors of SMF (P<0.01). PLS-PM revealed that nutrient cycling contributed the most to SMF (r = 0.517, P<0.001), exceeding the effects of nutrient supply (r = 0.201, P<0.05) and carbon storage (r = 0.299, P<0.001). Soil bacterial diversity indirectly enhanced SMF by promoting nutrient supply (r = 0.361, P<0.05) and strengthening nutrient cycling processes, highlighting the role of microbially mediated nutrient transformations in driving SMF variation.

Conclusion The nutrient contents in farmlands of northern Loess Plateau counties are relatively low, and shows significant spatial heterogeneity. The soil multifunctionality is primarily driven by microbial-mediated nutrient cycling, with bacterial diversity and nitrogen status acting as key regulatory factors.