Abstract
Building upon a theoretical framework, this study utilized 126 field survey questionnaires from farmers in Shouyang County, Shanxi Province, China, coupled with corresponding farmland soil heavy metal monitoring data, to investigate the extent of heavy metal pollution and its mechanistic relationship with farmers' behavior. The single-factor pollution index (P(i)), Nemerow composite pollution index (P(N)), and geographical detector were employed to assess pollution levels and elucidate the underlying mechanisms linking farmer practices to soil heavy metal accumulation. Analysis revealed that the mean concentrations of Cu, Ni, Cr, Pb, Cd, and Zn (25.54, 31.47, 98.50, 16.63, 0.16 and 76.92 mg/kg, respectively) in the farmland soil exceeded the background values for soil elements in Shanxi Province, whereas As (1.92 mg/kg) levels were lower. Assessment using P(i) indicated that Cr, Pb, Cd, Ni, Cu, and Zn (1.78, 1.13, 1.55, 1.05, 1.07 and 1.21, respectively) were predominantly in a state of mild pollution. Similarly, the P(N) (1.50) suggested an overall mild level of composite heavy metal pollution in the soil. Geographical detector(Geo-Detector) analysis demonstrated that the explanatory power (q-value) of interactions among factors-including agricultural film and fertilizer application intensity, farmland fragmentation degree, per capita annual household income, farmland area, and years engaged in farming-on soil heavy metal accumulation was significantly enhanced compared to that of individual behavioral factors. While individual farmers' behaviors are associated with heavy metal accumulation, the interaction effects among multiple behaviors constitute the dominant factor influencing localized accumulation in farmland soil. Consequently, local authorities should enhance farmers' requisite knowledge, skills, and practices for mitigating soil heavy metal accumulation through strategies such as promoting large-scale farming, implementing agricultural input reduction initiatives, and intensifying technical and environmental protection training. The Geo-Detector exhibits significant advantages in identifying nonlinear influencing factors and analyzing factor interactions, yielding more comprehensive insights compared to conventional linear models.