Abstract
The national soil risk screening values (RSVs) and intervention values (RIVs) have limited applicability in karst regions with high geochemical backgrounds, often leading to unnecessary remediation efforts. To address this issue, 125 paired soil-rice samples from Gejiu-Mengzi, Yunnan Province, China, were analyzed. We found that the RSV misclassification rates (mainly false positives) for cadmium (Cd) and arsenic (As) in neutral-to-alkaline paddies (pH > 6.5) ranged from 52.0-66.4%. Soil iron/manganese (Fe/Mn) oxides and pH were identified as critical factors controlling Cd/As bioavailability in rice. This decoupling is primarily governed by soil alkalinity and abundant Fe/Mn oxides, which immobilize metals despite high total concentrations. This also explains why there is such a high rate of false positives. By integrating multiple linear regression and species sensitivity distribution (SSD) models, we derived regional safety thresholds (STs) and hazard thresholds (HTs). The newly established STs (Cd: 4.54-7.12 mg·kg(-1); As: 91.43-92.30 mg·kg(-1)) were significantly higher than the national RSVs (Cd: 0.6-0.8 mg·kg(-1); As: 20-25 mg·kg(-1)). The HT for Cd (9.66-10.12 mg·kg(-1)) was 2.5-3.2 times greater than the RIV for Cd, whereas the HT for As (96.90-97.15 mg·kg(-1)) was slightly lower than the RIV for As (100-120 mg·kg(-1)). Applying the ST-HT system increased the soil quality assessment accuracy from 28.1-95.6% to 71.8-100%, drastically reducing the need for unnecessary remediation. This study not only serves as a region-specific framework for safe rice cultivation in karst areas of Yunnan, but also highlights the potential for integrating soil-plant system bioavailability with physiological tolerance mechanisms. Furthermore, we provide a crucial benchmark for evaluating innovative remediation strategies aimed at alleviating metal stress in plants.