Drivers of antagonistic spatial association patterns between soil and rice heavy metals in karst region: evidence from bioavailability and Pb isotopes

In Southwest China’s karst region, an antagonistic relationship between soil and rice heavy metal (HM) levels complicates risk identification and limits safe farmland use. This study aims to clarify the spatial association patterns between soil and rice HMs and to explore the underlying causes based on HM bioavailability and source tracing. Bivariate spatial autocorrelation was used to assess spatial association patterns and delineate risk zones. HM bioavailability was evaluated using CaCl 2 and EDTA extraction, and Pb isotopes were used for source apportionment. Soil exceedance rates for Cd, Pb, Cu, Zn, Cr, and Ni ranged from 7.94% to 92.1%, mainly in carbonate areas. However, only Cd, Pb, and Ni exceeded thresholds in rice (23.8%–30.2%), primarily in non-carbonate regions, showing a negative spatial correlation with soil HMs (Moran’s I = –0.12 to –0.31). Antagonistic clusters (High soil HMs–Low rice HMs and Low soil HMs–High rice HMs) covered a larger spatial extent (37.9%) than synergistic clusters (19.7%). CaCl 2 extraction outperforms EDTA in assessing HM bioavailability, revealing 1.5–4.9 times lower extractable concentrations in carbonate-derived soils than other samples. Pb isotopes revealed 69% of soil Pb originated from parent materials, whereas direct anthropogenic input and soil-to-rice transfer accounted for 42.4% and 14.7% of rice Pb, respectively. The antagonistic spatial patterns are primarily driven by low HM bioavailability in carbonate-derived soils and exogenous anthropogenic inputs to rice. These findings offer a scientific basis for targeted risk assessment and management of HM pollution in karst paddy fields.