Tuber crops cultivated in basalt-derived soils are influenced by naturally high geochemical backgrounds, which may elevate heavy metal(loid) levels and associated health risks. To clarify the geochemical controls governing metal accumulation, this study analyzed rock, soil, and tuber (sweet potato and yam) samples from the Qiongbei volcanic area of Hainan Island, China. Concentrations of eight heavy metal(loid)s (As, Cd, Cr, Cu, Hg, Ni, Pb, and Zn) and 22 nutrient-related indicators (N, P, K, SOC, S, Se, Fe, Mn, and their available fractions) were determined. Soil contamination and potential human health risks were evaluated using the pollution index and the health risk model. The results showed that 11.1–55.6% of soil samples exceeded pollution thresholds for Cr, Cu, Ni, and Zn, reflecting typical basaltic high-background characteristics. In contrast, heavy metal(loid) concentrations in tuber crops were relatively low and jointly regulated by parent material composition and soil nutrient status. Non-carcinogenic risks (HI) were below 1, indicating acceptable exposure levels, while carcinogenic risks were mainly associated with Cd, Cr, and Pb, with total carcinogenic risk (TCR) exceeding 1 × 10⁻⁴, suggesting potential health concerns. Strong correlations between soil nutrients (N, P, K, SOC, S, Se, Mn, and Fe) and plant uptake of As, Cd, Cu, and Cr indicate that nutrient availability plays a crucial role in controlling heavy metal(loid) bioavailability. The volcanic soils exhibited a “high total content–low bioavailability” pattern. Enhancing soil Se, SOC, available N, and slowly available K (SAK) can effectively reduce Cd and other high-risk metal accumulation in tuber crops. These findings elucidate the key geochemical processes influencing heavy metal transfer in volcanic agroecosystems and provide a scientific basis for safe agricultural utilization and health risk prevention in high-background regions.

Thermomagnetic-fraction geochemistry has shown promise beneath transported cover, but area-scale applications rarely combine explicit element-association analysis, statistical robustness tests and transparent target prioritization. We evaluate this workflow in the largely covered Jing’erquan Ni-Cu metallogenic area using 985 soil samples collected over approximately 140 km² at an average spacing of about 300 m (7.0 samples km⁻²). After removal of pre-existing strongly magnetic grains, 100 g aliquots were heated under oxygen-limited conditions at 650 °C for 45 min, and the newly generated magnetic fraction was separated at an instrument-current setting of 1 A. Twelve indicators were analyzed using descriptive statistics, correlation analysis, hierarchical clustering, principal component analysis (PCA), spatial clustering and a three-term composite score. Four rotated factors explain 75.1% of the variance; Cr and Ni load at 0.95 on F2, while TFe₂O₃, Zn, Co and Cu covary on F1. Pearson–Spearman matrix agreement (r = 0.990), winsorized-versus-original Pearson agreement (r = 0.999), and PCA factor-congruence coefficients (>0.998) show that the principal associations are not controlled by the largest Cu, Ni or Cr values. Alternative composite weights retain all eight named follow-up targets, although internal ranking changes when Cu is emphasized. Among them, JQ-1–JQ-4 are assigned high follow-up priority by the expert-informed score. The coincidence of JQ-1 with the known Jing’erquan Ni-Cu mining area provides an internal plausibility check for the workflow, whereas JQ-2–JQ-8 remain unverified follow-up targets. These results show that the thermomagnetic-fraction dataset delineates multielement surface geochemical anomalies in covered terrain and can serve as a complementary screening tool for subsequent Ni-Cu exploration.