Source apportionment of eleven trace metals across six São Paulo river basins (1978-2026): receptor modelling identifies geogenic misregulation and a monitoring blind spot for diffuse-source zinc.

Trace metal contamination of freshwater systems arises from a superposition of geogenic and anthropogenic sources that compliance monitoring cannot disentangle. This study applies Non-negative Matrix Factorisation (NMF) receptor modelling to 126,140 records of eleven trace metals (Al, As, Ba, Cd, Cu, Fe, Hg, Mn, Ni, Pb, Zn) at 114 monitoring stations across six water management units (UGRHIs) of São Paulo State, Brazil (1978-2026), operated by CETESB-one of the longest continuous total-recoverable multi-metal river monitoring records in Latin America. Four geochemically interpretable source factors are resolved: Fe-Al geogenic weathering (F1), Al soil erosion (F2), Fe-Mn redox/industrial effluents (F3), and Zn-Mn urban diffuse sources (F4). Mann-Kendall trend analysis (2000-2025) reveals two divergent trajectories: copper declines significantly (Cu, τ = - 0.709 , p < 0.001 ), consistent with CONAMA Resolution 357/2005 industrial-effluent enforcement, while the geogenic metals iron and aluminium also decline in a manner consistent with reduced erosional loading rather than effluent control, and median-based trends suggest rising urban-source metals (Zn, τ = + 0.727 ; Ni, τ = + 0.461 ). However, a censoring-invariant exceedance analysis introduced here demonstrates that these apparent increases are inflated by documented rises in analytical detection limits (Zn: 0.020-0.100 mg L - 1 in 2012): no metal exhibits a rising exceedance of its CONAMA Class 2 limit, and the fractions exceeding the limit for Zn and Ni in fact decline ( τ = - 0.477 and - 0.600 , both p < 0.001 ). System-wide aluminium (91.6%) and iron (88.2%) violation rates under CONAMA Class 2 are attributable primarily to geogenic Factors F1 and F2-tropical Latosol weathering-not to anthropogenic contamination, constituting a regulatory artefact with direct implications for environmental governance in high-weathering tropical geologies. Critically, the current Zn reporting limit (0.100 mg L - 1 ) sits above the chronic ecotoxicological threshold for sensitive freshwater species (10- 50 μ g/L ) and at 56% of the CONAMA limit, leaving the monitoring network structurally unable to resolve ecologically relevant Zn from tyre and road wear particles-a diffuse-source monitoring and governance gap absent from Brazil's current regulatory architecture. These findings position São Paulo's metal governance within the global pollution-transition pattern: measurable regulatory success for industrial point sources, alongside a detection-limit blind spot for emerging diffuse-source metals, with transferable methodological and policy implications for tropical regions where geogenic background geochemistry conflicts with standards calibrated to temperate-zone geology and where evolving detection limits can masquerade as concentration trends.