Contrasting land-use sources of microplastic and endocrine-disrupting compound pollution in a major coastal river...

Environmental monitoring in coastal megacities prioritizes land-based microplastics (MPs) and endocrine-disrupting compounds (EDCs) as key threats to marine ecosystems. This study investigated their occurrence profiles and spatial patterns in Shanghai's river network with instrumental analysis including liquid chromatography-tandem mass spectrometry and infrared spectroscopy. By integrating shape analysis with dual machine learning models (XGBoost and Random Forest), we disentangled complex land-based pollution sources and evaluated their marine export potential. MPs were detected in 84.3% of sites (1.4-3.9 items/L), dominated by fibers (86.7%), while EDCs (3.6-90.4% detection rates) were primarily nonylphenol (9.1 ng/L). Spatial interpolation revealed MPs and EDCs accumulated in urban and agricultural hotspots, respectively. Shape analysis quantified spatial similarity between pollutant distributions and land use via Euclidean distances (MPs: 0.17, EDCs: 0.19). It highlighted urban areas as MP hotspots and agricultural/green zones as EDC contributors. Our machine learning approach identified divergent land-use drivers: MPs were primarily linked to urban runoff from roads and buildings, while EDCs, including steroid and phenolic compounds were associated with agricultural land uses, including orchards and croplands. Despite overlapping urban sources, MPs and EDCs exhibited distinct diffusion mechanisms, with partial correlation between their concentrations. This study provides a novel framework for source apportionment in complex urban-rural interfaces, supporting targeted mitigation strategies for marine-bound pollutants.