Machine Learning Models for Forecasting Microplastic Dynamics in China’s Coastal Waters

Abstract

Understanding spatial-temporal microplastic (MP) patterns and regional drivers in China's coastal waters is crucial for pollution interventions. Based on selection criteria, this study synthesizes 1,146 validated data from 49 peer-reviewed studies across China’s four major seas (Bohai, Yellow, East China, and South China Seas). MP abundance showed a spatial gradient, with marine exhibiting lower concentrations than estuary/bay and coastal areas. Association rules suggest urban centers and industrial activities as potential causes. Notable trends highlight the complexity of microplastics type, as polyethylene terephthalate and polypropylene dominate. Machine learning and SHAP analysis revealed nonlinear drivers of MP pollution and ecological risks. In marine areas, total phytoplankton primary production correlated with MPs, potentially through biofouling interactions, while surface CO₂ indirectly influenced distribution via carbon cycle dynamics. Coastal and estuary/bay areas showed MP abundance correlations with scientific-technological innovation and higher education institutions, whereas the ecological risk aligned with wastewater treatment ratios and lengthen of urban sewage pipes, suggesting higher ecotoxicity from industrial discharge MPs. Ensemble modeling projected MP trends under different scenarios: economic and education development reduced MP concentrations, while industrial expansion and technology innovation increased pollution. The Pearl River Delta Economic Zone exhibited the highest MP levels, with coastal and estuary/bay regions displaying divergent pollution responses to anthropogenic pressures. Policy recommendations include integrating environmental criteria into technological innovation, optimizing wastewater management, and leveraging education for sustainable production. This study provides actionable insights for safeguarding marine ecosystems amid industrialization.