Environmental gradients explain nearshore microplastic distribution patterns: insights from machine learning models

Microplastic is a global concern recently, yet the factors associated with particle behavior after entering marine environments remain uncertain. Using three years of observations integrated with unsupervised and machine learning, with feature-family ablation, results show that microplastic distributions in a highly urbanized nearshore consistently align with chemical environmental gradients, particularly nitrogen, total phosphorus, and trace metals. Rather than isolating individual transport drivers, our results indicate that microplastic patterns co-vary with biogeochemical regimes, while the contribution of freshwater and hydrodynamic proxies remains limited under the spatial and temporal resolution considered for classifiers. Variables related to river proximity and runoff potential exhibited lower relative classification importance, reflecting proxy limitations rather than the ecological irrelevance of hydrodynamic processes. Overall, these findings support the interpretation of microplastics as quasi-passive tracers embedded within coastal chemical gradients, integrating signals of eutrophication, wastewater inputs, and industrial activities. By leveraging routinely monitored water-quality and nearshore gradients features, this framework provides a transferable approach for interpreting microplastic patterns after their entry into coastal waters.