Answers are grounded in this paper and cite their sources.
Microplastics (MPs) are widespread in freshwater environments, yet their influence on the bioavailability of organic contaminants in sediments under field conditions remains poorly understood. To fill this gap, a tree-based machine learning framework incorporating algorithm comparison, tree feature importance, and SHapley Additive exPlanations (SHAP) analysis was developed in this study. Three MP input schemes, including total abundance, particle size distribution, and polymer composition, were evaluated with sediment organic carbon and compound hydrophobicity as covariates to predict the measured bioavailable fractions of over 400 organic compounds. Baseline models using total MP abundance showed decision tree outperforming traditional polynomial fitting and interpolation methods (coefficients of determination R 2 = 0.67), while eXtreme Gradient Boosting (XGBoost) achieved the best performance with multi-type inputs (R 2 up to 0.78). Feature importance analysis indicated that fine MPs (30–100 μm) and polymers polypropylene, polyethylene terephthalate, polytetrafluoroethylene, polystyrene, and polyurethane were identified as the highest ranked MP-related factors of pollutant bioavailability. Stratified SHAP analysis revealed contrasting patterns, with size-based contributions stronger in the upper log K ow group and type-based contributions stronger in the lower log K ow group (median log K ow = 4.5). These findings link size and polymer resolved MP characteristics with measured pollutant bioavailability in field sediments, providing a quantitative basis for sediment risk screening in freshwater systems.
No takes yet. Share an insight, caveat, or question.
Huang et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: