How microplastics affect nitrogen removal in nature-based stormwater infrastructures: A machine learning and...

Area of Science:

• Environmental Science
• Water Quality Engineering
• Ecotoxicology

Background:

• Microplastics (MPs) are emerging contaminants impacting ecosystem functions.
• Their effects on nitrogen removal in nature-based stormwater infrastructures (NBSIs) are not well understood.
• NBSIs are crucial for managing stormwater runoff and improving water quality.

Purpose of the Study:

• To systematically investigate how microplastic properties, system characteristics, and environmental conditions affect nitrogen removal in NBSIs.
• To elucidate the stage-specific mechanisms of microplastic interference in nitrogen removal processes.
• To identify key factors regulating nitrogen removal performance in microplastic-contaminated NBSIs.

Main Methods:

• A meta-analysis of 19 published studies was conducted.
• Five machine learning models (XGBoost, RF, LightGBM, MLP, KAN) were trained and evaluated.
• SHapley Additive exPlanations (SHAP) analysis was used to determine variable importance.

Main Results:

• Microplastics significantly interfere with ammonium-nitrogen (NH₄⁺-N) removal, influenced by particle size, concentration, and polymer type.
• Nitrate-nitrogen (NO₃^--N) removal is co-regulated by environmental factors (pH, C/N ratio) and plant species.
• Total nitrogen (TN) removal is predominantly controlled by C/N ratio and pH (cumulative contribution > 60%), with microplastics primarily impeding nitrification.

Conclusions:

• Environmental regulation (C/N ratio, pH) has a stronger influence on TN removal than microplastic variables.
• Microplastics mainly disrupt the nitrification stage of nitrogen removal in NBSIs.
• XGBoost demonstrated the highest predictive accuracy (R² > 0.86), providing a robust model for NBSIs performance prediction.