Data-Driven Insights into Nanomaterial Impacts on Soil Nitrogen Processes.

Area of Science:

• Environmental Science
• Soil Science
• Ecotoxicology

Background:

• Nanomaterials (NMs) are increasingly used, raising concerns about their environmental impact.
• Soil nitrogen cycling, crucial for ecosystem health, involves microbially mediated nitrification and denitrification.
• Understanding NM effects on these processes is vital for ecological risk assessment.

Purpose of the Study:

• To assess the overall effects of NMs on soil nitrification and denitrification using meta-analysis and machine learning (ML).
• To identify key factors influencing NM impacts on nitrogen cycling processes.
• To develop a predictive and interpretable ML framework for NM risk assessment.

Main Methods:

• Meta-analysis to synthesize existing data on NM effects.
• Machine learning algorithms (XGBoost, SVM, RF) trained to predict changes in nitrification and denitrification rates.
• Model interpretability techniques (global and local) to identify key drivers and interactions.

Main Results:

• XGBoost model accurately predicted NM effects (nitrification R²=0.74, denitrification R²=0.73).
• NM concentration was the primary driver for both processes.
• Soil pH significantly influenced nitrification, while NM size was more critical for denitrification.
• Interactive effects of concentration, exposure time, and soil texture were significant.

Conclusions:

• A robust and interpretable ML framework was established for predicting NM impacts on soil nitrogen cycling.
• Key drivers of NM effects on nitrification and denitrification were identified.
• The study provides a theoretical basis for the ecological risk assessment of nanomaterials in soil environments.