Predicting the effect of silver nanoparticles on soil enzyme activity using the machine learning method: type, size,...

Area of Science:

• Environmental Science
• Soil Science
• Nanotechnology

Background:

• Silver nanoparticles (AgNPs) are increasingly used, raising concerns about their environmental impact.
• Soil enzymes are crucial indicators of soil health and function.
• Understanding AgNP interactions with soil ecosystems is vital for risk assessment.

Purpose of the Study:

• To predict the impact of silver nanoparticles (AgNPs) on soil enzyme activity using machine learning.
• To identify key factors influencing AgNP toxicity to soil enzymes.
• To provide insights into the regularity of soil enzyme responses to AgNPs.

Main Methods:

• Utilized machine learning models: Artificial Neural Network (ANN) optimized with Genetic Algorithm (GA), Gradient Boosting Machine (GBM), and Random Forest (RF).
• Employed Partial Dependency Profile (PDP) analysis to determine the influence of AgNP properties and experimental conditions.
• Simulated AgNP effects across varying doses, types, sizes, and exposure times.

Main Results:

• Polyvinylpyrrolidone-coated AgNPs (PVP-AgNPs) exhibited the most significant inhibitory effect (49.5%) on soil enzyme activity.
• ANN models predicted that enzyme activity initially decreases then increases with increasing AgNP size.
• ANN and RF models indicated that soil enzyme activity decreases before 30 days of exposure to uncoated AgNPs, then recovers and slightly declines thereafter.
• ANN model identified the importance order of factors as: dose > type > size > exposure time.
• RF model identified sensitive ranges for AgNP exposure: 0.01-1 mg/kg (dose), 50-100 nm (size), and 30-90 days (exposure time).

Conclusions:

• Machine learning models effectively predict soil enzyme responses to AgNPs.
• AgNP dose and type are primary drivers of soil enzyme activity changes.
• AgNP size and exposure duration also play significant roles in modulating enzyme responses.
• This research offers valuable predictive insights into the ecological risks of AgNPs in soil environments.