Related Experiment Video
Updated: Dec 26, 2025
08:21
Isolation and Screening from Soil Biodiversity for Fungi Involved in the Degradation of Recalcitrant Materials
Published on: May 16, 2022
5.6K
Artificial intelligence models to predict acute phytotoxicity in petroleum contaminated soils
Dmitrii Shadrin1, Mariia Pukalchik1, Ekaterina Kovaleva2
1Center for Computational and Data-Intensive Science and Engineering, Skolkovo Institute of Science and Technology, 143026, Moscow, Russia.
Ecotoxicology and Environmental Safety
|March 13, 2020
Summary
Total petroleum hydrocarbons (TPH) impact soil health. Machine learning models accurately predicted TPH phytotoxicity in Sakhalin soils, using barley root elongation as a key indicator.
Area of Science:
- Environmental Science
- Soil Science
- Ecotoxicology
Background:
- Environmental pollutants, particularly total petroleum hydrocarbons (TPH), exert complex effects on soil ecosystems.
- Understanding the phytotoxicity of TPH is crucial for assessing soil contamination and developing remediation strategies.
Purpose of the Study:
- To model the acute phytotoxicity of TPH on eleven soil samples from Sakhalin Island under greenhouse conditions.
- To evaluate the predictive capabilities of machine learning models for TPH phytotoxicity.
- To investigate the influence of soil properties on TPH effects.
Main Methods:
- Soils were contaminated with crude oil at varying concentrations (3.0-100.0 g kg⁻¹).
- Barley (Hordeum vulgare) root elongation was measured as a primary ecotoxicity parameter.
- Artificial Neural Network (ANN) and Support Vector Machine (SVM) models were employed to predict TPH phytotoxicity.
- Model performance was validated using Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Coefficient of Determination (R²).
Main Results:
- ANN and SVM models successfully predicted barley response to TPH contamination.
- Soil chemical properties (pH, LOI, N, P, K, clay, TPH) were significant predictors.
- The best model achieved an MAE of 8.44, RMSE of 11.05, and R² of 0.80.
Conclusions:
- Machine learning, specifically ANN and SVM, offers a robust approach for predicting TPH phytotoxicity in soils.
- Soil properties play a critical role in modulating the plant response to TPH contamination.
- This study provides valuable insights for environmental risk assessment and soil management in TPH-contaminated areas.
