A machine learning based approach for estimating site-specific partition coefficient Kd of organic compounds: Application to nonionic pesticides. (15th April 2023)

• Record Type: Journal Article
• Authors: Ma, Wankai; Wang, Meie; Jiang, Rong; Chen, Weiping
• Is Part Of: Environmental pollution. Volume 323(2023)
• Publication Date: 2023-04-15
• Subjects: Nonionic organic compound -- Fate and transport modeling -- Equilibrium concentration -- Soil organic carbon -- Applicability domain analysis; AD applicability domain -- C′ initial liquid phase concentration -- Ce liquid phase concentration at equilibrium -- CEC cation exchange capacity -- COPCs contaminants of potential concern -- HWSD Harmonized World Soil Database -- Kd solid-liquid partition coefficient -- Koc organic carbon normalized partition constant -- Kow octanol-water partition coefficient -- MAE mean absolute error -- MW molecular weight -- Qe solid phase concentration at equilibrium -- R2 coefficient of determination -- RF random forest -- RMSE root-mean-square error -- SHAP SHapley Additive exPlanations -- SOC soil organic carbon -- SWAT Soil and Water Assessment Tool -- XGBoost eXtreme Gradient Boosting; Pollution -- Periodicals; Pollution -- Environmental aspects -- Periodicals; Environmental Pollution -- Periodicals; Pollution -- Périodiques; Pollution -- Aspect de l'environnement -- Périodiques; Pollution -- Effets physiologiques -- Périodiques; Pollution; Pollution -- Environmental aspects; Periodicals; Electronic journals; 363.73
• DOI: 10.1016/j.envpol.2023.121297
• Languages: English
• ISSNs: 0269-7491

Abstract

Abstract: The partitioning coefficient K d for a specific compound and location is not only a key input parameter of fate and transport models, but also critical in estimating the safety environmental concentration threshold. In order to reduce the uncertainty caused by non-linear interactions among environmental factors, machine learning based models for predicting K d were developed in this work based on literature datasets of nonionic pesticides including molecular descriptors, soil properties, and experimental settings. The equilibrium concentration ( C e ) values were specifically included for the reason that a varied range of K d corresponding to a given C e occurred in a real environment. By transforming 466 isotherms reported in the literature, 2618 paired equilibrium concentrations of liquid-solid ( C e - Q e ) data points were obtained. Results of SHapley Additive exPlanations revealed that soil organic carbon, C e, and cavity formation were the most important. The distance-based applicability domain analysis was conducted for the 27 most frequently used pesticides with 15952 pieces of soil information from the HWSD-China dataset by setting three C e scenarios (i.e., 10, 100, and 1000 μg L −1 ). It was revealed the groups of compounds showing log K d < 0.06 and log K d > 1.19 were composed mostly of those with log K ow of −0.800 and 5.50, respectively. When log K d varied between 0.100 and 1.00, it was impacted by interactions among soil types, molecular descriptors, and C e comprehensively, which accounted for 55% of the total 2618 calculations. It could be concluded that site-specific models developed in this work are necessary and practicable for the environmental risk assessment and management of nonionic organic compounds. Graphical abstract: Image 1 Highlights: Machine learning models involving C e improved K d prediction of organic chemicals. The impact of C e was observed at each interval of K d range. Molecular, soil type, and C e impact K d for organic compounds frequently. Dominant effect of hydrophobicity happens when log K d < 0.06 and log K d > 1.19.