Ranking and classification of non-ionic organic pesticides for environmental distribution: a qsar approach

Abstract

The partitioning of pesticides into different environmental compartments depends mainly on the physico-chemical properties of the studied chemical. To rank pesticides according to their distribution tendency in various media, we propose a combination of two multivariate approaches: Principal Component Analysis and Hierarchical Cluster Analysis. In such explorative methods we take into account physico-chemical properties like the organic carbon partition coefficient (K _oc), the n-octanol/water partition coefficient (K _ow), water solubility (S _w), vapour pressure (VP) and Henry's law constant (H), being the more relevant to the determination of environmental partitioning.

The 54 studied pesticides of various chemical categories are thus ranked in 4 a priori classes according to their environmental behaviour (sorbed, soluble, volatile and non-volatile/medium class) and finally assigned to the defined four classes by different classification methods (CART, K-NN, RDA) using theoretical molecular descriptors. This QSPR approach allows a rapid indication of the environmental distribution of pesticides starting only from their molecular structure.

Keywords:

• Pesticides
• PCA
• Clustering
• Classification
• CART
• QSPR

Acknowledgements

We wish to thank Prof. R. Todeschini and Dr. V. Consonni for helpful discussions and the software for molecular descriptor calculation and modelling, and Dr. A. Di Guardo for the scientific reviewing of the article. This work was financially supported by BEAM Program of the Commission of the European Communities (EVK1-CT-1999-00012).