LSER model for predicting cuticle-water partition coefficients of pesticides

Partition coefficients between plant cuticles and water (K_cw) are important for investigating penetration and residual of pesticides in plant. Experimental determination of K_cw values for diverse pesticides was unrealistic. Therefore, it is necessary to develop an effective model for the prediction of K_cw values. In this work, 64 logK_cw values for 23 pesticides from previous literatures were collected, and a linear solvation energy relationship (LSER) model for the prediction of logK_cw was developed. This model exhibits good goodness-of-fit (R²_adj,tra = 0.79, RMSE_tra = 0.38), robustness (Q²_BOOT = 0.78) and external prediction performance (Q²_ext = 0.81). The developed model is appropriate for various pesticides with functional groups such as -X (Cl, F, I), >N－C(O)－NH₂, -OCH₂COOH, -NH-, -NH₂, >C=O, -O－C(O)－NH-, -CN, -S-, -S(O)(O)-. Mechanism analysis indicated that hydrophobic interactions (average relative contribution of 48%) and n/π-electron pairs interactions (average relative contribution of 9%) contributed to the increase of partition, while hydrogen bond accepting ability (average relative contribution of 26%) and polarizability (average relative contribution of 17%) had negative contribution to the partition of pesticides on plant cuticles. The pp-LFER model developed in this study can be used to predict logK_cw values of new pesticides, and revealed the partition mechanisms of pesticides between plant cuticles and water.