Multivariate modeling application to determine the behavior of picloram in soils: a laboratory trial

Abstract

ABSTRACT Knowledge about herbicide-soil interactions is essential for the safe use of these products. The objective of this study was to evaluate the potential of multivariate discriminant analysis to predict the picloram sorption and desorption capacity of soils and to determine soil groups based on the risk of environmental contamination. Physicochemical properties of 17 soils were subjected to principal component analysis (PCA). Thereafter, the soils were grouped by cluster analysis (CA). Discriminant analysis (DA) was used to determine the efficiency of the clusters. The CA generated three groups of soils, and the coefficients of sorption (Kfs) and desorption (Kfd) were 2.85 and 4.29 for G1, 0.93 and 11.71 for G2, and for G3, 0.36 and 30.76 mg1−1/n L1/n kg−1, respectively. The similarity of soil physicochemical properties in each group resulted in similar picloram sorption and desorption coefficients. Clay, organic matter, pH, and sum of bases showed the greatest influence on the sorption and desorption of picloram. Grouping soils by PCA and CA based on soil physicochemical properties is an alternative approach to determine the risk of environmental contamination by picloram in agricultural areas.