Multivariate statistical analysis of spatial variations in half-life time (DT50) of the pesticides metribuzin and MCPA in soils

Variation in the dissipation time (DT50) was studied for the herbicides MCPA and metribuzin in the unsaturated zone. Horizons A, B and C were sampled from nine sandy soil locations in Denmark. Variation is addressed both horizontally, i.e. variation between locations, and in relation to depth (sampling horizon). Results from several locations and depths are presented, and the effects of depth and soil type are discussed. The median DT50 value was metribuzin =117 days (range 42 to >100 days) and MCPA= 12 days (range 4 to 49 days). Three types of statistical analysis was applied 1/ ranking test for identifying possible correlation between dissipation time and horizon 2/ principal component analysis (PCA) for reduction of the dimensionality of the dataset 3/ canonical correlation analysis revealing possible correlation between dissipation variables and variables identified in the PCA. Variations in DT50 were observed between field sites as well as vertically. This variation could be related to a few variables particularly cations (Ca2+, K+, Mg2+ and Na+) but also exchangeable H+, humic substances, and amorphous iron. All these variables seem related to soil sorption characteristics. Correlation to hydrolic conductivity (pF4.2) was also indicated. Good correlation between dissipation and the sorption components was shown.