A tree regression analysis of factors determining the sorption and retention of heavy metals by soil

This paper describes in detail the tree regression models fitted, using the CART software package, to data on the sorption by 14 soils of Cd, Cr, Cu, Ni, Pb and Zn from solutions containing all six, and on the retention of these metals by the soils when subsequently exposed to metal-free medium. Pb, Cu and Cr were sorbed and retained to a greater extent than Cd, Ni and Zn, and the data for the former three metals were fitted better than those for the latter three, in keeping with previous indications that the influence of soil properties on the sorption and retention of Cd, Ni and Zn is limited by the presence of the more readily sorbed metals. Importance values suggest that the sorption and retention behaviour of Cr depends mainly on Fe/Al oxides, the organomineral fraction, sand and chlorite; that of Cu on CECe, organomineral content and haematite content; and that of Pb on CECe, haematite content and Mn oxides content. Although the behaviour of Cd, Ni and Zn depends on that of the more readily sorbed metals, it also appears to be influenced by silt content and various clay minerals (haematite in the case of Cd, mica in that of Ni, and quartz, gibbsite and chlorite in that of Zn).