Modelling trace metal partitioning in forest floors of northern soils near metal smelters.

Trace metal (TM) mobility and toxicity varies with changing soil conditions. Geochemical models can account for the influence of soil characteristics on TM behaviour. We tested the effectiveness of the Stockholm humic model (SHM), and the NICA-Donnan model (NDM) to estimate partitioning coefficients (logKd) in 26 forest floor horizons of podzolic soils enriched in trace metals from deposition by metal smelters. We wanted to know if a consistent approach could be applied to model metal partitioning in forest floors without optimizing each individual soil. When optimized, the SHM reproduced the partitioning of Cd, Cu and Zn but not Pb. It was necessary to revise the affinity constants for the NDM to simultaneously simulate the partitioning of the four metals. Revised affinity constants for the NDM model based on a fixed definition of soil organic carbon, i.e., a fixed ratio of fulvic and humic acids per unit carbon, reproduced metal partitioning more effectively in an independent data set of 16 soils than the use of generic affinity constants available for these models. From the perspective of the applicability of these models to risk assessment, this result suggests geochemical models using affinity constants that have been verified and/or modified against multiple soils from a region can provide good estimates of metal partitioning on a regional scale.