Including the spatial variability of metal speciation in the effect factor in life cycle impact assessment: Limits of the equilibrium partitioning method.

In life cycle assessment (LCA), the potential terrestrial ecotoxicity effect of metals, calculated as the effect factor (EF), is usually extrapolated from aquatic ecotoxicological data using the equilibrium partitioning method (EqP) as it is more readily available than terrestrial data. However, when following the AMI recommendations (i.e. with at least enough species that represents three different phyla), there are not enough terrestrial data for which soil properties or metal speciation during ecotoxicological testing are specified to account for the influence of soil property variations on metal speciation when using this approach. Alternatively, the TBLM (Terrestrial Biotic Ligand Model) has been used to determine an EF that accounts for speciation, but is not available for metals; hence it cannot be consistently applied to metals in an LCA context. This paper proposes an approach to include metal speciation by regionalizing the EqP method for Cu, Ni and Zn with a geochemical speciation model (the Windermere Humic Aqueous Model 7.0), for 5213 soils selected from the Harmonized World Soil Database. Results obtained by this approach (EFEqPregionalized) are compared to the EFs calculated with the conventional EqP method, to the EFs based on available terrestrial data and to the EFs calculated with the TBLM (EFTBLMregionalized) when available. The spatial variability contribution of the EF to the overall spatial variability of the characterization factor (CF) has been analyzed. It was found that the EFsEqPregionalized show a significant spatial variability. The EFs calculated with the two non-regionalized methods (EqP and terrestrial data) fall within the range of the EFsEqPregionalized. The EFsTBLMregionalized cover a larger range of values than the EFsEqPregionalized but the two methods are not correlated. This paper highlights the importance of including speciation into the terrestrial EF and shows that using the regionalized EqP approach is not an acceptable proxy for terrestrial ecotoxicological data even if it can be applied to all metals.