Due to diffuse atmospheric fallouts of process particles enriched by metals and metalloids, polluted soils concern large areas at the global scale. Useful tools to assess ecotoxicity induced by these polluted soils are therefore needed. Earthworms are currently used as biotest, however the influence of specie and earthworm behaviour, soil characteristics are poorly highlighted. Our aim was therefore to assess the toxicity of various polluted soils with process particles enriches by metals and metalloids (Pb, Cd, Cu, Zn, As and Sb) collected from a lead recycling facility on two earthworm species belonging to different ecological types and thus likely to have contrasted behavioural responses (Eiseina hortensis and Lumbricus terrestris). The combination of behavioural factors measurements (cast production and biomass) and physicochemical parameters such as metal absorption, bioaccumulation by earthworms and their localization in invertebrate tissues provided a valuable indication of pollutant bioavailability and ecotoxicity. Soil characteristics influenced ecotoxicity and metal uptake by earthworms, as well as their soil bioturbation.

Predicting the transfer of contaminants in soils is often hampered by lacking validation of mathematical models. Here, we applied Hydrus-2D software to three agricultural soils for simulating the 1900–2005 changes of zinc and lead concentration profiles derived from industrial atmospheric deposition, to validate the tested models with plausible assumptions on past metal inputs to reach the 2005 situation. The models were set with data from previous studies on the geochemical background, estimated temporal metal deposition, and the 2005 metal distributions. Different hypotheses of chemical reactions of metals with the soil solution were examined: 100% equilibrium or partial equilibrium, parameterized following kinetic chemical extractions. Finally, a two-site model with kinetic constant values adjusted at 1% of EDTA extraction parameters satisfactory predicted changes in metal concentration profiles for two arable soils. For a grassland soil however, this model showed limited applicability by ignoring the role of earthworm activity in metal incorporation.