Estimating bioaccessibility, phytoavailability and phytotoxicity of contaminant arsenic in soils at former sheep dip sites

Recognition that the bioavailability of soil As (As) is influenced by its soil dynamics has initiated research into development of more accurate, site-specific soil guideline values, departing from the assumption that the total soil As content is bioavailable. With the aim of deriving predictive models, the relationship between soil properties and As bioavailability (bioaccessibility and phytotoxicity) was examined on a set of naturally contaminated sheep dip soils (n = 30). Sampled soils were extensively characterised, bioaccessibility was estimated through an in vitro procedure, and soil As toxicity and availability to plants were evaluated using an early growth wheat bioassay. The in vitro bioaccessibility was consistently less than the total soil As content. Arsenic bioaccessibility was negatively correlated to soil iron (Fe), manganese (Mn) and aluminium (Al) contents, and it was positively related to the soil As loading. The in vitro extractable soil As concentrations were successfully modelled using linear combinations of soil As content, soil Fe and Mn determinations and soil pH. Differences in As phytotoxicity, expressed in terms of effective toxic concentration (EC50), between soils were directly related to soil Fe, Mn and Al contents. Available soil phosphorous (P) exerted an ameliorating effect on As toxicity, with the available soil As/P ratio representing the single best predictor of plant growth suppression. Plant P nutrition appeared to influence the relative selectivity for As and P by wheat, with greater selectivity for P demonstrated under P deficient conditions. Plant As uptake, its distribution, and also the plant nutrient status were all adversely affected by increasing soil As exposure. Co-contamination by Zn corresponded to a substantial elevation in proportion of the plant As allocated in shoots. Plant As levels exhibited a saturation-dependent relationship with increasing soil As. The best linear predictors of plant As levels in the non-toxic range were RHIZO-extractable and effective soil As concentrations, the latter based on the diffusive gradients in thin films (DGT) technique. Despite the complexity of soil As dynamics, large proportions in the variances exhibited by the two measures of bioavailability were explained using a small set of readily-available soil properties.