Ecological risk assessment of contaminated soil.

This review has described three cases of ecological risk assessment. The cases include two heavy metals (Cu and Zn) and an anthropogenic organic chemical (DDT). It concludes that there are at least two major constraints hampering the use of laboratory tests to predict effects under natural field conditions. One key issue is bioavailability, and another is suboptimal conditions or multiple stresses in the field such as climatic stress (drought, frost), predators, competition, or food shortage. On the basis of the presented case studies, it was possible to answer three essential questions often raised in connection to ecological risk assessment of contaminated sites. 1. To what extend does soil screening level (SSL) estimate the risk? The SSL are generally derived at levels corresponding to the lowest observed effect levels in laboratory studies, which often is close to the background levels found in many soils. In the cases of zinc and especially DDT, the SSL seemed quite conservative, whereas for copper they resemble the level at which changes in the community structure of soil microarthropods and the plant community have been observed at contaminated sites. The SSL correspond as a whole relatively well with concentrations where no effects or only minor effects were observed in controlled field studies. However, large variation in field surveys can often make it difficult to conclude to what extent the SSL corresponded to no-effect levels in the field. 2. Do bioassays represent a more realistic risk estimate? Here, there is no firm conclusion. The zinc study in UK showed a better relationship between the outcome of ex situ bioassays and field observations than the SSL. The latter overestimated the risk compared to field observations. However, this would be species dependent, as the sensitivity to metals may vary considerably between recognized test species, even within the same group of organisms, such as Folsomia candida and Folsomia fimetaria or Eisenia fetida and Lumbricus sp. Furthermore, it was demonstrated that bioassays were not useful for predicting general species diversity in the field as they are strongly influenced by natural variation and other factors not related to contamination. In the case of copper, bioassays with springtails and black bindweed seemed to underestimate the risk compared to the Cu concentrations at which significant changes in the community structure of soil microarthropods and plants have been observed at the contaminated site, and this was also the case for the DDT-contaminated soils. Here, bioassays with DDT-contaminated soils showed generally very low toxicity, with EC10 values considerably higher than the levels where clear effects on single species as well as community structure have been detected in the present field study. 3. Is it possible to make sound field surveys or do we lack suitable reference situations? Large natural variation caused by other factors than contaminants were observed in most cases, and this may have particularly hampered the conclusions made in the field surveys. These factors included pH, private and military traffic, age of vegetation, shading effects, and variations in light insensitivity as well as quantity and quality of organic matter. It was therefore concluded that field studies should always be interpreted in concert with similar data from a reference situation. Conclusions should therefore be made with caution in situations where important soil conditions vary between control plots and the contaminated sites. The cases also showed that indices focusing on species richness were unreliable. Estimates of evenness or dominance were recommended instead, and most authors concluded that multivariate analysis of community structure was a sensitive and useful method superior to single-species field data. This review concludes that there is a need for a tiered approach in ecological risk assessment of contaminated soils. Generic soil screening levels are needed as a first tier. Higher tiers of ecological risk assessment should, however, contain some kind of site-specific assessment. It is furthermore important to organize the various studies in a framework or decision support system that is transparent and useful for all stakeholders. A weight of evidence approach may be an obvious choice to deal with these uncertainties. The TRIAD approach, which incorporates and categorizes information in a triangle - chemistry, toxicology, and ecology - is an appropriate tool for handling conceptual uncertainties.