Sediment Toxicity Assessment: Rationale for effect classes (5 pp)
- First Online:
- Cite this article as:
- Ahlf, W. & Heise, S. J Soils Sediments (2005) 5: 16. doi:10.1065/jss2005.01.127
Dedicated to Prof. Dr. Ulrich Förstner on his 65th birthday
Background, Aims and Scope
The main challenge in the application of biotests for hazard assessment of sediments is a transparent interpretation, especially if combinations of biotests are applied and an integrated assessment of all results is wanted. For decision making, assigning results of biotesting into different classes that indicate increasing hazards can be a useful tool. In this paper we suggest an approach on how to set up a site-independent classification system for sediment toxicity tests.
About 250 sediment samples were collected from two rivers and evaluated. The bioassays applied are used in a standardized procedure for all samples over years. The test battery include toxicological endpoints like nematode growth and reproduction, algae growth, bacteria activity, and luminescence inhibition.
Results and Discussion
The classification system described emphasizes the following steps: the assessment of test-specific response spans for each applied biotest and the subsequent interpretation of inhibition values in terms of toxicity, estimation of accuracy and uncertainty of different test systems, and the use of cluster- and k-means analyses in order to identify occurring pattern within a large biotest data base, followed by a ranking. The outcome of the procedure were 5 effect classes with an increasing potential of hazard.
The effect classes represent the result of a logical procedure of deduction. The procedure is transparent giving an increased degree of certainty. The site independent ranking of 5 classes could be used for sediment monitoring and facilitate fast interpretation.
Recommendation and Perspective
A large database comprising differently contaminated sediments is the precondition for setting up effect classes, because it is essential to know the range of effects and the variability of the test methods. The proposed procedure could then provide the basic rules for designing an expert system.