The goal of this report is to come up with recommendations on how to calculate the ecotoxicity effect indicator (termed ecotox effect indicator) for use in the OMNIITOX base model (BM). The ecotox effect indicator is used together with the input from the fate modelling to calculate a characterisation factor for the chemical in question.
Within the OMNIITOX project consortium it have been decided that the ecotox effect indicator for the BM should be able to work on minimum three measured EC50 acute laboratory test data. The main reason for this decision is that the BM should be able to work on a significant number of chemicals today and be in accordance with the most likely data availability in the near future as defined by the proposed EU chemicals policy REACH. The focus in this report is therefore on an ecotox effect indicator that is able to work on only three EC50 values and only the freshwater pelagic compartment is dealt with here.
For about a year discussions have been going on within the ecotox task force established within the OMNIITOX consortium, especially about which estimation principle to chose, i.e. whether it should be no-effect based (PNEC) or effect based (e.g. median or geometric mean). As a starting point existing approaches used within LCIA has been described and evaluated. Reports on these two issues (estimation principle and evaluation of existing approaches) are enclosed with this report as possible sources for background information
In this report further theoretical elaboration of effect based average approaches (arithmetic mean, geometric mean and median) and the non-effect based approach (PNEC, here only as lowest EC50 in the dataset) are done focusing on their statistical robustness and the possibility to relate the effect indicator (based on a measure of effect rather than a no-effect measure) to damage on the endpoint, the ecosystem. The approaches are also tested for their robustness in estimating HC50 in a practical test on datasets from eleven different substances comprising seven different toxic modes of action (TMoA).
On the basis of the theoretical considerations and the results of the practical test of the different approaches, the following recommendations are given for the estimation principle of the ecotox effect indicator:
• The indicator is based on the GM-trophic calculated as the geometric mean (HC50) of three EC50 values, one from each of the three trophic levels, primary producers, primary consumers and secondary consumers comprising three different taxa, i.e. algae, crustacean (invertebrates) and fish
• If more than one EC50 value from each trophic level is available then the GM-trophic is calculated on the basis of the geometric means for each trophic level (GM-trophic-levels). The GM-trophic-levels is calculated as the geometric mean of the geometric means at the genus level (GM-genuses) which again are calculated as the geometric means of the geometric means at species level (GM-species). GM-species is calculated as the geometric mean of the single EC50 values for each species
• As limit values around the GM-trophic, the lowest EC50 value is used as the lower limit and the highest EC50 value as the upper limit in data sets with only three EC50’s values, i.e. one from each trophic level
• If more than one EC50 value from each trophic level is available then the limit values around the GM-trophic are based on the three GM-trophic-level values, i.e. the lowest GM-trophic-level value is used as the lower limit and the highest GM-trophic-level value as the upper limit
It is recommended to use EC50chronic values when possible but as only acute data will be available in most cases, the use of best estimate assessment factors are recommended to extrapolate from acute to chronic values. Even though there is a need for research in this area, an acute to chronic ratio of 10 between HC50acute and HC50chronic is recommended as a starting point.
For several reasons (i.a.. the fact that one of the main applications of the ecotox effect indicator is LCIA where the results are used in a comparison between substances), it is recommended only to use test results from laboratory tests, fulfilling certain standard conditions, e.g. standard organism and test duration restrictions.
The ability of a geometric mean to represent the toxicity of very toxic substances and very sensitive species has not been dealt with yet, and further research is needed here. However, it may be anticipated on the basis of the results from the practical test of different average approaches on substances with different TMoA, that the GM-trophic with its limit values at least to some degree accounts for very toxic substances if representative toxicity data are available.