Ecotoxicity data

During both steps of the effects assessment it is very important to evaluate data with regard to their adequacy and completeness. This is particularly important for well studied existing substances where there may be a number of test results available beyond the base-set. This section puts forward general guidelines on data evaluation of ecotoxicity data. The term adequacy is used here to cover the reliability of the available data and the relevance of that data for environmental hazard and risk assessment.

3.2.1.1 Completeness of data New substances

For new substances data equivalent to those foreseen in Annex VII A to Directive 67/548/EEC will be available: the base-set. This base-set comprises short term toxicity data for algae, Daphnia and fish for the aquatic compartment. Data for bacteria (respiration inhibition test) are also part of the base-set. These data are used for assessing the effects on microbial activity in a STP (see section 3.4). The base-set testing package contains relatively little data which are of relevance to the terrestrial and atmospheric compartments: further but nevertheless limited data are foreseen at level 1 and 2.

Existing substances

The quantity of data available for existing substances varies considerably. Regulation 793/93 requires that for priority substances at least the base-set according to Annex VII A to Directive 67/548/EEC is provided before the risk assessment process begins. However, for many substances more information will be available which can be used in the assessment.

The base-set ensures that short-term effect data are available for fish, Daphnia, algae and bacteria. Within a trophic level, a number of short term investigations may be available for several non-standard organisms. In addition, long-term toxicity investigations may also be available with several species, standard organisms as well as non-standard organisms. For the derivation of the PNEC these organisms should be assigned to appropriate trophic levels (see Appendix IV and section 3.3.1). Multi-species tests, investigations with model ecosystems and semi-field tests, are rare for most substances although in recent years more work has been done in this area (Hill et al., 1994; Knacker and Morgan, 1994).

3.2.1.2 Adequacy of data

The adequacy of a test can be considered to be defined by two basic elements:

• reliability: covering the inherent quality of a test relating to test methodology and the way that the performance and results of the test are described;

• relevance: covering the extent to which a test is appropriate for a particular hazard or risk assessment.

Reliable, relevant data can be considered valid for use in the risk assessment.

The assessment of data adequacy involves therefore a review of individual data elements with respect to how the study is conducted and how the results are interpreted and a critical selection (and rejection) of data in its proper context and in accordance with the purpose of the assessment.

New substances

The tests for new substances must be carried out in accordance with the EU test guidelines as laid down in Annex V to Directive 67/548/EEC, or if no EU guidelines are available or they are not applicable, in accordance with internationally recognised guidelines, preferably those of the OECD (OECD, 1993b). They must also be conducted in accordance with the principles of good laboratory practice as set out in Council Directive 87/18/EEC.

Existing substances

The risk assessment for existing substances starts with collecting all available information by the manufacturers, importers, and rapporteur. Any new tests carried out for risk assessments under Regulation 793/93 should be conducted according to the methods laid down in Annex V to Directive 67/548/EEC, or if no EU guidelines are available or they are not applicable, in accordance with internationally recognised guidelines, preferably those of the OECD (OECD, 1993b). They must also be conducted following good laboratory practice according to Directive 87/18/EEC.

This information will probably contain data which have been generated prior to the requirements of GLP and the standardisation of testing methods. However, these data may be used for the risk assessment, if valid conclusions can be drawn from them. This means that the data, and the test methods used to generate them, must be evaluated in order to determine whether they are of sufficient quality for use in risk assessment. Such an evaluation will require the use of expert judgement, but the determination of data as being valid or not valid must be both justified and transparent. The requirements of the standardised test methods and GLP principles should be regarded as a reference when evaluating the available tests. In order to allow any judgement on the reliability of a study sufficient information must be available.

Whenever studies carried out according to current methods (e.g. EC, OECD, or EPA) are available, greater weight should normally be attached to them (Ahlers et al., 1992; Water Quality Institute, Denmark). On this background, the criteria for the reliability of a test are concentrated to the question whether sufficient information on the test is available and whether the investigations were carried out according to the generally accepted standards. The following possibilities exist:

• A complete test report is available or the test has been described in sufficient detail and the test procedure is in accordance with generally accepted scientific standards.

These data are considered valid and can be used for risk assessment;

• The validity of the data cannot fully be established or the test differs in some respects from the test guidelines and the generally accepted scientific standards. Experts must decide in each case whether the test will be taken into consideration in the risk assessment or is regarded as not valid;

• It is obvious that the data are not valid because critical pieces of information are not available (e.g. it is not possible to establish the identity of the test substance).

These data are considered not valid and should not be used for the risk assessment. However, they may be used to design an appropriate test.

In principle, the same criteria apply for tests reported in published literature. The amount of information presented will provide the basis to decide on the quality of a report. In general, publications in peer reviewed journals are preferable. High quality reviews may be used as supporting information. Summaries or abstract publications may also supply supporting material.

In the cases, where differing results from similar studies were obtained or an extensive data set is available for an individual species or a taxonomic group, it may be possible to use the distribution of these data to draw general conclusions regarding the toxicity to that species.

Results from field studies may also be available. These studies can vary in experimental design: from indoor microcosms to outdoor macrocosms like experimental streams (Hill et al., 1994). Field studies may provide a better insight into the effects (including indirect effects), as well as routes of exposure (e.g. bioavailability, biodegradation) of the chemical considered. At present, there are no internationally accepted guidelines for field tests.

However, some general guidance has been given for field studies with aquatic ecosystems (SETAC, 1991; SETAC, 1992).

Relevance of data

In order to evaluate the relevance of the available data, it is necessary to judge, inter alia, if the appropriate endpoints are studied under relevant conditions and if the substance tested is representative for the substance as supplied. To be able to assess the latter it is necessary that the substance is properly identified and any significant impurities are described.

Interpretation of data

In some cases, information on the dose (concentration) - response (effect) relationship is not known since the corresponding data are not reported in the testing protocols or publications. The duration of a test may be different from that of standard tests.

Sometimes, the test parameters may not be comparable to those used in standard tests, for example investigations of photosynthesis, of behaviour, investigations on a cellular or a subcellular level. Expert judgement must therefore be used to determine whether such data can be interpreted for use in the assessment.

Short-term L(E)C50 and long-term NOEC values are used in the effects assessment. However, results from ecotoxicological studies may be reported in a different way. In Table 13 guidance is given with respect to the derivation of L(E)C50 and NOEC values.

In assessing long term aquatic toxicity tests with very hydrophobic organic chemicals such as PCBs, QSARs may be helpful. Due to their low water solubility, long term tests with such chemicals are difficult to perform as stable test concentrations are difficult to maintain. Also, it may take very long to reach steady state in the test organisms due to their low elimination rate. By comparing the test result with the "minimum toxicity" using the log Kow of the compound, insight can be gained into the validity of the test (see Chapter 4 on the "Use of QSARs").

Further details on the evaluation of the adequacy of data are to be found in Appendix III.

Special guidance for metals and metal compounds, petroleum substances and ionisable substances is given in Appendix VIII, IX and XI, respectively.