Critical review of methods to derive water quality guidelines for toxicants and a proposal for a new framework
Supervising Scientist Report 135
Supervising Scientist, 1998
ISBN 0 642 24338 7
- SSR135 - Critical review of methods to derive water quality guidelines for toxicants and a proposal for a new framework (PDF - 406 KB)
The aims of this document were to review and critically assess the available methodologies for deriving water quality guidelines (WQGs) and to substitute real toxicity data into these methods in order to recommend which method or methods were the most suitable for use by ANZECC.
There are two principle approaches to determining WQGs. The original method called the assessment factor method divided the lowest toxicity value by an assessment factor, the magnitude of which was based on the number, character and quality of the available toxicity data. The more data, and the more realistic they were, the lower the magnitude of the assessment factor. Typical assessment factors used are 10, 100 and 1000. The aim of such methods is to protect all species from lifetime exposures to toxicants. This type of approach is used by a variety of countries including Australia, New Zealand, USA, Canada, Denmark, The Netherlands and South Africa and the OECD has recommended it. A new approach-statistical extrapolation methods-has been developed since 1984. They use toxicity for all species that are available and fit a particular distribution to the data and from this calculate the concentration that should protect any percentage of species. This type of approach has also been adopted by many countries-in fact all the above mentioned countries except Australia, New Zealand and Canada use both the assessment factor and statistical extrapolation techniques. The USA, The Netherlands, OECD and South Africa use the statistical extrapolation techniques in preference to the AF methods unless there is insufficient data whereas Denmark prefers the AF method to the statistical extrapolation techniques.
Of the various versions of the AF method and the statistical extrapolation method the current ANZECC AF method and the Aldenberg and Slob method used by The Netherlands were chosen for detailed analysis. The critical evaluation of these two methods revealed that both had limitations. The major weaknesses of the assessment factor method are the arbitrary nature of the AFs, the questionable validity of the acute to chronic ratios, the assessment factors are too small to provide protection to all species and the method is not transparent ie. estimating the level of protection provided is not possible. The major weaknesses of the Aldenberg and Slob method are the limited number of WQGs that can be derived by this method, the questionable suitability of using no observed effect (NOEC) data, its relative complexity, and the validity of several of the assumptions may be compromised.
Comparison of the two methods in terms of the Precautionary Principle revealed that both methods had a number of precautionary elements. Despite this neither method was, in toto, precautionary because they both failed to account for the toxicity of mixtures, accumulation of toxicant in the animal tissue, and transfer of chemicals between the various compartments of the environment.
The comparison of 40 WQGs derived using real toxicity data by both methods indicated that the Aldenberg and Slob method derived WQGs that were significantly lower and offered a significantly greater degree of protection than the AF method. However, the WQGs derived by the Aldenberg and Slob method were not consistently lower than those derived by the AF method.
It was therefore concluded that it was not possible to state with any degree of confidence which of methods (Aldenberg and Slob or AF) was better.
The framework suggested for the derivation of WQGs is based upon this conclusion-it uses both methods (ie. the current ANZECC AF and the Aldenberg and Slob methods) to determine an estimate of the WQG for a chemical and then chooses the lower value as the WQG. Depending on the quality, quantity and type of toxicity data available a number of different methods are used to derive different types of WQGs (see Figure 1). There is a different level of confidence in each type of WQG that they provide the desired degree of environmental protection.
It is also proposed that three types of WQGs be derived based on the type, quality and quantity of the toxicity data available. For chemicals for which there is adequate suitable data Level I and Level II WQGs would be derived. There is a higher degree of confidence that Level 1 WQGs deliver the desired degree of environmental protection than Level II WQGs. However, in both cases there is sufficient confidence that the WQG should deliver the desired degree of protection. When the quality, quantity and type of toxicity data is not adequate then there is a low degree of confidence that the resulting WQG will provide the desired degree of protection. Such WQGs are termed 'interim' WQGs.
In the section comparing the precautionary natures of the AF and Aldenberg and Slob methods it was highlighted that a major limitation was that the toxicity of mixtures, accumulation of toxicant in the animal tissue, and transfer of chemicals between the various compartments of the environment were not considered. It was felt that given the current level of knowledge it was only possible to incorporate the toxicity of mixtures into a new method for deriving and using WQGs. Thus it is recommended that when water-resource managers use the WQGs the toxicity of mixtures be considered.