Accounting for local conditions

Metals

In the case of metals, the first option is to test the total (particulate plus dissolved) metals concentration against the guideline value.

More commonly, since it is generally the dissolved fraction that is bioavailable rather than any particulate forms, the fraction that passes through a 0.45 µm filter membrane is compared to the guideline value.

Another option to determine the bioavailable fraction in a sample is to use metal speciation measurements or modelling. This may be necessary, for example, if a hardness-modified guideline value is still exceeded. Labile metals that relate to a bioavailable fraction can be determined using Chelex® columns (Bowles et al. 2006), anodic stripping voltammetry or diffusive gradients in thin-films (DGT) gel samplers (Batley et al. 2004, Huynh & Vink 2016). Refer to ANZECC & ARMCANZ (2000) Section 8.3.5.16 for information on chemical speciation measurement and modelling techniques.

If a biotic ligand model (BLM) has been developed for a metal, you can use the BLM to predict the speciation of the metal in the environmental medium. The complexity of the BLMs requires the prior existence of an intensive calibration effort and multi-parameter mathematical calculations for implementation. BLMs are currently applicable to a limited range of metals and do not specifically handle metal mixtures, even though mixtures are commonly present at contaminated sites.

Organic toxicants

pH will be a potential modifier of bioavailability for those organic toxicants with ionisable functional groups. Other solution parameters may also affect solubility and partitioning of the more hydrophobic organics.

Few studies to date have looked at speciation and bioavailability of organic contaminants in any detail. There is potential for many of these to be associated with colloidal particulates, and separations based on ultrafiltration may be worth investigation.

The use of passive samplers represents best practice in monitoring of dissolved organics in waters (e.g. Vrana et al. 2005, Ghosh et al. 2014). Passive samplers can identify sources of contaminants where extremely low levels have to be detected or when the input and presence of the contamination is not constant.

Some surface waters contain concentrations of toxicants that may naturally exceed the DGVs. If this is the case, then you can use the referential approach to derive new guideline values based on natural background or baseline data. This applies mostly to metals and metalloids.

In highly mineralised areas, the background may involve mainly mineralised biologically unavailable metal forms.

The very few organic toxicants where background concentrations might be elevated include some phenols and globally distributed chemicals, such as DDT and other persistent organic pollutants (POPs), with rare exceptions, such as hydrocarbon-rich fluid seeps.

Gathering background data is always recommended — at least in the initial stages of a water quality management program — to establish whether or not concentrations of toxicants are naturally high.

Toxicant concentrations may vary seasonally, which is why you need confidence in the best estimate of background concentrations. To increase confidence in the estimate, you should gather data at least monthly for at least 2 years. In all respects, data requirements and collection are the same as for physical and chemical (PC) stressors.

Until the minimum data requirement has been established, you should compare the test site median with reference to the DGV. For those months, seasons or flow periods that constitute logical time intervals or events to consider and characterise background water quality, the 80th percentile of background data (from a minimum of 10 observations) should be compared with the DGV. Use the 80th percentile value as the new guideline value for this period if it exceeds the DGV.

In naturally mineralised catchments where background concentrations of certain inorganic compounds can be high and well above DGVs, there is often little evidence for enhanced tolerance by the ecosystem. It may be naturally stressed by the elevated toxicant concentrations and unable to tolerate substantial further increases in concentration. If this is the case, then the current understanding of the system should capture the status of the ecosystem under baseline conditions and its ability to tolerate increased contaminant concentrations. This is particularly important if the local biotic assemblage (plants, animals and microbes in the aquatic ecosystem) has naturally reduced biodiversity.

If background toxicant values fall consistently below DGVs, then sampling intensity at these sites could be reduced after a suitable period (e.g. 2 years).

DTA might be usefully applied to the background or reference waters to confirm that there is no toxicity (van Dam & Chapman 2001). If toxicity is observed, you could apply toxicity identification and evaluation (TIE) to try to determine the compound classes causing toxicity (Burkhard & Ankley 1989, USEPA 1992, 1993, Burgess et al. 2013).

Particularly where guideline values are derived from extrapolation of a species sensitivity distribution (SSD), it is possible that they may be near or below the analytical practical quantitation limits (PQLs) or even the method detection limits (DLs), which are often around one-fifth of the PQL.

It is not reasonable to automatically raise a guideline value to the analytical PQL if this is likely to result in environmental harm.

If such chemicals are detected with reasonable assurance, it may indicate that the guideline value is exceeded.

Sometimes it may be possible to introduce effluent controls to ensure that calculated ambient levels in the receiving ecosystem do not exceed guideline values. In other cases, low guideline values may encourage development of more sensitive analytical techniques. The measurement of chemical concentrations at levels below those currently available may well be feasible in the near future.

If the guideline value is below an analytical PQL or DL, biological data (e.g. DTA, field biological monitoring) provide the only certainty that the environment is protected. If testing of ambient waters indicates a toxic effect, management action should be initiated. This could include the use of TIE techniques, which can assist in identifying the unmeasured toxicant source.

The issue of guideline values being below the PQLs of commonly used analysis methods raises the possibility that a laboratory using a method with a relatively high level of detection may deliver only nondetectable results even if the true concentrations are at or above the guideline value. Water authorities may reserve the right to question and reject nondetectable results if there is concern that an appropriate methodology has not been used.

Given the possibility of analytical error around the PQL — resulting in false positives — it may be better to invest in analytical methods with lower PQLs if available, or use biological data (e.g. DTA) in preference to the chemistry line of evidence than to rely on a cheaper analysis method with a PQL that is too high.

Increases in water hardness reduce the toxicity of some metals. The hardness corrections used in the ANZECC & ARMCANZ (2000) guidelines followed those used by the US Environmental Protection Agency. All the toxicity test endpoint values were adjusted to a standard hardness value of 30 mg CaCO3/L, which was considered appropriate for the generally soft waters present in Australia and New Zealand. These corrections were largely based on acute toxicity data for fish but were being applied to modify guideline values based on chronic toxicity data.

Updated guidance on guideline-value derivation by Batley et al. (2018) and Warne et al. (2018) advised that no hardness adjustment should be undertaken for copper for chronic toxicity, but that hardness adjustments should still be incorporated for other hardness-sensitive metals (cadmium, chromium, lead, nickel, zinc) until otherwise advised.

If the water being examined has a hardness greater than 30 mg CaCO₃/L, then it is appropriate to modify the DGV for that hardness.

Competition with hydrogen ions, or complexation by DOC or alkalinity (bicarbonate), can reduce metal bioavailability.

Methods for dealing with these modifying factors involve:

• modifying guideline values for the site-specific solution parameters, or
• measuring or modelling the bioavailable fraction of the total dissolved metal fraction and then comparing that with the DGV.

Modelling could include application of the BLM (discussed later) or geochemical models of solution speciation.

One of few examples of a DOC correction for a guideline value in Australia was described for uranium by van Dam et al. (2017). However, that correction was for a site-specific guideline value, for which many site and species-specific data were required to develop the guideline value and associated correction.

Guideline values have been derived for both freshwater and marine water, for appropriate use in each system. Few toxicity data exist for estuarine organisms, which live in a mix of freshwater and salt water, as well as for organisms that live in hypersaline water. In most cases, it will be necessary to make best estimates of the appropriate guideline value for such ecosystems.

Exercise care when comparing figures for freshwater and marine water guideline values. We do not recommend making such comparisons to estimate salinity effects in brackish or hypersaline water bodies without closely examining the supporting data, because different types of datasets were used to calculate each guideline value.

For metals that are affected by hardness or other key toxicity modifying factors (e.g. pH, DOC, alkalinity), it is possible to use the freshwater DGVs for salinities up to 2 ppt (2000 mg/L) by applying the hardness algorithms in Warne et al. (2018) or other such corrections as may be published in the future for certain metals. For higher salinities, the marine DGVs for metals are recommended unless available evidence suggests otherwise. For example, the decisions should take into account the relative reliability of marine and freshwater DGVs, and whether there is or appears to be a difference in toxicity between freshwater and marine water.

For freshwater, the additional stress of higher toxicant concentrations on top of elevated salinity levels may result in mixture toxicity. The nature of this mixture toxicity cannot be readily predicted at present.

If you are uncertain about the appropriate guideline value to use for an estuary or a freshwater ecosystem with elevated salinity, then we recommend that you include ecosystem receptor lines of evidence in the assessment and consider developing a locally relevant guideline value.

In recent years, an alternative methodology for guideline-value derivation internationally has been to use biotic ligand models (BLMs) for metals (including copper, nickel, silver, zinc and manganese) to predict toxicity after incorporating the effects of the water quality modifying factors.

The central notion of a BLM is that organisms possess a key binding site (biotic ligand) that interacts with the toxic metal and other cationic components of the water, and that the degree of accumulation of the toxic metal at the binding site is directly related to the observed toxicity.

By defining and determining equilibrium constants for the reactions of the biotic ligand with key ionic species, including those of the metal of interest, BLMs take a wide range of ionic interactions into account. This allows the models to describe metal interactions with organisms over a range of water chemistry conditions. The interactions that the models take account of include:

• solution complexation
• competition by protons and base cations (Na⁺, Mg²⁺, K⁺, Ca²⁺)
• binding with DOC over a range of pH conditions.

These derivation procedures are complex and a suite of BLM-binding equations must be developed for acute and chronic toxicity data for key taxa groups (i.e. fish, invertebrates and algae). This can allow bioavailability corrections to be made for the ecosystem, rather than individual species, enabling bioavailability principles to be applied to derive water quality guideline values.

Development of BLMs has so far been based almost entirely on data from laboratory studies, although many of the validation tests have been performed on field collected water samples.

The US Environmental Protection Agency has used a BLM to derive a water quality criterion for copper (USEPA 2007). The US implementation of a BLM for copper requires that water quality information be available for a site prior to being able to calculate a numeric guideline value. More recently, a BLM has been used to derive predicted no-effect concentration (PNEC) values for copper in Europe (De Schamphalaere & Janssen 2008, Bass et al. 2008). This differs from the US approach in that a single environmental quality standard (EQS, expressed as ‘bioavailable metal’) is derived from the PNEC, and the BLM can be modified based on local exposure water quality conditions to provide a site-specific biological exposure concentration for that metal. The derivation of the EQS uses the BLM to provide normalised toxicity data as part of the derivation process, and the local bioavailability corrections are derived from BLM predictions of local ecosystem sensitivity.

BLM calculations can be used to:

• adjust toxicity values to a ‘normalised’ set of water quality conditions prior to deriving water quality guideline values, so that the sensitivity of all species is expressed under the same water chemistry conditions
• make site-specific adjustments to DGVs based on site-specific water quality conditions.

For copper, the US and European BLMs differ in their origin and incorporate different predictive equations, due to differences in the approaches used (i.e. being based principally on acute or chronic toxicity information). As such, they predict different responses to water quality modifiers, thus requiring specific consideration as to the appropriateness of their use.

A recent report by Hickey et al. (2016) investigated these options and compared them to an approach for zinc proposed by Environment Canada (2016), which generated relationships between modifiers and guideline values and provided look-up tables for categorised modifiers (considered a reasonable short-term option).

Regardless, existing US and European BLMs can be used as part of a consideration of metal bioavailability under the local water composition for Australian and New Zealand waters.

It would be considered appropriate to use existing BLM based approaches (if a suitably validated one exists for the metal and exposure period of interest for the ecosystem) for a site-specific assessment to derive a site-specific guideline value based on local water quality composition, and to use this as an additional line of evidence alongside the DGV.

Existing BLMs have not been validated for Australian and New Zealand conditions and, consequently, have not been incorporated into DGV-derivation processes. Research has been undertaken to validate the European nickel BLM for Australian and New Zealand conditions but the results are yet to be published and the method yet to be proposed for implementation.

At this stage, the use of BLMs is recommended only for site-specific assessments. You will need to consider what water quality variables (e.g. pH, hardness, alkalinity, DOC) act as important toxicity modifiers for the metal of interest when adapting the guideline value for local conditions. The process of normalising toxicity data — or correcting guideline values — to a standard water quality condition for the site of interest is complicated by the fact that toxicity modifiers may have:

• different response relationships for different taxonomic groups
• response relationships that depend on the duration of test exposures (whether acute or chronic).

Alternatives to application of a BLM-modifier approach are to censor the toxicity data in relation to the modifiers — to bound variability which might be associated with that parameter — or in some cases to exclude data where there is a high likelihood that strong effects will be occurring (e.g. DOC).

Many chemicals, such as pesticides, are released into the environment as part of a commercial formulation.

Guideline values for such chemicals should be based on toxicity tests where the test organism was exposed to a reasonably pure form of the active ingredient (AI) — typically more than 70% — such as a technical grade form of the chemical.

In practice, you should compare the toxicities of the commercial formulation and the AI because other chemicals present may modify the toxicity of the AI.

When refining a guideline value based on the AI for local conditions, it will be important to take into account the toxicity of the commercial formulation to protect the environment from it. Any formulation toxicity data must have gone through a screening and quality assurance process before it can be used in this comparison.

If the toxicities of the commercial formulation and the AI differ by greater than a factor of 3, then it would be appropriate to correct the guideline value by the difference. The resulting formulation-corrected guideline value might be larger or smaller than the guideline value of the AI.

If you only have qualitative information on the relative toxicity of the AI and the commercial formulation, then you should make a statement on whether the guideline value is likely to provide sufficient or insufficient protection.

Express guideline values for AIs as a concentration of the AI (e.g. x µg AI/L).

What is bioaccumulation?

According to Gobas et al. (2009), bioaccumulation is a process in which the concentration of a chemical in an organism achieves a level that exceeds that in the respiratory medium (e.g., water for a fish or air for a mammal), the diet, or both.

Bioaccumulation occurs by two main pathways: bioconcentration and biomagnification. Bioconcentration involves the absorption of a chemical substance exclusively from the ambient environment, i.e. through respiratory and dermal surfaces only (Arnot and Gobas 2006). Biomagnification relates exclusively to dietary intake of chemicals. It is defined as the process whereby pollutants are transferred from food to an organism resulting in higher concentrations in the organism compared with the source (Mann et al. 2011). Biomagnifying toxicants may also exhibit trophic magnification. Trophic magnification occurs when the concentration of the toxicant in organisms increases as it passes up two or more trophic levels in a food chain (Gobas et al. 2009).

Why do bioaccumulative toxicants require additional consideration?

Bioaccumulative–relative to non-bioaccumulative–toxicants may pose an increased risk of toxic effects to populations, due to an additional body burden and/or duration of exposure resulting from internal concentration of the toxicant. Bioaccumulative toxicants also present a risk of intergenerational effects, by maternal transfer of accumulated toxicants to offspring (Niimi 1983, Wanget al. 2015). Trophic magnification of toxicants threatens entire food webs, including higher trophic levels (i.e. top predators), within which tissue toxicant concentrations can be at their highest (Kelly et al. 2009). Some toxicants have the potential to biomagnify to concentrations that risk harmful effects across trophic levels (e.g. DDT, perfluoroalkylacids (PFAAs), dioxins, methyl-mercury). Trophic magnification threatens not only aquatic taxa, but also associated terrestrial taxa, e.g. birds that feed upon fish.

The DGVs are largely derived from single-species toxicity data that only account for direct effects of toxicants observed at the end of relatively short exposure durations. For bioaccumulative toxicants, such data are unlikely to capture toxicity caused as a result of longer-term bioaccumulation. The effects of bioaccumulative toxicants are best assessed using chronic field, mesocosm or microcosm data, or chronic laboratory toxicity data from tests carried out over multiple generations. Long-term chronic tests are best able to capture toxic effects of toxicants that bioaccumulate in tissues and manifest over longer timescales; multigenerational tests are best able to capture effects involving reproductive toxicity or maternal transfer of toxicants; and microcosm/mesocosm tests are best for detecting effects resulting from trophic magnification.

Which toxicants are bioaccumulative?

Bioaccumulation is typically measured in the field (Arnot and Gobas 2006). Toxicants considered ‘bioaccumulative’ for water respiring organisms (e.g. fish and macroinvertebrates) are those with log BAF (bioaccumulation factors) values ≥ 4. [Biomagnifying toxicants are those with a biomagnification factor (BMF) > 1 and trophically magnifying toxicants are those with a TMF (trophic magnification factor) > 1 (Gobaset al. 2009).]

BAFs are derived from field measurements of the ratio of the steady state chemical concentrations in an aquatic water-respiring organism and the water determined from sampled organisms exposed to a chemical in the water and in their diet (Gobaset al. 2009). They are considered to be reliable indicators of bioaccumulation. However, literature BAF values are often unavailable for toxicants (Arnot and Gobas 2006). Such toxicants may be deemed ‘potentially bioaccumulative’ based on indirect measures, e.g. (Gobaset al. 2009) where their:

• log K_ow (octanol–water partition coefficient) ≥ 4, or
• log BCF (bioconcentration factor) ≥ 4.

However, the risks of applying these inferential approaches should be noted, especially for emerging contaminants such as per- and poly-fluoroalkyl substances (PFAS). Inferring a toxicant’s potential to bioaccumulate from its log K_ow, for example, assumes the toxicant concentrates in lipid tissue. This is true of many historic persistent organic pollutants, but does not apply to all toxicants. PFAS have a much higher affinity for proteins, hence their bioaccumulation potential cannot be reliably inferred from their log K_ow values (Ng and Hungerbühler 2014). Similarly, log BCF values, by ignoring dietary intake, significantly underestimate the considerable bioaccumulation potential of many biomagnifying chemicals, e.g. perfluorooctane sulfonate (PFOS) (Gobaset al. 2009).

Field measurements of bioaccumulation, e.g. BAFs, BMFs and TMFs, should therefore be considered the most reliable indicators of a toxicant’s bioaccumulation potential (above inferences based upon indirect measures such as log K_ow and log BCF).

What is the GV approach for bioaccumulative toxicants?

If a toxicant has the potential to bioaccumulate, and the dataset used to derive its DGV did not include a significant proportion of (a) long-term mesocosm/field effects data, or (b) multigenerational tests where both bioaccumulation and ecologically relevant effects were measured (e.g. > 30% of the dataset and for > 3 taxa groups), we recommend using the 99% species protection DGV for slightly to moderately disturbed ecosystems instead of the 95% species protection DGV. Similarly, the 95% species protection DGV could be used for highly disturbed ecosystems instead of the 90% species protection DGV. A lower level of protection than this for highly disturbed sites is not recommended, due to the potential risks associated with bioaccumulative toxicants. However, a lower level of protection (e.g. 90% species protection DGV) could potentially be adopted as an interim target for a specified indicator only if it was evident and agreed by stakeholders, including regulators, that the 95% species protection DGV could not be attained in an acceptable timeframe (also see guidance on Level of protection for how to apply guideline values to highly disturbed ecosystems).

The recommended approach has no mechanistic basis with regards to bioaccumulation, and is recommended solely as a precautionary measure to minimise the risks from bioaccumulative substances. A more mechanistically-based approach may involve, for example, the future development of tissue residue guidelines (ANZECC & ARMCANZ 2000, pp. 8.3–17 to 20).

If, through a rigorous weight of evidence process, it can be shown that aquatic and terrestrial species are not experiencing harmful direct and indirect effects from the toxicant due to bioaccumulation at concentrations of the DGV that would normally have been applied based on the agreed ecosystem condition (e.g. the 95% species protection DGV for slightly to moderately disturbed ecosystems), and the community values are being protected, then this should be sufficient to support a corresponding change back to that DGV. Such decisions should always be made in consultation with regulators and the community and in accordance with the guidance on levels of protection for aquatic ecosystems.

Finally, the over-riding principle of continual improvement dictates that, where the concentration of a contaminant is below the appropriate guideline value, then the over-riding objective should be to continue to improve, or at least maintain, water quality (i.e. not to allow increases in concentration up to the guideline value).

For additional context and guidance on how to deal with bioaccumulation, refer to:

• Warne et al. (2018) toxicant guideline value derivation method
• ANZECC & ARMCANZ (2000) Section 8.3.3.4 background to incorporating bioaccumulation into guidelines values.

Science has progressed since ANZECC & ARMCANZ (2000), and some of the guidance may be outdated. Further guidance on guideline values that incorporate the impacts of bioaccumulative substances may need to be developed in the future.

Transient or short-term exposures of aquatic biota to toxicants can occur for various reasons, such as:

• highly volatile or hydrophobic toxicants that are rapidly lost from the water column
• toxicants with very short half-lives
• toxicants present in episodic or intermittent discharges (e.g. mine water discharges, stormwater discharges, pesticides in waterways draining agricultural land).

The current DGVs may not be relevant to such exposures but would at least be likely to be over-protective.

Short-term exposure DGVs have not as yet been derived for Australia and New Zealand. Such DGVs are focused at ensuring that short-term acutely toxic events do not occur and, as such, are often derived from acute EC/LC50 data.

Concentrations at which acute toxicity is likely to occur may not necessarily bear resemblance to the concentrations that should protect against transient exposure. This is because some toxicants are known to cause delayed effects after transient exposures have ceased.

Where site-specific transient exposures are known to occur, and the DGVs are considered to be inappropriate, Batley et al. (2018) and Warne et al. (2018) provided guidance on how to derive short-term guideline values.

Importantly, the exposure regime used to derive the toxicity data should be as similar as possible to the exposure regime being considered. This includes how frequently the pulses occur, the likely period needed for recovery and the cumulative effect of repeated pulses. If pulses are frequent with minimal recovery period it may be more appropriate to use the DGVs.

A useful example of the development of transient exposure guideline values is provided in the case study for the assessment of mine water discharges from Ranger uranium mine.

Additional details of research completed to develop these guideline values can be found in Hogan et al. (2013), Sinclair et al. (2014) and Prouse et al. (2015).

Whilst site-specific guideline values should be developed for specific issues and locations, it is not always going to be practicable or necessary.

Local ecosystems or species are sometimes deemed so ecologically, economically or culturally important that we need assurance the guideline values will protect them enough.

A good example of this is the need to ensure that the highly valued aquatic ecosystems of Kakadu National Park are protected from the impacts of uranium mining. Because of these high ecological and cultural value ecosystems, much effort has been invested in deriving site-specific guideline values for toxicants of concern based on toxicity data from local species in local waters.

An example is provided in the case study for the assessment of mine water discharges from the Ranger uranium mine. Additional details can be found in van Dam et al. (2010), Hogan et al. (2013), Harford et al. (2015) and van Dam et al. (2017).

When incorporating local species toxicity data, take care when excluding or including the original data used in the DGV derivation (if it exists).

This decision often becomes a trade-off between sample size (number of values used to derive the DGV) and relevance of the data to the ecosystem and species of interest. Considerations include whether the species and type of water quality used to assess toxicity of the toxicant in question are sufficiently representative of those for the ecosystem of interest. The requirements for data quality and quantity, including taxonomic representation, always need to be met. Seek expert guidance to help with making decisions about inclusion or exclusion of data.

Refer to relevant background information provided in ANZECC & ARMCANZ (2000) Section 8.3.5.8.

Even though the DGVs are based on data from toxicity tests for single toxicants, it is well known that toxicants rarely occur in isolation. In most circumstances, a particular toxicant will be present in combination with numerous other toxicants, and their combined presence may alter their toxicity. There are numerous terms for the various joint toxicities of mixtures, but at their simplest, mixtures of toxicants (Rand 1995) can result in:

• additive toxicity (sum of the toxicity of the individual components)
• greater than additive toxicity (also known as ‘synergism’), or
• less than additive toxicity (also known as ‘antagonism’).

The brief guidance provided here is generally reflective of that provided in ANZECC & ARMCANZ (2000) Section 8.3.518.

A thorough review of the science of mixture toxicity and associated methods for its assessment is beyond the scope of the Water Quality Guidelines but is likely to be necessary in the future if updated guidance on dealing with multiple stressors is to be provided.

A number of reviews on mixture toxicity and associated methods of assessment since the ANZECC & ARMCANZ (2000) guidelines (e.g. de Zwart & Posthuma 2005, Cedergreen 2014, Liu et al. 2017) suggest that its guidance is still applicable.

Cedergreen (2014) concluded that true synergistic interactions between chemicals are rare, and that addressing the cumulative rather than synergistic effect of co-occurring toxicants was the most important step in assessing the effects of toxicant mixtures.

Similarly, Liu et al. (2017) found that statistically significant deviations from additivity are not necessarily biologically relevant for metal mixtures and, as such, recommended to first use a relatively simple method for effect prediction of un-investigated metal mixtures.

Two types of models are most often used to predict mixture toxicity:

• concentration addition model of joint action — applied to mixtures with toxicants that have the same mode of action
• response addition model of independent action — applied to mixtures with toxicants that have different modes of action.

If we assume toxicity conforms to the concentration addition model (it is additive), then significant combined effects could be expected if all toxicants in a mixture are present at close to their DGVs.

You can determine whether or not a mixture exceeds a collective water quality guideline value using the formula (modified from Vighi & Calamari 1996):

TTM = Σ(C_i / WQG_i)

where TTM is the total toxicity of the mixture, C_i is the concentration of the ‘i’th toxicant in the mixture and WQG_i is the guideline for that toxicant.

If TTM exceeds 1, the mixture has exceeded the collective water quality guideline value.

The ANZECC & ARMCANZ (2000) guidelines stated that this method should only be applied to mixtures of up to 4 components. It is now recommended that this method be applied irrespective of the number of components in the mixture.

A number of studies have shown that the concentration addition model consistently predicts higher toxicity (and hence is more environmentally protective) than the response addition model (e.g. Faust et al. 1994, Backhaus et al. 2000, Chevre et al. 2006). This has led to the accepted practice of using the concentration addition model as a first step in estimating the toxicity of mixtures.

A more accurate estimate can be provided using a 2-step method (e.g. Altenburger et al. 2004, De Zwart and Posthuma 2005), which applies the concentration addition model to chemicals with the same mode of action, but then uses the response addition model to estimate the toxicity of groups of chemicals with different modes of action.

The best experimental method to account for the toxicity of mixtures is DTA of the mixture, which may be an effluent or ambient water. Refer to details in our guidance on DTA.