tissue residue approach for chemical mixtures

e 7, Number 1—pp. 99–115
Integrated Environmental Assessment and Management — Volum � 2010 SETAC 99
Tissue Residue Approach for Chemical MixturesScott Dyer,*y Michael St J Warne,z Joseph S Meyer,§ Heather A Leslie,k and Beate I Escher#,yyyProcter & Gamble, 11810 East Miami River Road, Cincinnati, Ohio, 45201, USAzCentre for Environmental Contaminants Research, Commonwealth Scientific and Industrial Research Organisation (CSIRO), Adelaide,South Australia, Australia§Department of Zoology and Physiology, University of Wyoming, Laramie, Wyoming, USA and ARCADIS U.S., Inc., Lakewood,Colorado, USAkInstitute for Environmental Studies, Free University of Amsterdam, The Netherlands#Department of Environmental Toxicology (Utox), Swiss Federal Institute of Aquatic Science and Technology (Eawag), Dubendorf,SwitzerlandyyThe University of Queensland, National Research Centre for Environmental Toxicology (Entox), Brisbane, Queensland, Australia

(Submitted 23 December 2009; Returned for Revision 10 March 2010; Accepted 9 June 2010)

EDITOR’S NOTEThis paper represents the one of six review articles generated from a SETAC Pellston Workshop entitled ‘‘The Tissue

Residue Approach for Toxicity Assessment (TRA)’’ (June 2007, Leavenworth, Washington, USA). The main workshopobjectives were to review and evaluate the science behind using tissue residues as the dose metric for characterizing toxicresponses and to explore the utility of the TRA for mixtures, guidelines or criteria, and ecological risk assessment.

Specia

lSerie

s

ABSTRACTAt the SETAC Pellston Workshop ‘‘The Tissue Residues Approach for Toxicity Assessment,’’ held in June 2007, we discussed

mixture toxicology in termsof the tissue residue approach (TRA). This article reviews the literature related to the TRA formixtures

of chemicals and recommends a practical, tiered approach that can be implemented in regulatory or risk assessment

applications. As with the toxicity of individual chemicals, addressing mixture toxicity by means of the TRA has a number of

significant advantages. Early work provided a theoretical basis and experimental data to support the use of TRA for mixtures;

later work provided a field-based validation of the integration. However, subsequent development has been hindered by the

lack of mixture toxicity data expressed in tissue or preferably target-site concentrations. We recommend a framework for

addressing the toxicology of mixtures that integrates the TRA and mixture toxicology in a 3-tier approach. Tier I uses

concentration addition (CA) to estimate the toxicity of mixtures regardless of the mechanism of action of the components.

However, the commonapproach that uses a bioaccumulation factor (BAF) to predict TR from the exposure–water concentration

for organics must be modified slightly for metals because, unlike organics, the BAF for a metal changes as 1) the aqueous

exposure concentration changes, and 2) the concentration of othermetals changes. In addition, total tissue residues of ametal

are not a good predictor of toxicity, because some organisms store high concentrations ofmetals internally in detoxified forms.

In tier I, if the combination of measured concentrations in the mixture exceeds that predicted to produce adverse effects or

above-reference levels, it is necessary to proceed to tier II. Tier II is a mixed model that employs CA and independent action to

estimatemixture toxicity. Tiers I and II estimate the toxicity ofmixtures to individual species. In tier III, the TRA is integratedwith

themultisubstance potentially affected fraction (ms-PAF)method to derive TR levels that are protective of a selected percentage

of species in aquatic communities (e.g., hazardous concentration for 5% of the species [HC5]). Integr Environ Assess Manag

2011;7:99–115. � 2010 SETAC

Keywords: Tissue residue approach Mode of toxic action Internal concentration Concentration additionIndependent action

INTRODUCTIONToxicity tests, in which single species are exposed to

individual toxicants, are the predominant means used toassess the potential impact of substances in the aquaticenvironment. In the environment, however, multiple speciesare invariably exposed to mixtures of substances, although alimited number of chemicals may dominate. These mixturesmay vary in terms of their complexity (i.e., number of com-ponents), timing of their occurrence (e.g., simultaneous orsequential or combinations of both), and types of chemicals

* To whom correspondence may be addressed: [email protected]

Published online 8 July 2010 in Wiley Online Library

(wileyonlinelibrary.com).

DOI: 10.1002/ieam.106

and their toxicokinetics and toxicodynamics, including mech-anisms of action (MeOA) (Escher et al. 2011).

Because organisms are exposed to mixtures of chemicals,each exerting their own toxic effects, toxicity data forindividual chemicals on individual species are likely tounderestimate the hazard and risk posed to organisms in thereal world. It is therefore crucial to conduct toxicity tests inwhich organisms are exposed to mixtures and to use theresulting data to generate a realistic estimate of the hazardand risk posed by mixtures in natural environments. Inaddition, approaches are needed to implement and deriveguidelines to protect the environment and human healthagainst mixtures of chemicals. Environmental quality guide-lines have been derived almost invariably for individualchemicals. However, there are exceptions, notably theProvince of Quebec Canada (Ministere de l’Environment du

100 Integr Environ Assess Manag 7, 2011—S Dyer et al.

Quebec [MEQ] 2001), Denmark (Syberg et al. 2009), andSwitzerland (Chevre, Loepfe, Singer, et al. 2006; Chevre,Loepfe, Fenner, et al. 2006), where water quality guidelines(WQGs) for mixtures of pesticides have been developed.In their review on the use of mixture toxicity assessment inthe European Chemical Regulation REACH and the Euro-pean Water Framework Directive, Syberg et al. (2009)concluded that ‘‘it is scientifically feasible and regulatorallypracticable to integrate a more holistic mixture toxicityapproach into both legislations.’’

As with most toxicity testing, researchers conducting mix-ture toxicity tests have generally expressed the toxicity interms of concentrations in the ambient media (e.g., water,soil, and sediment). Whereas the vast majority of mixturesassessments for field assessments use ambient-based methods,they suffer from the temporal and spatial fluctuations in res-ponse to diverse environmental factors that create exposurevariability. Hence, mixture assessments for field applicationsmay lack verification for suites of chemicals that have a widearray of physicochemical properties as well as susceptibility tobiodegradation. Tissue residues provide an opportunity toestimate potential toxicity to organisms because internal dosehas been measured.

Because the vast majority of the currently available mixturetoxicity data are based on aqueous concentrations, a keycomponent of this review is to present methods that enableexisting data to be converted to tissue-based concentrations.This will permit aqueous-based concentration data to be usedin the new tissue residue (TR)-based framework. However, itwill usually be preferable to use measured TR data rather thanestimates whenever appropriate TR data are available, exceptfor organisms that store high concentrations of detoxifiedforms of metals, if the TR data do not differentiate betweentoxic and detoxified forms (Adams et al. 2011).

The challenge for environmental toxicologists and regu-lators is to assess the extent to which consideration of thetoxic effects of mixtures is needed and how this might beachieved to adequately protect the environment (McCartyand Bogert 2006). In the present work we review mixture–toxicity concepts and attempts to incorporate the tissueresidue approach (TRA) for chemical mixtures; we alsoexplore how internal concentrations as dose metrics mightimprove the assessment of mixture toxicity. Additionally, wedevelop and present a new framework for determining thetoxicity of mixtures using the TRA.

The framework is universally applicable to all chemicalmixtures (i.e., chemicals with the same mechanisms of action,chemicals with different mechanisms of action, includingmixtures of metals, metalloids, and organics). Comparativelyspeaking, the development of the framework for organicchemicals was easier than for metals, in part because somecrucial pieces of information related to TRs of metals are notyet available. The framework is intended as a general app-roach to advance regulatory and risk-assessment science and

Table 1. The 4 types of joint action for mixture

Similar joint action

Noninteractive Simple similar (dose or concentrationaddition, CA)

Interactive Complex similar

to encourage researchers to address knowledge gaps in che-mical mixtures. However, this framework does not addressthe testing of preparations that are mixtures of substancesprepared for sale as products.

TERMINOLOGY AND MIXTURE CLASSIFICATIONBliss (1939) developed a scheme for classifying the types of

interactions possible between chemical components inmixtures, which was subsequently expanded by Plackettand Hewlett (1952). The types of joint action (Table 1) arebased whether components of a mixture interact (i.e., affectthe biological activity of the other components) or not (i.e.,noninteractive) and whether or not they have the same site ofaction and the same MeOA.

The classification scheme of Plackett and Hewlett (1952)was originally developed for pharmaceuticals where admin-istered dose is the dose metric. The same classification wassubsequently adopted by ecotoxicologists (Konemann 1981),who expressed the concentration of chemicals as the con-centration in the ambient media (e.g., mg/L in water or mg/kgin soil) instead of using the term ‘‘dose’’ (i.e., a specifiedquantity of chemical to be applied).

This classification is a very helpful general schema.However, it has limitations when it comes to differentiatingbetween mechanism of toxic action (MeOA) and mode oftoxic action (MoOA) (for definitions, see Escher et al. 2011).This is because interactions can occur during the toxicokineticor toxicodynamic phases, which affect mixture toxicity differ-ently. The definition of target site can vary in complexity(from target tissue to receptor) and may have dose-dependentirregularities (Borgert et al. 2004).

Simple similar joint action

In simple similar joint action, all of the chemicals in themixture have the same MeOA and do not affect each other’sbiological activity in the organism. This type of joint action isalso called concentration addition (CA) because the effects ofthe components of the mixture can substitute for each other.Each mixture component contributes similarly to the toxicityand thus they can be treated as a single toxicant. The com-bined effect of the components is equal to the sum of theconcentrations of each chemical expressed as a fraction of itsown individual toxicity. This fraction is commonly referred toas the toxic unit (TU), the concentration of compound i inrelation to a benchmark effects concentration (ECp) of com-pound i (Brown 1968; Sprague 1970). The TU is calculatedusing

TUi ¼Ci

ECpi

ð1Þ

where the subscript denotes component i of a mixture, Ci isthe aqueous concentration of component i in the mixture, andECpi is the aqueous concentration of component i acting

s developed by Plackett and Hewlett (1952)

Dissimilar joint action

Independent (independent action, IA also called responseaddition, RA)

Dependent

Tissue Residue Approach for Chemical Mixtures— Integr Environ Assess Manag 7, 2011 101

individually that causes a given toxic effect (e.g., LC50,EC20).

In practical application, Ci could equally well apply to theconcentration in any compartment of the environment or anorganism (e.g., soil, sediment, fish muscle) and ECpi could bethe concentration of the component in the respectivecompartment acting individually that causes a given toxiceffect. Given that CA applies to compounds with the samemode of action, the effect concentrations need to relate to thesame compartment and should use the same units (e.g., mg/kg, or mmol/kg) and the same endpoints (Cedergreen andStreibig 2005).

If not experimentally available for organic chemicals, theinternal effect concentration for the same endpoint IECpi canbe calculated from the product of the aqueous concentration-based ECpi and the bioconcentration factor with Equation 2(McCarty and Mackay 1993):

IECpi ¼ ECpi � BCFi: ð2Þ

This equation assumes a linear relationship between the
aqueous-based effect concentration and the bioconcentrationfactor. Assuming that the components of the mixture do notinteract in the toxicokinetic phase, the internal toxic unitsITUi should be equivalent to TUi:
ITUi ¼ICi

IECpi

: ð3Þ

Although the ITU is theoretically equal to the externalexposure-based TU, its derivation from experimental internalconcentrations has the advantage that it is independent fromuncertainty and time-related factors in BCF values.

The sum of toxic units TUmix for a mixture that conformsto concentration addition (based on aqueous concentrations)is determined using Equation 4 and analogously the internaltotal toxicity of such a mixture is defined by Equation 5:

TUmix ¼Xn

i¼1

TUi ð4Þ

ITUmix ¼Xn

i¼1

ITUi: ð5Þ

Because marked deviations from the simple models of CAare often caused by compounds that interact in thetoxicokinetic phase, i.e., during the bioconcentration step,the likelihood that the internal total toxic unit (ITUmix)model applies to mixtures is greater than for the total toxicitymodel (discussed below).

When calculating the (I)TUmix, it is essential to use (I)ECpi

values measured for the same endpoint (e.g., lethality,immobilization, reproductive impairment) for each chemicalin the mixture. The (I)TUmix must be calculated using eitheracute or chronic toxicity data, not a combination of the two,as the effect levels for acute and chronic exposure varytoo much to permit accurate characterization of the mixturetoxicity. However, in the near future, mixture models thatincorporate a time component and that are more elaboratethan the (I)TUmix may be developed, as this is an emergingarea of scientific research (e.g., Lee and Landrum 2006;Ashauer et al. 2007).

An example of the calculation of the ITUmix of a mixtureby CA follows. If a mixture consists of x mg/kg of compound

1 and y mg/kg of compound 2 which correspond to 0.5 and0.2 of the ITU based on acute 96-h ILC50 values, respec-tively, the ITUmix of the mixture would be 70% (i.e.,0.5þ 0.2) of the 96-h ILC50.

In concept, this derivation would also apply to metals.However, because 1) the BCF generally decreases as theaqueous concentration of a metal increases (McGeer et al.2003) and 2) metals can compete with each other for uptakeinto organisms, one would have to know the exact relation-ship between BCF and aqueous metal concentration and thedegree of metal–metal competition for uptake to validly applyEquation 2. Within that context, the ITU and ITUmix areequivalent to the terms tissue residue ratio (TRR) and totaltissue residue ratio (TTRR), respectively, which are describedin the ‘‘Tier I for metal mixtures’’ section in this article.Internal effect concentrations are merely surrogates for targettissue concentrations, which are the biologically effective dose(Paustenbach 2000). Ultimately we should be dealing withmixtures in terms of target tissue concentrations.

Independent joint action

Chemicals that elicit toxicity by independent joint action(IA) act at different target sites (i.e., have different MeOAs)and do not affect each other’s potency. This type of jointaction is also referred to in the toxicology literature as res-ponse addition (RA).

The total toxicity of a mixture that conforms to IA (BRmix)can be expressed mathematically as

BRmix ¼ 1�Y

1�BRið Þ: ð6Þ

where the subscript i indicates chemical i in the mixture andBRi is the biological response for chemical i of the mixture.BRi can take values between 0 (no difference from control) to1 (100% effect). As with mixtures exhibiting CA, the BRmix

of a mixture that conforms to IA must not be calculatedusing toxicity data based on different endpoints or differentexposure durations (i.e., acute and chronic).

An example of a calculation of the BRmix of a mixtureacting by IA follows. If a mixture consists of x mg/L ofcompound 1 (and alone this would elicit a 50% effect) and ofy mg/L of compound 2 (which corresponds to a 20% effect),the BRmix of the mixture would elicit a 60% effect (i.e., 1–[1–0.5][1–0.2]). If the toxicity does not conform to IA, it canbe either synergistic (greater toxicity than IA or CA) orantagonistic (less toxic than IA and CA).

Equation 6 does not require modification to address IA interms of the TRA (i.e., BRmix). Because the concept is definedexclusively on an effect level (BR) and not on an effectconcentration level, the model does not require that theeffects are derived from dose-response curves that are basedon concentrations in the same compartment, in contrast, thedosimetry can be related to different target sites of eachcomponent.

Complex similar and dependent joint actions

For these types of joint action, at least 1 chemical in themixture affects the biological activity of at least 1 otherchemical in the mixture. The TRA has great potential toprovide a better rationalization of these mixture toxicityconcepts as demonstrated in the conceptual flow diagram for

Figure 1. Classification of mixture toxicity according to mechanism of toxic

action, type of interaction between mixture components, and whether the

interaction occurs in the toxicokinetic or toxicodynamic phase.


assessing mixture toxicity and classifying the responses(Figure 1).

Chemical interaction in mixtures often takes place duringthe toxicokinetic phase. This leads to a complex interaction,but the toxicokinetic effect does not fit easily into thetraditional classification scheme devised by Plackett andHewlett (1952), shown in Table 1. Toxicokinetic interactionoccurs when chemicals in a mixture modify the biologicalactivity of chemical i by affecting, for example, the rates ofabsorption, metabolism, or elimination of chemical i or bycompeting with chemical i for binding at the target site.When the chemicals in the mixture engaging in suchinteractions elicit toxicity through the same MeOA, themixture toxicity is referred to as complex similar action. Aclassic example is the use of piperonyl butoxide with theorganophosphorus insecticide chlorpyrifos. In this case,piperonyl butoxide inhibits cytochrome P450 enzyme activ-ity, thereby reducing the rate of metabolic conversion ofchlorpyrifos to the more toxicologically active chlorpyrifosoxon (e.g., see El-Merhibi et al. 2004). The result is apparentsynergism, where less parent insecticide is needed to achieve asufficient dose of the active oxon analogue and thus a giveneffect. A similar example is the induction of metabolicenzymes by atrazine, in the toxicokinetic phase, leading toapparently synergistic effects in mixtures with organophos-phates (Pape-Lindstrom and Lydy 1997). Another example isinteractions such as depletion or induction of cytoprotectivefactors (glutathione) and alterations in tissue repair (Freidig

et al. 2001). The competition among metals for binding atuptake sites on the surface of an organism is yet anotherexample of this type of interaction (Niyogi and Wood 2003).

If chemicals do not interact during the toxicokinetic phase,the next decision point in the flow chart is the question ofwhether the 2 compounds have the same target site. Only ifthey do have the same target site will the TRA becomeimportant. Depending on interactions during toxicodynamics,the effect of the mixture will be complex similar (for inter-actions in the toxicodynamic phase) or CA (if no interactiontakes place). There are only a few examples of interactiveeffects during the toxicodynamic phase, among them thesynergistic effects of uncouplers that form mixed dimersinside the membrane, which have a higher permeability, andthus higher intrinsic toxicity, than that of the singlecompounds (Escher et al. 2001).

In summary, deviation from the CA or IA models can beexplained by either toxicokinetic or toxicodynamic inter-actions, or both, among components in a mixture. The toxi-codynamic interactions include those processes that do notdirectly affect the metabolism of a xenobiotic, but that doaffect the response or susceptibility of a tissue to toxic injury.Unfortunately, other than direct testing, predictive models forsuch situations are lacking (De Zwart and Posthuma 2005).

Limitations of the Plackett and Hewlett joint actionclassification

A number of limitations to the joint action classificationscheme identified by Plackett and Hewlett (1952) wererecently emphasized by De Zwart and Posthuma (2005), ofwhich users should be aware. The limitations arise predom-inantly from the fact that the scheme has a mathematicalrather than biological origin. The first limitation is that the4 types of joint action identified by Plackett and Hewlett(1952) are often not distinct and identifiable for complexmixtures. This can occur because toxicity tests simply mea-sure the total toxicity of the mixture, which could contain anycombination of CA, IA, complex similar and dependent jointactions.

A second limitation is that the scheme assumes that everychemical has only 1 target site. From a biological point ofview, this is a gross oversimplification. As stated in theaccompanying article (Escher et al. 2011), any single organismcontains multiple potential sites of action for any toxicant,such as acetylcholinesterase inhibitors (Pope et al. 2005), andthe effect on each of these may vary because of a number offactors, including the abundance of receptor sites and varyingactivities of detoxifying processes. The toxicity to an organismis thus the integration of the effects of the toxicant on each ofthe target sites. This concept was developed by Ashford(1981) for humans and first translated to ecotoxicology by DeZwart and Posthuma (2005). However, probably because of alack of toxicological knowledge of target sites and biochemicalprocesses, it has not been widely adopted in addressingmixtures in ecotoxicology.

A third limitation is that the various joint action classeshave distinct requirements regarding the MeOA of mixturecomponents. For example, CA requires that the componentshave the same MeOA. However, at the molecular level, this isoften unknown, yet it is often claimed without proof thatcertain chemicals act by CA or IA. In addition, the predictionsfor CA and IA are often not significantly different from each


other statistically, making it impossible to differentiate aposteriori between the 2 concepts. Often one can merelycompare whether the experiment is consistent with the hypo-thesis. A related limitation is that even when some infor-mation on the MeOA of a chemical is known under certainconditions, the MeOA is not constant. For example, theMeOA can vary with concentration, species, and exposureduration and other environmental factors. All organic chemi-cals exert part of their toxicity by the baseline MeOA (vanWezel et al. 1996), but in addition they may have a morespecific MeOA. Which MeOA is exerted depends on theconcentration at the target sites. The period of exposure alsocan modify the MeOA (Baird et al. 1990). Finally, MeOA canbe taxa-specific. For example, herbicides have a specificMeOA to plants but typically exhibit baseline toxicity tovertebrates. This flexibility in the MeOA exerted by compo-nents of a mixture has the effect of making the type of jointaction that occurs almost mixture specific and difficult topredict a priori.

A limitation to the implementation of the IA model is thatit requires the whole concentration-response relationship tobe available in order to determine the biological responsethat corresponds to a particular concentration. Unfortunately,very few ecotoxicology publications publish either the fullconcentration-response curve or an equation summarizing thecurve.

Subsequent developments in aquatic ecotoxicology ofmixtures

Compared with the wealth of ecotoxicity data available forindividual chemicals, there is a paucity of mixture toxicitydata. Research into the toxicity of mixtures has been largelypiecemeal, with the exceptions of the systematic work carriedout by Hermens and colleagues (e.g., Deneer et al. 1988;Deneer 2000; Hermens et al. 1984, 1985; van Leeuwen et al.1996); Broderius and colleagues (Broderius 1991; Broderiusand Kahl 1985; Broderius et al. 1995); Grimme and colle-agues (e.g., Altenburger et al. 2000; Backhaus et al. 2000a,2000b; Faust et al. 1994; Grimme et al. 1996); and Belden,Lydy, Cedergreen, and colleagues (Cedergreen and Streibig2005; Belden et al. 2007; Trimble et al. 2009). These studiesshowed collectively for most cases that mixtures containingchemicals with the same MeOA exert toxicity consistent withCA, whereas mixtures of chemicals with different MeOAsexert toxicity consistent with IA.

Hamers et al. (1996) proposed a 2-step mixed model foraddressing the toxicity of mixtures. This model applied CA tochemicals with a baseline MeOA and IA to all other MeOAs.A 2-step mixed model based correctly on the classification ofPlackett and Hewlett (1952) was proposed independently byJunghans (2004), Altenberger et al. (2004), and De Zwart andPosthuma (2005) and is consistent with the model developedby Olmstead and LeBlanc (2005). In this second model, thefirst step is to estimate the combined toxicity of componentsthat have the same MeOA, and then, if necessary, to estimatethe combined toxicity of components or groups of compo-nents that have different MeOA using the IA approach. Inhindsight, this may seem obvious, but it very elegantlyresolved problems that researchers had faced for years. It nowpermits researchers to estimate the toxicity of mixtures,which contain some components with the same MeOA andgroups of components with different MeOAs. This mixed

model approach has since been incorporated into the Dutchframework for ecological risk assessment of contaminatedland (Jensen and Mesman 2006), fully described by Posthumaet al. (2008), and has been used to assess the toxicity ofmixtures (Escher et al. 2005; Junghans et al. 2006; Ra et al.2006).

De Zwart and Posthuma (2005) also showed how CA, IA,and the mixed model approaches could be combined withspecies sensitivity distribution (SSD) methods to assess therisk posed by mixtures or potentially derive environmentalquality guidelines for mixtures.

Linking the TRA to mixture toxicity

Operationally, the MeOA of organic chemicals depends oninternal tissue concentrations. Concentrations above a thresh-old for a MeOA will elicit effects consistent with thatparticular MeOA, whereas concentrations below the thresh-old will contribute to baseline toxicity, or a narcotic MeOA.This applies equally to individual organic chemicals andmixtures. The proportional contribution to the toxicity of amixture by baseline toxicity or by various specific MeOAsdepends on the relative number of components that exerttheir toxicity by baseline or more specific MeOAs. Thus,mixtures exclusively containing components at concentra-tions below their elicited MeOA will result in baselinetoxicity, otherwise known as narcosis-type activity (Centrefor Ecotoxicology and Toxicology of Chemicals [ECETOC]2001; Deneer et al. 1988; Verhaar Wezel and Opperhuizen1995; Verhaar et al. 1995; van Loon et al. 1997). Forexample, an organochlorine insecticide present in tissues at 1/100th the LC50 may not be at a sufficient concentration toelicit a neurotoxic effect on fish, yet its presence in the fishmay contribute to a generalized, narcotic mode of action.Hence, classifications of substances into specific modes ofaction are typically based on studies in which acute and/orchronic toxicity thresholds have been exceeded (ECETOC2001). Warne and Hawker (1995) used this concept and thecritical concentration and critical volume hypotheses ofbaseline toxicity (e.g., Abernethy et al. 1988; Warne et al.1991), which are variants of the TRA approach, to developthe funnel hypothesis.

The funnel hypothesis argues that the more componentscontained in an equitoxic mixture (a mixture in which eachchemical has the same TU), the smaller the concentration ofeach component will have to achieve a chosen measure oftoxicity. Therefore, increasingly, the components will act onlyby their baseline mechanism of action and thus should beconcentration additive. The distribution of equitoxic mixturetoxicity data predicted by the hypothesis is presented inFigure 2. Warne and Hawker (1995) collected toxicity datafor 104 equitoxic mixtures composed of 182 chemicals withnarcosis, oxidative uncoupling, receptor-mediated, and reac-tive MeOAs. These data included 7 test species, includingbacteria, crustacea, amphibians, and fish, covered a variety ofmeasures and sublethal and lethal endpoints of toxicity, andincluded both acute and chronic values. Warne and Hawker(1995) found that these data conformed to the hypothesis(Figure 2A). Similarly, toxicity data for 973 mixtures, whichincluded organic–organic, metal–metal, and metal–organicmixtures, conformed to the predictions of the hypothesis(Ross 1996). Further, the funnel hypothesis held true for eachof these different types of mixtures. A review stated that ‘‘the

Figure 2. (A) Variation in toxicity of mixtures with the number of components in the mixture predicted by the funnel hypothesis. A cTEI (corrected toxicity

enhancement index) value of 0 (horizontal line) equates to additivity; cTEI values of 2 and �2 equate to a mixture toxicity 3 times that of concentration addition

and one-third that of concentration addition. (B) Observed variation in toxicity of mixtures with the number of components. Figure reprinted with permission

from Warne and Hawker (1995), Ecotoxicology and Environmental Safety, 31: 23–28. �1995 Elsevier.


literature is quite large that supports this hypothesis’’ and‘‘importantly, this thought process transcends modes ofaction’’ (ECETOC 2001).

Figure 2B shows the variation in the toxicity of mixtures asa function of the number of components in equitoxicmixtures. When equitoxic mixtures have few components,the maximum deviation from CA can be relatively large; asthe number of components in equitoxic mixtures increases,the maximum deviation from CA decreases until the toxicityof the mixtures conforms with CA. Data for mixtures withmore than 20 components are not substantial, but theavailable data indicate that when mixtures contain approx-imately 30 or more components, they conform to CA. Inequitoxic mixtures with 30 components that conform to CA,each component is present at 1/30th of their own individualtoxicity value (e.g., EC50 or LC50). We have therefore used1/30th of the acute lethal toxicity threshold as an operationallydefined and pragmatic point at which compounds irrespectiveof their MeOA will exert their toxicity by narcosis. A limitationof this hypothesis is that at present, it only applies to equitoxicmixtures whereas in all likelihood, mixtures in the environ-ment will not be equitoxic. However, the underlyingprinciple that chemicals present at low concentrations willact predominantly by their baseline mechanism of action,means that the predictions of the funnel hypothesis shouldapply to nonequitoxic mixtures. In addition, the hypothesiswas developed for short-term exposures and mortality, but itcan be expected to be applicable for sublethal effects andchronic exposure, too, as generally mixture toxicity conceptshold for acute and chronic exposure.

Jager et al. (2007) used the work of Ashford (1981) as theirmechanistic model for toxicant–organism interactions. Thisapproach states that a stimulus (for our purposes, a toxicant)interacts with a site of action (target site), and this exerts aneffect on the subsystems in which the target sites are located.Thus, Jager and colleagues moved away from a focus on theambient environment to concentrate on the intraorganismenvironment, consistent with the TRA. They also discussed theimplications of their work in terms of assessing the toxicity ofmixtures. Therefore, their work expands the theoretical under-pinnings for the TRA to assessing the toxicity of mixtures.

However, there are also limits to implementing the TRA inmixture toxicity assessment. Although the mixture toxicity ofreactive electrophilic chemicals can be described well bymixture toxicity concepts of CA and IA (Chen and Yeh 1996;Richter and Escher 2005), it will not be possible to useinternal concentrations for these fast-reacting chemicals as isdiscussed in more detail in the accompanying paper, whichreviews the mechanisms and modes of toxic action in relationto the TRA (Escher et al. 2011).

From aqueous to internal concentrations

The vast majority of the available mixture toxicity data hasbeen based on ambient (e.g., aqueous, sediment, or soil) ratherthan tissue-based concentration, and this is likely to remain tobe the case for some time. Thus, to use these historical data,methods capable of converting ambient concentrations to tissueconcentrations need to be developed and used. For organicchemicals, the tissue concentrations that correspond to aqueousconcentrations can be estimated by multiplying the aqueousconcentration by a measured bioconcentration factor (BCF) or acalculated BCF, which uses the octanol–water partitioncoefficient (KOW) or various lipid–water partition coefficients(e.g., triolein–water partition coefficient, Ktw, or liposome–water partition coefficient, Klipw). For organic chemicals withbaseline toxicity, it is preferable to use partition coefficients forlipids found in biological membranes or artificial biologicalmembranes rather than KOW, if available (Warne et al. 1991;Vaes et al. 1998; Escher et al. 2000). These estimated tissueconcentrations can then be plotted against the biologicalresponse to derive tissue concentration-response relationshipsand hence IECp (internal effect concentrations for p% ofmaximum effect) values.

For some compounds (e.g., highly hydrophobic and poorlymetabolized chemicals), relating aqueous concentrations tointernal residues will need to address other routes of chemicaluptake by aquatic organisms (e.g., diet). A variety of empiri-cal and modeling-based approaches are available for estimat-ing the bioaccumulation potential of chemicals via multipleexposure routes (water, diet) and these are summarized in theaccompanying review paper (Sappington et al. 2011).

Figure 3. Proposed use of internal lethal concentrations (ILC50) in the field

risk assessment of mixture. The amount of shading in the right-hand fish in

each panel indicates how much chemical accumulated from the laboratory

would be needed to elicit the adverse effect, whereas the unshaded portion of

the fish indicates how much of the chemical(s) was (were) accumulated from

the prior field exposure. Figure reprinted with permission from vanWezel et al.

(1996), Ecotoxicology and Environmental Safety, 35: 236–241. �1996

Elsevier.


Laboratory mixture toxicity studies

To our knowledge, experimental mixture toxicity studiesusing the TRA approach have been done so far exclusivelywith baseline toxicants. As early as 1988, Opperhuizen andSchrap explored the bioaccumulation and lethal bodyburdens for 2 PCB congeners and in acute toxicity testingof guppies their results indicated concentration addition ofinternal concentration. Based on this work, van Wezel et al.(1996) conducted mixture toxicity experiments by exposingfathead minnow to mixtures of dichlorobenzenes and mea-sured the tissue concentration of the chemicals to determinethe lethal body burden (LBB). They then tested whether theobserved toxicity on a LBB basis conformed to the CA modelof joint action and found that the CA approach accuratelypredicted the LBB toxicity of these mixtures.

A third study on the internal concentration-based mixturetoxicity of baseline toxicant was performed by Landrum et al.(2003). These workers investigated the toxicity of 4 poly-cyclic aromatic hydrocarbons (naphthalene, phenanthrene,fluorine, pyrene) to the freshwater amphipod Diporeia spp.,whose external concentration-based 28-d LC50 values variedby orders of magnitude, whereas the 28-d ILC50 rangedonly from 5.8 to 12.3 mmol/kg, and the 28-d ILC50 of theequipotent mixture of the 4 components came to 6.1 mmol/kg,confirming the CA model of joint action.

In addition to toxicity testing of the whole organism, thehypothesis of concentration addition of internal concentra-tions was also tested with an in vitro system that works withisolated energy-transducing membranes from the photosyn-thetic bacterium Rhodobacter sphaeroides (Escher et al. 2002).Although this is an in vitro system, it has the advantage ofbeing constituted of the isolated target sites only, thebiological membrane. In this system, CA of 3 different typesof mixtures of up to 6 baseline toxicants could be confirmedon the basis of membrane concentrations.

From laboratory to field: estimating body residues ofcomplex mixtures

Analyzing the body residues of all components in complexmixtures individually is very difficult (e.g., Booth et al. 2007).Some work has been done on the indirect estimation of bodyresidues of complex mixtures, e.g., in water, soil, andsediment media, and there is also data on mixture residuesin fish (Dyer 2005).

The implications of concentration addition of internalconcentration in terms of conducting ecological risk assess-ments were discussed by van Wezel et al. (1996). Theseinvestigators proposed to use the TRA concept to cope withmixture toxicity in a field situation. They postulated that afish from an uncontaminated site will have an ILC50 of 2 to8 mmol/kgwet weight based on previous work by McCarty andMackay (1993), and a fish that is preexposed with pollutantsin the environment will show 50% lethality at a lowerconcentration than an unexposed fish would, with thedifference between the ILC50 values corresponding to thebody burden of chemicals from the field (Figure 3). Thevalidity of this approach was demonstrated by van Wezel andJonker (1996) on the example of sticklebacks exposed for 2months to contaminated sediments and then brought back tothe laboratory and tested for their sensitivity toward 1,2,4-trichlorobenzene. Although some sites showed differences inthe additional internal concentration of 1,2,4-trichloroben-

zene to elicit lethality, differences in temperature and salinitylimited the immediate applicability of the approach.

The validation of concentration addition for organiccontaminants by van Wezel et al. (1996) helped justifyinvestigations that use passive samplers to estimate internalconcentrations. Passive samplers typically consist of hydro-phobic material such as in semipermeable membrane devices(SPMDs) (Sodergren 1987; Huckins et al. 1990), thin films(Wilcockson and Gobas 2001), Empore discs (Verhaar et al.1995; van Loon et al. 1997; Verbruggen et al. 1999), andsolid-phase microextraction (SPME) fibers (Parkerton et al.2000; Verbruggen et al. 1999; Leslie, ter Laak, Busser, et al.2002; Leslie, ter Laak, Oosthoek, et al. 2002; Leslie,Hermens, Kraak, 2004; van der Wal et al. 2004). They havebeen used to mimic and predict the uptake into biota ofchemicals known not to be extensively metabolized. Withthese samplers, it is possible to perform partition-drivenextractions of environmental samples and estimate internalconcentrations in organisms. The kinetics of uptake is depen-dent on the thickness of the hydrophobic layer and thehydrophobicity of the chemical. The uptake rate decreaseswith increasing thickness of the absorbing layer and withincreasing hydrophobicity of the chemical. In the case ofnarcosis-acting chemicals, the molar concentration in the cellmembrane determines the effect (Bernhard and Dyer 2005).This makes passive sampling for narcotic chemicals veryuseful, as body residues on a molar basis can be translated in astraightforward manner to expected toxic effect (Verbruggenet al. 1999; Leslie, ter Laak, Oosthoek et al. 2002). This isbecause there is a relatively small range of molar concen-trations in an organism that leads to effects (Abernethy et al.


1988; Warne et al. 1991; McCarty and Mackay 1993). Totalbody residues have been estimated with SPME fibers (e.g.,Verbruggen et al. 1999; Leslie, Hermens, Kraak, 2004;Parkerton et al. 2000). The relationship between contaminantconcentration of a mixture in SPME fibers and in aquaticinvertebrates has been demonstrated for different chemicals(Leslie, ter Laak, Busser, et al. 2002; Leslie, Hermens, Kraak,2004).

Real-world example of the application of the TRA to mixturetoxicity: Ohio Fish Tissue Residue Study

The work of Warne and Hawker (1995) showed atheoretical basis for using CA in assessing the ecotoxicologyof complex mixtures, and the work of van Wezel’s groupprovided laboratory-based evidence (van Wezel and Jonker1996; van Wezel et al. 1996). However, field-based eco-epidemiological support was lacking. The key reason was thatfew datasets contained information on both contaminantconcentrations in field populations and the ecological status ofaquatic communities over a geographically extensive area.Dyer et al. (2000) were the first to address this knowledgegap. They compiled a dataset, describing aquatic habitatquality, fish species and community integrity, and contami-nant residues in fish from 1010 locations throughout Ohio,USA. Several additive toxicant interaction methods were usedto predict the toxicity of organic and metal mixtures to arange of in-field fish community responses (e.g., index ofbiotic integrity, IBI, percentage fish with deformities, finerosions, lesions, and tumors), and these were compared withobserved values. For organics, additivity of organic contam-inants in fish residues was based on 3 methods: 1) calculatedscreening values that incorporate BCF and water qualitycriteria, 2) literature-based assessment of residues andtoxicological responses, and 3) addition of molar units as adirect test of narcosis theory. The toxic tissue screeningconcentration (TSC) is a product of the US EnvironmentalProtection Act (USEPA) water quality criterion andbioconcentration factor per respective chemical: TSC¼AWQC�BCF, where AWQC is the ambient water qualitycriteria. The TSC is a product that has been used as a toxicityreference value in screening level ecological risk assessmentsand is interpreted as a TR in aquatic biota above whichadverse ecological effects may occur. It was used by Dyeret al. (2000) as the denominator for the addition of toxic units(i.e., TU¼ fish tissue concentration/TSC). Because the goal ofAWQC is to protect 95% of taxa (Stephan et al. 1985), TUsshould be protective of 95% of taxa. Exceedence of 1 and 10TUs could be indicative of chronic (e.g., growth, reproduc-tion) and acute effects (e.g., mortality) to more than 5% ofspecies, respectively, assuming an acute to chronic toxicityratio of approximately 10.

The second method for establishing TSCs used a largeliterature review database (3400 records) to compile pub-lished papers relating measured whole body, wet weight TRsto adverse toxicological or ecological effects (Shephard 1998).Several endpoints were used in the derivation of adverseeffects residue values: mortality, reproduction, growth, beha-vior, and morphological changes. The 5th percentile of therank-ordered adverse effect residues for each chemical wasdetermined and used as the alternate TSC. The 5th percentileresidue was selected because it represented a comparable app-roach to the other TSC, yet based on measured residue and

adverse effects relationships. As stated, the third method usedfor organic contaminants was the simple addition of molarunits per tissue or body mass. The effect of lipid normal-ization of molar unit addition was also investigated.

Three approaches were also used to calculate metal TUs.The first (TU¼ tissue concentration/TSC) used a TSC basedon the geometric mean of measured BCF values and ambientwater quality criteria set at a hardness of 50 mg/L calciumcarbonate. The second approach used the 5th percentile ofliterature data for the alternate TSC, and the third approachwas based on comparing metal residue levels in fish occurringin Ohio locations with excellent index of biotic integrityscores (IBI >45). This last approach is only applicable formetals (see Adams et al. 2011).

The study found that the concentration addition of toxicunits (Figure 4) is a conservative estimate of risk that mayhave little ecological relevance. For instance, the organic toxicunit-based methods for organic mixtures showed that asubstantial fraction of the fish sampled (14%–�30%) shouldhave come from populations experiencing some type ofchronic effect. However, all of the samples were found tohave approximately �0.2 mmol total chlorinated organics/kgor <6 mmol/kglipid, indicating the estimated chronic thresh-olds (0.2–0.8 mmol/kg and 5 mmol/kglipid) for narcosis orbaseline toxicity had not been exceeded. This conclusionappeared to be confirmed by the lack of correlation of eitherTUs or mmol/kg with the index of biotic integrity (IBI) fromthe sampled sites and similarly the percentage of fish havingdeformities, fin erosions, lesions, and tumors. The lack ofcorrelations also indicated that an ecosystem-level threshold(HC5) of 0.25 mmol/kglipid (Verhaar et al. 1995) may beoverly conservative. Similar relationships of metals TSCs andin-field IBI scores and DELTs (index for deformities, finerosions, lesions and tumors in fish) were not found (i.e., nosignificant correlations). This is particularly true when thetoxic units were derived from water quality criteria intendedto be protective of 95% of aquatic taxa and when the numbersof chemicals added together increase. Second, in situations inwhich low concentrations of organic tissue contaminants arepresent, the addition of molar units appears to be a reasonableapproach for organic mixture risk assessments, even forchemicals with diverse modes of action, thus supporting thefunnel hypothesis (Warne and Hawker 1995).

Estimating chronic mixture toxicity from acute effects

Chronic toxicity data are limited and tissue-based chronicmixture toxicity data are very rare indeed. The limitation inthe amount of chronic toxicity data leads to the need to useacute to chronic ratios (ACR, defined as the ratio of an acuteLC to a chronic NOEC) to derive tissue-based qualitystandards for prolonged exposure. Acute to chronic ratiosdepend on the MeOA of the toxicant in question (Roex et al.2000). Mixtures of compounds with the same mode of actionare expected to have similar ACRs as the individual mixtureconstituents. In the case of nonpolar organic chemicals thathave a baseline MeOA, an ACR of 10 has been advocated(van Leeuwen et al. 1992), whereas for volatile nonpolarorganic chemicals an ACR of 4.5 to 5 has been recommended(Di Toro et al. 2000; McGrath et al. 2004). Because of thereversibility of effects with these compounds, the criticalbody residues are relatively constant (Rozman and Doull2001). However, for chemicals that are less reversible or

Figure 4. Distribution of tissue residues (organics only) and toxic units (TU) from (a) organic contaminants and (b) metals in fish collected from 1010 sites in Ohio

from 1990 to 1996. Reprinted with permssion from Environmental Science and Technology, 34: 2518–2524. �2000 American Chemical Society.


irreversible or that have specific MeOAs, alternative dosemetrics such as time integral of concentration or critical targetdepletion rates need to be used; in such cases, the timedependence of toxicity is implicitly dependent on dose(Verhaar et al. 1999).

Unfortunately, the use of ACRs for metals and mostorganics is only applicable on a chemical and species-specificbasis. The use of ACRs for untested taxa is dubious because

many metals are internally regulated (i.e., essential metals).Consequently, assuming a factor of 10 to estimate chronicresponses from acute tissue concentrations may overpredictor underpredict potential effects (Adams et al. 2011).Even so, in the following section we provide an approachin which field data may be used to derive thresholds, usefulfor predicting screening-level mixture assessments formetals.


MIXTURES FRAMEWORKWe propose and discuss a new framework for addressing

the toxicity of mixtures based on TRs. Other frameworks thataddress the toxicity of mixtures have recently been proposedby Posthuma et al. (2008), whose comprehensive treatise onmixture toxicity is useful for comparative purposes. Reflect-ing the bulk of the mixture toxicity research that has beenconducted to date, all the methods described in Posthumaet al. (2008) are for water and sediment toxicity assessments.Given the topic of the SETAC Pellston TRA workshop, wehave focused on developing a framework that addresses thetoxicity of mixtures from a TRA.

Organisms and ecosystems in the environment will in manycases be exposed to complex mixtures composed of chem-icals, some with the same MeOA and others with differentMeOAs. This framework attempts to address this complexitywhile acknowledging that the majority of research into thetoxicity of mixtures has focused on baseline toxicants. Ourproposed framework differs from that described by Posthumaet al. (2008) in 2 main ways: it deals with TR-based mixturetoxicity, and it has an improved mechanism for addressingthe toxicity of metal mixtures. The proposed frameworkwas developed with the aim of being applied to caged orcollected organisms in which TRs of metal and organiccontaminants have been measured; or for organic chemicals,where tissue concentrations have been estimated by biomi-metic techniques.

The framework has 3 tiers that operate in a manner similarto the tiered approach adopted in ecological risk assessments,in that the amount of effort and expenditure reflects theseverity of the situation and the framework should beconsidered within such a site investigation or remediationsetting. Having completed a tier, users need to make a cost-benefit-based decision on whether it is worthwhile to proceedto the next tier, or whether they should begin to address thesituation via remediation or some management strategies toreduce the toxicity of the mixture. The framework is arecommended approach; however, users can enter or exit theframework at any point that they desire, depending on theirparticular situation. Given the tiered structure, the first tier isconservative. Mixtures found to pose an unreasonable riskby this method may proceed to the more environmentallyrealistic and time consuming second or third tiers. Tiers I andII model the toxicity of mixtures to individual species. A keydifference from the first 2 tiers is that tier III models thetoxicity of mixtures to higher levels of biological organization,such as communities and ecosystems.

Numerous scenarios may require an estimation of the riskof mixture toxicity, hence use of the framework. Onescenario would be a site (e.g., a lake or a river reach) that isbeing assessed because either 1) it is suspected of beingcontaminated or 2) there are observed biological effects (e.g.,missing species, reduced species abundance, organisms withdeformities, lesions, and tumors), or both. In the latter case,it is necessary for the assessor to understand the potentialof toxicants and habitat quality to affect taxa presence orabundance and to determine whether either chemicals orhabitat effects may be the primary cause.

One proceeds through the framework if it has beendetermined that mixtures of chemicals were or are present.This can be resolved by 1) monitoring of chemicals in water,sediment, tissue and passive samplers; 2) results of a biolo-gically directed assessment (e.g., whole effluent toxicity,

WET, also called direct toxicity assessment, DTA) and toxi-city identification and evaluation (TIE) procedures; and 3)plausible source and impact relationships (e.g., downstreamsites from a known discharge point). For all situations, there isreason to understand the linkage of each component in themixture to causality. Hence, the execution of the frameworkis likely to be a high-tier assessment process only to be cond-ucted when risk management schemes are needed to protector remediate a defined resource.

Description of the framework

The first tier uses CA to estimate the toxicity of mixturesregardless of whether the components are inorganic or organicchemicals, and regardless of the MeOA of the components.Using the CA approach is consistent with the method adop-ted in the Australian and New Zealand Water QualityGuidelines (ANZECC and ARMCANZ 2000) and thatrecommended by Syberg et al. (2009) and the scientificliterature (discussed below). Given the diversity of possibleMeOAs among chemicals in mixtures likely to be experiencedin the environment, it is unlikely that the requirements of theCA model (that components of the mixture all have the sameMeOA) will be met. However, a number of laboratory andfield-based studies have shown that use of CA is a realisticworst case and is therefore appropriate to use in the first tierof the proposed framework. The CA model is conservative in2 ways. First, Deneer (2000), Faust et al. (1994), Warne andHawker (1995), and Ross (1996) found that approximately10% to 30% of mixtures (regardless of the type of chemical,but focusing predominantly on organic chemicals) wereantagonistic or synergistic, with each type of joint actionbeing equally frequent and the remaining 70% to 90% wereadditive, based on aqueous concentration toxicity data.Similar values but with higher percentages of antagonisticand synergistic mixtures (i.e., 44% antagonistic, 27% additive,and 29% synergistic) were found in a recent review byNorwood et al. (2003) on the toxicity of mixtures of metals.Recently, Belden et al. (2007) showed that for 207 pesticidemixtures the CA model accurately predicted the biologicalresponses within a factor of 2 for 88% of the studies. Thus, byassuming CA, the toxicity of at least 70% (based on Norwoodet al. 2003) or between 85% and 95% (based on Deneer 2000;Faust et al. 1994; Warne and Hawker 1995; Belden et al.2007) of mixtures would be estimated accurately or over-estimated and only 5% to 30% of mixtures would have theirtoxicity underestimated. Second, a series of papers (e.g., Faustet al. 1994; Backhaus et al. 2000a, 2000b; Dyer et al. 2000;Junghans et al. 2006; Chevre et al. 2006) showed that CAoverestimated effects and yielded slightly higher estimates ofthe toxicity of mixtures than IA when chemicals had differentMeOAs. Because the CA method is likely to overpredictthe toxicity of most mixtures, its use in tier 1 should beenvironmentally protective.

If, after having used the CA model, the toxicity of themixture is determined to have exceeded the appropriatethreshold, it will be necessary to advance to tier II, using morerigorous methods that properly employ CA and IA principlesacross all known components. Hence, the results of tier Iprovide a pragmatic and conservative approach for therelatively rapid assessment of effects in the measured biotaand other members within the aquatic community. Descrip-


tions on the derivation of data and integration of data sourcesfor tiers I and II are described below.

The terminology that we will use in this framework issomewhat different from the standard terminology used inmixture toxicology. We have deliberately done this so thatthe terminology is consistent with that used in the accom-panying review on metals (Adams et al. 2011). Because ofdifferences in the toxicological behavior of metals and organicchemicals, the tier I methodology is substantially different forthese 2 types of chemicals, but the scheme for metals issimply a modification of the more general scheme used fororganics to account for their different toxicological behavior.Thus, the scheme for organic chemicals will be described first.

Tier I organics scheme

Field monitoring of organics to derive TR thresholds istypically not undertaken. Further, many organics of interestare xenobiotics, hence the lack of natural background con-centrations. Notable exceptions to this include lipids and fats,hormones (e.g., estrogen, testosterone), fatty acids, alcohols,polycyclic aromatic hydrocarbons (PAHs), all componentsin oil, dioxins, and chlorophenols and natural brominatedcompounds in the sea that have a similar structure to poly-brominated flame retardants.

The CA equations are to be used to estimate the internaltotal toxicity of all organic chemicals (ITUmix) by summingup the ITUi values for all organic chemicals in the mixture(Eqn. 3). As stated earlier, the ITUmix must be calculated foreither short-term or long-term exposure, and acute andchronic toxicity data must not be mixed. The denominator inthe equation to calculate the ITUs (Eqn. 4) could in principlebe for any percentage toxic effect (e.g., 90%, 80%, 50%, 20%,5%). However, by convention and analogy to externalconcentration-based toxicity measures, the internal medianlethal concentration (ILC50) is used to assess the acutetoxicity of mixtures, although equivalent effect concentrationdata (i.e., IEC50) would also be appropriate, whereas theinternal chronic 10% effect concentration (IEC10) or internalchronic no observed effect concentration (INOEC) is used forestimating the chronic toxicity of mixtures (EuropeanCommission 2003). Whether in the laboratory or in the field,the smaller the difference between the TR concentration of amixture component and the selected internal toxic effectconcentration (e.g., IEC50, ILC50)—that is, the closer theITU is to 1—the greater the probability of the selected toxiceffect occurring.

If the ITUmix at a site is greater than or equal to 1, thismeans that, based on CA, it is estimated that the mixtureof organic chemicals would exert the selected toxic effect.For example, if the denominator is IEC50 (immobilization),an ITUmix value of greater than or equal to 1 indicates thatgreater than or equal to 50% of the test organisms will beimmobilized. Alternatively, if WQGs were used in the deno-minator and the ITUmix equals or exceeds unity, this meansthat protective criteria have been equaled or exceeded,respectively. Consequently, the desired percentage of speciesor less than the desired percentage of species, respectively,would be protected (assuming a species sensitivity distribu-tion (SSD) method was used to derive the WQG). Inpractical terms, this might mean that many species expectedto be present at a site might not be present or might bepresent in reduced numbers because the species used for

biomonitoring is a tolerant species (see discussion on theselection of the species for biomonitoring in the next section).Consequently, further investigation is warranted, and tier IIproceeds.

Tier I for metal mixtures

A modification of the organic chemicals mixture assess-ment scheme is necessary for metals. The reasons for this areoutlined below and described in detail in Adams et al. (2011).Aquatic and terrestrial organisms have a variety of mecha-nisms of decreasing the free internal or target tissue metalconcentration via sequestration (e.g., binding to metallothio-neins, storage in metal-rich granules) and elimination.According to the generalized model of metal toxicity, freeinternal metal concentration is maintained relatively low overwide ranges of exposure concentrations until just beforedeath, at which point the metal concentration at the targettissue increases rapidly and toxicity ensues (Figures 7 .18 and9.3 in Luoma and Rainbow 2008). Thus, based on internalmetal concentrations it is only possible to indicate when no tominimal toxic effects will occur and when toxic effects have ahigh certainty of occurring. In between these 2 concentrationsit is not possible to make inferences with any degree ofcertainty. This situation is markedly different from that fororganic chemicals, for which monotonic internal concentra-tion response relationships are the norm. The other keydifference is that all metals occur naturally and therefore theirbackground internal concentrations need to be taken intoaccount. In contrast, most organic contaminants are xeno-biotics and thus the background internal concentration canbe assumed to be zero. The above differences mean thatthe background internal concentration must be accountedfor in calculating the TRR values (i.e., the metal equivalent ofthe ITUs), 2 denominators are used to calculate the 2 sets ofTRR values, and 2 sets of TTRR values (i.e., the metalmixture (see sections below for definitions). Even so, wepresent here a method for establishing screening thresholdsbased on monitored internal residue concentrations fromspecies that occur at both contaminated and reference(minimally impacted) sites.

The species for which TR data should be collected must bea tolerant species. This is necessary so that the species canaccumulate metal over a much wider range of environmentalconcentrations than could a more sensitive species beforeexperiencing toxic effects, with the subsequent rapid increasein tissue concentrations. In addition, the selected speciesshould be relatively tolerant of a wide array of environmentalconditions because this will permit sampling of data from awider variety of reference sites.

It is preferable to obtain TR data for the selected speciesfrom a range of diverse reference sites. However, the frame-work will still function with data from as few as 1 site, butwith greater uncertainty.

The final requirement is that relationships be establishedbetween the sensitivity of the indicator species and other lesstolerant species (i.e., those likely to be affected by themixture). One way of approaching this is to establish species-to-species relationships. For example, tolerance indices fordiverse invertebrate taxa have been created by numerousinvestigators in several locations including England (Wright etal. 1993), Australia (Gray 2004), and regions within theUnited States (Wisconsin: Hilsenhoff 1987; North Carolina:


Lenat 1993; Ohio: DeShon 1995; Florida: Barbour et al.1996). Similar approaches have been used for fish (Karr1981). In addition, interspecies correlations and relationshipsbased on laboratory toxicity data have been published (Mayeret al. 1987; Dyer et al. 2006). Recently 2 new techniqueswere developed to predict toxicity using these relationships(Raimondo et al. 2007; Morton et al. 2008). Alternatively,empirical species-to-community relationships could be estab-lished, in which the accumulation of metal in the tolerantspecies is used to predict impairment of structural attributesof an invertebrate or fish community (e.g., richness, diversityor abundance of various taxonomic groups). This approach isdiscussed in greater detail in the accompanying article on theTRA for metals (Adams et al. 2011).

The background concentrations of metals are obtainedfrom reference sites at which human influence is minimal(Rankin 1989; Dyer et al. 2000), using protocols that havebeen established to help researchers to identify such sites.Because background concentrations are highly variablebetween sites and even vary temporally at 1 site, the metaltissue concentrations in the indicator species will also vary. Toreflect this variation, a percentile of the distribution of metaltissue concentrations for the indicator species should be usedas the estimate of the background concentration. The 90thpercentile of the metal TR concentrations at the referencesites (TRref90) is an example of an estimate of backgroundconcentration (Adams et al. 2011), but other percentilesmight be more appropriate for specific situations.

Figure 5 illustrates a possible distribution of TR concen-trations for a single component of a mixture from an indicatororganism collected from both reference and nonreferencesites. To assess the toxicity of a mixture of metals, a similarfigure for each metal in the mixture being examined would berequired. The potential for adverse population-level orcommunity-level effects caused by exposure to a mixture ofn metals can be predicted semiquantitatively using a 3-zonedecision algorithm. The relationship between the TR in theindicator species and an effect level measured in a moresensitive species (e.g., population density) or a community-level metric (e.g., species richness or diversity) would have to

Figure 5. Illustrative relationship of the cumulative frequency of occurrence

of metal tissue residues of an indicator species collected at reference sites and

nonreference sites for a single compound.

be developed from field data collected on a local, regional ornational basis, as appropriate (e.g., Fig. 17 .11 in Luoma andRainbow 2008; see also Adams et al. 2011).

The low zone is a range of metal TRs that correspond to noor minimal impacts to the higher-sensitivity species or no orminimal change in the community-level metric. For example,the 10th percentile of TR values from nonreference sites ispresented as the TRlow in Figure 5. The high zone (e.g.,concentrations greater than the 75th percentile of TR valuesfrom nonreference sites in Figure 5) is a range of metal TRsthat correspond to certain toxic impacts (e.g., reduced scopefor growth, lower abundances, for the more sensitive species,or decreased species richness or diversity of the community).In between the low and high zones is the range of TRs inwhich there is potential for toxic effects to occur but theuncertainty is high.

Therefore, the total TR ratio of a mixture of metals(TTRR) must be estimated twice—first to determine whetherit is greater or less than the TRlow, and again to determinewhether it is greater or less than the TRhigh. This is done usingthe following equations:

TTRRlow ¼Xn

i¼1

TRi�TRref90i

TRlowi�TRref90i; ð7Þ

TTRRhigh ¼Xn

i¼1

TRi�TRref90i

TRhighi�TRref90i; ð8Þ

where TRi is the tissue residue concentration of chemical i infield-collected organisms, and TRref90i is the 90th percentiletissue residue concentration of chemical i from referencesites (i.e., estimate of the background concentration). InEquation 7, TRlowi is the highest concentration of chemical iin tissue collected from nonreference field sites that are notadversely impacted and TTRRlow is the cumulative tissueresidue ratio when the TRi values for each of the n metals arecompared to their respective TRlow values (after adjusting forbackground metal uptake). This TTRRlow value is analogous tothe TRlow value (the 10th percentile concentration) that isillustrated as the boundary between the no impact and thepotential impact zones in Figure 5. In Equation 8, TRhighi is thelowest concentration of chemical i in tissue collected fromnonreference field sites that are always adversely impacted, andTTRRlow is the cumulative tissue residue ratio when the TRi

values for each of the n metals are compared to their respectiveTRhigh values (after adjusting for background metal uptake).This TTRRhigh value is analogous to the TRhigh value that isillustrated as the boundary between the potential impact andthe certain impact zones in Figure 5.

If the assessment of the toxicity of either the metals or theorganic chemicals resulted in a value of �1, further assess-ment is automatically required. However, if both theTTRRlow value for metals and the ITUmix value for organiccomponents was <1, it is still possible for the overall toxicityof the metals-plus-organics mixture to exceed 1 and thusrequire further assessment. In this situation the TTRRlow andITUmix values should be summed. If they exceed 1, furtherassessment is required.

If the assessment of the toxicity of either the metals or theorganic chemicals resulted in a value of �1, further assessmentis automatically required. However, if both the TTRRlow valuefor metals and the ITUmix value for organic components was<1, it is still possible for the overall toxicity of the metals-


plus-organics mixture to exceed 1 and thus require furtherassessment. In this situation the TTRRlow and ITUmix valuesshould be summed. If they exceed 1, further assessment isrequired.

It should be noted that estimating the toxicity of metalmixtures is highly controversial. Approaches that have beenadopted range from the pragmatic use of concentrationaddition and/or independent action principles for assessingpotential causality (Dyer et al. 2000) to the need formechanistically describing how metals interact and induceecological impacts (see Adams et al. 2011).

Tier II

The CA-based toxicity estimate of the mixture in tier I willin most cases overestimate the potential toxic effects ofcomplex mixtures observed in the field, because in laboratorystudies only 5% to 15% of mixtures were greater than CA(Faust et al. 1994; Warne and Hawker 1995; Ross 1996;Deneer 2000). In fact, as mixtures become more complexthey tend to act additively, as was discussed above in thecontext of the funnel hypothesis. A more accurate estimate ofthe toxicity of the mixture than that obtained from tier I isobtained by using the 2-step model discussed above thatintegrates both CA and IA models. An excellent descriptionof the mathematics and caveats to the 2-step model is providedin Posthuma et al. (2008). A flow diagram of tier II for organicchemicals is provided in Figure 6. The framework for tier II

Figure 6. Flow diagram of tier II of the mixture framework that evaluates the po

unknown MeOA cannot be considered in this scheme but could be treated as b

presented in Figure 6 would also be used for tier III inves-tigations except at the community level (see the Tier III section).

The recommended stepwise process for tier II for mixturesof organic chemicals is set out as follows:

S

ten

as

tep 1: Obtain TR concentration data or suitable surrogatedata. These could be estimated TR concentration dataestimated from passive samplers or from traditionalambient media-based toxicity data (e.g., toxicity dataexpressed in terms of water concentrations).
S tep 2: Compare the measured or estimated TR concen-trations with the threshold for baseline toxicity, this beingthe acute IEC50 divided by 30 (the number 30 was chosenwith respect to the funnel hypothesis). Based on the resultsof the funnel hypothesis (Warne and Hawker 1995),chemicals with TR concentrations that are lower than thisthreshold have a high probability of exerting their toxicityby baseline toxicity. For such chemicals, the toxicity of themixture is determined using the CA approach. Chemicalsthat have TR concentrations greater than the threshold forbaseline toxicity can be assumed to exert their toxicity bytheir specific MeOA. S tep 3: Such chemicals should therefore be divided intotheir various MeOAs. Descriptions of the various MeOAsfor organics are provided in the accompanying review onmodes of toxic action in TRA (Escher et al. 2011). TheITUmix of the mixture of chemicals belonging to eachMeOA is then determined using CA. Toxicity data for the
tial effects due to mixture residues of organic chemicals. Compounds with

eline MeOA in an initial tier, particularly if the chemical is organic.


same toxic endpoint and exposure duration should be usedfor the CA calculations.
S tep 4: The ITUmix for each MeOA is compared to thevalue that corresponds to 1/30th of the IEC50 (i.e.,ITUmix¼ 1/30). If the ITUmix equals or exceeds this valuefor a MeOA, those chemicals will exert their toxicity bytheir specific MeOA. If the ITUmix for a MeOA is less than1/30, the group of chemicals with this specific MeOAwillexert their toxicity by baseline toxicity. S tep 5: Those chemicals with specific MeOAs that occur atsufficiently low concentrations in a mixture that they exerttheir toxicity by the baseline MeOA are combined using CAwith the ITUmix of chemicals that only have a baseline MeOA. S tep 6: If any MeOAs are at sufficiently high concentrationsto exert their toxicity by their specific MeOA, these groupsare combined using IA with those chemicals that exert theirtoxicity by any MeOA.
Depending on the outcomes of step 4, either step 5 or 6will provide the final estimate of the toxicity of the mixturebeing examined.

In concept, an analogous procedure could be used formetals, except baseline toxicity would not be assumed ifthe ITUmix for each MeOA is less than 1/30. Therefore, theITUmix for all the MeOAs would be directly incorporated inthe IA calculation, instead of first being filtered through thebaseline toxicity evaluation. However, a major drawback isthat the MeOAs for many metals are not known or onlypoorly known. Niyogi and Wood (2003) presented a 3-mechanism model for the 6 best-studied cationic metals (Ag,Cd, Cu, Ni, Pb, and Zn).

Tier III (extrapolation of the mixed model for single speciesto communities)

Tier II permits the assessment of the toxicity of mixtures toindividual species in terms of TRs. However, in ecotoxicologywe are not concerned exclusively with individuals orindividual species but rather with the ongoing viability ofpopulations, communities and ecosystems. Therefore, infor-mation on the effects of mixtures on individual species mustbe aggregated together in a manner that permits assessment ofthe effects on communities. The method for doing this iscalled the multiple substance potentially affected fraction(ms-PAF) and was first proposed by Hamers et al. (1996) andis explained in detail by Traas et al. (2002). Briefly, the ms-PAF uses species sensitivity distributions as the responsevariable for aquatic communities and then uses IA mathe-matics for the joint action of multiple substances (e.g.,chemicals) exposed to an aquatic community. This methodincorporates the 2-step mixture model method and speciessensitivity distribution (SSD) methods of Altenburger et al.(2004), Junghans (2004), and De Zwart and Posthuma(2005). The ms-PAF method has subsequently been used toassess the risk associated with mixtures, e.g., the toxicity ofmixtures of organic and inorganic compounds in Dutchwaterways and that of metals in Dutch soils (NetherlandsNational Institute for Public Health and the Environment, theRijksinstituut voor Volksgezondheid en Milieu [RIVM] 1997,2000, respectively). The method has also been incorporatedinto the mixture toxicity framework developed by Posthumaet al. (2008). However, all the work to date using ms-PAF hasused toxicity data based on ambient environmental concen-

trations (e.g., aqueous or soil concentrations) rather thantissue concentrations. A recent study by Jager et al. (2007)demonstrated how toxicity data of organic chemicals can bedecomposed into 2 factors: 1) the potency of the givenchemical (representing mainly the toxicokinetics) and 2) thevulnerability of the species (representing solely the toxicody-namics). It is then possible to derive general SSDs that areonly based on the distribution of vulnerability of the differentspecies. This generalized SSD approach is consistent with thecentral tenet of the TRA in ecotoxicology. Although notdemonstrated in Jager’s paper, this approach can in principlebe modified for metals, for which even closely related speciescan have different sequestration or elimination strategies thatcomplicate the relationships underlying the SSDs, but whichwould be accounted for by the species vulnerability factor.

Thus, the goal in tier III is a mixed model that first usesCA, followed by IA to calculate the toxicity of mixtures toderive TR concentrations that are protective of a selectedpercentage of aquatic communities. Very similar formulas areused in the ms-PAF method to determine CA and IA as in tierII of the framework. However, when calculating CA, thetoxicity of each component of a mixture will be expressed interms of internal hazard units (IHU) rather than in terms ofITUs (as in Eqn. 3). The IHU for a particular combination ofchemical and species is the ratio of the TR-based NOEC forthat chemical to that species divided by the mean TR-basedNOEC of that chemical to all species for which there are data.The IHU values for each chemical in a mixture are thenentered separately into a SSD method. The concentration ofeach chemical in the mixture is then converted to IHU valuesand then chemicals with the same MeOA have their IHUvalues summed. Again, it might be possible to extend thisapproach to metals, as proposed by De Zwart et al. (2006).The IA toxicity of mixtures of chemicals that have differentMeOAs is calculated using Equation 6, except that thebiological response (BR) is replaced by the TR-based PAF foreach chemical. Details of how this is performed are providedby Posthuma et al. (2008).

CONCLUSIONSThis review illustrates principles by which mixtures of

similarly and dissimilarly acting organic and metal compoundscan be assessed for toxicity by using internal residue con-centrations. Further, we provide a framework by which thetoxicity of mixtures can be estimated for single species as wellas multiple species and communities. Even so, much work isneeded to test both these principles and the framework inthe field, as well as to assess their accuracy in understandingcausality. Such experiments will require collaboration ofacademic, government, and industry scientists, all of whomhave the goal of appropriately assigning risk and prioritizingremediation efforts based on sound scientific approaches.Currently very topical issues such as endocrine disruptors,nanomaterials, pharmaceuticals, and consumer productchemicals all fit within the constructs presented in thisreview.

Acknowledgment—We thank Jim Meador, Lynn McCarty

and two anonymous reviewers for suggestions that improvedthe manuscript. The authors thank the numerous sponsorsfor their generous financial contributions that supported thisworkshop.


REFERENCESAbernethy SG, Mackay D, McCarty LS. 1988. Volume fraction correlation for

narcosis in aquatic organisms: The key role of partitioning. Environ Toxicol

Chem 7:469–481.

Adams W, Blust R, Borgmann U, Brix K, DeForest D, Green A, Meyer S, McGeer J,

Paquin P, Rainbow P, Wood C. 2011. Utility of tissue residues for predicting

effects of metals on aquatic organisms. Integr Environ Assess Manag 7:75–98.

Altenburger R, Backhaus T, Boedeker W, Faust M, Scholze M, Grimme LH. 2000.

Predictability of the toxicity of multiple chemical mixtures to Vibrio fischeri:

Mixtures composed of similarly acting chemicals. Environ Toxicol Chem

19:2341–2347.

Altenburger R, Walter H, Grote M. 2004. What contributes to the combined effect

of a complex mixture? Environ Sci Technol 38:6353–6362.

Ashauer R, Boxall ABA, Brown CD. 2007. New ecotoxicological model to simulate

survival of aquatic invertebrates after exposure to fluctuating and sequential

pulses of pesticides. Environ Sci Technol 41:1480–1486.

Ashford JR. 1981. General modes for the joint action of mixtures of drugs.

Biometrics 37:457–474.

Backhaus T, Altenburger R, Boedeker W, Faust M, Scholze M. 2000a. Predictability

of the toxicity of a multiple mixture of dissimilarly acting chemicals to Vibrio

fischeri. Environ Toxicol Chem 19:2348–2356.

Backhaus T, Altenburger R, Boedeker W, Faust M, Scholze M. 2000b. Predictability

of the toxicity of a multiple mixture with combinations of environmental

chemicals. Science 272:489–492.

Baird DJ, Barber I, Calow P. 1990. Clonal variation and general responses of Daphnia

magna to toxic stress. 1. Chronic life history effects. Funct Ecol 4:399–407.

Barbour MT, Gerritsen J, Griffith GE, Frydenborb R, McCarron E, White JS, Bastian

ML. 1996. A framework for biological criteria for Florida streams using benthic

macroinvertebrates. J North Am Benthol Soc 15:185–211.

Bernhard MJ, Dyer SD. 2005. Fish critical cellular residues for surfactants and

surfactant mixtures. Environ Toxicol Chem 24:1738–1744.

Belden JB, Gilliom RJ, Lydy MJ. 2007. How well can we predict the toxicity of

pesticide mixtures to aquatic life? Integr Environ Assess Manag 3:364–372.

Bliss CI. 1939. The toxicity of poisons applied jointly. Ann Appl Biol 26:585–615.

Booth AM, Sutton PA, Lewis CA, Lewis AC, Scarlett A, ChauW,Widdows J, Rowland

SJ. 2007. Unresolved complex mixtures of aromatic hydrocarbons: Thousands

of overlooked persistent, bioaccumulative, and toxic contaminants in mussels.

Environ Sci Technol 41:457–464.

Borgert CJ, Quill TF, McCarty LS, Mason AM. 2004. Can mode of action predict

mixture toxicity for risk assessment? Toxicol Appl Pharmacol 201:85–96.

Broderius SJ. 1991. Modelling the joint toxicity of xenobiotics to aquatic organisms:

Basic concepts and approaches. In: Mayes MA, Barron MG, editors. Aquatic

toxicology and risk assessment 14th Ed. Philadelphia (PA): American Society for

Testing and Materials. p 107-127.

Broderius SJ, Kahl MD. 1985. Acute toxicity of organic chemical mixtures to the

fathead minnow. Aquat Toxicol 6:307–322.

Broderius SJ, Kahl M, Hoglund MD. 1995. Use of joint toxic response to define the

primary mode of toxic action for diverse industrial organic chemicals. Environ

Toxicol Chem 14:1591–1605.

Brown VM. 1968. The calculation of the acute toxicity of mixtures of poisons to

rainbow trout. Water Res 2:723–733.

Cedergreen N, Streibig JC. 2005. Can the choice of endpoint lead to contradictory

results of mixture-toxicity experiments? Environ Chem Toxicol 24:1676–1683.

Chevre N, Loepfe C, Singer H, Stamm C, Fenner K, Escher BI. 2006. Including

mixtures in the determination of water quality criteria for herbicides in surface

water. Environ Sci Technol 40:426–435.

Chevre N, Loepfe C, Fenner K, Singer H, Escher BI, Stamm C. 2006.

Wirkungsbasierte Qualitatskriterien fur Pestizide in Oberflachengewassern

der Schweiz. Ein konsistentes Konzept fur Einzelstoffe und Mischungen. Gas

Wasser Abwasser 297–307.

Chen CY, Yeh JT., 1996. Toxicity of binary mixtures of reactive toxicants. Environ

Toxicol Water Qual 11:83–90.

Deneer JW. 2000. Toxicity of mixtures of pesticides in aquatic systems. Pest Manag

Sci 56:516–520.

Deneer JW, Sinnige TL, Seinen W, Hermens JLM. 1988. The joint acute toxicity to

Daphnia magna of industrial organic chemicals at low concentrations. Aquat

Toxicol 12:33–38.

DeShon JE. 1995. Development and application of the invertebrate community

index (ICI). In: Davis WS, Simon T, editors. Biological assessment and criteria:

Tools for water resource planning and decision making. Boca Raton (FL): Lewis,

p 217–243.

De Zwart D, Posthuma L. 2005. Complex mixture toxicity for single and multiple

species: Proposed methodologies. Environ Toxicol Chem 24:2665–2676.

De Zwart D, Dyer SD, Posthuma L, Hawkins CP. 2006. Predictive models attribute

effects on fish assemblages to toxicity and habitat alteration. Ecol Appl

16:1295–1310.

Di Toro DM, McGrath JA, Hansen DJ. 2000. Technical basis for narcotic chemicals

and polycyclic aromatic hydrocarbon criteria. I. Water and tissue. Environ

Toxicol Chem 19:1951–1970 .

Dyer SD, White-Hull CE, Shepard BK. 2000. Assessments of chemical mixtures via

toxicity reference values overpredict hazard to Ohio fish communities. Environ

Sci Technol 34:2518–2524.

Dyer SD, Versteeg DJ, Belanger SE, Chaney JG, Mayer FL. 2006. Interspecies

correlation estimates predict protective environmental concentrations.


[ECETOC] European Centre for Ecotoxicology and Toxicology of Chemicals. 2001.

Aquatic toxicity of mixtures. Technical Report 80. Brussels (BE): ECETOC, 64 p.

El-Merhibi A, Kumar A, Smeaton T. 2004. Role of piperonyl butoxide in the toxicity

of chlorpyrifos to Ceriodaphnia dubia and Xenopus laevis. Ecotoxicol Environ

Saf 57:202–212.

Escher BI, Schwarzenbach RP, Westall JWC. 2000. Evaluation of liposome–water

partitioning of organic acids and bases. I. Development of sorption model.


Escher BI, Hunziker RW, Schwarzenbach RP. 2001. Interaction of phenolic

uncouplers in binary mixtures: Concentration-additive and synergistic

effects. Environ Sci Technol 35:3905–3914.

Escher BI, Eggen RIL, Vye E, Schreiber U, Wisner B, Schwarzenbach RP. 2002.

Baseline toxicity (narcosis) of organic chemicals determined by membrane

potential measurements in energy-transducing membranes. Environ Sci

Technol 36:1971–1979 .

Escher BI, Bramaz N, Eggen RIL, Richter M. 2005. In-vitro assessment of modes of

toxic action of pharmaceuticals in aquatic life. Environ Sci Technol 39:3090–

3100.

Escher BI, Ashauer R, Dyer S, Hermens J, Lee J-H, Leslie HA, Mayer P, Meador J,

Warne M. 2011. The crucial role of mechanisms and modes of toxic action for

tissue residues of organic chemicals. Integr Environ Assess Manag 7:28–49.

[EU] European Commission. 2003. European Commission Technical Guidance

Document in Support of Commission Directive 93/67/EEC on Risk

Assessment for New Notified Substances Commission Regulation (EC) 1488/

94 on Risk Assessment for Existing Substances, and Directive 98/8/EC of the

European Parliament and of the Council Concerning the Placing of Biocidal

Products on the Market. Office for Official Publications of the European

Communities. Available from: http://ecb.jrc.it/

Faust M, Altenberger R, Boedeker W, Grimme LH. 1994. Algal toxicity of binary

combinations of pesticides. Bull Environ Contam Toxicol 53:134–141.

Freidig AP, Hofhuis M, Van Holstijn I, Hermens JLM. 2001. Glutathione depletion in

rat hepatocytes: A mixture toxicity study with a,b-unsaturated esters.

Xenobiotica 31:295–307.

Gray BJ. 2004. Australian river assessment system: National guidelines for mapping

AusRivAS Macroinvertebrate scores. Monitoring river health initiative. Technical

report 38. Canberra (AU): Department of the Environment and Heritage.

Grimme LH, Faust M, Boedeker W, Altenburger R. 1996. Aquatic toxicity of

chemical substances in combination: Still a matter of controversy. Hum Ecol

Risk Assess 2:426–433.

Hamers T, Aldenberg T, van de Meent D. 1996. Definition report—indicator effects

for toxic substances (Itox). National Institute of Public Health and the

Environment. Bilthoven (NL): RIVM. Report 607128 001. Available from:

http://www.rivm.nl/bibliotheek/rapporten/607128001.html


Hermens JLM, Canton H, Janssen P, de Jong R. 1984. QSARs and toxicity studies of

mixtures of chemicals with anaesthetic potency: Acute lethal and sub-lethal

toxicity to Daphnia magna. Aquat Toxicol 5:143–154.

Hermens JLM, Leeuwangh P, Musch A. 1985. Joint toxicity of mixtures of groups of

organic aquatic pollutants to the guppy (Poecilia reticulata). Ecotoxicol Environ

Saf 9:321–326.

Hilsenhoff WL. 1987. An improved biotic index of organic stream pollution. Great

Lakes Entomol 20:31–39.

Huckins JN, TubergenMW,Manuweera GK. 1990. Semipermeablemembrane devices

containing model lipids—A new approach to monitoring the bioavailability

of lipophilic contaminants and estimating their bioaccumulation potential.

Chemosphere 20:533–552.

Jager T, Posthuma L, de Zwart D, van de Meent D. 2007. Novel view on predicting

acute toxicity: Decomposing toxicity data in species vulnerability and chemical

potency. Ecotoxicol Environ Saf 67:311–322.

Jensen J, Mesman M, editors., 2006. Ecological risk assessment of contaminated

land. Decision support for site specific investigations. Bilthoven (NL): RIVM.

Report 711701047, Available from: http://www.rivm.nl/bibliotheek/rapporten/

711701047.html

Junghans M. 2004. Studies on combination effects of environmentally relevant

toxicants: Validation of prognostic concepts for assessing the algal toxicity of

realistic aquatic pesticide mixtures. PhD thesis. Bremen (DE): Univ Bremen.

Available from: http://deposit.ddb.de/cgi-bin/dokserv?idn¼975465317&dok_var¼d1&dok_ext¼pdf&filename¼975465317.pdf

Junghans M, Backhaus T, Faust M, Scholze M, Grimme LH. 2006. Application and

validation of approaches for the predictive hazard assessment of realistic

pesticide mixtures. Aquat Toxicol 76:93–110.

Karr JR. 1981. Assessment of biotic integrity using fish communities. Fisheries

6:21–27.

Konemann H. 1981. Fish toxicity tests with mixtures of more than two chemicals: A

proposal for a quantitative approach and experimental results. Toxicology

19:229–238.

Landrum PF, Lotufo GR, Gossiaux DC, Gedeon ML, Lee JH. 2003. Bioaccumulation

and critical body residue of PAHs in the amphipod, Diporeia spp.: Additional

evidence to support toxicity additivity for PAH mixtures. Chemosphere

51:481–489.

Lee JY, Landrum PF. 2006. Development of a multi-component damage assessment

model (MDAM) for time-dependent mixture toxicity with toxicokinetic

interactions. Environ Sci Technol 40:1341–1349.

Lenat DR. 1993. A biotic index for the southeastern United States: Derivation and

list of tolerance values, with criteria for assigning water-quality ratings. J North

Am Benthol 12:279–290.

Leslie HA. 2003. Biomimetic solid phase microextraction: Estimating body residues

and predicting baseline toxicity in aquatic organisms. PhD thesis. Utrecht (NL):

Utrecht Univ.

Leslie HA, ter Laak TL, Busser FJM, Kraak MHS, Hermens JLM. 2002.

Bioconcentration of organic chemicals: is a solid-phase microextraction fiber

a good surrogate for biota? Environ Sci Technol 36:5399–5404.

Leslie HA, ter Laak TL, Oosthoek AJP, Busser FJM, Kraak MHS, Hermens JLM. 2002.

Biomimetic solid-phase microextraction to predict body residues and toxicity of

chemicals that act by narcosis. Environ Toxicol Chem 21:229–234.

Leslie HA, Hermens JLM, Kraak MHS. 2004. Chronic toxicity and body residues of

the nonpolar narcotic 1,2,3,4-tetrachlorobenzene in Chironomus riparius.

Environ Toxicol Chem 23:2022–2028.

Leslie HA, Kraak MHS, Hermens JLM. 2004. Baseline toxicity of a chlorobenzene

mixture and total body residues measured and estimated with solid-phase

microextraction. Environ Toxicol Chem 23:2017–2021.

Luoma SN, Rainbow PS. 2008. Metal contamination in aquatic environments.

Cambridge (UK): Cambridge Univ.

Mayer FL, Deans CH, Smith AG. 1987. Inter-taxa correlations for toxicity to aquatic

organisms. Gulf Breeze (FL): USEPA, Report EPA/600/X-87/332. 59 p.

McCarty LS, Borgert CJ. 2006. Review of the toxicity of chemical mixtures: Theory,

policy, and regulatory practice. Reg Toxicol Pharmacol 45:119–143.

McCarty LS, Mackay D. 1993. Enhancing ecotoxicological modeling and

assessment. Environ Sci Technol 27:1719–1728.

McGeer JC, Brix KV, Skeaff JM, DeForest DK, Brigham SI, AdamsWJ, Green A. 2003.

Inverse relationship between bioconcentration factor and exposure

concentration for metals: Implications for hazard assessment of metals in

the aquatic environment. Environ Toxicol Chem 22:1017–1037.

McGrath JA, Parkerton TF, Di Toro DM. 2004. Application of the narcosis target lipid

model to algal toxicity and deriving predicted-no-effect concentrations. Environ

Toxicol Chem 23:2503–2517.

[MEQ] Ministere de l’Environment du Quebec. 2001. Criteres de qualite de l’eau de

surface au Quebec (Water quality criteria for Quebec). Quebec (PQ): Ministere

de l’Environment du Quebec.

Morton R, Warne MStJ, Correll RL. 2008. Simultaneous prediction of toxicity of

multiple chemicals to multiple species using multivariate functional

relationships. Envirometrics 19:765–784.

Niyogi S, Wood CM. 2003. Effects of chronic waterborne and dietary metal

exposures on gill metal-binding: Implications for the biotic ligand model.

Hum Ecol Risk Assess 9:813–846.

NorwoodWP, Bormann U, Dixon DG, Wallace A. 2003. Effects of metal mixtures on

aquatic biota: A review of observations and methods. Hum Ecol Risk Assess

9:795–811.

Olmstead AW, LeBlanc GA. 2005. Toxicity assessment of environmentally relevant

pollutant mixtures using a heuristic model. Integr Environ Assess Manag

1:114–122.

Opperhuizen A, Schrap SM. 1988. Uptake efficiencies of two polychlorobiphenyls

in fish after dietary exposure to five different concentrations. Chemosphere

17:253–262.

Pape-Lindstrom PA, Lydy MJ. 1997. Synergistic toxicity of atrazine and

organophosphate insecticides contravenes the response addition mixture

model. Environ Toxicol Chem 16:2415–2420.

Parkerton TF, StoneMA, Letinski DJ. 2000. Assessing the aquatic toxicity of complex

hydrocarbon mixtures using solid-phase microextraction. Toxicol Lett 112/

113:273–282.

Paustenbach DJ. 2000. The practice of exposure assessment: A state-of-the-art

review. J Toxicol Environ Health B Crit Rev 3:179–291.

Plackett RL, Hewlett PS. 1952. Quantal responses to mixtures of poisons. J Roy Stat

Soc B 14:141.

Pope C, Karanth S, Liu L. 2005. Pharmacology and toxicology of cholinesterase

inhibitors: Uses andmisuses of a commonmechanism of action. Environ Toxicol

Pharmacol 19:433–446.

Posthuma L, Richards S, De Zwart D, Dyer S, Sibley P, Hickey C, Altenberger R. 2008.

Mixture extrapolation approaches. In: Solomon K, Brock T, de Zwart D, Dyer S,

Posthuma L, Richards S, Sanderson H, Sibley P, Van den Brink P, editors.

Extrapolation practice for ecotoxicological effect characterization of

chemicals. Boca Raton (FL): Taylor & Francis, p 137–185.

Ra JS, Lee BC, Chang NI, Kim SD. 2006. Estimating the combined toxicity by two-

step prediction model on the complicated chemical mixtures from wastewater

treatment plant effluents. Environ Toxicol Chem 25:2107–2113.

Raimondo S, Vivian DN, Barron MG. 2007. Web-based Interspecies Correlation

Estimation (Web-ICE) for acute toxicity: User manual. Version 1.1. Gulf Breeze

(FL): USEPA, Report EPA/600/R-07/071.

Rankin ET. 1989. The Qualitative Habitat Evaluation Index (QHEI): Rationale,

methods and application. Columbus (OH): Ohio Environmental Protection

Agency. Division of Water Quality Planning and Assessments.

Richter M, Escher BI. 2005. Mixture toxicity of reactive chemicals by using two

bacterial growth assays as indicators of protein and DNA damage. Environ Sci

Technol 39:8753–8761.

[RIVM] Rijksinstituut voor Volksgezondheid en Milieu (Netherlands National

Institute for Public Health and the Environment). 1997. National

Environmental Outlook 4. Alphen aan den Rijn (NL): Samson-Tjeenk Willink.

[RIVM] Rijksinstituut voor Volksgezondheid en Milieu (Netherlands National

Institute for Public Health and the Environment)., 2000. National

Environmental Outlook 5. Alphen aan den Rijn (NL): Samson-Tjeenk Willink.

Available from: http://www.rivm.nl/bibliotheek/rapporten/mono00001.html

Roex EWM, van Gestel CAM, vanWezel AP, van Straalen NM. 2000. Ratios between

acute aquatic toxicity and effects on population growth rates in relation to

toxicant mode of action. Environ Toxicol Chem 19:685.


Ross HLB. 1996. The interaction of chemical mixtures and their implications on

water quality guidelines. Honours thesis. Univ Technology, Sydney, NSW,

Australia. 167 p.

Rowland SJ, Volkman JK. 1982. Biogenic and pollutant aliphatic hydrocarbons in

Mytilus edulis from the North Sea. Mar Environ Res 7:117–130.

Rozman KK, Doull J. 2001. The role of time as a quantifiable variable of

toxicity and the experimental conditions when Haber’s c�t product

can be observed: Implications for therapeutics. J Pharmacol Exp Ther

296:663–668.

Sappington KG, Bridges TS, Bradbury SP, Erickson RJ, Hendriks AJ, Lanno RP,

Meador JP, Mount DR, Salazar MH, Spry DJ. 2011. Application of the tissue

residue approach in ecological risk assessment. Integr Environ Assess Manag

7:116–140.

Shephard BK. 1998. Presented at National Sediment Bioaccumulation Conference.

EPA 823-R- 98-002. Washington (DC): US Environmental Protection Agency.

Office of Water, 2-31-2-52 [http://nepis.epa.gov/Exe/ZyPURL.cgi?Dockey¼20003ZBH.txt].

Sodergren A. 1987. Solvent filled dialysis membranes simulate uptake of pollutants

by aquatic organisms. Environ Sci Technol 21:855–859.

Sprague JB. 1970. Measurement of pollutant toxicity to fish. Water Res 4:3–32.

Stephan CE, Mount DI, Hansen DJ, Gentile JH, Chapman GA, Brungs WA. 1985.

Guidelines for deriving numerical national water quality criteria for the

protection of aquatic organisms and their uses. Report PB85–227049.

Duluth (MN): US Environmental Protection Agency.

Syberg K, Jensen TS, Cedergreen N, Rank J. 2009. On the use of mixture toxicity

assessment in REACH and the Water Framework Directive: A review. Hum Ecol


Traas TP, van de Meent D, Posthuma L, Hamers T, Kater BJ, de Zwart D, Aldenberg T.

2002. The potentially affected fraction as a measure of ecological risk. In:

Posthuma L, Suter II, GW, Traas TP, editors. Species sensitivity distributions in

ecotoxicology. Boca Raton (FL): Lewis, p 315–344.

Trimble AJ, Weston DP, Belden JB, Lydy MJ. 2009. Identification and evaluation of

pyrethroid insecticide mixtures in urban sediments. Environ Toxicol Chem

28:1687–1695.

Vaes WHM, Urrestarazu-Ramos E, Verhaar H, Hermens JLM. 1998. Acute toxicity of

nonpolar versus polar narcosis: Is there a difference? Environ Toxicol Chem

17:1380–1384.

van der Wal L, Jager T, Fleuren RHLJ, Barendregt A, Sinnige TL, van Gestel CAM,

Hermens JLM. 2004. Solid-phase microextraction to predict bioavailability and

accumulation of organic micropollutants in terrestrial organisms after exposure

to a field-contaminated soil. Environ Sci Technol 38:4842–4848.

van Leeuwen CJ, van der Zandt PTJ, Aldenberg T, Verhaar H, Hermens JLM. 1992.

Application of QSARs, extrapolation and equilibrium partitioning in aquatic

effects assessment. I. Narcotic industrial pollutants. Environ Toxicol Chem

11:267–282.

van Leeuwen CJ, Verhaar HJM, Hermens JLM. 1996. Quality criteria and risk

assessment for mixtures of chemicals in the aquatic environment. Hum Ecol


van Loon WMGM, Verwoerd ME, Wijnker FG, van Leeuwen CJ, van Duyn P, van de

Guchte C, Hermens JLM. 1997. Estimating total body residues (TBRs) and

baseline toxicity of complex organic mixtures in effluents and surface waters.

Environ Toxicol Chem 16:1358–1365.

van Wezel AP, Jonker MTO. 1996. Use of the lethal body burden in the risk

assessment of field sediments: Influence of Pimephales promelas

temperature and salinity. Aquat Toxicol 42:287–300.

van Wezel AP, deVries DAM, Sijm DTHM, Opperhuizen A. 1996. Use of the lethal

body burden in the evaluation of mixture toxicity. Ecotoxicol Environ Saf

35:236–241.

Verbruggen EMJ, van LoonWMGM, Tonkes M, van Duin P, SeinenW, Hermens JLM.

1999. Polyacrylate coated SPME fibers as a tool to simulate body residues and

target concentrations of complex organic mixtures for estimation of baseline

toxicity. Environ Sci Technol 34:324–331.

Verhaar HJM, Busser FJM, Hermens JLM. 1995. A surrogate parameter for the

baseline toxicity content of contaminated water: Simulating bioconcentration

and counting molecules. Environ Sci Technol 29:726–734.

Verhaar HJM, De Wolf W, Dyer S, Legierse KCHM, Seinen W, Hermens JLM. 1999.

An LC50 vs time model for the aquatic toxicity of reactive and receptor-

mediated compounds. Consequences for bioconcentration kinetics and risk

assessment. Environ Sci Technol 33:758–763.

Warne MStJ, Hawker DW. 1995. The number of components in a mixture

determines whether synergistic, antagonistic or additive toxicity

predominate: The Funnel Hypothesis. Ecotoxicol Environ Saf 31:23–28.

Warne MStJ, Connell DW, Hawker DW. 1991. Comparison of the critical

concentration and critical volume hypotheses of non-specific toxicity for

individual compounds. Toxicology 66:187–195.

Wilcockson JB, Gobas FAPC. 2001. Thin-film solid phase extraction to measure

fugacities of organic chemicals with low volatility in biological samples. Environ

Sci Technol 35:1425–1431.

Wright JF, Furse MT, Armitage PD, Moss D. 1993. New procedures for identifying

running-water sites subject to environmental stress and for evaluating sites

for conservation, based on the macroinvertebrate fauna. Arch Hydrobiol

127:319–326.

tissue residue approach for chemical mixtures

Documents