2665 Environmental Toxicology and Chemistry, Vol. 24, No. 10, pp. 2665–2676, 2005  2005 SETAC Printed in the USA 0730-7268/05 $12.00 + .00 Hazard/Risk Assessment COMPLEX MIXTURE TOXICITY FOR SINGLE AND MULTIPLE SPECIES: PROPOSED METHODOLOGIES DICK DE ZWART* and LEO POSTHUMA National Institute for Public Health and the Environment, Laboratory for Ecological Risk Assessment, P.O. Box 1, NL-3720 BA Bilthoven, The Netherlands ( Received 9 December 2004; Accepted 23 March 2005) Abstract—Methods for the assessment of ecological risks associated with exposure to defined mixtures of toxicants are reviewed and formalized for single-species toxicity. Depending on the modes of action of toxicants in a mixture, these methods apply either the model for concentration additivity (CA) or the model for response additivity (RA). For complex mixtures, the present paper advocates the use of a new, two-step, mixed-model approach as a logical extension of model selection: Mixture toxicity for individual modes of action is evaluated with the CA model, and the toxicities of different modes of action are combined using the RA model. Using comparable mixture toxicity strategies in combination with the concept of species-sensitivity distributions, we develop a method to address and predict the risk for direct effects on the composition of species assemblages and biodiversity. The data needed for modeling can be obtained from existing databases, and lack of data can, in part, be addressed by the use of toxicity patterns in those databases. Both single- and multiple-species methods of mixture risk prediction are useful for risk management, because they allow ranking of polluted sites and affected species as well as identification of the most hazardous contaminants, at least in a comparative way. Validation of the proposed methods is feasible but currently limited because of a lack of appropriate data. Keywords—Mixture toxicity Species assemblages Ecological risk assessment Toxic mode of action INTRODUCTION Most ecotoxicological studies focus on exposure and effects of single compounds [1]. However, organisms in a polluted environment generally are exposed to many pollutants. With the exception of effect-based diagnostic bioassay interpretation (e.g., whole-effluent toxicity), most regulatory methods for the management of chemical compounds are based on single-sub- stance risk evaluations in combination with some basic toxi- cological models to predict the joint effect of chemical mix- tures on single species [2–4]. These models have been tested rigorously [5] and been found to be reliable for mixture studies testing high doses of a few constituents. Most real-world en- vironmental exposures, however, are to low doses and to a more complex range of chemicals ([6];http://www.toxicology. org / memberservices / meetings / MixturesWhitePapers. doc). The extrapolation from single-species mixture toxicity to in situ risk for an assemblage of species exposed to a mixture of pollutants adds complexity. The nature of the chemicals in the mixture, the variability of exposure routes, and the ranges of sensitivities of the receptor organisms are all crucial factors that determine the type and intensity of responses. The theo- retical developments for multispecies risk assessment are weak, and the collection of experimental data lags far behind. Nonetheless, methods to evaluate the combined risk of mul- tichemical exposure to multispecies systems are urgently need- ed given the common occurrence of complex mixtures in the environment [6,7]. Validated methods for mixture risk assess- ment might be beyond current reach, but a relative ranking of contaminated sites and the most affected groups of species, as well as identification of the most hazardous compounds, are all important for informed management decisions. * To whom correspondence may be addressed (d.de.zwart@rivm.nl). The present paper aims to review the current conceptual approaches of mixture toxicity evaluation for single species and the technical procedures for mixture risk calculation; to explore, discuss, and define technical procedures for estimating the risks of complex toxicant mixtures for multispecies bio- logical systems; and to explore limitations, technical feasibility (input data), and options for validation. SINGLE-SPECIES EFFECT PREDICTION FOR TOXICANT MIXTURES Concepts and limitations Methods to assess the joint action of components in a mix- ture of toxicants largely are based on the conceptual ground- work laid by Bliss [2] and are mathematical rather than bio- logical in nature. Plackett and Hewlett [3] expanded the scheme of Bliss with the possible types of interactions that can occur between chemical components of mixtures (Table 1). For the noninteractive or independent types of joint action, it is assumed that the chemicals in the mixture do not affect the toxicity of one another. Two different models are available depending on the toxic modes of action (TMoA) of the chem- icals in the mixture. The modeling approach commonly known as concentration addition (CA) relates to simple similar action (SSA) and concerns mixtures of chemicals with the same TMoA. The modeling approach called response addition (RA) is related to independent joint action (IJA) and is used to predict the combined effect of toxicants with dissimilar TMoA. No models other than empirical observations are available for the prediction of interactive joint action (either for similar or dissimilar TMoA), where the constituents of the mixture in- fluence each other in the expression of their toxicity. The models of CA and RA commonly are used for the