FORUM Indirect Effects in Aquatic Ecotoxicology: Implications for Ecological Risk Assessment BENJAMIN L. PRESTON* Pew Center on Global Climate Change 2101 Wilson Boulevard, Suite 550 Arlington, Virginia 22201, USA ABSTRACT / Understanding toxicant effects at higher levels of biological organization continues to be a challenge in ecotoxicology and ecological risk assessment. This is due in part to a tradition in ecotoxicology of considering the di- rect effects of toxicants on a limited number of model test species. However, the indirect effects of toxicity may be a significant factor influencing the manner in which ecosys- tem structure and function respond to anthropogenic stres- sors. Subsequently, failure to incorporate indirect effects into risk assessment paradigms may be a significant source of uncertainty in risk estimates. The current paper ad- dresses the importance of indirect effects in an ecotoxico- logical context. Laboratory, mesocosm, and whole ecosys- tem research into indirect effects is reviewed. The implications of indirect effects for ecological risk assess- ment and potential areas of profitable future research are also discussed. Since its first use in the late 1970s, the term ecotoxi- cology has been defined generally as the study of the effects of anthropogenic toxicants on ecological sys- tems (Truhaut 1977). This implies an emphasis on toxicant impacts on large-scale ecological phenomenon such as ecosystem structure (e.g., species abundance and diversity) and function (e.g., productivity and nu- trient cycling). However, in practice, ecotoxicologists frequently focus on toxicity at smaller scales of biolog- ical organization, such as physiological mechanisms of toxicity and the responses of discrete endpoints in a single species to toxicant exposure (Clements and Kiffney 1994). This disparity represents a continuing fundamental challenge in ecotoxicology. Although there must necessarily be a connection between these two scales (Calow 1994), our ability to extrapolate from toxicity at the individual level to that at the ecosystem level is currently limited (Barnthouse and others 1987, Cairns and McCormick 1992, Chapman 1995, Newman 1995, Power and Adams 1997). As a result, ecotoxicolo- gists find themselves in the precarious position of at- tempting to develop estimates of environmental risk sufficient to make informed management decisions with only a fraction of the information that is needed or at least a lack of certainty regarding what information is needed. The source of this dilemma has frequently been attributed to the scientific methods that have tradition- ally been utilized in ecotoxicology, namely standard- ized single-species toxicity bioassays. Several authors have questioned the utility of such toxicity assays in understanding toxicant effects at the ecosystem level (Cairns 1983, 1989, Clements and Kiffney 1994, Chap- man 1995), and there are certainly significant uncer- tainties associated with the ecological relevance of such data (US EPA 1998, Preston 2000). Toxicity assays are frequently conducted with a small body of test organ- isms, despite the significant difference in sensitivity that may exist among species and the diversity of species in natural ecosystems. There is a bias in toxicity databases for acute toxicity data, which is of questionable utility in predicting the consequences of sublethal exposures over extended time scales. Estimates of toxicity may also vary significantly with environmental conditions, con- founding site-specific risk assessments (Spruage 1995, Preston and others 1999a, 2000). However, methods have been developed to address these challenges. Sub- lethal endpoints, including biomarkers of toxicity, have increased our ability to detect stress responses at low toxicant concentrations prior to impacts on ecosystem structure and function (Chen and others 1999; Karouna-Renier and Zehr, 1999; Hassanein and others 1999). Methods for adjusting toxicity estimates based upon environmental conditions such as pH and water hardness are commonly employed (US EPA 1986), and ecotoxicologists are developing a firmer understanding of interspecies variation in toxicant sensitivity (Versteeg and others 1999). Despite these advances, a fundamental obstacle to risk assessment at higher levels of biological organiza- tion remains. That obstacle is the prevalent assumption that in natural environments, individual organisms function as discrete units, and thus, if one can rigor- KEY WORDS: Ecotoxicology; Indirect effects; Ecological risk assess- ment; Interspecific interactions *email: prestonb@pewclimate.org DOI: 10.1007/s00267-001-0023-1 Environmental Management Vol. 29, No. 3, pp. 311–323 © 2002 Springer-Verlag New York Inc.