Journal of Aquatic Ecosystem Stress and Recovery 6: 159–177, 1998. 159 c 1998 Kluwer Academic Publishers. Printed in the Netherlands. Pesticide toxicity endpoints in aquatic ecosystems David Simon, Stuart Helliwell & Kevin Robards School of Science and Technology, Charles Sturt University, P.O. Box 588, Wagga Wagga 2678, Australia Received 19 March 1997; accepted in revised form 1 December 1997 Key words: ecosystems, endpoints, pesticide, review, toxicity Abstract To adequately protect aquatic ecosystems from impact by anthropogenic perturbations it is necessary to distinguish what is safe from what is not. This review examines approaches to this problem in relation to primary and secondary effects of pesticides. Understanding nutrient – plankton and plankton – plankton interrelationships on both spatial and temporal scales is important if secondary or indirect effects are to be assessed. Before defining or measuring a toxicity endpoint, consideration must be given to whether to use single species or multispecies tests. Each has its strengths and weaknesses and is reviewed. In single species testing, toxicity endpoints can be more clearly defined but extrapolation of effects to an ecosystem is more difficult than with multispecies testing and can often lead to incorrect conclusions. Interpretation of multispecies testing results are challenging and numerical analysis techniques including methods whose objectives are inference, classification and ordination are required. Conceptual and fuzzy logic modelling techniques promise a solution to the interpretation of multispecies tests. 1. Introduction Exposure of aquatic ecosystems to pesticides results from both the direct application of pesticides and the accidental release of pesticide polluted water from, for example, rice bays or cotton farm holding dams (e.g. Bowmer, 1982a & 1982b; Scribner et al., 1987; Thoma & Nicholson, 1989; Peterson & Batley, 1993a; Korth et al., 1994). Several assessments have been made (Boreham and Birch, 1987; Raven and George, 1989) of the ecological effects of accidental introduc- tions of pesticides into aquatic systems. Nevertheless, the pollution of both soil and waterways by agricul- tural chemicals is difficult to observe unless there is a gross accidental or intentional spill. Primary effects such as fish kills can be detected but secondary effects may go unnoticed. Ideally, one desires an answer to the question: Does a particular concentration of pes- ticide or pesticide combination pose a threat to the ecosystem? This is distinct from the question of toxic- ity to individual species and requires that the toxicity of the chemical must be defined in terms of its effect at the cell, organism, population and community level (Holcombe et al., 1989) and implies that its actual or predicted environmental exposure concentration must be quantified. The classic ecotoxicology approach to testing aquatic toxicity is to measure the direct effect in simple experiments using death, more often than not, as the endpoint (Cairns, 1983). Both single and mul- tispecies studies have been used in defining endpoints and assessing pesticide toxicity. The endpoint of a given toxicity that defines the boundary between acceptable and too-much must be identified. For example, Barron and Woodburn (1995) distinguish acute studies in which mortality is the prin- cipal endpoint from chronic studies with sublethal tox- icity endpoints (e.g., growth, reproduction, behaviour, non-lethal pathology). The determination of an end- point encompasses the defining of effect. However, to be of use an endpoint must describe an observable effect, whether direct (e.g. the fish died) or indirect (e.g. poor reproduction due to poor growth), that can be defined and measured. Touart (1994) also suggested that the definition of the endpoint fitted into one or more of three groups; ecological (relative to the ‘health’ of the ecosystem), socioeconomic and/or regulatory