An ecologically relevant exposure assessment for a polluted river using an integrated multivariate PLS approach R. Smolders * , W. De Coen, R. Blust Laboratory for Ecophysiology, Biochemistry and Toxicology, University of Antwerp (RUCA), Groenenborgerlaan 171, B-2020 Antwerp, Belgium Received 10 November 2003; accepted 20 April 2004 ‘‘Capsule’’: Multivariate analysis indicates how multiple lines of evidence can be integrated to provide a holistic overview of the impact of effluent discharges on the receiving aquatic ecosystem. Abstract A case study is presented where an integrated, ecologically relevant exposure assessment is presented for a polluted lowland river. Using partial least squares regression of latent structures (PLS), an analysis of the impact of two effluents on physico-chemical water quality measures, macroinvertebrate and diatom communities, and in situ bioassay responses with four different test species are combined into an integrative exposure assessment. Bioassays focussed on growth and condition related endpoints, because they are key functional processes of organisms and populations. Integrating these multiple lines of evidence, we were able to discriminate among the impact of both effluents, link changes in physico-chemical water quality with bioassay endpoints and ecological quality of the ecosystem, and address the importance of integrating all information into one exposure assessment framework. The bioassays under field conditions indicated that most endpoints measured are a reflection of ecological effects rather than pollution related effects, or at least a combination of both ecological and toxicological effects. Factors such as food availability clearly influenced the outcome of in situ bioassays and ecological information was essential to explain observed discrepancies when trying to extrapolate bioassay results from the laboratory to the field. Ó 2004 Elsevier Ltd. All rights reserved. Keywords: Ecological relevance; PLS; Effluent; Pollution gradient; Multiple lines of evidence 1. Introduction More and more, the initial framework of ecotoxicol- ogy, describing single-source–single-effect stressors un- der controlled laboratory conditions, is being replaced by an ecosystem approach, which focuses on holistic and integrated studies at a basin or watershed level (Cormier et al., 2000; Isnard, 1998; Serveiss, 2002; Smolders et al., this issue). The discharge of effluents and associated toxic compounds into aquatic systems represents an ongoing environmental problem due to their possible impact on communities in the receiving aquatic ecosystem (Canivet and Gibert, 2002). Espe- cially in densely populated and industrial areas, point and non-point sources of anthropogenic chemicals and metals have polluted streams and rivers with highly complex mixtures of chemicals and other anthropogenic perturbations to a degree where life in streams and rivers is often severely impacted (Chapman, 2002; Demuynck et al., 1997; McCarty and Power, 1997). Karr and Chu (1997) argue that the first most important step for the protection of an aquatic ecosystem is to identify biological endpoints, and impact assessments * Corresponding author. Tel.: C32-3-2180349; fax: C32-3- 2180497. E-mail address: roel.smolders@ua.ac.be (R. Smolders). 0269-7491/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.envpol.2004.04.024 Environmental Pollution 132 (2004) 245–263 www.elsevier.com/locate/envpol