Transfer of PCBs from bottom sediment to freshwater river fish: A food-web modelling approach in the Rhˆ one River (France) in support of sediment management C. Lopes a , H. Persat b , M. Babut a,n a IRSTEA, UR MALY, 3 bis Quai Chauveau — CP220, F-69336 Lyon, France b UMR 5023 — LEHNA, Universite´ Claude Bernard Lyon 1, 69221 Villeurbanne, France article info Article history: Received 24 November 2011 Received in revised form 3 April 2012 Accepted 4 April 2012 Available online 22 May 2012 Keywords: PCB Food-web bioaccumulation model Bayesian inference Risk assessment Sediment management abstract Since 2005, restrictions have been because of fish consumption along the Rhone River because of high polychlorobiphenyl (PCB) concentrations, which have resulted inadverse economic consequences for professional fisheries in affected areas. French environmental authorities have expended considerable efforts to research sediment remediation strategies and development of sediment quality guidelines designed to protect the health of humans consuming Rhˆ one River fish. Here we: (1) develop a bioaccumulation food-web model that describes PCB concentrations in three common freshwater fish species of the Rhˆ one River, using Bayesian inference to estimate the input parameters; (2) test the predictive power of the model in terms of risk assessment for fish consumption; and (3) discuss the use of this approach to develop sediment quality guidelines that protect the health of humans consuming Rhˆ one River fish. The bioaccumulation model predictions are protective for human consumer of fish and are efficient for use in risk assessment. For example, 85% of the predicted values were within a factor of 5 of measured CB153 concentrations in fish. Using sensitivity analyses, the major role played by sediment and diet behaviors on bioaccumulation process is illustrated: the parameters involved in the respiratory process (contamination from water) have little impact on model outputs, whereas the parameters related to diet and digestion processes are the most sensitive. The bioaccumulation model was applied to derive sediment concentrations compatible with safe fish consumption. The resulting PCB sediment thresholds (expressed as the sum of seven PCB indicator congeners) that are protective for the consumption of the fish species ranged from 0.7 to 3 ng/g (dw). & 2012 Elsevier Inc. All rights reserved. 1. Introduction Persistent organic pollutants, such as polychlorinated biphenyls (PCBs), are widespread in the environment and can adversely affect human health and wildlife. Food-web bioaccumulation in aquatic ecosystems has been studied because of its relevance for environ- mental toxicology and risk assessment in an attempt to implement regulatory measures. In this context, food-web bioaccumulation models can be considered tools for researchers and environmental managers (Arnot and Gobas, 2006; Kelly et al., 2007). Levels of hydrophobic contaminants in fish have been shown to be highly variable at inter-individual (Kiriluk et al., 1995; Dufour et al., 2001), seasonal (De Laender et al., 2010b), and species (Kiriluk et al., 1995) levels, and between ecosystems (Kidd et al., 1998). This variability can be caused by environmental factors, such as the basal contamination level and pathways for the ecosystem considered (either atmospheric or point sources) (Stapleton et al., 2001; Stapleton et al., 2002), or from physiological factors, such as fish lipid concentrations (Borg ˚ a et al., 2004), age (Larsson et al., 1991), and metabolic rates (Burtnyk et al., 2009). Stable isotope methods applied to ecotoxicological studies have revealed that the variability in fish contamination levels can be related to the trophic position of the fish (Kiriluk et al., 1995; Kidd et al., 1998; Kidd et al., 2001) or to feeding habits (Kidd et al., 2001; Lopes et al., 2011). As a conse- quence, a food-web bioaccumulation model usually must include a large number of parameters, determined from the literature or from laboratory experiments, site-specific measurements, or empirical relationships. Linkov et al. (2001) showed that the use of point estimates in a bioaccumulation food-web model may significantly overestimate risk. Many studies have used least squares methods to estimate parameters and obtain the ‘‘best’’ match between estimates and measurements (Lundstedt-Enkel et al., 2005). More complicated approaches, such as Bayesian inference or linear inverse modelling, Contents lists available at SciVerse ScienceDirect journal homepage: www.elsevier.com/locate/ecoenv Ecotoxicology and Environmental Safety 0147-6513/$ - see front matter & 2012 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.ecoenv.2012.04.007 n Corresponding author. E-mail address: marc.babut@irstea.fr (M. Babut). Ecotoxicology and Environmental Safety 81 (2012) 17–26