Ion method [39]. Measuring MT concentrations using the cadmium thiomolybdate saturation assay of Klein et al. [37] allows the determination of MT binding places including the ones initially occupied by Cu, Cd and Zn. Measured MT concentrations in liver were compared to Cu, Cd and Zn concentrations measured in those tissues. In order to investigate differences in detoxification capacities among the three fish species, the theoretical level of MT needed to bind all the present Cd, Cu and Zn in the liver was calculated for the three species. Based on the specific binding capacity of the thiol groups of MT for the three metals (7 mol Cd, 12 mol Cu and 7 mol Zn/ mol MT, Kito et al. [38]) the following equation was used: MTtheoretical = [Cd]tissue/7 + [Cu]tissue/12+[Zn]tissue/7. The ratio MTtheoretical/MTmeasured (further referred to as MTt/MTm) provides an indication of how efficient an organism is inducing MTs after exposure to metals.Statistical analysisANOVA (Analysis of Variance) and Duncan’s multiple range tests were used to determine the significance of the differences between the different sampling sites. If the data were not normally distributed or showed heterogeneous variances, they were log transformed 
to get homogeneous variances and a normal distribution. Multiple linear regressions were used to construct empirical models to relate exposure concentrations to the tissue metal concentrations and the tissue MT concentrations. All statistical 
analyses were performed using the software package STATISTICA (get BTZ-043 StatSoft Inc., Tulsa).ResultsTable 1 summarizes the results of the water quality characteristics from the different sampling sites. General water quality characteristics all met the Flemish and international water quality standards. Total Cd and Zn levels clearly followed a gradient between site 2 and site 7. For copper no pattern was observed. Cd and Zn levels exceeded quality standards at all sites, including site 1 upstream of the pollution source but except the reference site, Table 2 gives the results of the sediment and pore water analysis. Cadmium and Zn concentrations in the sediment did not follow the same gradient as for the dissolved metals but also exceeded sediment quality standards. Highest levels of both metals were measured more downstream of the pollution source. At the downstream sites increased copper levels were measured although nowhere the quality standards were exceeded. For each species fish of comparable size were selected 15755315 for analysis (total length: roach: 13.9 6 2.8 cm; gudgeon: 11.1 6 1.6; perch: 14.0 6 5.6 cm). Roach could be captured in sufficient numbers at all sites. Gudgeon however, was absent at site 2 and only one specimen was captured at sites 3 and 4. Perch was absent at site 2. In total, metal and MT levels in the liver could be measured in 45 roach, 23 perch and 33 gudgeons captured in the pollution gradient. Figure 2 shows the results of the Cd, Cu and Zn concentrations in the liver of the three fish species caught at the different sampling. The average wet weight/dry weight ratios for the liver were 3.63 6 0.39, 3.85 6 0.31, 3.05 6 0.60 for Roach, Perch andCondition indicesThe Licochalcone A biological activity relative condition factor (K) was calculated according to the formula K = W/W’ [40,41], where the observed weight of an individual fish (W in g) is compared to its expected weight (W’) based on its observed length (L). This expected weight is derived from the length-weight regressions W = a Lb, with a and b as fitting c.Ion method [39]. Measuring MT concentrations using the cadmium thiomolybdate saturation assay of Klein et al. [37] allows the determination of MT binding places including the ones initially occupied by Cu, Cd and Zn. Measured MT concentrations in liver were compared to Cu, Cd and Zn concentrations measured in those tissues. In order to investigate differences in detoxification capacities among the three fish species, the theoretical level of MT needed to bind all the present Cd, Cu and Zn in the liver was calculated for the three species. Based on the specific binding capacity of the thiol groups of MT for the three metals (7 mol Cd, 12 mol Cu and 7 mol Zn/ mol MT, Kito et al. [38]) the following equation was used: MTtheoretical = [Cd]tissue/7 + [Cu]tissue/12+[Zn]tissue/7. The ratio MTtheoretical/MTmeasured (further referred to as MTt/MTm) provides an indication of how efficient an organism is inducing MTs after exposure to metals.Statistical analysisANOVA (Analysis of Variance) and Duncan’s multiple range tests were used to determine the significance of the differences between the different sampling sites. If the data were not normally distributed or showed heterogeneous variances, they were log transformed to get homogeneous variances and a normal distribution. Multiple linear regressions were used to construct empirical models to relate exposure concentrations to the tissue metal concentrations and the tissue MT concentrations. All statistical analyses were performed using the software package STATISTICA (StatSoft Inc., Tulsa).ResultsTable 1 summarizes the results of the water quality characteristics from the different sampling sites. General water quality characteristics all met the Flemish and international water quality standards. Total Cd and Zn levels clearly followed a gradient between site 2 and site 7. For copper no pattern was observed. Cd and Zn levels exceeded quality standards at all sites, including site 1 upstream of the pollution source but except the reference site, Table 2 gives the results of the sediment and pore water analysis. Cadmium and Zn concentrations in the sediment did not follow the same gradient as for the dissolved metals but also exceeded sediment quality standards. Highest levels of both metals were measured more downstream of the pollution source. At the downstream sites increased copper levels were measured although nowhere the quality standards were exceeded. For each species fish of comparable size were selected 15755315 for analysis (total length: roach: 13.9 6 2.8 cm; gudgeon: 11.1 6 1.6; perch: 14.0 6 5.6 cm). Roach could be captured in sufficient numbers at all sites. Gudgeon however, was absent at site 2 and only one specimen was captured at sites 3 and 4. Perch was absent at site 2. In total, metal and MT levels in the liver could be measured in 45 roach, 23 perch and 33 gudgeons captured in the pollution gradient. Figure 2 shows the results of the Cd, Cu and Zn concentrations in the liver of the three fish species caught at the different sampling. The average wet weight/dry weight ratios for the liver were 3.63 6 0.39, 3.85 6 0.31, 3.05 6 0.60 for Roach, Perch andCondition indicesThe relative condition factor (K) was calculated according to the formula K = W/W’ [40,41], where the observed weight of an individual fish (W in g) is compared to its expected weight (W’) based on its observed length (L). This expected weight is derived from the length-weight regressions W = a Lb, with a and b as fitting c.
