Contaminant levels in the European eel (Anguilla anguilla) in North Rhine-Westphalian rivers

PCDDs, PCDFs and indicator PCBs

Mean concentrations for the sum of PCDDs and PCDFs are generally below the EQS of Directive 2013/39/EU of 3.5 pg WHO₂₀₀₅ PCDD/PCDF TEQ/g. Exceptions are the values for the Rhine sites Rees and Emmerich where mean concentrations reached 4.7 and 5.4 pg WHO₂₀₀₅ PCDD/PCDF TEQ/g, respectively (significantly different from all other sampling sites). In contrast, mean concentrations of the sum of PCDD/PCDF + dl-PCB were nearly always above the EQS of 6.5 pg WHO₂₀₀₅ PCDD/PCDF +dl-PCB TEQ/g and also above the maximum level of 10 pg WHO₂₀₀₅ PCDD/PCDF + dl-PCB TEQ/g in EU Regulation 1259/2011. Highest mean TEQ concentrations were determined for eels from Emmerich (44.7 pg WHO₂₀₀₅ PCDD/PCDF+dl-PCB TEQ/g), Rees (39.4 pg WHO₂₀₀₅ PCDD/PCDF+dl-PCB TEQ/g), the Rur (39.6 pg PCDD/PCDF+dl-PCB WHO₂₀₀₅ TEQ/g) and the Ruhr (32.2 pg WHO₂₀₀₅ PCDD/PCDF+dl-PCB TEQ/g). Only eels from the Sieg had a mean concentration below the EQS (6.3 pg WHO₂₀₀₅ PCDD/PCDF+dl-PCB TEQ/g). These low concentrations cannot be attributed to the relatively low lipid contents of the Sieg eels as lipid normalised values are still low. The Rhine sites Rees and Emmerich and the Rur differed significantly from most other sites, but not from the Wupper, Ruhr, Niers and Lippe.

Similar TEQ concentrations were reported from Belgian waters [5] and from the Elbe in Germany [19]. Dioxin and dl-PCB levels were lower in the Loire (mean concentration 4.4 pg WHO₂₀₀₅ PCDD/PCDF + dl-PCB TEQ/g, [7]) and in Irish rivers (0.49 to 4.9 pg WHO₁₉₉₈ PCDD/PCDF + dl-PCB TEQ/g [27]).

TEQ concentrations for dl-PCBs were 8 to 16 times higher than TEQ concentrations for PCDD/PCDFs. The highest ratio was found for the Wupper with a mean value of 18.5. In eels from European waters, dl-PCB contamination is usually of higher importance than PCDD/PCDF contamination. High concentrations of dl-PCBs were e.g. also found in eels from Flemish waters [5], from the catchment area of the Elbe [19] and the Tiber in Rome [28]. In contrast, McHugh et al. [27] reported dl-PCB concentrations between 0.17 and 1.24 pg WHO₁₉₉₈ dl-PCB TEQ from relatively clean Irish rivers and therefore ratios between dl-PCBs and PCDD/PCDF of 0.1 to 4.3. Knutzen et al. [29] found PCDD/PCDF concentrations between 5.0 and 22.9 pg WHO₁₉₉₈ PCDD/PCDF TEQ and dl-PCB concentrations between 1.4 and 3.9 pg WHO₁₉₉₈ dl-PCB TEQ in Norvegian fjords which had been contaminated by discharges of waste water from a magnesium factory in the past.

All values recorded were well above the value of 4 pg PCDD/PCDF + dl-PCB WHO₁₉₉₈ TEQ/g which was determined by Palstra et al. [30] as a threshold value for adverse effects on the reproductive system of eels.

For indicator PCBs, the mean concentrations exceeded the maximum level for human consumption of 300 ng/g at most sampling sites. Exceptions were the eels from the Schwalm and the Sieg with 215 and 165 ng/g, respectively. Values in the Rhine eels ranged from 253 ng/g in Bonn to 982 ng/g in Emmerich. They are well below the concentrations found in Rhine eels in 1995 and 1998/2000 by Heinisch et al. [31]. All results in their paper are presented only as graphs and therefore approximations. Values for eels in Bonn were over 500 ng/g in 1995 and even over 600 ng/g in 1998/2000. Concentrations in Emmerich were about 1,500 in 1995 and 1,200 ng/g in 1998/2000, respectively. This comparison with our dataset documents a decrease of PCB pollution reflecting the general reduction of PCBs present in the atmosphere as well as a lower availability of PCBs in the Rhine after changes in production processes in major industrial sites in North Rhine-Westphalia.

By far the highest PCB concentrations were determined for eels from the Rur (mean concentrations of 1,630 ng/g). The catchment area of the Rur has a mining history. According to Detzel et al. [32], hydraulic fuel used in the mining industry has been a major source for PCB release into the environment. Additionally, PCB concentrations in the Rur were influenced by the emissions of paper industry upstream the sampling site. The high PCB concentrations in eels from the Rur were also described by de Boer et al. [33] who investigated the contamination of Dutch eels over a period of 30 years. For the Dutch part of the Rur, they reported concentrations of the sum of indicator PCBs + PCB 118, which decreased between 1982 and 2006 from 44,372 to 7,087 ng/g lw and therefore considerably more than in the Rhine (mean concentrations of 1,630 ng/g ww for Rur eels from this study correspond to 6,668 ng/g lw). Generally, PCB concentrations in eels from North Rhine-Westphalian waters were in the range of values found in other European waters impacted by industry and dense human population. Tapie et al. [20] determined the sum of indicator PCBs + PCB 118 in eels from the Gironde values ranging between 1,141 and 5,746 ng/g lw. High concentrations were as well reported from eels in a Belgian canal which reached up to 8,700 ng/g ww [34] and from Flemish waters, where eels reach concentrations between 78 and 4,812 ng/g lw [35].

Concentrations of PCDD/PCDFs, indicator PCBs and dl-PCBs were highly positively correlated with lipid content (p < 0.001) and also with age (PCDD/PCDFs p = 0.02, indicator PCB p = 0.004, dl-PCBs p = 0.003). But while PCDD/PCDFs and dl-PCB were positively correlated with length (p < 0.001 and p = 0.001, respectively), indicator PCBs were not. Similarly, indicator PCBs were not correlated with weight (PCDD/PCDFs p = 0.012, dl-PCBs p = 0.003).

Dioxins, PCBs and other contaminants with endocrine disrupting properties may contribute to the decline of the eel populations [5]. Several studies tried to relate the contaminant burden of eels to enzymatic activity and transcriptomic responses. Pujolar et al. [21] found a correlation for the up-regulation of the transcription of genes associated with detoxification and a down-regulation of the transcription of genes associated with the oxidative phosphorylation pathway in eels with a PCB burden comparable to many North Rhine-Westphalian sites. Kammann et al. [36] investigated the ethoxyresorufin-O-deethylase (EROD) activity in eels from the Elbe and found that the EROD activity corresponded to PAH metabolite concentration in eel bile and to pollution levels from the sampling sites.

HCB

Mean HCB concentrations exceeded the EQS of 10 μg/kg ww at most sampling sites. At the Sieg, all eels met the EQS (mean value 3.4 μg/kg) and at the Berkel and Schwalm, more than 50% of all eels sampled complied with the EQS (mean values 6.2 and 8.8 μg/kg, respectively). Along the Rhine, mean HCB concentrations increased from 23.2 μg/kg at Bonn to 50.2 μg/kg at Emmerich (27.8 μg/kg at Düsseldorf and 35.7 μg/kg at Rees). Differences between Bonn and Emmerich were significant. Also, HCB concentrations differed significantly between the Rhine sites and the other rivers. The increasing HCB concentrations in eels along the Rhine can at least be partly attributed to former production processes of the chemical industry based in North Rhine-Westphalia. Although HCB concentrations were comparatively high in the Rhine, eels sampled in 2009 demonstrate a decrease of HCB over time which had already been detected in data compiled by Heinisch et al. [37] on HCB concentrations in Rhine eels from 1990/1991, 1995 and 1998/2000. As mentioned above, results in their paper are presented only as graphs and therefore approximations. They show a decrease from approximately 180 μg HCB/kg in 1990 to 125 μg/kg in 2000 for eels from the Rhine site Emmerich. In our study, HCB concentrations in eels from this site were further reduced by about 60% to 50 μg/kg. The reduction of the HCB contaminant levels in eels is also due to the fact that production of HCB in the chemical industry in the upper part of the Rhine catchment area ceased in the 1990s and that the use of HCB as a fungicide was banned in Germany in 1981.

Comparable values to the HCB concentrations in the Rhine or even higher concentrations have been determined in eels from Dutch waters. De Boer et al. [33] reported mean HCB concentrations of 22 to 243 μg/kg lw. The latter value originated from eels from the Dutch part of the Rhine from the year 2004. The corresponding mean concentrations from this study for Emmerich, south of the Dutch border, of 178 μg/kg lw and for Rees, further south of the border, 155 μg/kg lw are in good agreement with their results. At other European waters, HCB concentrations in eels were far lower, e.g. in Italian lagoons (1 to 4 μg/kg lw [6]) or Scottish waters where HCB levels often were below the quantification limit [23].

As expected from its lipophilic nature, HCB was positively correlated with lipid content (p < 0.001). It was also positively correlated with length (p = 0.006), but not with the other biometric parameters.

Sum of PBDE

The PBDE concentrations always exceeded the EQS of 0.0085 μg/kg ww for the substances established by EU Directive 2013/39/EU by a factor of 10³ to 10⁵ (range of mean concentrations 9.2 to 241.9 μg/kg). Highest concentrations were found in eels from the Lippe, where an industrial complex with mainly chemical industry has been established for a long time. Means for the sampling sites Lippe (241.9 μg/kg), Rees (74.6 μg/kg) and Emmerich (76.2 μg/kg) - which are the Rhine sampling sites downstream of the confluence of the Lippe - differed significantly from all other sampling sites. Sühring et al. [22] investigated BDE-47, BED-66, BDE-99, BDE-100, BDE-153, BDE-154 and BDE-183 in yellow and silver eels from the Elbe and in silver eels from the upper part of the Rhine and found mean values of 8.9 and 8.3 μg/kg for yellow eels and silver eels of the Elbe and 21.3 μg/kg for the Rhine. PBDE concentrations in the range of 10 to 100 μg/kg ww (sum of BDE-28, BDE-47, BDE-49, BDE-66, BDE-85, BDE-99, BDE-100, BDE-153, BDE-154, BDE-183 and BDE-209) have been reported from a number of Belgian waters [38], with a maximum value of 5,811 μg/kg lw (for comparison, Lippe 242 μg/kg ww =1,242 μg/kg lw). In eels from 60 Flemish sites sums for 14 BDE congeners (mainly congeners 47, 99 and 100) ranged between 12 and 1,400 μg/kg with a mean of 110 μg/kg [35]. PBDE concentrations from Irish waters were lower and ranged from 1.0 to 7.1 μg/kg ww for the sum of 11 PBDE congeners [27].

In this study, the sum of 6 PBDEs was significantly positively correlated with the lipid content (p = 0.006) and weakly negatively correlated with the condition factor (p = 0.05) but not with any other of the biometric parameters.

There is still few evidence for possible effects of PBDEs at environmental concentrations, but Tomy et al. [39] found evidence for lower plasma levels of thyroxine in Lake Trout with BDE concentrations similar to those found in eel from the Lippe and the Rhine sites Rees and Emmerich. Kierkegaard et al. [40] investigated the effects of BDE-209 on Rainbow Trout over a period of 120 days. They detected significantly reduced numbers of leucocytes and lymphocytes and an increase of liver weight. But they used high doses resulting in a 5 to 25 times higher body burden than the sum of PBDEs measured in the eels from North Rhine-Westphalia.

PFOS

Also, the PFOS concentrations often exceeded the EQS of 9.1 μg/kg ww with mean concentrations between 8.3 (Berkel) and 49.0 μg/kg (Niers). High concentrations were also found in the Rur (42.3 μg/kg). Mean values from the Niers and the Rur differed significantly from most other rivers but not from the Rhine eels and those from the Wupper. Only five eels were collected from the Niers, so the mean value for this site should be interpreted with caution. Nevertheless, the high PFOS concentration could be taken as an indication of the high proportion of municipal waste water present in the Niers. In North Rhine-Westphalia, the main source for PFOS is the plating industry. Until 2008, PFOS was widely used as fire extinguishing agent [41]. PFOS is also often present in consumer goods. As mentioned above, the Rur has been affected by paper industry emissions. PFOS has been used in the production process of special papers which explains the high PFOS concentrations in eels from the Rur. For eels, only data on PFOS contents in the liver have been published which are of limited value for the interpretation of tissue concentrations. Roland et al. [15] determined mean liver concentrations between 31 and 390 μg/kg from eels collected from Belgian waters. Hoff et al. [16] measured mean hepatic PFOS concentrations of 1,387 μg/kg ww in eels from four heavily polluted Belgian water bodies and related them to the high concentrations to fluorochemical production units in the vicinity of the sampling sites. In this study, they found a strongly significant correlation between liver PFOS concentrations and serum alanine aminotransferase activity which they interpreted as indication for induction of hepatic damage. In North Rhine-Westphalia, a survey on PFOS in the muscle tissues of several fish species, with eel being one of them, was conducted [41]. The median concentration of all fishes from surveillance monitoring was 7.1 μg/kg, and the median concentration from impacted sites was 24.4 μg/kg. Hloušková et al. [26] investigated PFT levels in a mixed sample consisting mainly of bream (Abramis brama), European chub (Squalius cephalus) and Roach (Rutilis rutilus) and found a median concentration of 5.7 μg/kg. The results of both studies indicate that the PFOS concentrations measured in eels during this study are comparatively high.

In this study, PFOS did not show any significant correlation with any of the biometric parameters.

Mercury

For mercury concentrations, in accordance with results from other European waters, all eels from North Rhine-Westphalia had concentrations below the EU limit for mercury in fish for human consumption (1.0 mg/kg ww) but above the EQS of 20 μg/kg ww for secondary poisoning. Highest concentrations were measured in eels from the Wupper (mean concentration 0.314 mg/kg) and the Rhine (Rees 0.273 mg/kg, Emmerich 0.195 mg/kg, Bonn 0.210 mg/kg and Düsseldorf 0.165 mg/kg) and lowest concentrations for eels from the Sieg (0.069 mg/kg) The mean concentrations from the Wupper eels and from the Rhine eels at Rees differed significantly from most other sampling sites but not from each other. Eels in two Spanish rivers had comparable mean mercury concentrations between 0.155 and 0.533 mg/kg [42], and in Portuguese lagoons, eels contained between 0.55 and 0.285 mg Hg/kg [43]. Noël et al. [25] reviewed further data from the studies in other countries including Poland, the Czech Republic, Germany, UK, Slovenia and Hungary and reported a range of values between 10 and 800 μg/kg. In their study of French rivers, they found mean concentrations of 0.199 mg/kg and reported a strong positive correlation between Hg concentration and the condition factor (p < 0.0001) but no correlation between Hg levels and body weight or length. In this study, mercury concentrations were significantly positively correlated with age (p = 0.029). Also, there was a weak positive correlation with length (p = 0.049) and weight (p = 0.055) but not with the condition factor. After a review of experimental data on Hg effects on fish, Sandheinrich and Wiener [44] estimated that the threshold value for negative effects of Hg on fish is between 0.3 and 0.7 mg Hg/kg in the whole body homogenates which are at least about a quarter lower than the concentrations in the filet. Therefore, a deleterious effect of the Hg concentrations measured in the eels from this study is unlikely.

PCB congener patterns

Based on the eel data for PCB contamination, the Rhine sites Bonn and Emmerich and the Rur were chosen for a comparison between PCB congener patterns in eel and in the suspended matter from corresponding sampling sites. In the Rhine and the Rur, PCB 153 comprised 40% to 43% of the sum of indicator PCBs in eel, PCB 138 accounted for 23% to 28% and PCB 180 for 12% to 16%. PCB 101 and PCB 52 had a slightly higher proportion in eels from Emmerich (16% and 8%, respectively) compared to eels from Bonn and from the Rur (7% to 9% and 3% to 5%). PCB 28 did not exceed 1% (Figure 2). The pattern of PCBs in suspended matter varied slightly with PCB 138 being the most dominant congener in the Rhine while in the Rur, PCB 153 had a slightly higher percentage than PCB 138. PCB 28 had mean percentages of 8% to 9% in suspended matter and therefore higher concentrations than in eels. According to Tapie et al. [20], PCB 153 is not regarded as being metabolisable and therefore an indicator for bioaccumulation. Müller et al. [45] examined PCB concentrations in eels and in sediment samples in the urban water bodies of the city of Berlin, Germany. They found similar PCB patterns in eels and in the sediment with PCB 138 and 153 dominating in eels and in the sediment while PCB 28 was more prominent in sediment samples compared to eel tissue. On the whole, PCB patterns in suspended matter and in biota found in this study were similar to PCB patterns in eels recorded from many European waters and also from North American water bodies reflecting the commercial PCB mixtures used in the past. Examples are data for eels from the Loire [7], from Scottish water bodies [23] and also for the American eel Anguilla rostrata from the Hudson River [46].

The bioconcentration ratio between contaminant content of suspended matter and eel tissue was calculated between PCB concentration of suspended matter expressed as dry weight and PCB concentration of eel expressed as dry weight and as lipid weight. The dry weight ratio for the Rhine at Bonn was 39 (76 for lw), at Emmerich 66 (116) and in the Rur 54 (92). Harrad and Smith [47] reported concentration ratios between sediment (normalised for the carbon content of the sediment) and eel tissue for individual PCB congeners in the range of <1 and 10. The ratios found in this study are far higher which might be explained by the 10 to 10² higher PCB concentrations in North Rhine-Westphalian rivers compared to the River Severn investigated by Harrad and Smith [47]. As mentioned above, Müller et al. [45] also determined PCB concentrations in eels and in sediment samples from corresponding sampling sites. They expressed all eel data per wet weight (ww) and found a ratio of approximately 5. The water bodies examined had comparable PCB levels to the suspended matter of the Rur. In this study, the dry weight of the eels investigated was approximately 30% of the wet weight. Based on this estimate, eels from Berlin waters had an eel to sediment ratio of approximately 15 which is still below the values found in this study.

Considering the 12 dl-PCB congeners, PCB 118 was clearly dominating with 50% to 67% of the sum of dl-PCBs. PCB 105, 156 and 167 were present with 5% to 15%. All other PCB congeners were present in negligible concentrations (Figure 3). This congener pattern is quite common in eels from European waters and has also been reported, e. g. from Belgium [5] and Portugal [48].

PBDE congener patterns

PBDE patterns in eel and in suspended matter were compared for the Rhine sites Düsseldorf and Emmerich and the Lippe (Figure 4). Lippe eels had very high PBDE concentrations, and as a consequence, eels at the Rhine sites below the confluence of the Lippe had an elevated PBDE content. Therefore, the sampling sites at Düsseldorf upstream of the confluence of the Lippe and at Emmerich downstream of the Lippe were chosen to track a possible influence of the Lippe contamination on the Rhine. The comparison was limited to BDE-28, BDE-47, BDE-99, BDE-100, BDE-153, BDE-154 and BDE-209 as these were the only BDE congeners determined in suspended matter.

At all sites, BDE-47 was the dominant congener with 60% to 62% at the two sites at the Rhine and 74% in the Lippe. BDE-100 was present with 28% to 30% in the Rhine and 17% in the Lippe. The other congeners were present in minor concentrations. A similar congener pattern with BDE-47 dominating and BDE-100 as the second most important congener was reported from the eel populations from the Scheldt [38], from the Elbe [49] and by Sühring et al. [50] for eel populations from several European rivers. In contrast, Sühring et al. [50] found equal concentrations of BDE-47 and BDE-100 in yellow eels from North America. They related this difference to the continuous exposition of American eels to technical penta- and octa-PBDE mixtures which had been banned in Europe in 2004. In contrast to the biota PBDE pattern, the Rhine suspended matter comprised almost exclusively of BDE-209 (95% to 97%), reflecting the continuous use of deca-BDE as fire protection agent in consumer goods. In the suspended matter of the Lippe, BDE-209 accounted for only 25%, and BDE-99 and BDE-47 were present with 35% and 24%, respectively. This conspicuous PBDE pattern, which can be related to the influence of local chemical industry, did not lead to a change in the PBDE pattern in eels. Roosens et al. [51] investigated PBDE patterns in eels and in the sediment from various locations in Flemish water bodies. They also found high BDE-209 concentrations in the sediment and only rarely found elevated BDE-209 concentrations in eels. They attributed this discrepancy to a poor uptake of BDE-209 from the sediment, a low solubility in water and a high degradation in the fish. The debromination of BDE-209 to lower brominated congeners has been described by several authors, e. g. [40],[52]-[54] and seems to be a major source of penta-PBDEs in fish.

Ratios between PBDE concentrations in eel tissue and in suspended matter ranged between 0.6 (Düsseldorf) and 2.6 (Lippe) for eel dry weight and 1.5 (Düsseldorf) and 4.9 (Lippe) for eel lipid weight. These ratios are considerably lower than the ratios calculated for PCBs despite the fact that the sum of PBDEs was significantly positively correlated with the lipid content. The low ratios can be attributed to the poor transfer of BDE-209 congener from the suspended matter/sediment to the eels. This is in agreement with the higher eel to suspended matter ratios in the Lippe. According to Mariottini et al. [55], BDE-47 is considered to have the highest bioconcentration factor of PBDE congeners. In the Lippe suspended matter, BDE-47 was present in approximately the same concentrations as BDE-209.

Figure 5 shows the pattern of BDE-28, BDE-47, BDE-49, BDE-66, BDE-71, BDE-99, BDE-100, BDE-153, BDE-154 and BDE-209 in eel tissue for all sampling sites. The pattern resembles the profile depicted in Figure 4 although additional BDE congeners were included in the calculation. At all sites, BDE-47 dominated the BDE pattern, with BDE-100 being the second most important congener. Together, these two congeners accounted for 85% to 94% of the BDEs measured.

Sampling of eels

In order to set up eel management plans in 2009 and 2010, the North Rhine-Westphalian government commissioned the Rhineland Fishery Association to carry out a survey of eel populations in ten major rivers. Eel populations were monitored using point abundance sampling with electrofishing equipment from a boat (electrofishing equipment EFKO FEG 8000, Leutkirch, Germany 8 KW, up to 600 V, direct current, 60 points with a distance of 3 m between points). The length of each eel caught was measured before the eels were released again. As part of the survey, a subsample of 124 eels was collected from 13 sites in 10 rivers (Sieg, Wupper, Ruhr, Lippe, Erft, Rur, Niers, Schwalm, Berkel and four sites - Bonn, Düsseldorf, Rees and Emmerich - at the Rhine, Figure 1) for biometrical measurements and for inspection of the health status. Sites were generally situated in the downstream regions of the rivers, apart from the sites at the Rhine which covered the North Rhine-Westphalian stretch of the river. Usually, ten to eleven eels were taken, but as there were only small numbers of eels at the sampling sites at the Niers, the Schwalm and the Berkel five eels were taken at the Niers and eight eels at the Schwalm and the Berkel. The eels collected roughly represented the size-class distribution of the population at the sampling site. The eels were killed immediately by over narcotisation with 300 mg Benzocain/l (Merck, Darmstadt, Germany), and transported frozen to the lab of the North Rhine-Westphalian State Agency for Nature, Environment and Consumer Protection (LANUV NRW). The sampling and killing was carried out by a licensed fish biologist. In the lab, the length, weight and sex were determined, and the condition factor was calculated (condition factor C _F = weight (g) × 100/length (cm)³). Heads were dissected and sent to the Thünen Institute of Fisheries Ecology, Hamburg, Germany, where the age was determined using otolith preparation following ICES [57]. The developmental stage was determined using the eye index according to Durif et al. [58]. The eels were with very few exceptions in stages 1 to 3. Eels from the Schwalm and from the lower Rhine were predominantly in stage 3 which is according to Durif et al. [58], the stage just before the onset of metamorphosis. At all other sampling sites, stages 1 to 3 were present in about equal numbers. It was decided to take tissue samples of the yellow eels for the analysis of contaminants. Five eels which were judged to be silver eels according to their eye index were excluded from further analysis (one each from the Ruhr, the Erft and the Rhine sites Bonn, Düsseldorf and Rees). Therefore, 119 eels altogether were chemically analysed. The left filet of each specimen was sampled, including the skin (at least 100 g; from a large specimen, the middle part of the filet was chosen), and stored frozen.

Analysis of eels

In 2011, the tissue samples were analysed for PCDDs, PCDFs, PCBs, PBDEs, HCB, mercury, perfluorooctanoic acid (PFOA) and PFOS by Eurofins GfA Lab Service (Hamburg, Germany). Prior to further treatment for analysis of organic contaminants, eel samples were lyophilized and homogenised. For dry matter determination, lyophilised subsamples were subjected to a drying procedure at 103°C for at least 4 h. Dry matter content was calculated taking into account the dry matter content after lyophilisation as well as the dry matter after the drying procedure.

PCDD/PCDFs, dl-PCBs and indicator PCBs

In brief, samples were analysed for 17 2,3,7,8-substituted congeners of PCDDs and (PCDFs, 12 dl-PCBs and the 6 indicator PCBs 28, 52, 101, 138, 153 and 180 using high-resolution gas chromatography and high-resolution mass spectrometry (HRGC/HRMS) analogue to a method described by Neugebauer et al. [59]. For each native PCDD/PCDF and PCB congener to be determined, the corresponding ¹³C₁₂- internal standard was added to the samples before the extraction procedure. After spiking, the samples were Soxhlet-extracted overnight with appropriate organic solvents for ultratrace-analyses (e.g. toluene). Lipid determination was performed gravimetrically in these crude extracts as extractable lipids. Subsequently, these extracts were subjected to a cleanup procedure using a multi-column chromatography system (involving carbon-on-glass fibre or carbon-on-celite for PCDD/PCDF and PCB). Further ¹³C₁₂-labelled internal standards were added to the extracts for the determination of the recovery of the internal standards added before. Analyses were performed by HRGC/HRMS (Waters Autospec, Eschborn, Germany or DFS mass spectrometers, Thermo Fisher Scientific, Bremen, Germany). For each substance, two isotope masses were measured. The quantification was carried out by means of isotope dilution analysis with the use of internal and external standards. In addition to values for individual congeners, calculations of the TEQ according to the WHO-system 2005 [60] were carried out by taking into account the quantification limit for non-quantifiable compounds (upperbound procedure).

PBDEs

Samples were analysed for 24 congeners of PBDEs (including BDE-17, BDE-28, BDE-47, BDE-49, BDE-66, BDE-71, BDE-77, BDE-85, BDE-99, BDE-100, BDE-119, BDE-126, BDE-138, BDE-153, BDE-154, BDE-156, BDE-183, BDE-184, BDE-191, BDE-196, BDE-197, BDE-206, BDE-207 and BDE-209) using gas chromatography and mass spectrometry (GC/MS) analogue to a method by Päpke et al. [61].

A mixture of ¹³C₁₂-labelled internal standards was added to the samples before the extraction procedure. After spiking, the samples were extracted by means of Soxhlet extraction using a mixture of appropriate polar and non-polar solvents for ultratrace-analyses (e.g. hexane/acetone). After extraction, a cleanup procedure was performed using concentrated sulfuric acid additionally to column chromatography involving activated silica gel resp. alumina. Further ¹³C₁₂-labelled internal standards were added to the extracts for the determination of the recovery of the internal standards added before. PBDEs were analysed by means of GC/MS. For each substance, two isotope masses were measured. The quantification was carried out by means of isotope dilution analysis with the use of internal and external standards.

As many of the BDEs analysed were always or almost always below the limit of quantification, only the results for BDE-28, BDE-47, BDE-49, BDE-66, BDE-99, BDE-100, BDE-153, BDE-154, BDE-183 and BDE-209 are presented in this publication.

PFOS/PFOA

A mixture of ¹³C₁₂-labelled internal standards was added to the samples before the extraction procedure. After spiking, the samples were extracted by means of ultrasonic extraction with appropriate polar solvents (e.g. methanol) for ultratrace-analyses (e.g. nanograde). After extraction, the cleanup was carried out involving carbon black. A further ¹³C₁₂-labelled internal standard was added to the extracts for the determination of the recovery of the internal standards added before. The analysis was performed using liquid chromatography and tandem mass spectrometry (LC/MS-MS). For each substance, two isotope masses were measured. The quantification was carried out with the use of internal and external standards. As values for PFOA rarely exceeded the limit of quantification, only the results for PFOS are presented in this publication.

HCB

Before extraction, the eel samples were dried by thorough mixing with anhydrous sodium sulfate and homogenised. A mixture of ¹³C₁₂-labelled internal standards was added to the samples before the extraction procedure. After spiking, the samples were extracted with a mixture of appropriate polar and non-polar solvents for ultratrace-analyses (e.g. hexane/acetone). The extraction was followed by a cleanup procedure using a column system (involving florisil resp. alumina). Further ¹³C₁₂-labelled internal standards were added to the extract for the determination of the recovery of the internal standards added before. The measurement was taken by means of HRGC/HRMS. For each substance, two isotope masses were measured. The quantification was carried with the use of internal and external standards.

Mercury (Hg)

Eel homogenate was digested by means of microwave using nitric acid. After reduction of mercury compounds by Tin(II)-chloride, the total mercury content was determined by means of cold vapor atomic absorption spectrometry (CV-AAS) following DIN EN 13806.

Sampling of suspended matter

Sampling and analysis of suspended matter was carried out by LANUV NRW as part of a separate surveillance monitoring program.

Suspended matter was collected using a flow through pump (Carl Padberg Cepa Z 61, Lahr, Germany) which was deposited in the water approximately 0.5 m below the surface. The time necessary to collect sufficient material was estimated by measuring the turbidity of the water (WTW IQ Sensor Net, System 184 with sensor with VISOTurb 700/Q, Weilheim, Germany). Usually, the pump collected material over 21 to 24 h with a flow rate of 1,000 to 1,100 l/hr. The material was weighed in the field and transported refrigerated to the lab where it was freeze-dried, ground by a grinding mill equipped with a zirconium dioxide mortar and pestles (KM1, Retsch, Haan, Germany) and sieved for the 63-μm fraction.

Data for suspended matter were available as annual means of usually between two and five individual samples per site. As sampling sites were sampled in different years, values for suspended matter were calculated as means for all annual means available between 2005 and 2012 (3 to 8 values).

Analysis of suspended matter

Quality assurance

Statistical analysis

Statistical analysis was performed using the statistical software SAS 9.2 and SPSS 19. The Shapiro-Wilk test demonstrated that the data of the contaminants were not normally distributed. For differences between mean contaminant concentration (least squares means) of sampling sites, an analysis of covariance with age and lipid content as covariates and Tukey's test were performed (α =0.1). Significant differences between sampling sites were checked using the Kruskall-Wallis test. For correlations between biometric parameters and contaminants Pearson's correlation coefficients were used. As n >100, an approximately normal distribution was assumed and significance levels could be calculated. To test the influence of biometric parameters on contaminant concentration, an analysis of covariance was performed between each biometric parameter and the contaminants after eliminating the influence of the sampling sites.

To analyse the differences of mean contaminant concentrations in different parasite classes, an analysis of variance was performed followed by Tukey's test. To compare the parasite patterns of the sampling sites, a Monte Carlo simulation of Fisher's exact test was used. Logistic regressions between parasite infection and biometric parameters or pollutants were calculated. For all the tests, apart from Tukey's test on differences on mean contaminant concentration of sampling sites, a significance level of α =0.05 was set.