Levels of Polybrominated Diphenyl Ether (PBDE) Flame Retardants in Animals Representing Different Trophic Levels of the North Sea Food Web Environmental Science and Technology 1oct02

Jan P Boon,*,† Wilma E Lewis, #,† Michael R. Tjoen-A-Choy,† Colin R. Allchin, ‡ Robin J. Law,‡ Jacob de Boer, § Cato C. Tenhallers-Tjabbes, ,† and Bart N. Zegers†

Royal Netherlands Institute for Sea Research (NIOZ), P.O. Box 59, 1790 AB Den Burg, Texel, The Netherlands, Centre for Environment, Fisheries and Aquaculture Science (CEFAS), Remembrance Avenue, Burnham-on-Crouch, Essex CMO 8HA, UK, and Netherlands Institute for Fisheries Research (RIVO), P.O. Box 68, IJmuiden, The Netherlands

* Corresponding author phone: +31 222 369 466; fax: +31 222 319 674; e-mail: boon@nioz.nl. 
† Royal Netherlands Institute for Sea Research (NIOZ). 
‡ Centre for Environment, Fisheries and Aquaculture Science (CEFAS). 
§ Netherlands Institute for Fisheries Research (RIVO).
# Present address: TNO-MEP, Dept. for Ecological Risk Studies, P.O. Box 57, 1780 AB Den Helder, The Netherlands.

The levels of individual PBDE congeners were investigated in the invertebrate species whelk (Buccinum undatum), seastar (Asterias rubens), and hermit crab (Pagurus bernhardus), the gadoid fish species whiting (Merlangius merlangus) and cod (Gadus morhua), and the marine mammal species harbor seal (Phoca vitulina) and harbor porpoise (Phocoena phocoena). These species are all important representatives of different trophic levels of the North Sea food web. All six major PBDE congeners detected (BDEs 28, 47, 99, 100, 153, and 154) are most prevalent in the commercial Penta-BDE formulation. There is no evidence for the occurrence of the Octa-BDE formulation in the North Sea food web, since its dominant congener, BDE183, was never detected. BDE209, the main congener (>97%) in the Deca-BDE formulation, was detected only in a minority of the samples and always in concentrations around the limit of detection. Since BDE209 is often the major BDE congener in sediments from the area, the main reason for its low concentrations in biota from the North Sea seems to be a relatively low bioaccumulation potential. This can either be due to a low uptake rate of the very large molecule or a relatively rapid excretion after biotransformation. Since all invertebrates investigated are sentinel species, they are highly representative for the area of capture. The highest lipid-normalized concentrations of PBDEs in the invertebrates occurred near the mouth of the river Tees at the East coast of the UK. The geographical distribution of the PBDEs can be explained by the residual currents in the area. The direction of these currents differs between the summer and the winter season as a result of the presence or absence of vertical summer stratification of the deeper waters north of the Dogger Bank. Summer stratification results in the development of a density-driven bottom water current formed after the onset of vertical stratification of the water column in May leaving the UK coast near Flamborough Head toward the Dogger Bank. In winter, the residual currents run in a more southerly direction and follow the UK coastline. The distribution pattern of the PCBs and p,p'-DDE in the invertebrates was entirely different from that of the PBDEs, which could be expected, since the use of these organochlorines in western Europe peaked in the 1960s and 1970s but has been forbidden more than two decades ago, whereas the production and use of the penta-BDE formulation is of a more recent origin. The higher trophic levels of the North Sea food web were represented by the predatory gadoid fish species whiting and cod and the marine mammal species harbor seal and harbor porpoise. The lipid-normalized levels of the six major PBDE congeners in fish were similar to the levels in the invertebrates, but a biomagnification step in concentrations of generally more than an order of magnitude occurred from gadoid fish to marine mammals. Based on the limited number of samples, no differences could be observed between harbor seal and harbor porpoise. In summary, the results in three species of sentinel invertebrates from a network of stations covering a major part of the North Sea basin showed that the estuary of the river Tees at the UK East coast is a major source for tri- to hexa-PBDEs. Throughout the food-chain, the most marked increase in (lipid-normalized) levels of all six PBDE congeners occurred from predatory (gadoid) fish to marine mammals, agreeing with the transition from gill-breathing to lung-breathing animals. This has serious consequences for the route of elimination of POPs, since their elimination from the blood into the ambient seawater via the gill-membrane is no longer possible.

Introduction

Brominated flame retardants are added to many plastics and printed circuit boards of electronic household equipment and textile and polyurethane foam in furniture and cars for obvious safety reasons. Among the brominated flame retardants (BFRs), there are three commercial formulations that contain the diphenyl ether skeleton. In order of increasing overall bromination, these formulations are named Penta-BDE, Octa-BDE, and Deca-BDE. The global market demands in 1999 were 54 800 metric tons for the Deca-BDE, 8500 tons for the Penta-BDE, and 3825 tons for Octa-BDE (ref: www.bsef.com).

PBDEs are very hydrophobic (log K_ow values 4-10) and resistant to degradation. The water solubility and vapor pressure of PBDEs decrease with increasing degree of bromination, whereas hydrophobicity increases (1-3). Two major congeners of Penta-BDE, i.e., BDEs 47 and 99 (numbering is according to the IUPAC nomenclature for PCBs), showed even higher bioaccumulation factors than PCBs of similar hydrophobicity, despite a larger molecular size of the brominated compounds (4). The presence of these congeners and BDE100 in tissues of the deep-sea foraging sperm whales indicates that these PBDEs have become globally dispersed chemicals (5). The temporal trends of the major congeners of the penta-BDE formulation still showed a sharp increase between 1978 and 1998 in lake trout (Salvelinus namaycush) from the Great Lakes (6). The levels in Beluga whale (Delphinapterus leucas) from the St. Lawrence estuary collected between 1997 and 1999 were on average 20 times higher than the levels in animals collected a decade earlier (7), irrespective of sex. Mixed temporal trends were observed in sediments of several UK rivers entering the North Sea or the Irish Sea (8). In the Rhine-Meuse delta, the levels in yellow eel (Anguilla anguilla) from the "Haringvliet" decreased since 1984, and those in eel from the German/ Dutch river Roer, a tributary of the Meuse, since 1993 (8). In Japan, the levels in sea bass and grey mullet from Osaka bay have been rapidly decreasing since 1990 (9). Less is known about the situation for the other two technical PBDE formulations, i.e. the Octa- and the Deca-BDE formulation. Octa-BDE was reported to be absent in freshwater fish from Virginia (10). BDE209 constitutes > 95% of the commercial deca-BDE formulation. The only occasion where this congener and BDE 183 have been reported in relatively high concentrations in wildlife concerns eggs of pererine falcons breeding in Sweden (11). The peregrine falcon (Falco peregrinus) belongs to the top-predators of the terrestrial food-web, since this bird of prey feeds mainly on other birds. The uptake of BDEs 183 and 209 in blood of humans working in a recycling factory of electronic household equipment has also been demonstrated (12). Thus, the assumption that BDE209 cannot bioaccumulate due to its large molecular size is not always correct. This is an important observation, since the concentrations of BDE209 are often much higher than those of the other PBDE congeners in suspended particulate matter and sediments of several continental European, UK, and Irish rivers (8).

Although data on PBDE levels in the estuaries of several rivers entering the North Sea have been published (8), a survey of a number of PBDE congeners representative for the different commercial formulations in a number of animal species representing different trophic levels of the North Sea food web has not yet been carried out. For this purpose, the invertebrate species seastar (Asterias rubens), common whelk (Buccinum undatum), and hermit crab (Pagurus bernhardus) and the predatory gadoid fish species whiting (Meralangius merlangus) and cod ((Gadus morhua) were sampled between 52 and 58° N and 1°W-10°E, covering a large part of the North Sea basin and the Skagerrak, the connection between the North Sea and the Baltic Sea. The highest trophic levels were represented by harbor seals (Phoca vitulina) from the German Wadden Sea and harbor porpoises (Phocoena phocoena) stranded or by-caught along the Dutch coast. Since the three invertebrate species are sentinel, the geographical distribution of their PBDE levels can give an impression of the location of important input sources into the North Sea basin. For comparison with the PBDEs, the levels of the major PCB congeners CB101, 138/158, and 153 and p,p'-DDE were also analyzed.

Experimental Section

Sampling of Invertebrates and Fish. The majority of the samples were taken with a 5 m beam trawl with a mesh size of 3.5 cm n during cruise 64PE144 with the RV Pelagia in August-September 1999. The basic data on the position, sampling date, water depth, and temperature, and salinity at the surface and just above the sea-bottom, are given as Supporting Information in Table SI-1. Large differences in temperature and/or salinity between surface and bottom water are indicative for the occurrence of vertical stratification. The presence of a thermocline or a halocline severely inhibits the exchange rates between the sea surface and the sea bottom. The herring samples were caught with a pelagic net at 51°34' N and 2°47' E in the Southern Bight by the commercial fishing vessel TX 37 in October 2000.

Because it is very hard to obtain a good homogenate from entire animals for analysis of POPs, a number of tissues were selected. The following tissues were excised on board immediately after capture and frozen at -20 °C until further analysis: for the sea star the pyloric caeca (part of the intestinal system), for the hermit crab the soft abdomen; for the whelk the whole (soft) body; and for fish the liver and filet from one side of the backbone. No differentiation was made between the sexes except for the whelk. In this case only the males were taken because the females were used for research on levels and effects of organotin compounds. Samples of invertebrates and fish each contained material from five individual animals. From each station, two such samples were analyzed.

Samples of Marine Mammals. All samples of marine mammals came from beach-stranded animals or animals drowned in fishing nets. The samples of harbor porpoise originated from the southern North Sea and were obtained from Dr. Chris Smeenk and Drs. Marjan Addink of the Museum of Natural History "Naturalis" in Leiden, The Netherlands. The samples of harbor seals were obtained from Dr. Ursula Siebert of the Centre for Research and Technology in Büsum, Germany, and originated from Wadden Sea of Schleswig-Holstein (Germany). For cetaceans and harbor seals, samples of liver and blubber of individual animals were stored for analysis.

Extraction Method for the Determination of POPs. Tissue amounts corresponding to approximately 50 mg of lipid were extracted, using an ultra-Turrax method. After extraction with pentane and acetone, the sample was treated with sulfuric acid, and a silica cleanup was performed. The method has been described in more detail elsewhere as the "NIOZ method" (13).

Analysis of PBDEs. The levels of 15 individual PBDEs (Cambridge Isotope Laboratories, Inc. Andover, MA) were determined by GC/MS. The GC was a Hewlett-Packard 6890; the mass-selective detector a Hewlett-Packard 5973. GC specifications: split-splitless injection, split valve closed for 1.5 min. T_injector 270 °C. Column: stationary phase CP Sil-8, 25 m * 0.25 mm * 0.25 µm (Chrompack, NL). Carrier gas He; linear gas velocity 74 cm n s^-1, constant flow programmed. Oven temperature program: 90 °C (1.5')/20 °C min^-1/190 °C (0')/4.5 °C min^-1/270 °C ((5')/10 °C min^-1/320 °C ((10'). MSD specifications: negative chemical ionization (NCI) in the SIM mode at the m/z ratios of both bromine isotopes (79 and 81) and m/z ) 487 (for BDE 209 only). Ionization gas CH₄. T_{ion source} 210 °C; T_transferline 320 °C; T_quadrupole 160 °C.

The limit of detection (LOD), defined as a signal of three times the noise level, was established at 0.6 ng g-1 lipid for all BDE congeners except BDE209, when an amount of wet tissue containing 50 mg of lipid was extracted. Other amounts of lipid affect this LOD value proportionally. The LOD for BDE209 was 5 ng g-1 lipid.

Chromatographic Conditions for the Analysis of PCBs and p,p'-DDE. The levels of individual CBs and p,p'-DDE were determined by gas chromatography on a Carlo Erba 5300 (Italy) with electron capture detection (ECD). The specifications of the GC were as follows: 1 µL split-splitless injection, split valve closed for 4 min, _Tinjector ) 300 °C. Column: stationary phase CP Sil-8, 50 m * 0.25 mm * 0.25 µm (Chrompack, NL). Carrier gas H2; constant pressure. Oven temperature program: :90 °C ((2')/10 °C min^-1/215 °C ((10')/8 °C min^-1/275 °C ((17'). . The ECD was 300 °C and 90 mL N2 min^-1 was used as make up gas.

Quality Assurance. Before sample treatment, the internal standards CB112 and decabromobiphenyl (BB209) were added. Although BB209 has been produced in France until recently, no indication of its occurrence has been found in the present set of samples, which were run simultaneously without the addition of this internal standard. Moreover, PBBs were also not detected in any sample from a large set from Dutch coastal waters either (14). Our participation in the first interlaboratory study, which has been organized for PBDEs, showed a good performance for the samples of biota and sediment. Until now, no certified reference materials are available for the determination of PBDEs.

Extraction Method for the Determination of Tissue Lipid Contents. The levels of all POPs have been expressed on the basis of extractable lipids extracted from the same tissue as used for the POP analyses, since the POPs of interests are all very hydrophobic and thus they prefer to accumulate in fatty tissues (15, 16). However, the extraction method used for the determination of the different POPs, extracts especially the polar phospholipids incompletely. Therefore a separate extraction of the tissues was performed for the determination its total lipid content, using a dichloromethane/methanol/water solvent system (17). Especially in lean tissues, this method produced substantially higher values than the acetone/pentane/water solvent system used for the extraction of the POPs. For comparison, a table with data of the lipid contents as obtained with both methods is given as Supporting Information in Table SI-2.

Multivariate Statistical Analysis. The data of the six major PBDEs, three major PCBs, and p,p'-DDE were subjected to Principal Component Analysis (PCA) to find the underlying pattern and correlation structure of the data set. PCA was performed on the correlation matrix of the lipid-based concentrations. The data were 10log-transformed prior to the calculations to obtain a greater homogeneity of variance. The results have been graphically displayed in the form of a principal components bi-plot. From such figures, several things can be read.

A: The Position of Each Sample in the Plane. Each sample can be conceived as a point in a k-dimensional space, where k is the number of compounds taken into account. The first two principal components (PCs 1 and 2) span the plane on which the projected points show the highest variance. The position of each projected point (= individual sample) on this plane can be visualized in the form of a so-called covariance biplot (18). The x and y coordinates of each data point represent the scores for the first and second PC, respectively. The coordinates of each data point can be read in the bi-plot from the lower x-axis for the value of the first PC and the left y-axis for the value of the second PC.

B: The Vectors. The biplot also shows the values of the correlation (r) between each original-in this case the logarithm concentration of each compound^-and the first and second principal components by means of a vector. The x and y coordinates of the endpoint of each vector represent the loading. The squared length of each vector shows the goodness of fit and is equivalent to the fraction of the total variance that is explained by the sum of the first and second PC (In formula: (l_vector)²= R²). Thus, if a vector of a compound reaches the drawn circle of unit variance (circle R²=1), then all variance in the concentration is explained by this sum. The values for the individual correlations with the PCs plotted can be read by orthogonal projection of the endpoint of each vector on the upper X-axis in the plot for the first PC and on the right y-axis for the second second PC. The correlation r between the concentrations of two different compounds is equal to the product of the length of their vectors and the cosine of the angle R between them. Thus, the concentrations of two compounds of which the vectors touch the unit-circle (i.e. all variance can be explained by the sum of the first two PCs), and point in the almost the same direction (cosine a approaches 1), show a very high positive correlation. In contrast, orthogonal vectors indicate a zero correlation (cosine 90°= 0); a negative correlation is indicated by angles between 90 and 180°.

C: The Relation between the Location of the Datapoints and the Vectors. The orthogonal projection of each datapoint (= individual sample) on each vector--or its extension on the other side of the center of the plot--shows the relative concentration of the accompanying compound in the sample with respect to the mean value, which is represented by the center of the plot.

Results and Discussion

PBDE Congeners Detected. Of the 16 congeners analyzed, six were present as major compounds. The basic statistical data (mean, median, standard error, minimum-maximum values) of the levels of these congeners in the different species and tissues are given in Table 1. The general order of decreasing concentrations is BDE47 > BDE99, BDE100 > BDE153, BDE154 > BDE28. An exception to this rule is the sea star, where BDE100 > BDE47, BDE99. In contrast to the situation in the sea star, BDE100 is remarkably low in the harbor seal tissues, which might indicate the occurrence of a certain biotransformation capacity in this species. In the invertebrates, the amount of BDE47 is generally < 50% of SPBDE except for the shrimps, whereas in fish it is always > 50%. In harbor seal blubber the contribution of BDE47 to SPBDEs was somewhat higher than in harbor porpoise with values for respectively the mean ( SEM/median/min-max range of 63 ± 3.6%/66%/42-80% for harbor seal blubber and 51 ± 3.9%/49%/32-68% for harbor porpoise blubber (n) = 9 for both species). However, far fetched conclusions should not be drawn from these differences in relatively small data sets, since for another larger data set of harbor porpoise blubber, the values 60 ± 1.4%/59%/39-88% have been reported (19), which is very similar to our results for the harbor seal.

The congeners BDE66, BDE75, BDE77, BDE119, and BDE138 were present at concentrations around the limit of detection (LOD; 0.6 ng g^-1 lipid in a tissue containing 50 mg of extractable lipid). The congeners BDE71, BDE85, BDE183, and BDE190 were never detected.

BDE209 was occasionally present in concentrations just above its LOD of 5 ng g^-1 lipid. However, when the samples contained parts of the digestive system, a presence of BDE209 just above the detection limit cannot be interpreted as unambiguous proof for are al uptake by the organism. Instead, the levels may represent remainders of food present in the digestive system. Since BDE209 is a major compound in suspended particles and sediments from the North Sea and related environments (14), it does apparently not bioaccumulate to a high degree (20). Thus, either its large molecular size decreases the uptake rates, or a relatively rapid biotransformation increases its degradation rate. The literature presents evidence for both sides. A number of field studies showed that the bioaccumulative properties of BDE209 were much lower than those of the tetra- and pentabrominated congeners that dominate in the Penta-BDE formulation (21, 22). The only occasion where this congener has been reported (together with BDE183) in relatively high concentrations in wildlife concerns eggs of pererine falcons breeding in Sweden (11). The peregrine falcon (Falco peregrinus) belongs to the top-predators in the terrestrial food-web, since this bird of prey feeds mainly on other birds. The uptake of BDEs 183 and 209 in blood of humans working in a recycling factory of electronic household equipment has also been demonstrated (12). However, as a result of biotransformation processes, BDE209 had a half-life of less than 10 days in these humans. A relatively rapid metabolism of this compound has also been shown in rats (23). A slow but ^measurable uptake of BDE209 has also been demonstrated in laboratory studies with rainbow trout (24) and juvenile salmon (25). Laboratory studies with ¹⁴C-labeled BDE47 in marine invertebrates demonstrated the formation of polar metabolites, and thus a certain metabolic capacity toward PBDEs does exist even in marine invertebrates, although the fraction metabolized differed from barely detectable to about 50% between the species investigated (26). No clear relations between metabolic capacity and taxonomic group could be observed. Thus, it remains inconclusive whether the main reason for the absence of BDE209 is on the accumulation or on the elimination side of the bioaccumulation process.

TABLE 1: Lipid-Normalized Concentrations in ng g-1 of the Six Major PBDE Congeners Encountered in Animals of Different Trophic Levels of the North Sea Food Web^a

                       BDE28            BDE47               BDE100             BDE99             BDE154             BDE153
       Br subst.:    (2,2',4-)      (2,2',4,4'-)        (2,2',4,4',6-)     (2,2',4,4',5-)    (2,2',4,4',5,6'-) (2,2',4,4',5,5'-
                 mean/            mean/                mean/             mean/              mean/              mean/
     species     median (min-max) median   (min- max)  median (min- max) median  (min-max)  median  (min-max)  median (min-max)

                                                          Invertebrates
sea star (pc)    0.6/0.5 (0.4-1.1) 22/19    (3.4-56)   23/9.1  (2.2-82)  9.3/2.9 (1.0-49)   9.0/1.1 (0.4-41)   2.6/2.6 (0.3-4.8)
hermit crab(abd) 1.4/1.6 (0.8-1.7) 38/29    (8.6-118)  15/11   (2.8-40)  16/9.2  (1.7-71)   10/5.0  (1.5-56)   3.7/1.9 (0.9-18)
whelk (sp)       1.2/1.0 (1.0-1.6) 10/5.5   (2.6-30)   4.2/2.8 (1.5-9.7) 6.1/4.0 (2.9-17)   4.1/4.1 (2.0-6.3)  7.1/5.8 (1.7-13)
shrimp (wb)      2.6/2.6 (2.4-2.7) 37/37    (35-39)    6.9/6.9 (5.1-8.7) 5.1/5.1 (4.6-5.5)  3.9/3.9 (3.9-3.9)  <LOD    <LOD
                                                              Fish
herring
   liver         2.1/2.1 (1.6-2.5) 30/25    (19-52)    9.1/6.9 (5.6-17)  13/12   (8.0-21)   2.6/1.8 (1.5-4.4)  2.1/1.3 (1.1-3.9)
   filet         1.9/1.9 (1.2-2.4) 37/38    (23-47)    9.2/9.3 (6.3-12)  12/11   (9.9-17)   1.5/1.5 (1.3-1.9)  0.9/0.8 (0.6-1.3)
cod
   liver         6.7/6.2 (2.0-12)  133/99   (63-307)   40/33   (18-93)   15/8.8  (1.4-53)   6.4/4.4 (4.3-12)   0.7/0.7 (0.5-1.3))
   filet         2.7/2.1 (1.5-4.5) 43/34    (26-74)    13/12   (5.9-21)  6.3/4.1 (3.1-16)   3.9/3.9 (3.9-3.9)  <LOD    <LOD
whiting
   liver         3.6/3.6 (0.7-6.3) 70/74    (7.6-132)  16/15   (1.7-31)  15/14   (1.9-34)   4.5/3.8 (0.6-11)   1.4/1.2 (0.3-3.1)
   filet         1.8/1.8 (1.3-2.4) 26/28    (7.1-40)   8.6/10  (4.2-12)  9.0/9.6 (5.3-14)   3.3/3.3 (2.2-4.4)  <LOD    <LOD
                                                        Marine Mammals
harbor porpoise
   liver         26/17   (5.0-86)  1331/720 (1.2-4877) 562/285 (0.3-2142)715/402 (0.5-2494) 331/227 (0.2-1054) 185/111 (0.1-504)
   blubber       22/21   (7.6-36)  864/796  (245-1312) 242/228 (47-479)  406/350 (43-764)   178/100 (12-801)   149/60  (5.6-768)
harbor seal
   liver         16/16   (4.1-28)  1328/368 (95-5065)  83/22   (8.7-271) 454/38  (29-1580)  44/18   (2.6-163)  222/22  (7.5-692)
   blubber       9.7/2.9 (1.1-49)  1236/210 (57-9248)  82/25   (6.2-543) 396/57  (11-3065)  21/7.9  (2.4-83)   98/23   (3.4-720)

   a pc ) pyloric caeca; abd ) abdomen; sp ) soft parts; wb ) whole body. < LOD ) value < limit of detection.

The absence of BDE183 from any environmental compartment indicates that the environmental occurrence of the Octa-BDE mixture is presently still negligible in the North Sea and the Skagerrak.

The analyses were performed on different tissues of the same fish (liver and filet) and marine mammals (liver and blubber). The lipid physiology of herring and both gadoid species is quite different; herring stores its depot lipids mainly in muscle ("fatty fish", with a relatively lean liver), whereas the gadoid species whiting and cod use the liver for this purpose ("lean fish", with a very fatty liver and a lean muscle). In both gadoid species, the lipid-normalized PBDE concentrations were higher in the liver than in the filet. This is 11h illustrated for the ratio of PBDE concentrations in liver divided by the sum of the concentrations in liver+filet in Table 2. Here, equal concentrations in both tissue types result in a ratio of 0.5, but the actual mean and median values were 0.67/0.63 for whiting and even 0.79/0.79 for cod. Such a tissue preference for the liver was not found in herring and both marine mammal species. In the case of planar nonortho PCBs, PCDDs, and PCDFs, a preference for the liver after lipid normalization is attributed to the binding of such planar compounds with a high affinity to the phase-I biotransformation enzyme cytochrome P450 1A (27). However, since the levels of CYP1A expression in cod were at the low end of the range observed in different fish species from the North Sea (28), and certainly lower than those in marine mammals, we do not believe that this can explain the observed liver preference of PBDEs in gadoid fish species.

TABLE 2: Tissue Preference of PBDEs, Expressed as Ratios of the Lipid-Normalized Concentrations of BDE47 in Liver to (Liver + Filet) for Fish and Liver to (Liver + Blubber) for Marine Mammals^a 

species 	n  mean   SEM 	median 	min 	max

			Fish
herring 	4  0.445  0.057 0.408 	0.362 	0.604
whiting 	6  0.674  0.048 0.729 	0.518 	0.780
cod 		4  0.790  0.022 0.793 	0.736 	0.841

			Marine Mammals 
harbor porpoise 3  0.524  0.089 0.597 	0.347 	0.627
harbor seal 	3  0.427  0.087 0.354 	0.327 	0.600

a A ratio of 0.5 indicates equal concentrations of BDE47 in both tissues of the same species.

Biomagnification through the Food Chain. The major biomagnification step in the food chain occurs from fish to marine mammals; the lipid-normalized PBDE levels in blubber and liver were similar and generally more than an order of magnitude higher than in the invertebrates and fish. Surprisingly, there was not a clear difference between the three invertebrate species, the planktivorous herring, or the predatory gadoid species whiting and cod in the present study. In an earlier study, biomagnification related to trophic level was found for tetra- to hexa-PBDEs in the imaginary food chain copepods (mainly Calanus finmarchius)-- planktivorous fish (sprat (Sprattus sprattus), or small-and large herring (Clupea harengus))-predatory fish (salmon (Salmo salar)) from the Baltic and the North Atlantic off Iceland (29). The lipid-normalized levels of the major congeners (BDEs 47, 99, and 100) were up to two times higher in large herring than in zooplankton, whereas the levels in salmon were again 2-3 times higher than those in large herring. However, there may also be a size-dependent component in this field observation, since apart from trophic level, the bioaccumulation process also incorporates a size-dependency component (30). This is due to the fact that the ratio of the total surface area of an animal to its size gets smaller when size increases. The surface area of the gill-membrane to total animal volume also plays a role. Both factors result in lower elimination rates in larger animals.

Perhaps the mean and median values of the lipid-normalized PBDE concentrations are biased to higher values for the invertebrates of the present data set, because a number of stations between the UK coast and the Dogger Bank with relatively high exposure levels were present. This may have increased the levels in the sentinel invertebrates more than those in the rapidly migrating gadoids caught in the same area.

To illustrate the observed concentration ranges, the basic statistical descriptors for the data of all six major BDE congeners in the different species have also been given in Table 1. A large difference between median and mean indicates that the levels in the different samples were far from normally distributed. This was especially the case for the harbor seal data, where the mean concentrations are always much higher than the median values, due to the contribution of two animals with very high concentrations, as illustrated with a maximum value of 9248 ng g-1 lipid for BDE47 in blubber.

Geographical Distribution in Invertebrates. The results for the sea star Asterias rubens and the hermit crab Pagurus bernhardus are shown in Figure 1, and those for the common whelk in Figure 2. In general it can be said that the geographical trend is highly similar in the three species. The concentrations of BDEs in the abdomens of hermit crabs being slightly above those in the pyloric caeca of sea stars, which are higher than the concentrations in the soft parts of whelks.

Since these invertebrates do not migrate over large distances, they are more representative for the site of capture than fish and marine mammals. The highest levels of PBDEs in these invertebrates occurred near the English coast, especially at the latitude of the estuaries of the rivers Tyne and Tees, but also further south. These levels were at least an order of magnitude higher than those found along the coastline of continental Europe. Thus the PBDE levels in the major rivers there (Scheldt, Rhine/Meuse, Ems, Weser, and Elbe) do not appear to affect the levels of BDEs from the penta-BDE formulation in invertebrates living in the receiving waters of the North Sea to a marked degree. We were able to complement our data set from the Pelagia-cruise with the data of a sample of A. rubens taken by CEFAS directly in the main dredged channel at the mouth of the river Tees. The very high levels in this sample confirm that the Tees is a major source for the PBDE congeners of the penta-BDE formulation in the North Sea. The long-term residual currents of the North Sea run in a counterclockwise direction (3134). As a result, a residual current along the English coast runs south from Scotland to the Wash, at least in the winter season. In the late spring and summer (May-October) however, a density driven bottom water jet-like current develops along the 40 m depth contour (35, 36), which is close to the coastline in the area of the Tees estuary. This is caused by the development of vertical stratification in the water column caused by increasing air temperatures that warm the surface of the North Sea in the late spring and the summer season. Just like the main circulation, this bottom current runs south but it turns east to the Western-flank of the Dogger Bank at Flamborough Head. Thus water and suspended (dredged) particles from the Tees estuary can be transported either to the south (in winter) and to the Dogger Bank in the east (in summer). The data on the invertebrates illustrate that although atmospheric movements are considered as the main transport pathway for POPs on a global scale (37, 38), water-associated transport can be more important on the scale of regional seas. The identification of the river Tees as a significant source of penta-BDE for the North Sea implicates that the background levels due to the PBDEs released into the environment by the evaporation from flame retarded products in individual households can be overruled by the quantities released in large industrial processes. However, it is yet unknown whether the PBDE production facility at Newton Aycliffe or the application of Penta-BDE by the user industries along the river represents the major source (39). However, the Tees is probably not the only major source at the UK East coast, since in an earlier study, surface sediments from the Humber estuary also contained relatively high levels of BDEs associated with both the Penta-and Deca-BDE formulations in a survey of different European rivers (40). High levels of deca-BDE only were found in sediments of the Belgian/Dutch river Scheldt (40). Intermediate levels of deca-BDE were found in sediments from the UK river Thames, the French river Seine, and the Swiss/French/German/Belgian/Dutch Rhine-Meuse estuary. Low levels of all PBDE formulations were present in the German rivers Ems, Weser and Elbe and the Scandinavian rivers Göta, Glomma, Skaien, and Otra (40).

The levels in fish were measured in liver and filet of two gadoid species, whiting and cod. The geographical trends in concentrations of these rapidly migrating species did not correspond with the trends in the invertebrates any more, although a single sample of whiting filet taken directly at the mouth of the river Tees by CEFAS showed also very high levels (Stetra-hexa BDEs > 2000 ng g^-1 lipid; C. R. Allchin, unpublished data). The Tees has been identified as a major source for PBDEs in an earlier study too (39). The river is also dredged at regular intervals, and the sediment is released at a dispersive site in the outer estuary.

In the Skagerrak, the levels seem to increase from west to east; this might be due to an increasing proportion of Baltic influence.

Comparison of the PBDE Levels in Fish with Other ^Areas. The comparison of the levels obtained for the North Sea with other areas can be done on the basis of SPBDEs or a ^single congener. We have selected the congener that is always present in the highest concentrations, BDE47, for ^this purpose. The present levels ranged between 26 and 133 ng g-1 lipid in the different tissues of herring, whiting, and cod. Due to the limited number of data, we have neglected any effects of biomagnification in predatory fish compared to planktivorous fish in this paragraph (29). Sprat (Sprattus sprattus), herring, and salmon (Salmo salar) caught in the Baltic Sea in 1998, showed BDE47 values in the range of 6.8-45.8 ng g^-1 lipid (29), which is similar to the present range for the North Sea. The levels of BDE47 in yellow eel from the Rhine-Meuse estuary in 1999 were also similar to those in fish from the present study (8). The average levels in lake trout caught in the U.S./Canadian Great Lakes in 1997 were similar to those in fish from the North Sea in Lake Erie and Lake Huron, but the values for Lake Superior (225 ng g^-1) and Lake Ontario (380 ng g^-1) were about an order of magnitude higher (6). An even higher average concentration of 1590 ng g-1 lipid was found for coho and chinook salmon from Lake Michigan (41). The levels in 1999 in grey mullet and sea bass from Osaka Bay in Japan were similar to those in fish from the North Sea, but the historical levels in this coastal area have been more than an order of magnitude higher in the second half of the 1980s (9). Data from the North Atlantic Ocean office land that can be regarded as background levels due to long-distance transport, ^concern herring and salmon; these samples showed indeed lower levels than all coastal areas with a concentration range of BDE47 between 2.0 and 7.6 ng g^-1 lipid ^(29).  

FIGURE 3. Covariance bi-plots of the (¹⁰log-transformed) lipid based concentrations (in ng g^-1) of the six major PBDEs (BDEs 28, 47, 99, 100, 153, and 154), three major PCBs (CBs 101, 138, and 153), and p,p'-DDE. The lower x-axis and the left y-axis give the scores for the first and second PC, respectively. The upper x-axis and the right y-axis show the values of the correlation (r) between each vector and both PC’s. The values for the first and second PC can be obtained by orthogonal projection on the upper X-axis for the first PC and the right y-axis for the second PC. The numbers indicate the different species: 1 ) Asterias rubens; 2 ) Pagurus bernhardus; 3 ) Buccinum undatum; 4 ) Clupea harengus; 5 ) Gadus morhua; 6 ) Merlangius merlangus; 7 ) Phocoena phocoena; 8 ) Phoca vitulina.

Multivariate Statistical Analysis. For a comparison with the PBDEs, the levels of three major PCB congeners (CBs 101, 138, and 153) and p,p'-DDE as the dominant compound of the DDT family were also quantified in the samples. The bi-plot of the (¹⁰log transformed) lipid-based concentrations of the six major PBDEs, three major PCBs, and p,p'-DDE in the tissues of all species investigated is given in Figure 3. The vectors of the PBDE and the PCB congeners are clearly clustered, indicating a high covariance of the concentrations of the different congeners within the same class of chemicals. The vector of p,p'-DDE is closer to the PCB cluster, indicating closer resemblance in its environmental behavior with the PCBs than with the PBDEs. Thus, the environmental distribution of the PBDEs in the North Sea food web clearly differs from the older organochlorines. This is in contrast with the situation in the Great Lakes, where a high degree of covariation was found between the levels of PBDEs and PCBs in atmospheric samples (42) and in samples of salmonids (41). In the air samples, both groups of compounds were clearly associated with large industrial and urban centers of this area.

The datapoints of the samples of the different invertebrates, fish species, and marine mammals are also clearly clustered in the bi-plot. A plot of the value of the first PC against the concentration of BDE47 showed a high correlation between the two variables. This means that the lipid-normalized PBDE concentrations increase from left to right in the bi-plot.

Acknowledgments

The samples of the invertebrates and gadoid fish were taken during a cruise of the R.V. Pelagia that was funded in part by the EU-LIFE project "Harmful impact of TBT communicated" (HIC-TBT). We wish to express our gratitude to the crew of the "Pelagia". The herring samples were taken by the fishing vessel TX 37, for which we thank the owner Mr. Jan van der Vis and his crew. The samples of the harbor seals were obtained from Dr. Ursula Siebert of the Research and Technology Centre (FTZ) in Büsum, Germany. The samples of the harbor porpoise were obtained from Dr. Chris Smeenk and Drs. Marjan Addink of the Museum of Natural History "Naturalis" in Leiden, The Netherlands. This project was financed by the Bromine Science and Environmental Forum (BSEF), Brussels, Belgium. This is NIOZ contribution # 3645. 

Supporting Information Available

Tables of locations and basic characteristics of the sampling stations of the cruise with the RV Pelagia in the North Sea and the Skagerrak for all invertebrates, whiting, and cod (Table SI-1) and lipid contents of the samples as percentage of wet weight (Table SI- 2). This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited

(1) Darnerud, P. O.; Eriksen, G. S.; Jóhanneson, T.; Larsen, P. B.; Viluksela, M. Environ. Health Perspect. 2001, 109, ^49-68.(2) de Boer, J.; de Boer, K.; Boon, J. P. In New types of persistent halogenated compounds; Paasivirta, J., Ed.; Springer-Verlag: 2000; pp 61-95, ISBN 3-540-6583-6..

(3) WHO/IPCS. .Environmental Health Criteria 162, Polybrominated Diphenyl Ethers, listed.; World Health Organization, International Programme on Chemical Safety: Geneva, 1994; ISBN 92 4 157162 4.

(4) Gustafsson, K.; Björk, M.; Burreau, S.; Gilek, M. Environ. Toxicol. Chem. 1999, 18, 1218-1224.

(5) de Boer, J.; Wester, P. G.; Klamer, H. J. C.; Lewis, W. E.; Boon, J. P. . Nature 1998, 394, ^28-29.

(6) Luross, J. M.; Alaee, M.; Sergeant, D. B.; Whittle, D. M.; Solomon, K. R. Organohalogen Compd. 2000, 47, ^73-76.

(7) LeBeuf, M.; Trottier, S. Organohalogen Compd. 2001, 52, 22-26.

(8) de Boer, J.; Allchin, C. R. Organohalogen Compd. 2001, 52, 1317..

(9) Ohta, S.; Ishizuka, D.; Nishimura, H.; Nakao, T.; Aozasa, O.; Okumura, T.; Miyata, H. Organohalogen Compd. 2002, 52, 210-213.

(10) Hale, R. C.; La Guardia, M. J.; Harvey, E. P.; Mainor, T. M.; Duff, W. .H.; Gaylor, M. .O. Environ. Sci. Technol. 2001, 35, 4585-4591..

(11) Sellström, U.; Lindberg, P.; Häggberg, L.; de Wit, C. Brominated flame retardants (PBDEs) found in eggs of peregrine falcon (Falco peregrinus) breeding in Sweden; Rep. no. Bromsve 80402, Svenska Naturskyddsforeningen in cooperation with TCO-Utveckling AB; Stockholm, 2001.

(12) Sjödin, A.; Hagmar, L.; Klasson-Wehler, E.; Kronholm-Diab, K.; Jacobsson, E.; Bergman, A. Environ. Health Perspect 1999, 107, 643-648.

(13) de Boer, J.; Allchin, C. R.; Law, R. J.; Zegers, B. N.; Boon, J. P. Trends Anal. Chem. 2001, 20, ^591-599.

(14) de Boer, J.; van der Horst, A.; Wester, P. G. Organohalogen Compd. 2000, 47, ^85-88.

(15) Duinker, J. C.; Zeinstra, T.; Hillebrand, M. T. J.; Boon, J. P. Aquatic Mammals 1989, 15, 95-124.

(16) Boon, J. P. In Marine biology of polar regions and effects of stress on marine organisms; Gray, J. S., Christiansen, M. E., Eds.; John Wiley and Sons Ltd.: 1985; pp 493-512, ISBN 0 471 90465 1.

(17) Booij, K.; van den Berg, C. Bull. Environ. Contam. Toxicol. 1994, 53, 71-76.

(18) Gabriel, K. R. Biometrika 1971, 58, 453-467.

(19) Law, R. J.; Allchin, C. R.; Bennett, M. E.; Morris, S.; Rogan, E. Chemosphere 2002, 46, 673-681.

(20) Zegers, B. .N.; Lewis, W. .E.; Boon, J. P. Organohalogen Compd. 2000, 47, 229-232.

(21) Sellström, U.; Kierkegaard, A.; de Wit, C. A.; Jansson, B. Environ. Toxicol. Chem. 1998, 17, 1065-1072.

(22) Booij, K.; Zegers, B. N.; Boon, J. P. Organohalogen Compd. 2000, 47, 89-92.

(23) Klasson-Wehler, E.; Mörck, A.; Hakk, H. Proceedings of the 2nd international workshop on brominated flame retardants, May 14-16, Stockholm, 2001; pp 93-97.

(24) Kierkegaard, A.; Balk, L.; Tjärnlund, U.; de Wit, C. A.; Jansson, B. Environ. Sci. Technol. 1999, 33, 1612-1617.

(25) Boon, J. P.; van Zanden, J. J.; Lewis, W. E.; Zegers, B. N.; Goksøyr, A.; Arukwe, A. Mar. Environ. Res. 2002, in press.

(26) Magnusson, K. .Fate and effects of organic pollutants in marine fauna and sediments; Göteborg University: 2001; ISBN 91-628-4870-4.

(27) van den Berg, M.; De Jongh, J.; Poiger, H.; Olson, J. R. CRC Crit. Rev. Toxicol. 1994, 24, 1-74.

(28) van Schanke, A. Metabolic interaction of polychlorinated biphenyls and polycyclic aromatic hydrocarbons in the marine flatfish dab (Limanda limanda); Utrecht University: : 2001; ISBN 90-393-2927-7..

(29) Burreau, S.; Zebühr, Y.; Ishaq, R.; Broman, D. Organohalogen Compd. 2000, 47, 253-255.

(30) Sijm, D. T. H. M.; Van der Linde, A. Environ. Sci. Technol. 1995, 29, 2769-2777.

(31) Kautski, H. Deutsch. Hydrog. Z 1973, 26, 241-246.

(32) Kershaw, P.; Baxter, A. Deep Sea Res. 1995, 42, 1413-1448.

(33) ICONA. North Sea Atlas for Netherlands Policy and Measurement; Stadsuitgeverij: Amsterdam, 1992; ISBN 90-5366-047-X.

(34) Otto, L.; Zimmerman, J. T. F.; Furnes, G. K.; Mork, M.; Saetre, M.; Becker, G. Neth. J. Sea Res. 1990, 26, 161-238.

(35) Brown, J.; Hill, A. E.; Fernand, L.; Horsburgh, K. J. Est. Coast. Shelf Sci. 1999, 48, 343-355.

(36) Brown, J.; Fernand, L.; Horsburgh, K. J.; Hill, A. E.; Read, J. W. J. Plankton Res. 2001, 23, 105-116.

(37) Kalantzi, O. I.; Alcock, R. E.; Johnston, P. A.; Santillo, D.; Stringer, R. . L.; Thomas, G. . O.; Jones, , K. . C. . Environ. Sci. Technol. 2001,
35, 1013-1018.

(38) Lohman, R.; Ockenden, W. .A.; Shears, J.; Jones, K. C. Environ. Sci. Technol. 2001, 35, 4046-4053.

(39) Allchin, C. R.; de Boer, J. Organohalogen Compd. 2001, 52, 30-34.

(40) Kierkegaard,A.; Sellström,U.; Wideqvist,U.; Winberg,A.; Alsberg, T. .Reports of the results of the one-off survey DIFFCHEM-project; Oslo and Paris Commissions Rep. no. SIME 97/6/1-E; 1997. 

(41) Manchester-Neesvig, J. B.; Valters, K.; Sonzogni, W. C. Environ. Sci. Technol. 2001, 35, 1072-1077.

(42) Strandberg, B.; Dodder, N. G.; Basu, I.; Hites, R. A. Environ. Sci.Technol. 2001, 35, 1078-1083.  

Received for review December 6, 2001. Revised manuscript received May 1, 2002. Accepted July 11, 2002. ES0158298