2: Evidence for Bioaccumulation

• 2.1 Biota Concentration Data
  • 2.1.1 Information Evaluated in the Screening Assessment
  • 2.1.2 New Biota Concentration Data
  • 2.1.3 Synthesis of Biota Concentration Data
• 2.2 Bioconcentration, Bioaccumulation and Biomagnification Data
  • 2.2.1 Information Evaluated in the Screening Assessment
  • 2.2.2 New Bioaccumulation Data
  • 2.2.3 Model Predictions
• 2.3 Weight-of-Evidence Analysis
  • 2.3.1 Summary of Evidence
  • 2.3.2 Conclusion Respecting Bioaccumulation

The evaluation of new lines of evidence conducted in this section relies on the definitions of bioaccumulative and biomagnification summarized in the preceding section. Most importantly, the evaluation of whether these new lines of evidence indicate that decaBDE is bioaccumulative relies primarily on ratio-based methods for bioaccumulation assessment. From the definitions in Section 1.2, it can be argued that the degree of chemical accumulation in an organism is best characterized by a ratio comparing the concentration of the subject substance in an organism to the appropriate exposure concentration in the organism’s environment (i.e., to enable quantification of BCFs, BAFs, BMFs, TMFs or BSAFs).

Other evidence, such as the presence of high concentrations in top predators, is sometimes useful as a supporting line of evidence for ratio-based measures of bioaccumulation. However, it is recognized that the characterization of biota chemical concentrations as "high" is subjective and is based on professional judgment, and thus prone to alternative interpretation. The evidence provided by studies quantifying chemical residues in tissues is often confounded by the following factors:

1. In many monitoring and field studies, particularly those showing unusually high concentrations in preditor organisms, there is a lack of knowledge regarding levels of exposure. In such instances, quantification of BAF or BMF is impossible.
2. Although a high concentration of a chemical in a top predator could be due to trophic magnification, it could also be due to high geographically localized exposure to a particular chemical, as would be the case with scavenging from or living in waste dumps. It is also possible that uptake routes other than the food chain can be occurring, such as inhalation, or direct ingestion of plastic.
3. Particularly in the case of decaBDE, there are also difficulties respecting the potential contamination of samples and laboratory blanks. Interlaboratory studies have elaborated on this issue, noting that decaBDE is present in small dust particles in laboratories, causing high background values and potential for contamination on glassware. The issue of contamination for decaBDE is particularly problematic at trace levels of detection (de Boer and Wells 2004).

The next section summarizes the data respecting decaBDE biota concentrations considered in the PBDE screening assessment report (Canada 2006, Environment Canada 2006b). This summary is followed by an analysis of new data available after 2004 and an interpretation of their significance in relation to bioaccumulation and biomagnification of decaBDE.

• Norstrom et al. (2002) did not detect decaBDE in herring gull (Larus argentatus) eggs from the Great Lakes.
• Lichota et al. (2004) determined a total PBDE concentration (67% of which was decaBDE) of 0.777 mg/kg lipid measured in fat from Vancouver Island marmot (Marmota vancouverensis).
• European Communities (2002) and Law et al. (2003) summarized the results of many analyses for decaBDE in fish and marine mammals from Europe and elsewhere. It was noted that decaBDE was detected only occasionally, at a concentration close to the detection limit of the method used.
• DecaBDE (i.e., BDE209)was detected once at a concentration of 1.4 µg/kg wet weight (ww) in mussels from Japan (sampling conducted from 1981 to 1985) (Watanabe et al. 1987).
• Dodder et al. (2002) analyzed decaBDE concentrations in freshwater fish from the northeastern United States and found that all concentrations of decaBDE were below the analytical detection limit (ranged from 1.4 to 1.6 µg/kg ww).
• Ikonomou et al. (2002a) conducted sampling from ringed seal (Phoca hispida) blubber between 1981 and 2000, but none of the samples contained decaBDE concentrations in excess of the procedural blanks.
• Ikonomou et al. (2000, 2002a,b) analyzed decaBDE in marine biota collected from the British Columbia coast and Holman Island, Northwest Territories, between 1981 and 2000. The levels of decaBDE were equivalent to those of the procedural blanks.
• DecaBDE (i.e., BDE209) was detected in 18 of 21 analyzed eggs of peregrine falcons (Falco peregrinus) from Sweden, at concentrations from 28 to 430 µg/kg lipid (Lindberg et al. 2004).
• A study by de Boer et al. (2001) reported on sampling conducted in 1999 in the North Sea and the Tees estuary, UK, of marine mammals -- harbour seal (Phoca vitulina), harbour porpoise (Phocoena phocoena), white-beaked dolphin (Lagenorhynchus albirostris) and bottlenose dolphin (Tursiops truncatus) -- for levels of BDE209 in liver, muscle and blubber tissues. Most samples showed concentrations of BDE209 below detection limits (ranged from approximately 0.8 to 9.0 µg/kg lipid); however, a few samples showed elevated levels in specific tissues (e.g., up to 26 µg/kg lipid in the blubber of harbour seal, up to 160 µg/kg lipid in the liver of harbour porpoise, and up to 318 µg/kg lipid in white-beaked dolphin).
• Sampling was conducted (de Boer et al. 2004) to determine the occurrence of decaBDE in liver, muscle tissue and eggs of high trophic level bird species from the United Kingdom and the Netherlands. Levels of decaBDE were detected in 10 of 28 liver samples (range < 1.5 to 181 µg/kg lipid), 14 of 28 muscle samples (range < 4.2 to 563 µg/kg lipid) and 25 of 68 eggs (range < 1.8 to 412 µg/kg lipid). Concentrations in the Swedish peregrine falcon (Falco peregrinus) eggs, which were re-analyzed in the study, were all within 30% of those originally determined by Lindberg et al. (2004). Highest concentrations of decaBDE were measured in muscle tissue samples collected from United Kingdom heron (Ardea cinerea) and peregrine falcon (Falco peregrinus), and in eggs from Swedish peregrine falcon.

Further details on these studies are provided in Appendix A and in the Supporting Working Document for the Ecological Screening Assessment of Polybrominated Diphenyl Ethers (Environment Canada 2006b).

Recently, a large number of studies showing tissue concentrations of decaBDE in various global biota have been published. These studies are discussed below and are interpreted with reference to bioaccumulation and biomagnification as defined in earlier sections of this report.

• Voorspoels et al. (2006a) reported decaBDE concentrations in red fox (Vulpes vulpes) from Belgium. Samples of liver (n=30), muscle (n=33) and adipose tissue (n=27) were collected from foxes which were being processed for diagnostic screening of rabies infection. Concentrations of decaBDE, detected in less than half of the samples of each tissue type, ranged from <9.1 to 760 ng/g lipid for liver, from <3.9 to 290 ng/g lipid in muscle and from <3.7 to 200 ng/g lipid for adipose tissue. The authors concluded that these data confirmed "unambiguously" that decaBDE does "bioaccumulate" in red fox. However, in a follow-up paper by Voorspoels et al. (2007), decaBDE was not detected in the prey species of fox -- wood mice (Apodemus sylvaticus) and bank voles (Clethrionomys glareolus) -- meaning that biomagnification could not be determined. However, the authors noted that the amount of tissue that was available for analysis was low due to the small size of the rodents and this caused sometimes elevated limits of quantification, ranging from 7.3 to 17 ng/g lipid for BDE209.
• Voorspoels et al. (2006b), reported decaBDE concentrations in birds of prey from Belgium which were found dead or severely injured and had to be euthanized. Bird species included the common buzzard (Buteo buteo), sparrowhawk (Accipiter nisus), long-eared owl (Asio otus), barn owl (Tyto alba) and tawny owl (Stryx aluco). Tissues analyzed from each species included brain, fat, liver, muscle and adipose tissue. Levels of decaBDE were detected in nearly all serum samples (2-58 ng/g lipid for all species combined) and in some liver samples (below detection limit (BDL) - 190 ng/g lipid for all species combined), but not in any other tissues. Based on these findings, the authors concluded that either the exposure to decaBDE was low, or that this congener was poorly accumulated. In the follow-up paper by Voorspoels et al. (2007), decaBDE was not detected in prey species of buzzard (wood mice, Apodemus sylvaticus; and bank voles, Clethrionomys glareolus) or sparrowhawk (great tit, Parus major), meaning that biomagnification could not be assessed.
• Jaspers et al. (2006) analyzed liver and muscle samples from 7 species of aquatic birds and terrestrial predatory birds from Belgium -- grey heron (Ardea cinerea) and great crested grebe (Podiceps cristatus), both aquatic, and common buzzard (Buteo buteo), kestrel (Falco tinnunculus), sparrowhawk (Accipiter nisus), long-eared owl (Asio otus) and barn howl (Tyto alba), all terrestrial. decaBDE (i.e., BDE209) was detected in 1 of 2 muscle samples from barn owl at 68 ng/g lipid and not detected in muscle tissue of any other species (detection limit (DL)= 1-2.25 ng/g lipid). decaBDE (BDE209) was not detected in liver of heron, grebe or buzzard. The range of decaBDE concentrations in liver of the remaining species was 52-85 ng/g lipid. However, decaBDE was only detected in a few samples from the terrestrial birds, but not in the aquatic birds. The authors suggested that terrestrial bird species might be more exposed to decaBDE than aquatic species.
• Christensen et al. (2005) reported decaBDE residues in muscle and fat from coastal (n=6) and interior (n=6) grizzly bears (Ursus arctos horribilis) in British Columbia, Canada. decaBDE concentrations (i.e., of BDE209) ranged from approximately 1 to 2.77 mg/kg lipid in coastal bears and from approximately 0.5 to 41.71 mg/kg lipid in interior bears (concentrations read from graph), suggesting a higher exposure to decaBDE for bears feeding in interior environments where diets are predominantly terrestrial in nature. The authors also used a "bioaccumulation slope" method to characterize the observed bioaccumulation of decaBDE in coastal and interior bears. This analysis is evaluated further in Section 2.2.
• Verreault et al. (2004, 2005) analyzed decaBDE in two Norwegian arctic top predators, glaucous gull (Larus hyperboreus) and polar bear (Ursus maritimus), inhabiting Svalbard and Bjoroya, Norway. In gull plasma from Svalbard (n=27), decaBDE ranged from <0.05 to 0.33 ng/g ww while egg concentrations (n=20) ranged from below detection to 170 ng/g lipid. In Bjoroya, gull plasma (n=89) and egg (n=4) concentrations of 410 ng/g lipid and 23-53 ng/g lipid, respectively, were determined. In polar bear plasma from Svalbard (n=15), decaBDE was detected in only one sample, at a concentration of 0.1 ng/g ww. The authors claim that decaBDE was "bioaccumulative" to a limited degree. However, this claim was based solely on the occurrence of decaBDE in tissues without an evaluation of whether the concentrations are high relative to the environment or prey.
• Vorkamp et al. (2005) reported concentrations of decaBDE in eggs of peregrine falcon (Falco peregrinus) from southwestern Greenland collected from 1986 to 2003. Although decaBDE was detected in all samples (n=37) at concentrations ranging from 3.8 to 250 ng/g lipid, high concentrations were only found in two eggs from 1995 and 2002 and the median concentration was 11 ng/g lipid. The authors noted a significantly increasing temporal trend in decaBDE concentration in the falcon tissues. They also concluded that their results provided evidence of some uptake and bioaccumulation of decaBDE (ruling out the idea of a size threshold for membrane permeation of decaBDE). The authors characterized decaBDE concentrations as low in falcon eggs.
• SØrmo et al. (2006) reported analyses of decaBDE in biota from Svalbard, Norway. decaBDE (BDE209) was not detected in two calanoid copepod species (method detection limit (MDL) = 0.012-1.3 ng/g lipid) but was detected at a concentration of 7.22 ng/g lipid in ice amphipod (Gammarus wilkitzkii). The decaBDE concentration in polar cod (Boreogadus saida; n=7) ranged from 0.05 to 0.42 ng/g lipid with a mean of 0.2 ng/g lipid. decaBDE (i.e., BDE209) was detected in only 1 of 6 samples of ringed seal (Pusa hispida) blubber, at a concentration of 0.02 ng/g lipid (MDL = 0.014-0.75 ng/g lipid). Polar bear (Ursus maritimus) adipose tissue (n=4) contained 0.03-0.16 ng/g lipid of decaBDE with an average concentration of 0.09 ng/g lipid. DeWit et al. (2006) summarized additional data from Muir et al. (2006) and Skaare (2004) for polar bears in Svalbard. The BDE209 congener (decaBDE) was not detected in the Muir et al. (2006) study (MDL = 1 ng/g lipid) and was 1 ng/g lipid in the Skaare (2004) study.
• Chen et al. (2007) reported measurements of decaBDE concentrations in birds of prey from northern China including samples of muscle, liver and kidney from common kestrel (Falco tinnunculus; n=6), sparrowhawk (Accipiter nisus; n=11), Japanese sparrowhawk (Accipiter gularis; n=6), little owl (Athene noctua; n=6), scops owl (Otus sunia; n=6), long-eared owl (Asio otus; n=6), upland buzzard (Buteo hemilasius; n=3) and common buzzard (Buteo buteo; n=3). The common kestrel had the highest mean concentrations of decaBDE in muscle (2150 ng/g lipid), liver (2870 ng/g lipid) and kidney (483 ng/g lipid), followed by scops owl (mean range of 59-537 ng/g lipid in muscle, liver and kidney), and sparrowhawk (mean range of 83-249 ng/g lipid in muscle, liver and kidney). The concentration of decaBDE in tissues from the other species did not exceed 150 ng/g lipid. The authors concluded that decaBDE was especially elevated compared to other published reports and that this may be related to significant production, usage or disposal of decaBDE-containing products in China. They also suggested that exposure to decaBDE could be higher in terrestrial food chains than in aquatic food chains.
• Lam et al. (2007) assessed PBDEs levels in eggs of birds of southern China. The researchers examined eggs (n=5, for each species) of little egret (Egretta garzetta), black-crowned night heron (Nycticorax nycticorax), Chinese pond heron (Ardeola bacchus) and cattle egret (Bubulcus ibi) from Hong Kong, Ziamen and Quanzhou. Levels of decaBDE ranged from <0.5 (in black-crowned night heron from Quanzhou) to 99 ± 130 ng/g lipid (in Chinese pond heron from Xiamen). The higher abundance of BDE209 in birds from Xiamen appeared to correspond with high production of electronics in Xiamen. Exposure concentrations were not determined in this study.
• Johnson-Restropo et al. (2005) reported the findings of a monitoring study of marine fish -- teleosts, Atlantic stingray (Dasyatis sabina) and sharks (Rhizoprionodon terraenovae and Carcharhinus leucas) -- and dolphins (Tursiops truncatus) of coastal Florida. BDE209 (decaBDE) was not detected (MDL=0.022 ng/g lipid) in muscle samples from silver perch (Bairdiella chrysoura), striped mullet (Mugil cephalus), spotted seatrout (Cynoscion nebulosus) and red drum (Sciaenops ocellatus), and also not detected (MDL=0.022 ng/g lipid) in blubber samples from bottlenose dolphin. The measured concentrations of decaBDE in hardhead catfish (Arius felis) and Atlantic stingray were reported to be 4.5 ng/g lipid and 0.1 ng/g lipid, respectively, while for sharks, the decaBDE concentration ranged from 16.9 ng/g lipid in spiny dogfish (Squalus acanthias) to 778 ng/g lipid in bull shark (Carcharhinus leucas). decaBDE (i.e., BDE209) was the most abundant congener in sharks, suggesting either that there was preferential exposure to decaBDE or that preferential accumulation of decaBDE was occurring in shark species.
• Gauthier et al. (2008) conducted monitoring of bird eggs and reported on temporal trends (1982-2006) for PBDEs, and most notably, for BDE209 in pooled samples of herring gull (Larus argentatus) eggs from seven colonies on lakes Superior, Michigan, Huron and Ontario, and on the Detroit and Niagara rivers. BDE209 concentrations in 2006 egg pools ranged from 4.5 to 20 ng/g ww and composed 0.6-4.5% of the total PBDE concentrations measured. The authors noted mean doubling times for BDE209 concentrations in eggs of 2.1-3.0 years. For octa- and nonaBDEs, the mean doubling time was determined to be 3-11 years and 2.4-5.3 years, respectively. Based on this rapid increase in concentrations over time, they indicate that the potential for decaBDE to bioaccumulate appears to have been previously underestimated, but they did not quantify exposure concentrations. The researchers also suggested that congener patterns in egg samples suggested meta-position metabolic debromination of BDE209 and -207 to BDE197.
• Shaw et al. (2007) reported decaBDE concentrations in blubber of harbour seals (Phoca vitulina concolor) of the northwest Atlantic coast from 1991 to 2005. The BDE209 congener (decaBDE) was detected in 2 of 4 samples at concentrations up to 7.4 ng/g lipid.
• DeWit et al. (2006) summarized decaBDE levels in Arctic biota from many additional studies:
  • concentrations of 0.025 to 0.12 ng/g ww in moss from Norway (Schlabach et al. 2002);
  • 0.5, 0.1 and 0.8 ng/g lipid in grouse (Lagopus sp.), lynx (Lynx lynx) and moose (Alces alces) liver, respectively, from Norway (Mariussen et al. 2004);
  • detectable (but not quantified) amounts in eggs of birds of prey from northern Norway (Herzke et al. 2005); and
  • 3.6-33 ng/g lipid in the muscle tissue of blue mussels (Mytilus edullis) and 0.98-0.99 ng/g lipid in the liver of Atlantic cod (Gadus morhua) from Norway (Fjeld et al. 2004).
• Verreault et al. (2007) monitored decaBDE in egg yolk and plasma of male and female glaucous gulls (Larus hyperboreus) from the Norwegian Arctic. decaBDE (i.e., BDE209) was "virtually non-detectable" in egg yolk and plasma but the authors speculated that the presence of detectable concentrations of octa- and nonaBDEs in the samples suggested accumulation and subsequent in vivo debromination of decaBDE. Although the detection limit for decaBDE was not reported, detection limits for hepta- to nonaBDEs were reported to be 0.16 ng/g ww and are presumably the same for decaBDE.

Additional studies are also described in Section 2.2. These studies contained sufficient information for conducting a ratio-based evaluation of bioaccumulation or biomagnification.

The wealth of detected concentrations of decaBDE in a wide range of biota provides confirmation that this substance can accumulate in organisms. However, these data provide little evidence that decaBDE may be bioaccumulative. The main shortcoming is that they fail to compare the measured concentrations with either those determined in the environment in which the biota reside (e.g., as a BAF or BCF) or the prey which the biota consumes (e.g., as a BMF or TMF).

In the absence of a ratio-based assessment of whether decaBDE is bioaccumulative or biomagnifying, it is still possible to conduct a subjective judgment of whether the observed concentrations in top predators appear high, indicating potential trophic magnification.

Concentrations of decaBDE have been observed to be relatively high (e.g., exceeding 100 ng/g lipid) in several top predator species, including:

• birds of prey, especially kestrel, sparrowhawk and owl from China; peregrine falcon, sparrowhawk and kestrel in UK and Sweden; peregrine falcon in Greenland; and buzzard in Belgium;
• red fox in Belgium;
• sharks of coastal Florida;
• marine mammals such as harbour porpoise and white-beaked dolphin; and
• some marine bird species, including heron from Sweden and glaucous gulls from Byoroya and Svalbard, Norway.

Of these, the Greenland perigrine falcon results and the coastal Florida shark results may provide the strongest evidence that trophic magnification might be occurring, given the potential remoteness of the sites of sampling. However, Vorkamp et al. (2005) also note that the perigrine falcon (Falco perigrinus) subspecies of Greenland from which samples were taken migrates to Central and South America in the winter, following a route along the Atlantic coast of North America. As a result, there is a potential that the organisms were exposed to locationally specific contaminated environments that may explain their relatively high loading of decaBDE. Although high decaBDE concentrations are observed in Florida sharks, similarly high concentrations are not observed in teleost fish species from the same ecosystem which would likely make up the prey species for shark; trophic magnification is a potential explanation for this difference. Another explanation could be that the sharks are also exposed to contaminated refuse that they could be consuming.

For the remaining results, several factors tend to confound the findings with respect to whether decaBDE is bioaccumulative. These include

• the potential that the sampled species inhabit decaBDE hotspots close to industrialized areas; and
• the large number of non-detects within the same studies and species in which relatively high concentrations were observed.

In addition, there are many data that directly contradict a generalized conclusion that the concentration of decaBDE is relatively high in top predators. These include

• relatively low and often non-detected concentrations of decaBDE in marine fish species, even though some of these species such as salmon or cod could be considered top predators within their respective food webs (see Appendix A for summary of these data);
• relatively low and often non-detected concentrations of decaBDE in mammalian predators such as grizzly bear, polar bear and lynx which feed near the top of their respective food webs;
• relatively low and often non-detected concentrations of decaBDE in marine mammals which feed near the top of their respective food webs;
• relatively low and often non-detected concentrations of decaBDE in marine/aquatic birds which feed at the top of their respective food webs; and
• relatively low and often non-detected concentrations of decaBDE in birds of prey, often in the same studies as those where relatively high concentrations were also observed.

Thus, the evidence based on concentrations in top predators fails to provide a strong indication that decaBDE is bioaccumulative or biomagnifying in food webs. Observations for some species in certain locations suggest that decaBDE might be bioaccumulating or biomagnifying under certain conditions; however, without quantification of exposure one cannot strongly conclude that bioaccumulation or biomagnification is occurring.

This section first summarizes the data respecting decaBDE bioaccumulation considered in the PBDE screening assessment report (Canada 2006, Environment Canada 2006b), then examines new data available after 2004 and its significance.

• MITI (1992) determined that the BCF for carp ranged from <5 to <50 in their 6-week bioconcentration study (recalculated to <3000 by European Communities (2002)). Given the exceptionally low water solubility limit of decaBDE it is not expected that this substance will be appreciably taken up from the water phase by aquatic organisms. Stapleton et al. (2004) exposed juvenile common carp (Cyprinus carpio) to decaBDE (>98% purity as reported by Cambridge Isotope Laboratories) amended in food on a daily basis for 60 days (d) (940 ng/d/fish) followed by a 40-day period in which fish were fed clean food. decaBDE (i.e., BDE209) was not detected in whole fish tissues during the exposure or depuration periods (detection limit of approximately 1 µg/kg ww); however, several peaks were observed in the chromatograms of the exposed fish that were not observed in the control fish, suggesting transformation of decaBDE. Their results suggested that at least 0.44% of decaBDE was bioavailable in the form of its metabolites. This value could be higher if other metabolites were present which were not determined in these studies.
• Kierkegaard et al. (1999) dosed juvenile rainbow trout (Oncorhynchus mykiss) with food contaminated with Dow FR-300 (composition not determined in this study, but reported as 77.4% decaBDE, 21.8% nonaBDE and 0.8% octaBDE by Norris et al. 1973, 1974). A small proportion of the test material was taken up during the 120-day exposure phase of the experiment, amounting to about 0.02-0.13% based on the muscle concentrations of the total hexa- to decaBDEs present, or approximately 0.005% based only on the decaBDE concentrations in muscle and the mean dietary dose of Dow FR-300. The authors did not report the decaBDE concentration in food, meaning that a dietary BAF cannot be estimated from the study data.
• Metabolism studies using rats (Norris et al. 1973, 1974; El Dareer et al. 1987) suggested that decaBDE has a low bioaccumulation potential in mammalian species. For instance, Norris et al. (1973, 1974) dosed male and female rats with 1.0 mg of ¹⁴C-labelled decaBDE as a suspension in corn oil and found that around 90.6% of the administered ¹⁴C-labelled decaBDE was excreted in the feces within 24 hours (h), and by 48 h, all of the administered chemical had been excreted. Tissue accumulation studies in which rats were fed diets of decaBDE at a rate of 0.1 mg/kg body weight per day showed that bromine contents in various tissues were not significantly greater than those of the controls. The bromine content of the adipose tissue of decaBDE-dosed rats was found to be significantly increased at the p<0.03 level but not at the p<0.01 level when compared with the controls (Norris et al. 1973, 1974).

The bioaccumulation data from these studies are also summarized in Appendix B. Additional studies were also reviewed in the screening assessment but their findings did not allow estimates of bioaccumulation parameters. Rather, they are more relevant to the evaluation of debromination and metabolite formation, and are discussed in Section 3.

The bioaccumulation data from studies published after 2004 are discussed in this section and summarized in Appendix B.

Studies on Aquatic Species

Stapleton et al. (2006) exposed 45 juvenile rainbow trout (Oncorhynchus mykiss) to spiked food containing decaBDE (purity reported as 98.7%) for a period of 5 months. The 45 fish were randomly distributed to three experimental tanks while an additional 15 fish that were fed non-spiked food were kept in a separate tank. The concentration of decaBDE in the spiked food was 940 ng/g and the fish (average weight 91.2 g) were fed at a rate of 1% of their body weight per day. One fish from each tank was sacrificed for analysis at 9 time points throughout the 5-month exposure period. Blood was sampled from individual fish at the initiation of exposure and at three time points during the final 3 months of the study. Samples of blood serum, intestine, liver and carcass at each sampling time were each analyzed for decaBDE and lower brominated BDE congeners. An in vitro study using rainbow trout liver microsomes was conducted concurrently to confirm the metabolic capacity of rainbow trout tissues and potential metabolites.

The net uptake of decaBDE during the experiment was estimated at 3.2% based on the total burden of hepta- through decaBDE congeners present in the carcass, or 3.7% if the liver was included in the calculation. The decaBDE concentration on the final day of exposure was highest in the liver (342 ng/g ww) followed by the intestine (~60 ng/g ww, read from graph), serum (26-40 ng/g ww) and carcass (5.3 ng/g ww). The detection limit in this study was 1 ng/g ww. Several hepta-, octa- and nonaBDE congeners were also accumulated, potentially, as a result of decaBDE debromination (fish were dosed only with decaBDE, and background decaBDE concentrations in the control fish ranged from <0.5 ng/g to 0.5 ng/g). These debromination results are discussed further in Section 3.1.

Using concentration data from day 112, the reported lipid content of the test fish, and the known decaBDE concentration/lipid content in food reported by Stapleton et al. (2006), it is possible to estimate BMFs for decaBDE on its own and for the combined burden of decaBDE plus debrominated congeners. Table 2-1 summarizes the calculations.

It is expected that the higher lipid content in food (10.1% lipid) compared with carcass and liver (4.5% and 2.3%, respectively) may cause the ww BMFs to underestimate the bioaccumulation potential of decaBDE, and as a result, lipid-weight BMFs were calculated. For decaBDE on its own, the BMFs ranged from 0.02 (carcass) to 1.28 (liver) whereas for the total BDE burden, the BMFs ranged from 0.09 to 1.74.

Based on the findings of other studies (refer to Section 3), it is possible that decaBDE could be transformed to hydroxylated and/or methoxylated debrominated congeners. The Stapleton et al. (2006) study did not include these potential products as analytes. If they are formed in rainbow trout, then the reported net uptake of neutral BDEs only would underestimate the actual total uptake of decaBDE. It could be speculated that the much lower absorption of decaBDE observed for rainbow trout relative to that observed in rats (see Section 3.1) might be attributable to both neutral BDEs and hydroxylated/methoxylated BDEs. If metabolites other than hepta-, octa-, and nonaBDEs were being formed and persisting in the fish, then the BMFs calculated above would underestimate the total accumulation potential of decaBDE-related compounds.

Tomy et al. (2004) studied the uptake by juvenile lake trout (Salvelinus namaycush) of twelve tetra- to heptaBDEs (Wellington Laboratories, all purities >96%) plus decaBDE (technical grade, Great Lakes Chemical Corp., purity not provided) from spiked commercial fish food. Test fish were exposed to spiked food for 56 d followed by a 112-day elimination period. Seventy fish each were exposed to low and high concentrations (measured in food) of technical-grade decaBDE, and a non-exposed control group was also monitored (concentrations measured in food). Significant uptake of decaBDE was observed for both the low- and high-exposure treatments. For the low-exposure treatment, depuration of chemical during the elimination phase was non-detectable (slope not significant) and the absorption efficiency, half-life and BMF were not estimated. For the high-exposure treatment, the absorption efficiency was estimated at 5.2% with a half-life of 26 ± 5 d and a BMF ^[1] of 0.3. The authors also reported an absorption efficiency for decaBDE as low as 0.3%. Although lower brominated PBDE congeners appeared to be bioformed in the fish, it was not possible to include the debrominated congeners in the BMF estimates because similar congeners were also present in the spiked food.

Ciparis and Hale (2005) examined the bioavailability and accumulation of multiple PBDEs, including decaBDE, from sediments and biosolids to the aquatic oligochaete, Lumbriculus variegatus. Oligochaetes were exposed to either composted biosolids containing 1600 ng/g total PBDEs or artificial sediments spiked with technical Penta- and decaBDE formulations (1300 ng/g total PBDEs). The experimental protocol included a 28-day uptake phase followed by a 21-day elimination phase. Following solvent extraction, clean-up in a size exclusion column, and further purification using solid-phase extraction columns, decaBDE was quantified from substrates and tissues on a gas chromatography (GC) device equipped with a halogen-selective electrolytic conductivity detector with MDLs of 190 ng/g and 20 ng/g for tissues and substrates, respectively. Although significant accumulation of lower brominated PBDEs (especially BDE47 and BDE99) was observed with both biosolids and spiked sediments, uptake of decaBDE was minimal and it was not possible to estimate steady-state sediment BSAFs or kinetic parameters for decaBDE accumulation. The authors speculated that the bioavailability of decaBDE was limited by its high log K_ow (suggesting that desorption from sediment particles is minimal) and large molecular size, which may impede its transport across cell membranes.

Burreau et al. (2004, 2006) reported the results for three separate food web monitoring programs for PCBs and PBDEs. Burreau et al. (2004) sampled perch (Perca fluviatilis; n=120, 33 individuals), roach (Rutilus rutilus; n=23, 8 individuals) and pike (Esox lucius; n=51, 25 individuals) in the Lumparn estuary in the Åland archipelago in the Baltic Sea and analyzed muscle tissue composites from each species. ^[2]. Burreau et al. (2006) described monitoring studies for the Baltic Sea and the Atlantic Ocean (south of Iceland). The Baltic Sea study was conducted in 1998 and sampled zooplankton (Calanoid crustacea; n=3 net tows), sprat (Sprattus sprattus; n=6), herring (Clupea harengus; n=5) and Atlantic salmon (Salmo salar; n=10), while the Atlantic Ocean study was conducted in 1999 and sampled zooplankton (Calanoid sp.; n=10), small herring (Clupea harengus; n=6), large herring (Clupea harengus; n=10) and Atlantic salmon (n=10). The detection limits ranged from 140 to 148 pg/g. Given that 10 g of muscle were used, this represents concentration-based detection limits of 14-14.8 pg/g ww. Determinations of decaBDE were blank-corrected. Table 2-2 provides a summary of median observed concentrations in the Baltic Sea and North Atlantic Ocean biota samples.

To examine the potential for food web biomagnification of decaBDE in the biota data for each Baltic Sea food web (roach-perch-pike; zooplankton-sprat-herring-salmon), the authors conducted an analysis of trophic magnification. This involved a regression of lipid-normalized concentration vs. d¹⁵N according to the following model:

Where C is the biota concentration (lipid-normalized), A is a constant representing d¹⁵N at the base of the food chain and B represents the "biomagnification power" of the substance. A positive B-value indicates that biomagnification is occurring while a negative B-value indicates trophic dilution of chemical concentrations. B is similar to a TMF except that a TMF is expressed on an arithmetic, rather than logarithmic, basis and the TMF is based on a regression with trophic level (estimated from d¹⁵N) rather than the d¹⁵N content itself. The B-values for both food webs were not significantly different from zero, indicating that biomagnification of decaBDE did not appear to be occurring in these food webs. Failure to detect decaBDE in salmon from the Atlantic Ocean precluded a similar analysis for this food web.

Using the reported concentration data, it is also possible to estimate lipid-normalized BMFs for specific predator-prey combinations; these are summarized in Table 2-3. BMFs range from 0.03 to 5, depending on the predator-prey combination, suggesting that biomagnification was taking place in some predator-prey combinations. However, it is important to consider that the exact feeding relationships for these food webs are unknown, resulting in considerable uncertainty in these BMF estimates.

In evaluating the Burreau et al. (2004, 2006) studies, the United Kingdom (2007a) cautions that the relatively high levels of decaBDE in procedural blanks and low concentrations of decaBDE in biota samples create uncertainty in the overall biomagnification analysis. Currently, this appears to be a common issue with field studies of decaBDE in biota.

Law et al. (2006) conducted a field study of the trophic magnification of decaBDE in a pelagic food web of Lake Winnipeg. Samples of fish, plankton, mussels, sediment and water were collected from the south basin of the lake, offshore near Gimli, Manitoba. Muscle tissue from multiple fish species were collected between 2000 and 2002, including walleye (Stizostedion vitreum; n=5), whitefish (Coregonus clupeaformis; n=5), emerald shiner (Notropis atherinoides; n=5), burbot (Lota lota; n=5), white sucker Catostomus commersoni; n=5) and goldeye (Hiodon alosoides; n=3). Samples of net plankton (n=5; zooplankton and phytoplankton combined) were collected using horizontal tows with 160-µm nets (precise date not indicated in article). Mussels (Lampsilis radiata; n=5, muscle tissue retained for analysis) were collected by divers in 2002. Sediment grab samples were collected at 4 locations with only the surficial 2 cm of sediment retained. Water was sampled in 2004 using a Teflon column packed with XAD-2 absorbent. Each XAD-2 column was used to sample six 54-L samples from 324 L of water collected. Samples were pulled through an inline glass-fibre filter (1-μm pore size) and then onto a XAD-2 column.

All samples were analyzed for decaBDE (and several other chemicals) using gas chromatography-mass spectrometry (GC/MS), with additional analyses of organic carbon (OC) for sediments and lipid and d¹⁵N for biota. The d¹⁵N measurements were used to estimate trophic position. The detection limit of the analytical method used was 0.1 μg/kg for biota and sediment samples, and 15 pg/L for water. The decaBDE concentration, lipid and OC contents, and estimated trophic level of biota are summarized in Table 2-4.

Using the trophic levels estimated from d¹⁵N data, the rank order of the trophic levels in the pelagic food web was estimated to be mussel Þ zooplankton, whitefish Þ goldeye, white sucker Þ burbot, walleye (top predators). A regression of the lipid-normalized concentration of decaBDE vs. trophic level was used to estimate a TMF of 3.6 ^[3] (r²=0.46 p=0.0001) for decaBDE in the pelagic food web. Predator-prey BMFs (on a lipid-normalized basis) were also calculated using the biota data set. The estimated BMFs for decaBDE ranged from 0.1 to 34, depending on the predator-prey combination.

While estimated TMFs and BMFs are intended to provide a real-world indication of trophic magnification and biomagnification of decaBDE in an aquatic food web, it is important to consider uncertainties associated with this study.

Many of the concentrations of decaBDE in the Law et al. (2007) study were near the detection limit, increasing uncertainty in these determinations and raising the possibility of false positives. In addition, biomagnification was identified using lipid-normalized data; however, certain tissues were characterized by very low lipid concentrations (e.g., walleye, burbott and mussel muscle). Such low lipid contents result in extremely uncertain concentrations expessed on a lipid weight basis. When biomagnification is evaluated in this study on the basis of ww concentrations, biomagnification is not shown to occur.

There is further uncertainty respecting the appropriateness of lipid-normalization for decaBDE. This substance has been suggested to bind protein in some situations, although based on chemical structure, protein binding would not be expected. It is likely that this is a result of non-specific binding in blood plasma (i.e., lipids) transported to the liver (arterial and portal flow) (e.g., Han et al. 2007). Although there is some evidence that decaBDE could undergo non-specific binding in blood plasma causing preferential accumulation in liver, this has not been established definitively.

The United Kingdom (2007a) also highlighted the following issues regarding the Law et al. (2006) study:

• There is some uncertainty as to how well the d¹⁵N analysis characterizes trophic levels. The resulting trophic structure does not necessarily match that which would be expected based on species size and feeding characteristics (e.g., emerald shiner has a higher estimated trophic level than whitefish but based on size/life history, it would be expected to feed lower in the food web).
• There is uncertainty as to how well the samples represent the food web, since sample sizes were small in some cases, and they were collected at varying times between 2000 and 2002.
• Information regarding the actual feeding relationships of the analyzed species is lacking. It is unclear whether the species examined feed primarily on one another or whether other species that were not sampled could also be an important component of the food web.

In a study of in vivo and environmental debromination of decaBDE, La Guardia et al. (2007) monitored decaBDE concentrations in sediments and aquatic organisms in the receiving environment of a wastewater treatment plant (WWTP) located in Roxboro, North Carolina. All samples were extracted and purified using size-exclusion chromatography and then analyzed for PBDEs using GC/MS in electron capture negative ionization (ECNI) mode and electron ionization (EI) mode. Further study details are provided in Sections 3.1.2 and 3.2.1. In samples collected in 2002, decaBDE was detected in both sediments and tissues of sunfish (Lepomis gibbosus) and crayfish (Cambarus puncticambarus sp. c) collected immediately downstream of the WWTP outfall. The reported 2002 concentrations from this location in sediments, sunfish and crayfish were 1 630 000 mg/kg organic carbon (OC), 2880 mg/kg lipid and 21 600 mg/kg lipid, respectively. The much higher concentration in crayfish was attributed to the sediment-association of this species and the authors speculated that crayfish could form a link for the transfer of decaBDE from sediments to pelagic organisms.

Based on the La Guardia et al. (2007) results, estimated sediment BSAFs of 0.0018 for sunfish and 0.013 for crayfish can be estimated. These are well below the biomagnification (i.e., 1.7 to 3; refer to Section 1.2), suggesting that a combination of low sediment bioavailability and/or metabolic transformation is limiting the bioaccumulation and biomagnification of decaBDE in this system.

Wang et al. (2007) examined water, sediment and aquatic species collected from a small lake in Beijing, China, which receives effluent discharged from a large wastewater treatment plant. Samples were homogenized, extracted and analyzed using high-resolution gas chromatography / high-resolution mass spectrometry (HRGS/HRMS) using EI ion source. The researchers found that average accumulations of 12 PBDEs (total, tri- to heptaBDEs) and BDE209 were 6.33 and 237.01 mg/kg dw in sediments. BDE209 concentrations in lake water and effluent were below the analytical detection limit (not given for water; 1 mg/kg (ww or dw unknown) for sediment and biota). High concentrations of BDE209 were determined for lichen (1572 mg/kg dw), march brown (Limnodrilus hoffmeisteri; 11.37 mg/kg dw), coccid (114 mg/kg dw) and the zooplankton Monia rectirostris, Monia micrur and, Monia macrocopa (151.9 mg/kg dw). Average concentrations in common carp (Cyprinus carpio), java tilapia (Tilapia nilotica), leather catfish (Silurus meridionalis), crusian carp (Carassius auratus) and Chinese softshell turtle (Chinemys reevesii) were much lower, ranging from below detection to 19.32 mg/kg dw. Bioconcentration/bioaccumulation for BDE209 was not identified. In addition, the authors found no obvious biomagnification of PBDEs when they analyzed the relationship between PBDE concentrations and organism trophic level.

Xiang et al. (2007) sampled biota and sediment samples for PBDEs, including BDE209, from the Pearl River Estuary of China. In sediments they found that BDE209 was the dominant congener, ranging from 792 to 4137 ng/g OC in sediment samples (median 1372 ng/g OC). With respect to biota, they found non-detectable and measurable BDE209 concentrations in all biota species. Concentrations ranged up to 532.3 ng/g lipid in large yellow croaker (Pseudosiaena crocea; n=13, median=117.4 ng/g lipid), 623.5 ng/g lipid in silvery pomfret (Platycephalus argenteus; n=10, median=24.4 ng/g lipid), 38.4 ng/g lipid in flathead fish (Platycephalus indicus; n=17, median=0.0 ng/g lipid), 373.4 ng/g lipid in robust tongue fish (Cynoglossus robustus; n=8, median=0.0 ng/g lipid), 150.4 ng/g lipid in Bombay duck (Harpodon nehereus; n=9, median=0.0 ng/g lipid), 555.5 ng/g lipid in jinga shrimp (Metapenaeus affinis; n=10, median=0.0 ng/g lipid), 405.3 ng/g lipid in greasy-back shrimp (Metapenaeus crocea; n=10, median=30.3 ng/g lipid), and 88.5 ng/g lipid in mantis shrimp (Oratosquilla oratoria; n=9, median=42.47 ng/g lipid). The study notes that the high BDE209 concentrations in biota apparently resulted from elevated concentrations of BDE209 in local sediments. However, sediment BSAFs were calculated to range from 0 to 0.04 for BDE209 and trophic magnification was not deemed to be occurring based on the data shown in this study.

Eljarrat et al. (2007) reported the results of fish (n=29), sediment (n=6) and effluent (n=3) sampling conducted in November 2005 from the River Vero in Spain. They found high BDE209 concentrations in sediments (up to 12 459 ng/g dw) and fish -- barbel (Barbus graellsii) and carp (Cyprinus carpio), from non-detectable to 707 ng/g lipid -- downstream of an industrial park containing industries producing textiles and epoxy resins and involved with polymide polymerization. Using concentration measured in sediments and fish, the authors calculated sediment BSAFs for BDE209 of 0.0011 to 0.0013, thus suggesting that bioaccumulation was not occurring based on these data.

Marine Mammals and Terrestrial Species

Huwe and Smith (2007a,b) examined the dietary accumulation, debromination and elimination of decaBDE in rats. Sprague-Dawley rats (n=26) were dosed with a commercial decaBDE formulation (DE-83R 98.5% purity; 0.3 mg/g of diet) for a 21-day exposure period, which was followed by a 21-day elimination period. Following the 21-day exposure period, rats were sacrificed in groups of three for tissue analysis (liver, gastrointestinal (GI) tract, plasma, and remaining carcass) on days 0, 3, 7, 10, 14 and 21 of the elimination phase. Control rats were also sacrificed on these days to determine background PDBE concentrations (n=3 on day 0 and n=1 on all other days) and background values were subtracted from the PBDE determinations in the exposure group. Feces were collected from dosed rats daily during the exposure phase and pooled for analysis. Rat feed, feces and tissues were analyzed for a suite of PBDEs, including hepta- to decaBDE congeners.

Based on the analytical results, it was estimated that only 5% (or 3.6 mg) of the total decaBDE dose was retained in rat tissues following the 21-day dosing period, while approximately 50% was excreted to feces during this time. In addition to decaBDE, the authors concluded that one nonaBDE congener (BDE207) and two octaBDE congeners (BDE201 and BDE197) were derived from the uptake of decaBDE. However, the total burden of BDE207, -201 and -197 accounted for only 3% of the total decaBDE dose and 42% of the dosed decaBDE was unaccounted for in rat tissues and feces. The authors speculated that the formation of bound and/or hydroxylated metabolites which were not included in their analysis was a likely explanation for the incomplete mass balance of decaBDE.

Based on the observed carcass concentration of decaBDE a "BCF" (analogous to a BMF ^[4]) of 0.05 (on a ww basis, justified by the fact that the percent of lipid in food and carcass were similar) was calculated. It is not known whether decaBDE concentrations achieved steady state in rat tissues during the 21-day exposure period and, as a result, it is uncertain how well the reported BMF represents the potential steady-state value. It is also likely that the BMF based on a full accounting of decaBDE plus neutral, bound and or hydroxylated metabolites would be higher, but the concentrations of bound and hydroxylated metabolites were not reported. The study reported the half-life of decaBDE in rat tissue based on the observed elimination of decaBDE during the 21-day elimination period. First-order half-lives for decaBDE ranged from 3.9 d in plasma to 8.6 d in carcass. For the liver and plasma, second-order decay equations were found to represent the data well, with relatively rapid distribution phase half-lives of 0.7 and 1.2 d, respectively, but longer elimination phase half-lives of 20.2 and 75.9 d, suggesting potential persistence of decaBDE in rat tissues following chronic dosing. However, the authors cautioned that there was a high level of uncertainty in the second-order estimates.

Huwe et al. (2008) conducted a study of 29 male Sprague-Dawley rats to determine and compare the adsorption, distribution and excretion of PBDEs administered for 21 d as either a dust reference material or as a corn oil solution. The dust reference material (NIST Standard Reference Material 2585), containing a characterized and homogeneous composition of PBDEs, was mixed with rat chow. The corn oil solution contained commercial PBDE products DE-71, DE-79 and DE-83. These products were first dissolved in toluene and then mixed into a corn oil mixture. Daily doses were administered to the rats of either 1 or 6 µg/kg body weight (bw) of the dust/food or corn oil mixtures. The rats were randomly divided into five groups: controls (4 rats), low oil (4 rats), high oil (13 rats), low dust (4 rats) and high dust (4 rats). Most rats were dosed over a period of 21 d, and then were killed 24 hours after their last feeding. To assess whether rats were approaching a steady-state body burden, three groups of rats (n=3) from the high-oil-dose group were killed on days 3, 7 and 14 after dosing began. Sampling was conducted of the feed and oil mixtures, feces and tissues (epididymal fat, liver, kidney, brain, gastrointestinal tract and remaining carcass) for 15 PBDEs (BDE28/38, -47, -85, -99, -100, -138, -153, -154, -183, -186, -197, -203, -206, -207 and -209).

To determine whether steady-state body burdens were reached, only epididymal fat was sampled and analyzed. This analysis showed that all major tri- to octaBDEs had reached or were approaching steady state after 14 d, with no statistically significant differences between PBDE concentrations on days 14 and 21.

Retention of PBDEs in the body of rats was congener dependent and ranged from 4.0 to 4.8% of the dose for BDE209, and 10.1 to 22.6% for nonaBDEs, to approximately 69 to 78% for BDE47, -100, and -153, but did not generally differ between the dust and oil treatment groups. The study did not consistently detect nona- and decaBDEs in the adipose tissues above that of the controls. Urine contained less than 0.3% of any congener. Fecal excretion was the major route of elimination and was described in the study as that component of the dose not adsorbed. Fecal excretion was found to reach a steady state by day 2, with no statistically significant differences between mean concentrations in feces on days 2, 11 or 20. Excretion of BDE209 was approximately 68%, and ranged from 55.5 to 91.7% for the nonaBDEs. The amount of the BDE209 dose not adsorbed or excreted as parent compound ranged from 28% to 31.9%. Metabolic transformation products could have accounted for some portion of these percentages. Derived BCFs (analogous to BMFs as defined in this study) for adipose tissues were inversely related to the degree of bromination, and ranged from 7 to 24 for tri- to hexaBDE, 1 to 6 for hepta- to nonaBDEs and < 1 for decaBDE. For the liver tissues, BCFs for all PBDEs were below 1 except for BDE206 (BCF=2.4), and -207 (BCF=1.09). Hepatic Cyp2B1 and 2B2 mRNA expression increased in rats receiving the higher PBDE doses, suggesting potential effects of metabolic activity. The use of PBDE mixtures in this study made it impossible to determine whether metabolic debromination had occurred.

Kierkegaard et al. (2007) reported the findings of a 3-month feeding study with dairy cows. The study was originally undertaken to measure the long-term mass balance of PCBs; however, archived samples from two cows and feed were subsequently analyzed for a range of PBDEs. Over the 13-week study period, the cows were kept indoors and milk and feces were sampled once per week. The feed consisted of silage, concentrate and a mineral supplement which was not deliberately spiked with PBDEs. As a result, the PBDE concentration in food generally represented "background" contamination levels.

The milk and feces samples were pooled according to the following scheme to obtain a series of 5 composite samples representing discrete portions of the 13-week study period: three 3-week composites and two 2-week composites. In addition, one of the cows was slaughtered at the end of the 13-week study and samples of adipose tissue from 6 fat compartments as well as tissues from liver, kidney, heart and leg muscle were collected for analysis. Silage samples were retained for analysis at three intervals during the 13-week study, and concentrate and mineral samples were analyzed once during the study. All samples/composites of feed, milk, feces and tissue were analyzed for hepta- to nonaBDEs using hrms and were analyzed for decaBDE using LRMS in negative chemical ionization mode (limit of quantification = 0.4 to 150 pg/g lipid or dw).

DecaBDE (i.e., BDE209) was the dominant congener in all matrices except milk, suggesting that milk levels were influenced more by the existing burden of PBDEs in storage tissue rather than uptake from food. In addition, PBDE levels were higher in the adipose storage tissues than in organ tissues. Based on these observations, the authors proposed that the cows were in a state of PBDE elimination rather than accumulation and that the observed PBDE concentrations may have been influenced by exposure to PBDEs prior to initiation of the 13-week experiment. Although a mass-balance analysis of input and output fluxes of PBDEs was attempted, it was largely unsuccessful due to a large increase in octa-, nona- and decaBDE congeners in the second silage sample relative to the first and third samples. It was unclear whether this second sample accurately represented the cows’ exposure since the increase of PBDEs in feces did not appear to coincide with the greater increase in decaBDE concentration in feed. PBDE concentrations in concentrate and mineral supplements were much lower than in the silage and did not likely affect the overall mass balance conducted on the PBDEs in this study.

The Government of the United Kingdom (United Kingdom 2007a) conducted a critical analysis of the Kierkegaard et al. (2007) study and calculated dietary accumulation factors for cows from silage based on either adipose tissue or whole body and the average silage concentration or end silage concentration. They concluded that the high variation in silage PBDE concentration led to considerable uncertainty in the study results with respect to bioaccumulation. To estimate accumulation factors and BMFs on a lipid-normalized basis, they assumed a 4% lipid content in silage. Table 2-5 summarizes the calculated accumulation factors and BMFs for nona- and decaBDE congeners from the Kierkegaard et al. (2007) study.

The calculated BMFs and accumulation factors were well below 1 for decaBDE and only exceeded 1 for BDE207. For BDE207, it is uncertain whether the estimated values represent direct accumulation of BDE207 from food, or accumulation combined with bioformation as a result of the debromination of decaBDE. The United Kingdom (2007a) study proposed that, for chemicals such as decaBDE which undergo transformation once accumulated in organisms, BMF estimates should be based on the total burden of parent chemical and the metabolites resulting from accumulation and transformation of the parent chemical. In the case of the Kierkegaard et al. (2007) study, it is difficult to do so since the lower brominated PBDEs were also present in the feed and their presence in tissue could thus have resulted from both accumulation and bioformation. Furthermore, based on studies with rats, it is likely that alternative transformation pathways are also present in mammals which result in the presence of polar metabolites, bound residues and water-soluble residues. For a full accounting of the total chemical burden related to the accumulation of decaBDE, these would also have to be quantified.

Thomas et al. (2005) examined the absorption of decaBDE from diet by three captive juvenile grey seals (Halichoerus grypus). The captive seals were fed a constant diet of herring for 3 months prior to the initiation of the 3-month study (6 months total time). During the 3-month study, feeding with herring continued (1-2.5 kg/d), with all fish obtained from a single batch caught in the North Sea. The second month of the study involved a decaBDE exposure phase with the diet supplemented with 12 mg decaBDE per day, dissolved in a cod liver oil capsule. For the final month of the study, the diet was supplemented with the cod liver oil capsule only to measure elimination of accumulated decaBDE. Fish, blood and feces samples were collected and analyzed for PBDEs on a weekly basis throughout the 3-month period, while blubber biopsies were taken and analyzed for PBDEs 3 times during the study (beginning, 3 days after initiation of exposure phase, and end, after 29 days on a decaBDE-free diet).

The blood decaBDE concentration increased from non-detectable at the start of the exposure phase (day 28) to a maximum of approximately 1000 ng/g lipid (value read from graph) between 5 and 11 days after the end of the exposure phase. Concentrations of decaBDE in blubber ranged from non-detectable to 3.9 ng/g lipid on day 30 (3 days into the exposure phase) and from 3.4 to 7.4 ng/g lipid on day 83 (after 29 days on a decaBDE-free diet). The percentage of total ingested decaBDE estimated to be in blubber on day 30 ranged from 36 to 68% (for one of the three seals, decaBDE was not detected in blubber) and on day 83, from 11 to 15%.

During the elimination phase, the decaBDE half-lives in blood were estimated to be between 8.5 and 13 d, most likely due to a combination of metabolic transformation/elimination and transfer to blubber. The authors suggested that once stored in blubber, decaBDE was unlikely to be metabolized.

Based on the mass balance of measured input flux (in consumed diet) and output flux (in feces) of decaBDE, the authors determined an average absorption efficiency of 89% for decaBDE. The authors suggested that the high absorption efficiency called into question the theories regarding molecular size thresholds for chemical absorption. The relatively high apparent absorption efficiency was attributed to the following factors:

• Large fat reserves (i.e., blubber) which provide a reservoir for lipophilic chemicals.
• High food absorption efficiency in homeothermic carnivores such as seals, which creates strong a fugacity gradient in the gut, resulting in greater chemical uptake.
• Possible incomplete collection of feces; however, it was estimated that over a 24-hour period, not more than 10% of feces was lost.
• Degradation of decaBDE in the gut. The lower brominated PBDE congeners remained constant in feces despite the supplement with decaBDE during the second month, suggesting that degradation was not significant.

An additional explanation for the high apparent absorption efficiency which was not discussed by the authors is the potential formation and subsequent excretion of phenolic and/or methoxylated metabolites. Recent studies with Sprague-Dawley rats (Mörck et al. 2003, Huwe and Smith 2007a,b) have indicated that the formation of phenolic and methoxylated metabolites may be a more significant transformation pathway for accumulated decaBDE than debromination on its own. Thus, if phenolic or methoxylated metabolites were present but not analyzed in the feces, it is possible that the output flux was underestimated, resulting in an overestimate of the absorption efficiency of decaBDE. However, based on the high proportion of total ingested decaBDE estimated to be present in blubber on day 30 and 83 (i.e., up to 68%), the net absorption of decaBDE still appears to have been significant, and much higher than that observed in laboratory studies with fish.

By using the reported decaBDE concentrations in blood and blubber in conjunction with an estimated decaBDE concentration in food, it is possible to estimate blood-based and blubber-based BMFs for the Thomas et al. (2005) exposure study. Unfortunately, the authors did not report the lipid content of the herring consumed by the seals; however, based on values reported in the literature (e.g., Iverson et al. 2002, Jensen et al. 2007) a value of 10% lipid was used as a reasonable approximation for the BMF estimates summarized in Table 2-6. Note that the blood concentrations do not appear to have reached steady state during the exposure phase and it is likely that this also applies to the blubber concentrations. Furthermore, the blubber concentrations at day 30 and 83 are unlikely to represent maximum accumulation of decaBDE (maximum accumulation would be expected at or after the end of the exposure phase (at day 54). Thus, the blood-based and blubber-based BMFs underestimate the steady-state BMF to an unknown degree.

The calculation results indicate a relatively high (i.e., exceeding 1) BMF for blood and relatively low BMF for blubber. The relatively high blood-based BMF suggests that significant magnification could be occurring from the diet to blood and related tissues, indicating relatively high bioaccumulation potential. The blood BMFs reported by Thomas et al. (2005) may be somewhat uncertain because of other non-lipid constituents (proteins) that can attribute to the overall sorptive capacity of blood, thus making blood lipid normalization less accurate. The lower BMFs for blubber could be explained by the large storage capacity of fat tissues for lipophilic chemicals -- it is possible that only a small fraction of the potential steady-state concentration in blubber was reached during the 26-day exposure phase.

Sellström et al. (2005) analyzed decaBDE in soil and earthworm (species not identified), samples collected in 2000 from three research stations (with reference plots and sewage sludge amended plots) and two farms (reference and amended/flooded soils) in Sweden. Soil decaBDE concentrations at the various sites ranged from 0.015 to 22 000 ng/g dw and organic carbon content (based on loss-on-ignition data) ranged from 2.12 to 7.22%. Concentrations of decaBDE in worms ranged from 0.99 to 52 000 ng/g lipid. Using these data, Sellström et al. (2005) calculated site-specific soil BSAFs for co-occurring worm and soil samples. The estimated BSAFs ranged from 0.04 to 0.7 and averaged 0.3. Based on these results, the authors concluded that decaBDE was bioavailable in soils and could accumulate in earthworms, presenting an exposure pathway into the terrestrial food web. The authors did not observe any evidence of photolytic debromination in soils.

The soil BSAFs determined by Sellström et al. (2005) were all below the range that might provide evidence of decaBDE biomagnification (i.e., 1.7 to 3; refer to Section 1.2).

As part of their study of grizzly bears from British Columbia, Christensen et al. (2005) conducted a bioaccumulation analysis for decaBDE. Their method involved the estimation of a "bioaccumulation slope" based on a comparison of decaBDE concentration with the proportion of meat in the diet. The rationale was that if trophic magnification was causing an increase in decaBDE concentrations in prey above that in consumed vegetation, then the concentration of decaBDE in bears would increase as the proportion of prey-derived meat in their diet increased. A positive bioaccumulation slope indicates trophic magnification while a negative bioaccumulation slope indicates that trophic dilution was taking place. The authors indicated that the bioaccumulation slope for decaBDE was negative (albeit not significantly different from 0) but did not provide the value. The lack of a significant positive bioaccumulation slope for decaBDE suggests that decaBDE was not undergoing trophic magnification in the studied grizzly bear food webs based on a diet of meat. The study did not examine the potential significance of decaBDE exposure via consumed vegetation or inhaled air.

In their study of decaBDE in the marine food web of Svalbard, Norway, Sørmo et al. (2006) also attempted to estimate BMFs for predator and prey species. Unfortunately, the high number of non-detected concentrations precluded the estimation of BMFs for polar bear / ringed seal or ringed seal / polar cod combinations. The estimated BMF for polar cod / ice amphipod based on the mean concentrations in each species (except for ice amphipod for which there was only one sample) were 0.1 (wet weight basis) and 0.03 (lipid weight basis), indicating a lack of biomagnification for this predator-prey combination.

Published models exist for predicting bioaccumulation in aquatic food webs and biomagnification in terrestrial mammals. The BAF-QSAR model described by Arnot and Gobas (2003) has generic applicability to the Canadian environment and a modified version of this model was applied during the Government of Canada’s categorization of its Domestic Substances List. This model predicts both BAFs and BCFs for three representative fish trophic levels (low, middle and upper) in a generic aquatic food web based on a standard set of conditions found in the Canadian environment. For the terrestrial environment, Gobas et al. (2003) describe a terrestrial biomagnification model for adult male wolves (Canis lupus) based on the work of Kelly and Gobas (2003). It predicts BMFs for wolves based on chemical log K_oa (logarithm of octanol-air coefficient), chemical log K_ow (logarithm of octanol-water coefficient), and a set of parameters describing the lichen-caribou-wolf food web of Bathurst Inlet in the Canadian Arctic. Both of these models incorporate a metabolic rate constant as part of chemical elimination, allowing for metabolism correction of BAF, BCF and BMF predictions based on field or laboratory observations.

BAF and BCF predictions for fish were made for decaBDE using the BAF-QSAR model. Environment Canada’s review of log K_ow values for decaBDE revealed that a log K_ow of 8.7 reported by Wania and Dugani (2003) represents the most reliable value achieved analytically. For further rationale regarding the selection of log K_ow values, refer to Appendix C. Two prediction scenarios were conducted: the first with no correction made for metabolic transformation, and the second corrected for metabolism based on the laboratory observations of Tomy et al. (2004). Tomy et al. determined a half-life of 26 d in juvenile lake trout fed a diet containing decaBDE, which is consistent with a total elimination rate (k_T) of approximately 0.027/d. The Tomy et al. (2004) value was used to derive an in vivo-based metabolic rate constant (k_M) according to the method of Arnot et al. (2008b). In this method, when k_T is available, k_M is derived according to the following equation:

where:

The method of Arnot et al. (2008b) provides for the estimation of confidence factors (CFs) for the k_M to account for error associated with the in vivo data (i.e., measurement variability, parameter estimation uncertainty and model error). A CF of ±3.0 was calculated for the available BMF data.

Because metabolic potential can be related to body weight and temperature (e.g., Hu and Layton 2001, Nichols et al. 2006), the k_M was further normalized to 15°C and then corrected for the body weight of the middle trophic level fish in the Arnot-Gobas model (184 g) (Arnot et al. 2008a). The middle trophic level fish was used to represent overall model output as suggested by the model developer (personal communication from JA Arnot to Mark Bonnell, Environment Canada, 15 February 2008; unreferenced) and is most representative of fish weight likely to be consumed by an avian or terrestrial piscivore. After normalization routines, the k_M ranges from 0.02 to 0.17 with a median value of 0.06.

The BAF and BCF predictions for the middle trophic level fish for decaBDE are summarized in Table 2-7. All predicted BCF values are below 5000, which is expected given that uptake and elimination via the gill (which BCF accounts for) is limited and only important for substances with a log K_ow of approximately less than 4.0.

The predicted BAF, when corrected for metabolic transformation, ranges from 4056 to 161 618 depending on the rate of metabolism. The predicted BAF for an average k_M, which can be said to represent the typical fish metabolic potential in the Canadian environment, was calculated to be 29 386. When a default of no metabolism is used in the model, the BAF is several orders of magnitude higher than the BAF calculated with average metabolic rate potential. These results demonstrate the influence of both chemical partitioning behaviour (i.e., log K_ow) and metabolic transformation on the bioaccumulation potential of decaBDE.

The metabolism-corrected BAFs probably provide the best estimate of the bioaccumulation potential of decaBDE since metabolic transformation of decaBDE has been demonstrated or inferred in most laboratory studies. It is important to note that these corrected BAFs could underestimate the total chemical bioaccumulation related to decaBDE since they are based on parent chemical only and do not account for the additional presence of metabolites in tissues.

BMF predictions for wolves were made using a spreadsheet version of the Gobas et al. (2003) model. In addition to the range of potential log K_ow values described for the BAF-QSAR model in Appendix C, two log K_oa estimates were available: 15.27 (Tittlemier et al. 2002) and 18.423 (predicted by the QSAR model, KOAWIN). However, the BMF predictions do not vary significantly for log K_oa or log K_ow values in these ranges and systemic variation of log K_oa and log K_ow made little difference in the predicted output. Four prediction scenarios were conducted: the first with no correction for metabolic transformation, and the remaining three corrected for metabolism based on the laboratory observations of Huwe and Smith (2007a,b). They determined a range of potential half-lives for decaBDE in rats based on a combination of first-order and second-order approximations to elimination data for carcass, liver and blood plasma. For carcass, a first-order half-life of 8.6 d was found to best represent the elimination data. For blood plasma and liver, second-order models (i.e., with distribution and elimination phases) provided the better fit, with elimination phase half-lives (representing slower elimination from residual body stores) of 75.9 and 20.2 d, respectively. These half-lives were used to infer k_M values using a similar method as with the BAF-QSAR model, resulting in k_M ranging from 0.0086 to 0.08/d. Because these rates are based on rodent exposures, they also must be scaled to the body weight of the wolf (e.g., Hu and Layton 2001). This results in k_M values of 0.004 to 0.03 asuming a rat weight of 0.25 kg and the wolf model body weight of 80 kg (Hu and Layton 2001, Arnot et al. 2008b).

The terrestrial BMF predictions for decaBDE are summarized in Table 2-8. Dietary assimilation efficiency (E_D) of ~6% was calculated as outlined in Kelly et al. (2004) for substances with high log K_ow values. In the absence of a correction for metabolism, the predicted BMF was ~10 and when corrected for metabolism BMF predictions were <1. The metabolism-corrected BMFs probably provide the best estimate of the biomagnification potential of decaBDE since metabolic transformation of decaBDE has been demonstrated or inferred in most laboratory and captive feeding studies reviewed from the open literature. Thus, corrected BMFs suggest a lack of potential for biomagnification largely as a result of low dietary efficiency and some metabolism of decaBDE.

It is important to note that the corrected BMFs could underestimate the total chemical biomagnification related to decaBDE since they are based on parent chemical only and do not account for the additional presence of metabolites in tissues. Thus, if all metabolites were included in the BMF calculations, it is possible that all predicted BMFs might be higher.

This section provides a summary of evidence which is currently available regarding the bioaccumulation and biomagnification of decaBDE. It is intended to synthesize the existing state of the science regarding decaBDE biomagnification. The available evidence respecting the bioaccumulation of decaBDE is grouped into categories according to whether it is considered to provide equivocal support for, or does not support, the conclusion that decaBDE is "bioaccumulative" or may biomagnify in food chains. Unequivocal evidence supporting the conclusion that decaBDE is bioaccumulative and meets the criteria for bioaccumulation under the Persistence and Bioaccumulation Regulations, or is biomagnifying in food chains, was not available. Additional considerations and interpretations on the capacity for decaBDE to bioaccumulate and/or biomagnify are also summarized in a third section.

1. Equivocal ^[5] evidence with respect to whether decaBDE has significant potential to bioaccumulate or biomagnify in the environment:

• The findings of a Lake Winnipeg pelagic food web study by Law et al. (2006) led to an estimated TMF of 3.6 for a Lake Winnipeg food web, with many BMFs exceeding 1. These observations were based on parent decaBDE only. If both decaBDE plus metabolites were considered, it is possible that the TMF and BMFs might be higher. However, this study had a number of limitations that make the results less certain.
• A feeding study by Stapleton et al. (2006) using juvenile rainbow trout found tissue-specific BMFs which ranged up to 1.74 for decaBDE and neutral metabolites. Other metabolites which were not measured may have also been present. However, BMFs based on serum and carcass data did not exceed 1.
• Two studies by Burreau et al. (2004, 2006) on food webs of the Baltic Sea evaluated biomagnification in two Baltic Sea food webs and found that biomagnification of decaBDE did not appear to be occurring. Failure to detect decaBDE in salmon from the Atlantic Ocean precluded a similar analysis for this food web. In addition, the relatively high levels of decaBDE in procedural blanks and low concentrations of decaBDE in biota samples created uncertainty in the overall biomagnification analysis.
• In a feeding study using captive seals by Thomas et al. (2005), blood-serum-based BMFs exceeded 1 but blubber BMFs were well below 1. Blubber concentrations are unlikely to have reached steady state. The steady-state BMFs for blubber could be higher. However, this study reported an unusually high calculated dietary assimilation efficiency of 89%, which would not be expected for decaBDE.
• A feeding study with cows by Kierkegaard et al. (2007) found that all cow-silage BMFs were less than 1. However, a full accounting of parent decaBDE plus metabolites was not conducted. There was significant uncertainty in the results of this study due to the high variability of decaBDE concentrations in the silage feed.
• A feeding study using juvenile lake trout by Tomy et al. (2004) determined a BMF of 0.3, using the kinetic method based on parent decaBDE only. If metabolites were included in the BMF estimate, it may have been higher.
• Although high concentrations of decaBDE have occasionally been observed in birds of prey (especially in China and Europe), red fox, marine mammals and a few marine bird species, these high observed concentrations are confounded by the potential that the sampled species inhabit decaBDE hotspots close to industrialized areas of Europe and China. Thus, if environmental exposures were very high, organisms could still achieve very high concentrations in their tissues even though BAFs or BMFs were very low.
• Relatively high concentrations of decaBDE were observed in peregrine falcons in Greenland and in sharks of coastal Florida. Although these studies do not quantify BAFs or BMFs, they lend support that decaBDE appears to be bioaccumulating to relatively high levels in some top predator organisms.
• Using the BAF-QSAR model, metabolism-, weight- and temperature-normalized aquatic BAF predictions range from approximately 4000 to 16 200 and reflect the uncertainty associated with estimating metabolic rates in fish. The average BAF corrected for metabolism was calculated to be approximately 30 000. If the BAF-QSAR model also accounted for the body burden of metabolites, then the BAF estimates could potentially be higher than the range predicted for the parent compound.

2. Evidence that does not support ^[6] the conclusion that decaBDE has significant potential to bioaccumulate or biomagnify in the environment:

• Using the Gobas et al. (2003) model, metabolism-corrected terrestrial BMF predictions (based on a wolf food chain) are less than 1, mainly due to low dietary assimilation efficiency of decaBDE. It is possible that, if the BMF model also accounted for the body burden of metabolites, then the BMF estimates could be higher and potentially exceed 1. It should be noted that it is difficult to compare modelled BMF with any of the available measured BMFs because none of them are reported for the wolf; thus, any comparisons involve uncertainties due to different species with different metabolic rates and body weights.
• A feeding study using rats by Huwe and Smith (2007a,b) determined a BMF of 0.05, while a study by Huwe et al. (2008) obtained a derived BMF of less than 1. A full accounting of parent decaBDE and metabolites could have increased the BMF estimates.
• A study by Christensen et al. (2005) on grizzly bears suggested that biomagnification is not occurring.
• A study by Sørmo et al. (2006) for a marine food web in Svalbard, Norway, identified only very low concentrations and a very low BMF (0.03) for the polar cod / ice amphipod predator-prey relationship. Other BMFs could not be calculated because of the high frequency of samples in which decaBDE was not detected, suggesting a low potential for biomagnification in general in this food web.
• A study by Sellström et al. (2005) of the accumulation of decaBDE in earthworms from soil found that soil BSAFs were well below the threshold range (>1.7 to 3) that would indicate biomagnification above equilibrium conditions.
• Relatively low and often non-detected concentrations of decaBDE were found for a wide range of middle to top predators including some marine fish, mammalian predators (grizzly bear, polar bear and lynx), marine mammals, marine/aquatic birds and birds of prey (multiple studies, often the same ones as those where relatively high concentrations were also observed).
• Calculated sediment BSAFs determined by La Guardia et al. (2007), Xiang et al. (2007) and Eljarrat et al. (2007) were well below thresholds indicative of biomagnification above equilibrium conditions.
• There was minimal observed uptake of decaBDE by Lumbriculus variegatus exposed to decaBDE in biosolids and artificial spiked sediments (Ciparis and Hale 2005).

3. Additional considerations and evidence:

• A BCF study by MITI (1992) resulted in an estimated BCF<3000. Note that this BCF value was based on the estimated water solubility of decaBDE and the detection limit in tissues and therefore, is highly uncertain.
• There was low chemical uptake efficiency and evidence of metabolism in carp (Stapleton et al. 2004).
• There was low chemical uptake efficiency and evidence of metabolism in rainbow trout (Stapleton et al. 2006, Kierkegaard et al. 1999).
• Metabolism studies using rats (Norris et al. 1973, 1974; El Dareer et al. 1987) suggested that decaBDE has low bioaccumulation potential in mammalian species.

The existing evidence for the bioaccumulation of decaBDE does not support a conclusion of "bioaccumulative" as defined in the current Persistence and Bioaccumulation Regulations under CEPA 1999. There is no measured or experimental evidence considered to support a conclusion that decaBDE, as the parent compound, has a significant potential to bioaccumulate or biomagnify in the environment. The metabolism-corrected, model-predicted aquatic BAFs range from below the 5000 criterion to well above 5000, demonstrating the uncertainty associated with metabolism potential for decaBDE in fish. Although less relevant than BAF or BMF, experimental BCF measures are below the 5000 criterion. Nevertheless, the substance is shown to be accumulating at rapid rates to high levels in some wildlife species. decaBDE (i.e., BDE209) is also considered to be contributing to the bioaccumulation potential of total PBDEs as a result of metabolism to lower brominated forms (to be discussed later in this report).

^[1] BMF = aF/k_d where α is the absorption efficiency, F is the feeding rate on a lipid basis and kd is the total elimination rate constant
^[2] The disparity between the n of samples and the n of analysis is explained by the fact that only the samples in which the congener was detected are reported.
^[3] Note that this is the corrected value published in Law et al. (2007).
^[4] The authors used the term "BCF"; however, given that this value was based on a comparison between rat tissue and food concentrations, it is actually analogous to a BMF.
^[5] Studies cited here are considered by Environment Canada to provide uncertain evidence for decaBDE to bioaccumulate or biomagnify in the environment.
^[6] Studies cited here indicate that decaBDE does not have the potential to bioaccumulate or biomagnify in the environment; these studies include both reliable and less certain evidence.