Environ. Sci. Technol. 2010, 44, 2466–2473
Temporal Variations of Polybrominated Dibenzo-p-Dioxin and Methoxylated Diphenyl Ether Concentrations in Fish Revealing Large Differences in Exposure and Metabolic Stability ¨ FSTRAND,‡ P E T E R H A G L U N D , * ,† K A R I N L O ¨ R N , ‡,| A N D E R S B I G N E R T , § ANNA MALMVA AND LILLEMOR ASPLUND| Department of Chemistry, Umeå University, SE-901 87 Umeå, Sweden, Environmental Chemistry Unit, Department of Materials and Environmental Chemistry, Stockholm University, SE-106 91 Stockholm, Sweden, Contaminant Research Group, Swedish Museum of Natural History, Box 50007, 104 05 Stockholm, Sweden, and Department of Applied Environmental Science, Stockholm University, SE-106 91 Stockholm, Sweden
Received December 15, 2009. Revised manuscript received February 3, 2010. Accepted February 5, 2010.
The concentrations of polybrominated dibenzo-p-dioxins (PBDDs) and polybrominated methoxylated diphenyl ethers (MeOPBDEs) were investigated in perch (Perca fluviatilis) collected from a Baltic Sea background contaminated area between 1990 and 2005. No temporal trend was found, but large variations were observed s up to 5-fold and 160-fold differences in MeOPBDE and PBDD concentrations, respectively s between consecutive years, suggesting that retention of these compounds, particularly the PBDDs, is limited. Examination of the congener profiles using principal component analysis (PCA) and correlation analysis indicated that MeO-PBDEs without adjacent substituents (6-MeO-BDE47) or with two adjacent substituents (2′-MeO-BDE68 and 6-MeO-BDE90) are retained more than MeO-PBDEs with three adjacent substituents (6-MeO-BDE85 and 6-MeO-BDE99) and that 1,3,6,8-tetraBDD and 1,3,7,9-tetraBDD are retained more than the other PBDDs which have vicinal hydrogen. Debromination could explain the limited retention of 6-MeO-PBDE85 and 6-MeO-BDE99 and the absence of 2-MeOBDE123 and 6-MeO-BDE137, and cytochrome P-450 mediated oxidation could explain the limited retention of PBDDs containing vicinal hydrogen. The levels of organobromines, especially MeO-PBDEs, were found to covary with water conditions related to primary production, for example temperature, depth visibility, and inorganic nutrient concentrations, which also favor fish productivity. The results suggest natural production of MeO-PBDEs and PBDDs and imply that they fluctuate * Corresponding author phone: +46-90-7866667; fax: +46-907867655; e-mail:
[email protected]. † Umeå University. ‡ Department of Materials and Environmental Chemistry, Stockholm University. | Department of Applied Environmental Science, Stockholm University. § Swedish Museum of Natural History. 2466
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 7, 2010
considerably over time, as do common marine toxins in fish. Thus, assessments of human and environmental risk should consider both the average and peak concentrations of these contaminants in marine biota.
Introduction Baltic Sea ecosystems are influenced by nutrient enrichment and heavy pollutant loads (1) which lead to harmful algal blooms (2) and increased mortality and reproductive disorders in seabirds (2, 3) and fish (2, 4). The salinity of water in the Baltic Sea decreases with distance from the Atlantic, from 30 to 3 practical salinity units (psu), due to dilution with freshwater, creating stress for marine and freshwater fish that inhabit the area and increasing their susceptibility to toxic compounds. Furthermore, in several studies preceding the present investigation high levels (ng-µg/g lipids) of polybrominated dibenzo-p-dioxins (PBDDs), which have the potential to increase chemical stress, were detected in Baltic Sea fish and bivalves (5-7). PBDDs and polybrominated dibenzofurans (PBDFs) have been detected in many samples of abiotic origin. Buser reported high levels (up to 10%) of PBDD/Fs in pyrolysates of polybrominated diphenyl ethers (PBDEs) (8), and Thoma et al. detected PBDD/Fs in technical PBDE mixtures, but at lower concentrations (9). PBDD/Fs have also been detected in exhaust emissions from cars run on leaded gasoline (10), incinerator flue gases (11), soot and smoke from accidental fires (12), and emissions from the open burning of electronic waste (13). It is clear that there have been, and still are, anthropogenic emissions. It is therefore not surprising that PBDD/Fs, primarily highly brominated dibenzofurans congeners, have been detected in sediment from Swedish lakes (14) and in marine sediments from Japan (15), Korea, and Hong Kong (16). In the latter study, the congener patterns in samples from industrial regions where PBDEs were in use differed significantly from those in samples from remote locations. The congener patterns from the industrial regions were dominated by tetra-heptaBDFs, which are commonly formed by thermal stress, pyrolysis, or incomplete combustion of octa- and decaBDEs (17). PBDDs, and especially tetrabromodibenzo-p-dioxins (TeBDDs), were most abundant in samples from remote locations. It has been suggested that the PBDDs with low bromination are primarily of natural origin (7) because structurally related compounds have been found to be naturally produced, including bromophenols (BPs), hydroxylated polybrominated diphenyl ethers (OH-PBDEs), methoxylated PBDEs (MeO-PBDEs), hydroxylated PBDDs (OH-PBDDs), and methoxylated PBDDs (MeO-PBDDs) (18-20). Furthermore, MeO-PBDEs, OH-PBDEs, and PBDDs have been found in several primary production organisms, such as the red algae Ceramium tenuicorne and Polysiphona fucoides, brown alga Pilayella littoralis, the aquatic sponge Ephydatia fluviatilis, and the cyanobacteria Aphanizomenon flos-aquae and Nodularia sprumigena (5-7, 21-24). The present study was largely prompted by a finding that concentrations of PBDDs in blue mussels (Mytilus edulis) had increased over time (7). The aims of the study were to investigate whether the concentrations of PBDDs and MeOPBDEs are increasing in fish from the Baltic Proper, to study possible temporal variations and the factors that affect them, and to determine whether the concentrations of PBDDs and MeO-PBDEs are correlated, which could indicate a common source. 10.1021/es9038006
2010 American Chemical Society
Published on Web 02/24/2010
Materials and Methods Samples. Samples of perch (Perca fluviatilis) and flounder (Platichtys flesus) were collected using gill nets from a background location on the Swedish Baltic Proper coast (Kva¨do¨fja¨rden; 58° 1′ 59 N, 16° 51′ 0 E) within the Swedish environmental monitoring program. All samples were collected during late summer or autumn to ensure that the sampled individuals were well nourished and were not reproducing. The perch were also selected by age (2-years). Fish muscle (without skin and subcutaneous fat) was used for analysis. Composite samples were prepared from ten individuals in order to reduce the effect of individual variations in pollutant concentrations among fish. The perch from 1990-2004 were retrieved from the Swedish Environmental Specimen Bank. In addition, a sediment core sample was collected in 2005 by the Geological Survey of Sweden using a core sampler and was sectioned on board the sampling vessel. Top sediment was used in the present study. All samples were stored frozen (-80 °C) until analysis. Extraction and Clean-up. Two sets of fish samples were extracted, one for MeO-PBDE analysis and one for PBDD analysis. The samples were spiked with internal standards (ISs), extracted, and cleaned up as described elsewhere (6, 7) with minor modifications. Recovery standards (RSs) were added, and the samples were analyzed by gas chromatography/mass spectrometry (GC/MS). PBDDs were analyzed using electron ionization (EI) and high resolution mass spectrometric detection (HRMS; R > 10,000) while MeOPBDEs were analyzed using MS in the electron capture negative ion chemical ionization (ECNI) mode. The analyses were focused on the homologues that were found at highest concentrations in previous studies of biota from the region (6, 7, 25, 56), i.e. tetra- to hexa-MeO-PBDEs and di- to tetraBDDs. It is worth noting that previous screening of perch from the region did not detect any PBDFs (7). The methods used are further described in the Supporting Information. Statistical Analysis. Prior to statistical analysis all PBDD congeners for which results were not obtained for more than 25% of samples were excluded, and in the remaining data set all congeners that were not detected were assigned a value of half the detection limit (which was 0.005-0.05 pg/g fresh weight). The lipid contents (0.86% ( 0.15%) were uncorrelated to the total concentrations of PBDDs and MeO-PBDEs, respectively, and this parameter not considered in the modeling. Prior to principal component analysis (PCA) the individual PBDD and MeO-PBDE concentrations were normalized to the total concentration of each class of compounds, and the data were scaled to unit variance and mean-centered to weight all variables equally. PCA was used to study the covariation within and between the variables (fractions of individual PBDDs and MeO-PBDEs) and objects (perch samples from different years). Cross-validation was applied to determine the number of significant components and to calculate the fraction of the total variation in the data (normalized concentrations of pollutants) that could be predicted by the model (Q 2). A biplot was generated to visually represent the covariations between all variables and objects. Spearman rank correlation coefficients were also calculated for the perch data to identify statistically significant correlations between individual analyte concentrations in perch. This technique was chosen because the relationship between variables was expected to be monotonic but not necessarily linear. The results from the correlation analyses were compiled in a correlation matrix. The significance level (R) was set to 0.05.
Results and Discussion Levels and Temporal Variations of PBDDs and MeOPBDEs. The most abundant congeners in the perch samples
have previously been detected in biota from the Baltic Proper (7, 25). The most abundant target compounds detected were 6-MeO-BDE47, 2′-MeO-BDE68, 6-MeO-BDE85, 6-MeOBDE90, 6-MeO-BDE99 and 2,7/2,8-diBDD (DBDD), 1,3,7triBDD (TrBDD), 1,3,8-TrBDD, and 1,2,4,7- and 1,2,4,8tetraBDD (TeBDD) (Figures 1 and 2 and Table S1, Supporting Information). The lipid weight adjusted MeO-PBDE and PBDD concentrations (34 and 0.48 ng/g lipids) were of the same magnitude as previously reported for fish from the Baltic Proper area (7, 26). However, 2-MeO-BDE123 and 6-MeO-BDE137 were not detected in the perch samples although they have been detected at considerable concentrations (ng/g lipids) in mussels from the Baltic Proper (6) (Figure 3A). In contrast to the previously reported increase in PBDD concentrations in mussels over time (7), there was no obvious temporal trend in MeO-PBDE or PBDD concentrations between 1990 and 2005. Considerable variations in concentration were observed for both classes of compounds (Figure 2), with up to 5-fold and 160-fold differences for MeO-PBDEs and PBDDs, respectively, in consecutive years. The variations in PBDD concentration were much greater than the decreasing temporal trend for PCBs in the same area, as illustrated in Figure 2. The data support the hypothesis of natural formation because similar fluctuations in concentration have been reported for marine toxins, including BPs, paralytic shellfish poisoning (PSP), and diarrheic shellfish poisoning (DSP) toxins (27-29). Since such large variations can confound temporal trends, studies over longer time periods will be required to provide conclusive data. Relationships between PBDD and MeO-PBDE Concentrations and Patterns. The concentrations of both compound classes appeared to vary in a cyclical manner, and there seemed to be a weak covariation (Figure 2), but also major differences, including elevated levels of PBDDs in 1996 and of MeO-PBDEs in 1994. The absence of a strong, direct correlation might indicate different sources, but there could be an indirect link if PBDDs and MeO-PBDEs are formed as secondary products in the natural production of OH-PBDEs by enzymatic (30) or photolytic (31) coupling and biomethylation (32), respectively, or if they are transformed further, e.g. by debromination (33). There may also be differences in half-lives between the classes of PBDDs and MeO-PBDEs and between individual congeners that would influence their relative abundances after an exposure event. To fully evaluate the congener profiles and identify possible relationships between the patterns and total concentrations, the PBDD patterns were ranked according to their total concentration (Figure 3C), and some general features became apparent. TrBDDs predominated in samples with high total concentrations (75-80% contribution), and their proportional contribution decreased (down to 50%) with decreasing total concentration. Conversely, the TeBDDs made only minor contributions (5-10%) in high level samples but high contributions (30-40%) in low level samples. The contribution from the DBDDs was relatively constant at approximately 10% of the total concentration. The covariations in the PBDD and MeO-PBDE fractions of perch from 1990 to 2005 were investigated further using PCA. The resulting model had two significant principal components that explained 91% (63% and 29%) of the overall variance in the data and had moderate predictive power with a Q 2 of 0.60. The results were presented as a biplot (Figure 4) which demonstrated the expected separation of MeOPBDEs and PBDDs, due to the relatively high levels of MeOPBDEs in perch from 1991 and 1994 and the dominance of PBDDs in samples from 1996. Further inspection revealed a separation within each class and the clustering of two MeOPBDEs (6-MeO-BDE47 and 6-MeO-BDE85) and two PBDDs (1,3,6,8-TeBDD and 1,3,7,9-TeBDD) between the two groups VOL. 44, NO. 7, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2467
FIGURE 1. Chemical structures of MeO-PBDEs and PBDDs found in high abundances in perch (left and upper right, respectively) and of MeO-PBDEs found in high abundance in mussels (6) but not detected in perch (lower right). of compounds. Spearman rank correlation coefficients gave a similar picture (Table S2). All DBDDs and TrBDDs were strongly correlated (rs ) 0.86-0.97). The TrBDDs were correlated to the TeBDDs (rs ) 0.78-0.90) and to 6-MeOBDE47 and 6-MeO-BDE85 (rs ) 0.82-0.88). 1,3,6,8-TeBDD was correlated to 1,3,7,9-TeBDD (rs ) 0.77) and also to 6MeO-BDE47 and 6-MeO-BDE85 (rs ) 0.72-0.84). MeOBDE47 and 6-MeO-BDE85 were correlated (rs ) 0.88), and 2′-MeO-BDE68, 6-MeO-BDE90, 6-MeO-BDE99 were correlated (rs ) 0.61-0.83). All these correlations were statistically significant (Spearman critical value 0.54). Thus, both the PCA and the correlation analysis suggested considerable covariation between 1,3,6,8-TeBDD, 1,3,7,9-TeBDD, 6-MeOBDE47, and 6-MeO-BDE85. Influence of Exposure Routes and Metabolism on the PBDD Congener Distribution. It is possible that perch are exposed to PBDDs by different pathways. The most abundant PBDDs (1,3,7-TrBDD and 1,3,8-TrBDD) and MeO-PBDEs (2′-MeO-BDE68 and 6-MeO-BDE90) have significantly different lipophilicities. Their octanol-water partition coefficients (Log Kow) have been estimated to be 6.3 (1,3,7-TrBDD), 6.4 (1,3,8-TrBDD) (34), 7.2 (2′-MeOBDE68), and 7.4 (6-MeO-BDE90) (35). Food web transfer may be greater for the MeO-PBDEs, and direct uptake via 2468
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 7, 2010
the gills may be significant for the TrBDDs and even greater for the DBDDs with Log Kow values in the range 5.5 to 5.9 (35). All of the DBDDs and TrBDDs fulfill the criteria for classification as “swimmers” according to the definition of Wania, i.e. they have Log Kow
