Chemosphere 106 (2014) 36–43

Contents lists available at ScienceDirect

Chemosphere journal homepage: www.elsevier.com/locate/chemosphere

Evidence for photochemical and microbial debromination of polybrominated diphenyl ether flame retardants in San Francisco Bay sediment Lisa A. Rodenburg a,⇑, Qingyu Meng b, Don Yee c, Ben K. Greenfield c,1 a b c

Department of Environmental Sciences, Rutgers University, 14 College Farm Road, New Brunswick, NJ 08901, United States School of Public Health, Rutgers University, Piscataway, NJ 08854, United States San Francisco Estuary Institute, 4911 Central Avenue, Richmond, CA 94804, United States

h i g h l i g h t s  The analysis used a data set of 24 BDE congeners in 233 samples.  Factor analysis resolved five factors.  Two factors resembled the deca and penta commercial BDE formulations.  Two factors contained di and tribromo congeners indicative of debromination.  Photolytic and microbial debromination are thought to explain these two factors.

a r t i c l e

i n f o

Article history: Received 19 July 2013 Received in revised form 3 December 2013 Accepted 30 December 2013 Available online 28 January 2014 Keywords: Photolysis Debromination Monitoring Factor analysis California Flame retardant

a b s t r a c t Brominated diphenyl ethers (BDEs) are flame retardant compounds that have been classified as persistent organic pollutants under the Stockholm Convention and targeted for phase-out. Despite their classification as persistent, PBDEs undergo debromination in the environment, via both microbial and photochemical pathways. We examined concentrations of 24 PBDE congeners in 233 sediment samples from San Francisco Bay using Positive Matrix Factorization (PMF). PMF analysis revealed five factors, two of which contained high proportions of congeners with two or three bromines, indicating that they are related to debromination processes. One of the factors included PBDE 15 (4,40 -dibromo diphenyl ether, comprising 20% of the factor); the other included PBDE 7 (2,4-dibromo diphenyl ether; 12%) and PBDE 17 (2,20 ,4-tribromo diphenyl ether; 16%). The debromination processes that produce these congeners are probably photochemical debromination and anaerobic microbial debromination, although other processes could also be responsible. Together, these two debromination factors represent about 8% of the mass and 13% of the moles of PBDEs in the data matrix, suggesting that PBDEs undergo measurable degradation in the environment. Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction Brominated diphenyl ethers (BDEs) are flame retardant compounds that have been widely used in consumer products since the 1970s, exhibit elevated concentrations in seafood and indoor dust, and may pose human health hazards at environmentally relevant concentrations (Domingo, 2012). As a result, PBDEs have been classified as persistent organic pollutants (POPs) under the Stockholm Convention and are targeted for phase-out (United ⇑ Corresponding author. Tel.: +1 732 932 9800x6218; fax: +1 732 932 8644. E-mail address: [email protected] (L.A. Rodenburg). Current address: Environmental Health Sciences Division, School of Public Health, University of California, Berkeley, 50 University Hall #7360, Berkeley, CA 94720-7360, United States. 1

0045-6535/$ - see front matter Ó 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.chemosphere.2013.12.083

Nations Environment Program, 2009). Despite their classification as persistent, PBDEs undergo debromination via microbial and photochemical pathways that have been studied under laboratory conditions, but as yet there is little understanding of their importance for the environmental fate of PBDEs. The photolysis of PBDEs, studied with multiple light sources and media, exhibits characteristic pathways and breakdown products (Sanchez-Prado et al., 2006; Fang et al., 2008; Sanchez-Prado et al., 2012; Wei et al., 2013 and references therein). Several researchers (Fang et al., 2008; Sanchez-Prado et al., 2012; Wei et al., 2013) have demonstrated that during photolysis, removal of bromines in the ortho position predominates. PBDE 15 (4,40 -dibromodiphenyl ether) is a major photolysis product, with PBDE 17 (2,20 ,4-tribromodiphenyl ether) sometimes reported as a

L.A. Rodenburg et al. / Chemosphere 106 (2014) 36–43

minor product (Sanchez-Prado et al., 2012; Wei et al., 2013). Ahn et al. (2006) studied PBDE photolysis on clays, metal oxides, and sediments and concluded that the photolysis pathways were largely matrix independent. Although the regiospecificity of microbial debromination of PBDEs is more complicated, certain breakdown products are observed. Using a biomimetic system, Tokarz et al. (2008) noted that PBDEs 17 (2,20 ,4-tribromo diphenyl ether) and 28 (2,4,40 -tribromo diphenyl ether) were major microbial debromination products of high molecular weight PBDEs in anaerobic sediment. Tokarz et al. (2008) also indicated that although microbial debromination at the ortho positions is possible, removal of the meta and para bromines predominates, especially for heavy congeners. Similarly, Robrock et al. (2008) and Ding et al. (2013) found that debromination of PBDEs by several cultures preferentially removed bromines at the meta and para positions, with formation of PBDE 17 as a major product. La Guardia et al. (2007) observed evidence of microbial debromination of PBDEs in sewage sludge and near wastewater outfalls. They measured PBDE 17 but could not detect PBDE 15 in sewage sludge. Davis et al. (2012) also detected PBDE 17 in biosolids. In contrast, Lee et al. (2011) studied the debromination of PBDE by a coculture consisting of Dehalococcoides and Desulfovibrio species, and found that debromination at the ortho position is preferred, with significant amounts of PBDE 15 formed. This echoes the regiospecificity of the microbial dechlorination of polychlorinated biphenyls (PCBs): dechlorination at the meta and para positions is preferred, but chlorines at the ortho position can be removed by some strains of bacteria (Bedard, 2003). For the PBDEs, it is not clear which set of pathways predominates in the environment, and indeed different regiospecificity might be observed in different environments. The purpose of this work was to use factor analysis to examine the importance of these debromination processes in the environment. Data collected by the Regional Monitoring Program for Water Quality in the San Francisco Estuary (RMP) were analyzed using Positive Matrix Factorization (PMF) (Paatero and Tapper, 1994). The RMP data set is a good choice for this investigation because it includes measurements of 50 PBDE congeners using highresolution mass spectrometry in hundreds of sediment samples collected in San Francisco Bay (Oram et al., 2008; Klosterhaus et al., 2012). PMF is an advanced factor analysis method, described in detail by Paatero and Tapper (1994). Briefly, PMF defines the sample matrix as product of two unknown factor matrices with a residue matrix:

X ¼ GF þ E

ð1Þ

The sample matrix (X) includes n observed samples and m chemical species. F is a matrix of p chemical profiles. The G matrix describes the contribution of each factor to any given sample, while E is the matrix of residuals. The PMF solution (i.e., G and F matrices) is obtained by minimizing the objective function Q through the iterative algorithm: n X m X Q¼ ðeij =sij Þ2

ð2Þ

i¼1 j¼1

Q is the sum of the squares of the difference (i.e., eij) between the observations (X) and the model (GF), weighted by the measurement uncertainties (sij). PMF has several advantages over simpler factor analysis tools such as Principle Components Analysis (PCA). PMF allows only positive correlations, and because the model input includes a pointby-point uncertainty estimate, PMF allows the inclusion of missing or below detection limit data points, which are assigned an arbitrary concentration and then associated with a high uncertainty. PMF and other similar factor analysis methods have been used to

37

investigate the dehalogenation of PBDEs (Zou et al., 2013) as well as other POPs, including polychlorinated biphenyls (PCBs) (Magar et al., 2005; Bzdusek et al., 2006b; Rodenburg et al., 2010) and polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans (PCDD/Fs) (Barabas et al., 2004; Rodenburg et al., 2012). Like all factor analysis methods, however, PMF can only isolate the factors that make up the data set. It is up to the user to use her best judgment to understand what these factors mean.

2. Methods 2.1. Study site San Francisco Bay (the Bay, Fig. 1) is one of the largest urbanized estuaries in the world, with a surrounding population of 7 million people. Bay hydrology is driven primarily by the tidal influence from the Pacific Ocean and the freshwater inflow from the Sacramento and San Joaquin Rivers, which drain an area of about 150 000 km2 (Cloern, 1996), and enter the Bay at its northern end in Suisun Bay (Fig. 1). As a result of freshwater inflow and tidal mixing, the northern Bay (Suisun and San Pablo Bays) is relatively well flushed, whereas South Bay is less well flushed (Conomos, 1979). Treated wastewater discharge is released via 31 permitted outfalls distributed Baywide, totaling a dry weather permitted flow into the Bay of 788 MGD. Concentrations of many pollutants, including PBDEs, are highest in Central Bay, South Bay, and Lower South Bay sediment and tend to decline moving northward towards San Pablo and Suisun Bays (Oros et al., 2005; Davis et al., 2007; Klosterhaus et al., 2012). The Regional Monitoring Program for Water Quality in the San Francisco Estuary (RMP) is an ongoing monitoring partnership established in 1993 between a regulatory agency, over 70 local regulated entities, and other local stakeholders (e.g., environmental NGOs), administered by an independent scientific organization (Hoenicke et al., 2003). RMP sediment monitoring employs a stratified design, in which eight probabilistic sites are collected annually from each of five Bay segments following a rotating panel design, with seven additional samples collected from fixed monitoring stations (Lowe et al., 2004). Every sample incorporates two to three homogenized subsamples, each collected by 0.1 m2 KynarÒ coated stainless steel Young-modified Van Veen grab. Trace clean techniques are employed, including quadruple rinsing of all equipment between each sampling event (SFEI, 2012). Beginning in 2002, fifty PBDE congeners were measured in the sediment by Axys Analytical Services (British Columbia, Canada) using high resolution gas chromatography/mass spectrometry methods equivalent to EPA method 1614A (EPA, 2007).

2.2. Data matrix Of the 50 congeners measured, only 24 were above the detection limit in more than 50% of the 403 sediment samples. These 24 congeners were retained for PMF analysis. Only samples with fewer than four non-detects were included in analysis. This resulted in 24 congeners measured in 233 samples, with adequate detection limits only occurring in sampling years 2005 through 2010.

2.3. PMF analysis The data were analyzed using Positive Matrix Factorization (PMF), employing PMF2 software (YP-Tekniika KY Co., Helsinki, Finland).

38

L.A. Rodenburg et al. / Chemosphere 106 (2014) 36–43

Fig. 1. Map of the San Francisco Bay showing municipal wastewater outfalls.

PMF computes the error estimate (sij) for each data point (xij) based on the data point and its original error estimate. The present study utilizes the error model (EM) = 14 (Paatero, 2003):

Sij ¼ t ij þ uij

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi maxðjxij j; jyij jÞ þ v ij maxðjxij j; jyij jÞ

ð3Þ

where t is the congener- and sample-specific detection limit, u is the Poisson distribution (here designated as 0), v is the measurement precision, x is the observed data value, and y is the modeled value. The uncertainty matrix was constructed by assigning an uncertainty of 15% to all congeners except those below detection. For values below detection, a value of one-half the detection limit was assigned, and the uncertainty was set to 166% (Brown and Hafner, 2003). Congener and sample-specific detection limits were used to construct the detection limit matrix. 3. Results and discussion In the 233 samples analyzed by PMF, R24BDE congener concentrations ranged from 0.16 to 54 lg kg1 (mean = 4.3 lg kg1; median = 3.3 lg kg1). PBDE 209 was the most abundant congener in 216 of the 233 samples. Congeners with four or fewer bromines comprised between 0.3% and 28% of R24BDEs (mean = 9.9%; median = 9.5%). These low molecular weight congeners (for example, PBDEs 17 and 28) constitute less than 1% of penta PBDE formula-

tions and are not detectable in octa and deca formulations (La Guardia et al., 2006). Their abundance in the majority of samples suggests that debromination processes significantly impact PBDE congener patterns in Bay sediments. A matrix of Pearson’s correlation coefficients (Supporting information Table S1) reveals that PBDE 209 is correlated with PBDEs 206 and 207, indicative of the commercial deca PBDE formulation. PBDEs 47, 99, and 100 are also correlated, indicative of the penta formulation (La Guardia et al., 2006). PBDEs 15 and 17 are uncorrelated, indicating that they arise from different sources, processes, or locations. 3.1. Positive matrix factorization Determining the correct number of factors is always a central challenge of factor analysis. A weight of evidence approach, described in the Supplemental information text, determined that five factors were appropriate; remaining interpretations focus on the five factor solution. 3.2. PMF congener patterns Examining congener patterns, factor 5 is dominated by PBDE 209 with smaller contributions from PBDEs 206, 207, and 208 (Fig. 2). PBDE 209 is also dominant in factor 2, present but not

L.A. Rodenburg et al. / Chemosphere 106 (2014) 36–43

Fraction of total

0.30

Factor 1 (photolysis) 3.1% mass 4.9% moles

0.25 0.20 0.15 0.10 0.05 BDE85

BDE99

BDE49

BDE51

BDE66

BDE71

BDE85

BDE99

BDE49

BDE51

BDE66

BDE71

BDE85

BDE99

BDE49

BDE51

BDE66

BDE71

BDE85

BDE99

BDE100

BDE153

BDE154

BDE155

BDE203

BDE206

BDE207

BDE208

BDE209

BDE49

BDE51

BDE66

BDE71

BDE85

BDE99

BDE100

BDE153

BDE154

BDE155

BDE203

BDE206

BDE207

BDE208

BDE209

Factor 2 (background) 20% mass 21% moles

BDE209

BDE66

BDE71

BDE47 BDE47 BDE47

BDE47

BDE208

BDE51

BDE37 BDE37 BDE37

BDE37

BDE207

BDE49

BDE35 BDE35 BDE35

BDE35

BDE203

BDE37

BDE47

BDE28 BDE28 BDE28

BDE28

BDE206

BDE35

BDE17 BDE17

BDE155

BDE17

BDE28

BDE15 BDE15

BDE17

BDE15

BDE15

BDE17

BDE154

BDE15

BDE12 BDE12 BDE12

BDE12

BDE100

BDE8

BDE12

BDE8 BDE8 BDE8

0.3

BDE153

BDE7 BDE7 BDE7 BDE7

BDE8

0.4

BDE7

Fraction of total

0.00

0.2 0.1 BDE209

BDE208

BDE207

BDE206

BDE203

BDE155

BDE153

BDE154

BDE100

0.10

BDE153

0.15

BDE100

Fraction of total

0.0

Factor 3 (microbial debromination) 5.4% mass 7.8% moles

0.05

BDE209

BDE208

BDE207

BDE206

BDE203

BDE154

0.3

BDE155

Fraction of total

0.00

Factor 4 (penta-BDE) 21% mass 26% moles

0.2 0.1

Fraction of total

0.0 0.8 0.6

Factor 5 (deca-BDE) 51% mass 40% moles

0.4 0.2 0.0

Fig. 2. Congener patterns of the five factors resolved by the PMF analysis from the BDE data set.

dominant in factors 3 and 4, and a minor contributor to factor 1. Factors 1 through 4 contain tetra- or less-brominated congeners which are not abundant in commercial PBDE formulations; these likely represent debromination products of higher molecular weight PBDE congeners. Based on comparison with the congener patterns of the commercial PBDE formulations (La Guardia et al., 2006), factor 4 represents penta-BDE formulations such as DE-71 and Bromkal 70-5DE, because it is dominated by PBDEs 47, 99, and 100 (Fig. 2). However, factor 4 contains higher proportions of PBDEs 17 (1.8% of total R24BDEs in the factor) and 28 (2.6%) than would be expected from the technical mixtures, since Bromkal 70-5DE contains less than 0.2% of these congeners. This elevated proportion may indicate either debromination or weathering (for example, preferential transport of lower molecular weight congeners). Factor 5 contains 82% of all of the PBDE 209 mass in the data set and more than 75% of the PBDEs 206, 207, and 208 masses (Fig. 2), and thus represents the technical deca mixtures, such as Saytex 102E and Bromkal 82-0DE (La Guardia et al., 2006). Factors 2 and 4 contain moderately high proportions of PBDE 209 (Fig. 2), accounting for 14% and 4% of the PBDE 209 mass, respectively. Notably absent from the resolved factors is anything resembling the technical octa-BDE mixtures. Factor 2 contains congeners that are not present in any of the commercial PBDE mixtures, including dibromo congeners PBDEs 7 and 8 (Fig. 2), and 12% of PBDE 17 (as compared to less than 0.1% within the commercial mixtures), suggesting debromination of heavy congeners. However, the major constituents of this factor are PBDE 47, 99, and 100, which are associated with the

39

commercial penta PBDE mixtures, and PBDE 209. We hypothesize that factor 2 represents the weathered PBDE background in the sediments, with traces of both major commercial formulations (penta and deca) and some evidence of debromination. Weathering is a generic term that encompasses many processes, including debromination as well as physical processes such as mixing, partitioning, and preferential transport. The PMF factor analysis identifies congener patterns that co-vary. This can mean that they are produced from the same location or process, or that they are merely transported together. Thus factor 2 may represent BDE congeners that have been transported together despite the fact that they come from several sources and have undergone debromination as well as preferential transport, both of which may alter the original (source) congener patterns. This might indicate that factor 2 is associated with an aggregate or secondary source, such as storm water or treated wastewater. Alternatively, it could indicate PBDEs that entered the bay long ago and have had time to undergo a lot of mixing, transport, and degradation. Factors 1 and 3 contain lower molecular weight PBDE congeners that are not present in the commercial PBDE mixtures. Factor 3 represents 5.4% of the mass and 7.8% of the moles of PBDEs in the data set, and contains high proportions of PBDE 17 (16% of R24BDEs), PBDE 7 (12%) and PBDE 49 (11%) (Fig. 2). PBDEs 17 and 49 have been reported as products of microbial debromination of PBDE technical mixtures (Robrock et al., 2008; Tokarz et al., 2008), suggesting that factor 3 indicates microbial debromination. The abundance of PBDE 47 in factor 3 is inconclusive, as PBDE 47 is both a major congener in many commercial formulations (La Guardia et al., 2006) and a microbial debromination product (Robrock et al., 2008; Tokarz et al., 2008). Factor 1 represents about 3.1% of the mass and 4.9% of the moles of PBDEs in the data set. It contains 20% PBDE 15, which is mostly absent from the other factors (Fig. 2), consistent with the lack of correlation between PBDE 15 and 17. Factor 1 also contains PBDEs 47, 49, and 66. BDE 15 in Factor 1 has not been reported as a measurable constituent of the commercial formulations, and therefore likely indicates either photochemical or microbial debromination. The case for photolysis rests on the observations of several researchers (Fang et al., 2008; Sanchez-Prado et al., 2012; Wei et al., 2013) who noted PBDE 15 as a major product of photochemical debromination of heavier congeners. PBDEs 47, 49, and 66, present in Factor 1, are all important in the photochemical degradation pathway. Although photolysis is expected to be negligible within surface sediment, it can occur during transport while PBDEs are sorbed to suspended clay minerals (Ahn et al., 2006), or while in the gas phase or dissolved in water. In the sediments of the Daliao River Estuary, China, PBDE 15 was abundant at concentrations which were not well correlated with other congeners (Zhao et al., 2011). The authors suggested PBDE 15 formation during atmospheric transport, with atmospheric deposition as the main source to surface sediment. The interpretation of factor 1 as photochemical debromination is also supported by its relatively even distribution across the Bay, as described below. It is also possible that factor 1 indicates microbial debromination that occurs with different regiospecificity than in factor 3. This would imply different bacterial populations and/or environmental conditions for each factor. Since PBDE 15 is the major product of this process, ortho debromination would have to be prevalent in this pathway. As noted above, there is no correlation between PBDEs 15 and 17 across the data matrix. If both congeners represent debromination, this lack of association suggests different locations or transport processes. Since microbial debromination is most likely an anaerobic process, this implies that there are two sets of anaerobic zones distributed throughout the Bay, each producing a very different congener pattern. This seems unlikely. If

40

L.A. Rodenburg et al. / Chemosphere 106 (2014) 36–43

debromination was observed to occur in a few hot spots, this explanation would be more plausible. A third possibility is that factor 1 or 3 (or both) could indicate debromination by aquatic animals (Stapleton et al., 2004), plants (Wang et al., 2012; Huang et al., 2013), or aerobic bacteria (Deng et al., 2011). For many of these processes, the products have not been measured, or at least the regiospecificity has not been reported, so it is unknown whether they could produce PBDE 15, which requires ortho debromination. Like PBDEs, PCBs can also be degraded in fish, plants, and by aerobic bacteria (Buckman et al., 2006; Field and Sierra-Alvarez, 2008; Van Aken et al., 2010), sometimes resulting in dramatic changes in congener patterns. To date, however, none of these altered congener patterns has been observed in sediment (Imamoglu et al., 2004; Bzdusek et al., 2006a; Bzdusek et al., 2006b; Soonthornnonda et al., 2011; Praipipat et al., 2013). The total biomass of aquatic animals and plants is likely to be relatively low, compared to microbes, suggesting that they would have limited impact on PCB and PBDE congener patterns on an ecosystem-wide basis. In contrast, microbial dechlorination of PCBs is known to alter PCB congener patterns in sediment (Bzdusek et al., 2006a; Bzdusek et al., 2006b). Based on this evidence, we speculate that the most likely explanation of the PMF2 results is that factor 1 indicates photochemical debromination, and factor 3 indicates debromination by anaerobic bacteria, which either occurs in the sediments themselves or in sewers or other environments, with the products transported into and mixed throughout the Bay. Additional evidence is needed to confirm this conjecture, however. Factor analysis alone cannot definitively distinguish between photochemical and microbial debromination because both processes can produce BDEs 15 and 17.

3.3. Spatial distribution of factors Factor 1 (photolysis or ortho debromination) concentrations ranged from below detection limits (ND) to 0.61 ng g1 (Fig. 3 panel 1A). The relative contribution of factor 1 to the total (Fig. 3 panel 1B) was higher for the less urbanized northern Bay segments (San Pablo Bay and Suisun Bay; average 9% contribution to total) than southern Bay segments (Lower South, South, and Central Bays; averaging 2–3% contribution to total). This distribution is consistent with the hypothesis that factor 1 results from photolysis in the gas phase and subsequent atmospheric deposition. Specifically, the mass of photodebrominated PBDEs contributed from atmospheric deposition would be similar across the Bay, but the relative contribution would be reduced by greater inputs of PBDEs in the southern Bay. Factor 2 exhibited variable concentrations ranging from ND to 4.2 ng g1 (Fig. 3 panel 2A). The fractional contribution of factor 2 was greatest in San Pablo Bay and Central Bay (Fig. 3 panel 2B), consistent with our hypothesis that this factor represents a weathered PBDE background. Specifically, the largest PBDE sources likely occur in the South Bay region, with increased mixing and weathering with transport and dispersion away from original sources. Factor 3 (microbial debromination) maximum concentrations were 1.1 ng g1 (21% of the total PBDE mass; Fig. 3 panels 3A and 3B). The highest fraction of factor 3 was generally seen in South Bay and the lowest fraction in San Pablo Bay. South Bay receives a high proportion of stormwater and treated sewage outfalls, mostly via Lower South Bay, whereas the majority of inflow to San Pablo Bay originates from riverine input (Sacramento and San Joaquin Rivers, via Suisun Bay, and to a lesser extent Napa and Petaluma Rivers). The sewage outfall inputs and lower flushing rates may increase sediment anoxia in the South Bay

Fig. 3. Spatial distribution of the resolved factors. Left panels: fraction of the total BDE mass in each sample. Right panels: raw concentration.

L.A. Rodenburg et al. / Chemosphere 106 (2014) 36–43

(Cloern, 1996), potentially promoting microbial dehalogenation. Elevated factor 3 concentrations were not observed in Lower South Bay (where the major outfalls are located), suggesting that debromination prior to discharge within the sewers draining into Lower South Bay may not be a major contributor to microbial debromination products. Concentrations of factor 4 ranged up to 3.8 ng g1 and 76% of the R24BDEs (Fig. 3 panel 4A). Concentrations displayed no obvious spatial trend in the Bay, in keeping with the ubiquitous sources of fresh penta-BDE throughout the urban area. However, the fraction of total PBDEs contributed by factor 4 generally increased moving northward from Lower South Bay through Suisun Bay and the Rivers (Fig. 3 panel 4B). Factor 5 (deca-BDE) concentrations ranged up to 12.4 ng g1 (Fig. 3 panel 5A). Factor 5 was the predominant factor in 138 of the 233 samples, and was most abundant in the Lower South Bay (Fig. 3 panel 5B). This distribution was opposite that determined by loading estimates. Oram et al. (2008) considered loads of PBDEs 47 and 209 from the Sacramento–San Joaquin Delta, local tributaries, municipal wastewater, and atmospheric deposition. Oram et al. suggested that PBDE 47 (generally corresponding to factor 4) loads were dominated by wastewater, while PBDE 209 (corresponding to factor 5) loads were dominated by runoff from the Sacramento and San Joaquin Rivers. There are several possible reasons for the discrepancy between the PMF results and the mass budget of Oram et al. First, Oram et al. did not consider some PBDE sources that might be significant, including storm water runoff. More importantly, the PMF analysis considered the PBDE concentrations prevailing in the sediment during 2005–2010. These are likely to reflect decades of inputs, while the mass budget considered inputs over only a limited time period (water years 2005 and 2006). Changes in the use patterns of the various technical PBDE formulations and/or greater persistence of some PBDE congeners could account for some of the discrepancy. The rates of degradation of PBDEs in the environment are not well known. Oram et al. assumed that the degradation rates for PBDEs 47 and 209 were the same, with half-lives of 150 d in water and 578 d in sediment. The PMF results indicate that for PBDEs that have presumably resided in the sediment for several years, only about 13% of the detected moles of BDEs are products of debromination, and much of that debromination may have happened in the atmosphere. Thus the PMF results suggest that the half-lives invoked by Oram et al. are either too fast, or that the products of these degradation processes were not measured in the data set used for PMF analysis (i.e. either degradation processes other than debromination, or complete debromination to diphenyl ether). 3.4. Debromination and its implications This analysis suggests that congener patterns indicative of debromination of PBDEs are apparent in San Francisco Bay. It is not certain where or how the debromination occurred. It may be the result of photolysis occurring in the gas, dissolved, or suspended sediment phases, microbial debromination under anaerobic conditions, or even debromination by other biota. Debromination products are associated with factors representing 8% of the mass and 13% of the moles of PBDEs in the data matrix, suggesting that PBDEs undergo measurable degradation in the environment. Also, the dibromo congeners PBDE 7, 8, 12, and 15 were detected as debromination products, indicating that debromination has proceeded to an advanced stage. Several other researchers have observed evidence of PBDE debromination in sediments and other environmental compartments. La Guardia et al. (2007) reported PBDEs debromination in sewage sludge and near sewage outfalls. Wei et al. (2012) reported debromination in sediments from Arkansas, but identified the

41

products only by homologue. Salvado et al. (2012) noted decreasing sediment PBDE 209 concentrations and increasing lower molecular weight PBDEs with distance from the probable site of release (Gulf of Lion, Mediterranean Sea). Zhao et al. (2011) found that congeners that were believed to come from photolysis were the most abundant congeners in the sediments of the Daliao River Estuary, China. In a study similar to the present investigation, Zou et al. (2013) used PMF combined with eigenspace projection to investigate PBDE congener patterns in sediments cores from the Great Lakes. Using a data matrix of 10 congeners in 93 samples, they resolved five factors, thought to represent the commercial penta, octa, and deca formulations, and two factors thought to represent debromination products. The first of these was characterized by large contributions from PBDEs 66 and 85, while the other was characterized by a high proportion of PBDE 28. Unlike the present work, their study did not include congeners 7, 8, 12, 15, and 17, so it is difficult to compare their factors with those resolved in the present study. Nevertheless, it is noteworthy that two studies, one in a freshwater system and one in an estuary, identified two distinct debromination signals in sediment. In the future, we recommend that PBDEs 7, 15, and 17 be routinely monitored, since they appear to be markers for debromination processes. The present analysis did not find an octa signal in San Francisco Bay. PBDE 183, the major congener in the octa formulation, was not included in the data matrix because it was below detection in a majority of samples. However, the maximum contribution of PBDE 183 to the sum of PBDEs was 10%. PBDE 183 comprised more than 2% of the sum of PBDEs in only 11 of 344 samples. In addition, the data matrix did include PBDEs 153, 154, and 155, which are prominent in the octa formulation, yet PMF analysis did not resolve an octa factor. Thus we conclude that the octa formulation is not a major source of PBDEs to San Francisco Bay, possibly due to lower production of octa relative to penta and deca BDE formulations in the Americas (Birnbaum and Staskal, 2004). Also, both the octa and penta BDE formulations have undergone a gradual production bad in California, which was approved in 2003 and fully implemented in 2008. Anaerobic microbial debromination, which we speculate is associated with Factor 3, may have occurred in Bay sediments. Such a process has been shown to debrominate octa mixtures in laboratory microcosms (Lee and He, 2010). It is instructive to consider the similar process of bacterial dechlorination, since in both cases, anaerobic bacteria use the halogenated compound as an electron acceptor. Bacteria capable of dechlorination have been isolated from Bay sediments (Sun et al., 2000; He et al., 2006). Although some researchers have suggested that the threshold concentration for PCB dechlorination is around 40 ppm (Cho et al., 2003), well above the maximum of about 50 ppb reached in Bay sediments, others have shown evidence of PCB dechlorination with an enriched culture under substrate concentrations as low as 1– 5 ppm (Royal et al., 2003; Krumins et al., 2009; Payne et al., 2011). It is also possible that the threshold concentration for PBDE debromination is lower than the threshold for PCBs. Previous studies indicate microbial dechlorination in sewers (Rodenburg et al., 2010; Rodenburg et al., 2012), raising the possibility of microbial debromination in San Francisco Bay area sewers, with debromination products emitted to the Bay via stormwater or treated wastewater outfalls. As noted above, several researchers have seen evidence of PBDE debromination in sewage sludge and near sewage outfalls (La Guardia et al., 2007; Davis et al., 2012). The location of degradation is important. If microbial debromination occurs in Bay sediments, then PBDEs will be less persistent in the long run and will have less tendency to accumulate in sediments. In contrast, if microbial debromination occurs primarily in sewers (i.e., prior to discharge), then both the parent compounds and the

42

L.A. Rodenburg et al. / Chemosphere 106 (2014) 36–43

debromination products would accumulate in sediments and become problematic in the long term. Based on the increased percent contribution of factor 3 from Lower South Bay to South Bay, we hypothesize that the majority of microbial debromination occurs in Bay sediments. Examination of congener ratios in wastewater and treatment plant sludge would aid in confirming this. Regardless of the location of dehalogenation, the results indicate that PBDEs undergo measurable debromination in the environment. We cannot rule out the possibility that debromination leads to the fully debrominated diphenyl ether or to bromophenols (Bendig and Vetter, 2013), which were not measured in this data set. Thus our estimate that about 13% of the moles of PBDEs in the Bay have undergone debromination is a lower bound. This is a relatively large degree of transformation. By comparison, POPs such as PCBs and PCDD/Fs show no evidence of degradation in most aquatic systems. Rodenburg et al. (2010) demonstrate that as much as 19% of the PCBs emitted by permitted dischargers in the Delaware River, USA, basin were subject to dehalogenation, but the dechlorination products (primarily PCBs 4 and 19) are barely detectable in Delaware River sediments (Praipipat et al., 2013). These lower chlorinated congeners may be prevented from accumulating in sediments in part due to volatilization, aerobic degradation, or dissolution and advective export. PCBs 4 and 19 have lower octanol–water partition coefficients (log Kow = 4.84 for PCB 4 and 5.16 for PCB 19) (Hansen et al., 1999) than PBDE 17 (ranging 5.4–6.6) or PBDE 15 (reported at 5.48) (Wania and Dugani, 2003). Thus, dehalogenated PBDE congeners are more likely to accumulate in sediments than the more hydrophilic dehalogenated PCB congeners. The relative toxicity of PBDEs 7, 15, and 17 vs. parent compounds is poorly characterized. If the debromination products have equal or greater toxicity, this would be compounded by the greater environmental mobility and potential exposure to aquatic life associated with lower hydrophobicities (Arnot and Gobas, 2003). For example, PBDE 47 has a greater biota-sediment accumulation factor (BSAF) than PBDE 209 (La Guardia et al., 2012), so debromination of PBDE 209 to PBDE 47 will result in greater bioaccumulation of PBDEs in some organisms. This work has demonstrated that debromination is an important process affecting the fate of PBDE formulations, and that the debromination products accumulate in sediments. Further research is needed to determine the toxic impacts of the debromination products. Acknowledgments Sediment PBDE collection and analysis was performed by Applied Marine Sciences, SFEI, the East Bay Municipal Utility District, and Axys Analytical Services, and supported by the Regional Monitoring Program for Water Quality in the San Francisco Estuary. BG is supported by a USEPA STAR Fellowship. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.chemosphere. 2013.12.083. References Ahn, M.Y., Filley, T.R., Jafvert, C.T., Nies, L., Hua, I., Bezares-Cruz, J., 2006. Photodegradation of decabromodiphenyl ether adsorbed onto clay minerals, metal oxides, and sediment. Environ. Sci. Technol. 40, 215–220. Arnot, J.A., Gobas, F.A.P.C., 2003. A generic QSAR for assessing the bioaccumulation potential of organic chemicals in aquatic food webs. Quant. Struct.-Act. Relat. 22, 1–9. Barabas, N., Goovaerts, P., Adriaens, P., 2004. Modified polytopic vector analysis to identify and quantify a dioxin dechlorination signature in sediments. 2. Application to the Passaic River. Environ. Sci. Technol. 38, 1821–1827.

Bedard, D.L., 2003. Polychlorinated biphenyls in aquatic sediments: environmental fate and outlook for biological treatment. In: Haggblom, M.M., Bossert, I.D. (Eds.), Dehalogenation: Microbial Processes and Environmental Applications. Kluwer Academic Publishers, Boston. Bendig, P., Vetter, W., 2013. UV-induced formation of bromophenols from polybrominated diphenyl ethers. Environ. Sci. Technol. 47, 3665–3670. Birnbaum, L.S., Staskal, D.F., 2004. Brominated flame retardants: cause for concern? Environ. Health Perspect. 112, 9–17. Brown, S.G., Hafner, H.R., 2003. Exploratory Source Apportionment of Houston’s Clinton Drive Auto-GC 1998–2001 Data. Texas Commission on Environmental Quality, Petaluma, CA. Buckman, A.H., Wong, C.S., Chow, E.A., Brown, S.B., Solomon, K.R., Fisk, A.T., 2006. Biotransformation of polychlorinated biphenyls (PCBs) and bioformation of hydroxylated PCBs in fish. Aquat. Toxicol. 78, 176–185. Bzdusek, P.A., Christensen, E.R., Lee, C.M., Pakdeesusuk, U., Freedman, D.C., 2006a. PCB congeners and dechlorination in sediments of Lake Hartwell, South Carolina, determined from cores collected in 1987 and 1998. Environ. Sci. Technol. 40, 109–119. Bzdusek, P.A., Lu, J., Christensen, E.R., 2006b. PCB congeners and dechlorination in sediment of Sheboygan River, Wisconsin, determined by matrix factorization. Environ. Sci. Technol. 40, 120–129. Cho, Y.-C., Sokol, R.C., Frohnhoefer, R.C., Rhee, G.-Y., 2003. Reductive dechlorination of polychlorinated biphenyls: threshold concentration and dechlorination kinetics of individual congeners in Aroclor 1248. Environ. Sci. Technol. 37, 5651–5656. Cloern, J.E., 1996. Phytoplankton bloom dynamics in coastal ecosystems: a review with some general lessons from sustained investigation of San Francisco Bay, California. Rev. Geophys. 34, 127–168. Conomos, T.J., 1979. Properties and circulation of San Francisco Bay waters. In: Conomos, T.J. (Ed.), San Francisco Bay: The Urbanized Estuary. California Academy of Sciences, San Francisco, pp. 47–84. Davis, J.A., Hetzel, F., Oram, J.J., McKee, L.J., 2007. Polychlorinated biphenyls (PCBs) in San Francisco Bay. Environ. Res. 105, 67–86. Davis, E.F., Klosterhaus, S.L., Stapleton, H.M., 2012. Measurement of flame retardants and triclosan in municipal sewage sludge and biosolids. Environ. Int. 40, 1–7. Deng, D.Y., Guo, J., Sun, G.P., Chen, X.J., Qiu, M.D., Xu, M.Y., 2011. Aerobic debromination of deca-BDE: isolation and characterization of an indigenous isolate from a PBDE contaminated sediment. Int. Biodeterior. Biodegrad. 65, 465–469. Ding, C., Chow, W.L., He, J.Z., 2013. Isolation of Acetobacterium sp. Strain AG, which reductively debrominates octa- and pentabrominated diphenyl ether technical mixtures. Appl. Environ. Microbiol. 79, 1110–1117. Domingo, J.L., 2012. Polybrominated diphenyl ethers in food and human dietary exposure: a review of the recent scientific literature. Food Chem. Toxicol. 50, 238–249. EPA, U.S., 2007. Method 1614: Brominated Diphenyl Ethers in Water Soil, Sediment and Tissue by HRGC/HRMS. Washington, DC. Fang, L., Huang, J., Yu, G., Wang, L., 2008. Photochemical degradation of six polybrominated diphenyl ether congeners under ultraviolet irradiation in hexane. Chemosphere 71, 258–267. Field, J.A., Sierra-Alvarez, R., 2008. Microbial transformation and degradation of polychlorinated biphenyls. Environ. Pollut. 155, 1–12. Hansen, B.G., Paya-Perez, A.B., Rahman, M., Larsen, B.R., 1999. QSARs for KOW and KOC of PCB congeners: a critical examination of data, assumptions and statistical approaches. Chemosphere 39, 2209–2228. He, J., Robrock, K.R., Alvarez-Cohen, L., 2006. Microbial reductive debromination of polybrominated diphenyl ethers (PBDEs). Environ. Sci. Technol. 40, 4429–4434. Hoenicke, R., Davis, J.A., Gunther, A., Mumley, T.E., Abu-Saba, K., Taberski, K., 2003. Effective application of monitoring information: the case of San Francisco Bay. Environ. Monit. Assess. 81, 15–25. Huang, H.L., Zhang, S.Z., Wang, S., Lv, J.T., 2013. In vitro biotransformation of PBDEs by root crude enzyme extracts: potential role of nitrate reductase (NaR) and glutathione S-transferase (GST) in their debromination. Chemosphere 90, 1885– 1892. Imamoglu, I., Li, K., Christensen, E.R., McMullin, J.K., 2004. Sources and dechlorination of polychlorinated biphenyl congeners in the sediments of Fox River, Wisconsin. Environ. Sci. Technol. 38, 2574–2583. Klosterhaus, S.L., Stapleton, H.M., La Guardia, M.J., Greig, D.J., 2012. Brominated and chlorinated flame retardants in San Francisco Bay sediments and wildlife. Environ. Int. 47, 56–65. Krumins, V., Park, J.W., Son, E.K., Rodenburg, L.A., Kerkhof, L.J., Haggblom, M.M., Fennell, D.E., 2009. PCB dechlorination enhancement in Anacostia River sediment microcosms. Water Res. 43, 4549–4558. La Guardia, M.J., Hale, R.C., Harvey, E., 2006. Detailed polybrominated diphenyl ether (PBDE) congener composition of the widely used penta-, octa-, and decaPBDE technical flame-retardant mixtures. Environ. Sci. Technol. 40, 6247–6254. La Guardia, M.J., Hale, R.C., Harvey, E., 2007. Evidence of debromination of decabromodiphenyl ether (BDE-209) in biota from a wastewater receiving stream. Environ. Sci. Technol. 41, 6663–6670. La Guardia, M.J., Hale, R.C., Harvey, E., Mainor, T.M., Ciparis, S., 2012. In situ accumulation of HBCD, PBDEs, and several alternative flame-retardants in the bivalve (Corbicula fluminea) and gastropod (Elimia proxima). Environ. Sci. Technol. 46, 5798–5805.

L.A. Rodenburg et al. / Chemosphere 106 (2014) 36–43 Lee, L.K., He, J., 2010. Reductive debromination of polybrominated diphenyl ethers by anaerobic bacteria from soils and sediments. Appl. Environ. Microbiol. 76, 794–802. Lee, L.K., Ding, C., Yang, K.L., He, J.Z., 2011. Complete debromination of tetra- and penta-brominated diphenyl ethers by a coculture consisting of Dehalococcoides and Desulfovibrio species. Environ. Sci. Technol. 45, 8475–8482. Lowe, S., Thompson, B., Hoenicke, R., Leatherbarrow, J., Taberski, K., Smith, R., Stevens Jr., D., 2004. Re-design Process of the San Francisco Estuary Regional Monitoring Program for Trace Substances (RMP) Status & Trends Monitoring Component for Water and Sediment. SFEI, Oakland, CA, p. 86. Magar, V.S., Johnson, G.W., Brenner, R.C., Quensen, J.F., Foote, E.A., Durell, G., Ickes, J.A., McCarthy, C.P., 2005. Long-term recovery of PCB-contaminated sediments at the Lake Hartwell superfund site: PCB dechlorination. 1. End-member characteristics. Environ. Sci. Technol. 39, 3538–3547. Oram, J.J., McKee, L.J., Werme, C.E., Connor, M.S., Oros, D.R., Grace, R., Rodigari, F., 2008. A mass budget of polybrominated diphenyl ethers in San Francisco Bay, CA. Environ. Int. 34, 1137–1147. Oros, D.R., Hoover, D., Rodigari, F., Crane, D., Sericano, J.L., 2005. Levels and distribution of polybrominated diphenyl ethers in water, surface sediments, and bivalves from the San Francisco Estuary. Environ. Sci. Technol. 39, 33–41. Paatero, P., 2003. User’s Guide for Positive Matrix Factorization Programs PMF2 and PMF3. Part 1: Tutorial. Paatero, P., Tapper, U., 1994. Positive matrix factorization: a non-negative factor model with optimal utilization of error estimates of data values. Environmetrics 5, 111–126. Payne, R.B., May, H.D., Sowers, K.R., 2011. Enhanced reductive dechlorination of polychlorinated biphenyl impacted sediment by bioaugmentation with a dehalorespiring bacterium. Environ. Sci. Technol. 45, 8772–8779. Praipipat, P., Rodenburg, L.A., Cavallo, G.J., 2013. Source apportionment of polychlorinated biphenyls in the sediments of the Delaware River. Environ. Sci. Technol. 47, 4277–4283. Robrock, K.R., Korytar, P., Alvarez-Cohen, L., 2008. Pathways for the anaerobic microbial debromination of polybrominated diphenyl ethers. Environ. Sci. Technol. 42, 2845–2852. Rodenburg, L.A., Du, S., Fennell, D.E., Cavallo, G.J., 2010. Evidence for widespread dechlorination of polychlorinated biphenyls in groundwater, landfills, and wastewater collection systems. Environ. Sci. Technol. 44, 7534–7540. Rodenburg, L.A., Du, S.Y., Lui, H., Guo, J., Oseagulu, N., Fennell, D.E., 2012. Evidence for dechlorination of polychlorinated biphenyls and polychlorinated dibenzo-pdioxins and -furans in wastewater collection systems in the New York metropolitan area. Environ. Sci. Technol. 46, 6612–6620. Royal, C.L., Preston, D.R., Sekelsky, A.M., Shreve, G.S., 2003. Reductive dechlorination of polychlorinated biphenyls in landfill leachate. Int. Biodeterior. Biodegrad. 51, 61–66. Salvado, J.A., Grimalt, J.O., Lopez, J.F., de Madron, X.D., Heussner, S., Canals, M., 2012. Transformation of PBDE mixtures during sediment transport and resuspension in marine environments (Gulf of Lion, NW Mediterranean Sea). Environ. Pollut. 168, 87–95.

43

Sanchez-Prado, L., Lores, M., Llompart, M., Garcia-Jares, C., Bayona, J.M., Cela, R., 2006. Natural sunlight and sun simulator photolysis studies of tetra- to hexabrominated diphenyl ethers in water using solid-phase microextraction. J. Chromatogr. A 1124, 157–166. Sanchez-Prado, L., Kalafata, K., Risticevic, S., Pawliszyn, J., Lores, M., Llompart, M., Kalogerakis, N., Psillakis, E., 2012. Ice photolysis of 2,20 ,4,40 ,6pentabromodiphenyl ether (BDE-100): laboratory investigations using solid phase microextraction. Anal. Chim. Acta 742, 90–96. SFEI, 2012. 2010 Annual Monitoring Results. San Francisco Estuary Institute (SFEI), Richmond, CA. Soonthornnonda, P., Zou, Y.H., Christensen, E.R., Li, A., 2011. PCBs in Great Lakes sediments, determined by positive matrix factorization. J. Great Lakes Res. 37, 54–63. Stapleton, H.M., Letcher, R.J., Baker, J.E., 2004. Debromination of polybrominated diphenyl ether congeners BDE 99 and BDE 183 in the intestinal tract of the common carp (Cyprinus carpio). Environ. Sci. Technol. 38, 1054–1061. Sun, B., Cole, J.R., Sanford, R.A., Tiedje, J.M., 2000. Isolation and characterization of Desulfovibrio dechloracetivorans sp. nov., a marine dechlorinating bacterium growing by coupling the oxidation of acetate to the reductive dechlorination of 2-chlorophenol. Appl. Environ. Microbiol. 66, 2408–2413. Tokarz, J.A., Ahn, M.Y., Leng, J., Filley, T.R., Nies, L., 2008. Reductive debromination of polybrominated diphenyl ethers in anaerobic sediment and a biomimetic system. Environ. Sci. Technol. 42, 1157–1164. United Nations Environment Program, 2009. Stockholm Convention on Persistent Organic Pollutants, Amendments to Annexes A, B And C. United Nations Environment Program, Geneva, Switzerland. Van Aken, B., Correa, P.A., Schnoor, J.L., 2010. Phytoremediation of polychlorinated biphenyls: new trends and promises. Environ. Sci. Technol. 44, 2767–2776. Wang, S., Zhang, S.Z., Huang, H.L., Lu, A.X., Ping, H., 2012. Debrominated, hydroxylated and methoxylated metabolism in maize (Zea mays L.) exposed to lesser polybrominated diphenyl ethers (PBDEs). Chemosphere 89, 1295–1301. Wania, F., Dugani, C.B., 2003. Assessing the long-range transport potential of polybrominated diphenyl ethers: a comparison of four multimedia models. Environ. Toxicol. Chem. 22, 1252–1261. Wei, H., Aziz-Schwanbeck, A.C., Zou, Y.H., Corcoran, M.B., Poghosyan, A., Li, A., Rockne, K.J., Christensen, E.R., Sturchio, N.C., 2012. Polybromodiphenyl ethers and decabromodiphenyl ethane in aquatic sediments from southern and eastern Arkansas, United States. Environ. Sci. Technol. 46, 8017–8024. Wei, H., Zou, Y.H., Li, A., Christensen, E.R., Rockne, K.J., 2013. Photolytic debromination pathway of polybrominated diphenyl ethers in hexane by sunlight. Environ. Pollut. 174, 194–200. Zhao, X.F., Zhang, H.J., Ni, Y.W., Lu, X.B., Zhang, X.P., Su, F., Fan, J.F., Guan, D.M., Chen, J.P., 2011. Polybrominated diphenyl ethers in sediments of the Daliao River Estuary, China: levels, distribution and their influencing factors. Chemosphere 82, 1262–1267. Zou, Y.H., Christensen, E.R., Li, A., 2013. Characteristic pattern analysis of polybromodiphenyl ethers in Great Lakes sediments: a combination of eigenspace projection and positive matrix factorization analysis. Environmetrics 24, 41–50.