Environ. Sci. Technol. 2007, 41, 2703-2710
Time Trends in Sources and Dechlorination Pathways of Dioxins in Agrochemically Contaminated Sediments MINORI UCHIMIYA* AND SHIGEKI MASUNAGA Graduate School of Environment and Information Sciences, Yokohama National University, 79-7 Tokiwadai, Hodogaya, Yokohama 240-8501, Japan
Although polychlorinated dibenzo-p-dioxins and dibezofurans (PCDD/Fs) are considered recalcitrant toward biotic and abiotic degradation processes, laboratory studies indicated lateral dechlorination pathways (removal of 2,3,7,8substituted chlorines) as possible natural remediation strategies under highly reducing conditions prevailing in contaminated sediments. Previous principal component analysis (PCA) of PCDD/Fs in Japanese sediments left unidentified a factor characterized by penta- to octahomologues fully chlorinated at 1,2,6,9-positions (1,2,6,9pattern). In the present study, we reexamined PCDD/Fs in sediment cores from urban (Tokyo Bay) and remote (Lake Shinji) areas of Japan using positive matrix factorization (PMF) and revealed a lateral dechlorination fingerprint exhibiting the 1,2,6,9-pattern. Relative molar concentrations of putative lateral dechlorination products linearly increased with sediment depth, suggesting that decades of reaction resulted in the accumulation of hepta- and hexachlorinated lateral dechlorination products in the bottom sediment layers. Times required for in situ formation of dechlorination products were estimated to be at least 27.8 ( 17.9 year(mole %)-1 in Lake Shinji and 4.7 ( 0.5 year(mole %)-1 in Tokyo Bay (both for the formation of 1,2,3,4,6,7,9HpCDD) and are significantly longer than the dechlorination pathways observed in the laboratory.
Introduction Because polychlorinated dibenzo-p-dioxins and dibezofurans (PCDD/Fs) deposited decades ago are still found in deep sediment layers, PCDD/Fs are often considered recalcitrant toward biotic and abiotic degradation processes. However, dechlorination of PCDD/Fs by pure culture of anaerobic bacterium Dehalococcoides sp. (1), enrichment cultures from river sediments (2), model humic substances (3), and in historically PCDD/F-contaminated estuarine sediments (4) has been observed in the laboratory under highly reducing conditions. Dechlorination of PCDD/Fs proceeds via preferred lateral (removal of 2,3,7,8-substituted chlorines) and/ or peri (removal of 1,4,6,9-substituted chlorines) pathways (5). Careful assessment of the fate of PCDD/Fs after sedimentation is essential, as reductive dechlorination leads either to natural remediation or increased toxicity of the acutely toxic contaminant. * Corresponding author phone: +81-45-339-4346; fax: +81-45339-4352; e-mail:
[email protected]. 10.1021/es0627444 CCC: $37.00 Published on Web 03/08/2007
2007 American Chemical Society
The congener profiles of PCDD/Fs in Japan are uniquely impacted by the past use of PCDD/F-contaminated herbicides, particularly pentachlorophenol (PCP) and chloronitrophen (CNP) in rice paddies (6, 7). Using principal component analysis (PCA) and multiple regression analysis (MRA) on PCDD/Fs extracted from dated sediment layers from Lake Shinji and Tokyo Bay, Masunaga et al. determined congener-specific historical trends of primary PCDD/F sources, PCP, CNP, and combustion (6, 7). In a separate study, PCA of additional sediment and soil samples from Kanto Region, Japan afforded an unknown factor showing strong contribution to penta- to hepta-homologues fully chlorinated at 1,2,6,9-positions (1,2,6,9-pattern) (8). Significant shortcoming of PCA, causing “unknown” factors, is that factor loadings are abstract, orthogonal matrices and do not quantitatively represent congener profiles of sources/sink (9). Consequently, an original data set can be reproduced by MRA only if source profiles for the particular study site are available (6, 7). While significance of dechlorination pathways on the PCDD/F profiles of Japanese sediments has been suggested (10), no concrete evidence has been provided. Receptor models such as the positive matrix factorization (PMF) are powerful statistical tools for quantitatively resolving the number, chemical composition, and geographical/ temporal distribution of the chemical fingerprints simultaneously (9). Receptor models, traditionally employed to deduce point sources, are a useful tool for assessing the chemical fate of contaminants after sedimentation (11). For example, PMF provided a snapshot of in situ reductive dechlorination of polychlorinated biphenyls (PCBs) in river (12) and lake (13) sediments. Depth-dependent examination of PCB congener profiles revealed a gradual shift (toward deeper layers) from the original PCB mixture (first PMF factor) to its dechlorination products (second PMF factor) (12). The estimated in situ PCB dechlorination rates (4.3-11.6 years per chlorine removal) were less than one-tenth of the values observed in the laboratory (14). To our knowledge, previous study on the receptor model-based assessment of PCDD/F dechlorination pathways in the sediment cores is limited to the modified (allowing negative factor contribution) polytopic vector analysis (PVA) of Passaic River (New Jersey) samples (11). A dechlorination fingerprint with a highly positive 2,3,7,8-TeCDD component and a highly negative HpCDDs component suggested in situ peri dechlorination (11). This is the first of two papers in a series demonstrating ways to distinguish multiple sources and dechlorination pathways of PCDD/Fs using dynamic, spatiotemporal distribution of PMF-derived chemical fingerprints. In this study, sediment core PCDD/F profiles from both rural (Lake Shinji) and urban (Tokyo Bay; Figure 1) areas of Japan were analyzed using PMF. At both sites, PMF afforded a factor showing the previously unidentified 1,2,6,9-pattern (8). Relative molar concentrations of the main congeners in this factor linearly increased with depth, suggesting that the factor represented lateral dechlorination products of OcCDD and other highly chlorinated PCDDs. Our subsequent paper in this series will employ PMF to determine additional sources of dioxins causing localized contamination at Ichihara Anchorage in northeastern Tokyo Bay.
Materials and Methods Original Data Sets. Original data sets were obtained from literature sources (6, 15, 16). A sediment core from Lake Shinji was collected in 1994 near the mouth of River Hii (6). As shown in Figure 1 (open triangle), a sediment core from Tokyo Bay was collected in 1993 at the center of the bay (15). The VOL. 41, NO. 8, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2703
FIGURE 1. Sampling points in Tokyo Bay for a sediment core (open triangle) and surface sediments (circled numbers). average sedimentation rates at both sites were determined by the 210Pb and 137Cs methods (17). Surface sediments were collected in 1995 at seven sites shown in Figure 1 (circled numbers) (16). To prevent sample contamination/aging, sediment samples were immediately freeze-dried and stored under dark at -30 °C until analyses using HRGC/HRMS (see original reports for detailed analytical procedures) (6, 15, 16). Positive Matrix Factorization. The PMF is a PCA-based receptor model with nonnegativity constraints that involve solution of quantitative source apportionment equations by the oblique solutions in reduced dimensional space (18). The following linear algebraic equation addresses PMF (19): p
xij )
∑a
ikfkj
+ ij
(1)
k)1
where xij is the concentration of the jth congener in the ith sample of the original data set, aik is the contribution of the kth factor on sample i, fkj is the fraction of the kth factor arising from congener j, and ij is the residual between xij and the estimate of xij using p principal components. The objective of PMF is to minimize Q, the weighted sum of squares of differences between the PMF output and the original data set (19):
( )
2
p
n
Q)
m
∑∑ i)1 j)1
xij -
∑ k)1
sij
aikfkj
(2)
where sij is the uncertainty of the jth congener in the ith sample of the original data set containing m congeners and n samples. 2704
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 8, 2007
Computer software (20) based on Paatero’s PMF program (21) was used for all analyses. For more exact replications of the simulated data, predicted factor contributions were allowed to be slightly negative down to -0.1 (20). Pretreatment of Original Data Sets. Original data sets (in pg(g-dry sediment)-1) were first converted to % contribution to total tetra- to octachlorinated PCDD/F concentrations in each sample to normalize out the effects of dilution away from the source, thereby allowing precise recognition of the relative proportions of congeners in the fingerprint. This normalization procedure also enabled the recognition of samples having significant contribution from a fingerprint regardless of total PCDD/F concentration. Then, values below the detection limit (DL) were replaced with half of DL (9). Congeners with more than 15% below DL values were eliminated. The final data set contained 85 chromatographic peaks for tetra- to octachlorinated PCDD/Fs in both Lake Shinji core (12 layers) and Tokyo Bay (16 layers plus 7 surface sediments) samples. The PMF analysis of normalized data set afforded chemical fingerprint (fkj in eq 1) and contribution (aik) for each factor. For a given factor, the aikfkj values (% factor contribution to total PCDD/F concentrations) allow us to recognize samples having significant contribution from a fingerprint regardless of total PCDD/F concentration. Absolute factor contribution (in pg(g-dry sediment)-1) can then be determined using total PCDD/F concentrations. Sum of contributions for all factors canceled out negative contributions to better reproduce the original data set. Error Analyses. The coefficient of determination (COD) was used to evaluate the ability of PMF to reproduce the original data set. The COD provides the goodness of fit (r 2) between the observed and predicted concentration of each congener and equals 1.0 for a perfect fit (9).
Results and Discussion Lake Shinji. Lake Shinji (81 km2 surface area, 4.5 m mean depth) resides in a rural area of Shimane Prefecture, Japan. The Lake Shinji basin (1227 km2) consists mainly of forest (88%) with some agricultural fields (6). Rice paddies comprised 9% (114 km2) of the lake basin in 1985 (6). Figure 2 shows four factors obtained from the PMF analysis of the Lake Shinji sediment core PCDD/Fs profile. Figure 2A-D shows the fingerprints (fkj in eq 1), and circles in Figure 2E-H show the total tetra- to octachlorinated PCDD/Fs arising from each fingerprint. The OcCDD was present in more than 5-fold excess of all other congeners in all samples and was omitted from Figure 2A-D. By increasing the number of factors from three to four, COD of all congeners except 1,2,3,9-TeCDF improved to 0.6 and above. An additional (fifth) factor was similar to the fourth factor in Figure 2 and was not included. Results of diagnostic analyses are provided in Table S1 of the Supporting Information (SI). Contribution of each fingerprint is also given in % factor contribution to total PCDD/F concentrations in Figure S2 of the SI. Two factors showed distinct PCDD/F profiles for the synthetic byproducts of two herbicides widely applied to rice paddies in Japan between 1957 and 1995 (22): PCP (Figure 2A) and CNP (Figure 2B). Figure 2A is dominated by heptachlorinated (1,2,3,4,6,7,9- and 1,2,3,4,6,7,8-substituted) homologues and OcCDF. This congener profile characterizes the PCDD/F byproducts formed during the chlorination of phenol to form PCP (22). Figure 2B is dominated by the signature TeCDD byproducts of CNP synthesis from 2,4,6trichlorophenol and 4-chloronitrobenzene: 1,3,6,8- and 1,3,7,9-TeCDDs (22). Reference congener patterns used to identify PCP and CNP are given in Figure S1A-B of the SI. Circles (left y-axis) in Figure 2E and F show the time series of contributions associated with the factors attributable to PCP and CNP, respectively. Bars (right y-axis) show the annual
FIGURE 2. Chemical fingerprints (A-D) and total PCDD/Fs arising from each fingerprint (circles in E-H; left y-axis) in the dated sediment layers from Lake Shinji. In A-D, vertical lines separate PCDDs and PCDFs. Bars (right y-axes) represent the annual PCP (E) and CNP (F) consumption in Japan. consumption of PCP and CNP in Japan (22) for comparison. For both PCP and CNP, time trends in the PMF-derived factor contribution (circles) followed that of the reported national consumption (bars). While contribution of the PCP factor peaked to 10.6 ng‚g-1 in 1968, that of CNP peaked to 9.1 ng‚g-1 in 1976 (circles in Figure 2E and F). Indeed, CNP was introduced as a replacement of PCP in 1970s and was used in lower annual amounts (bars in Figure 2E and F) (22). As illustrated above, PMF resolves the original data set (sample
by congener) into two independent matrices that complement one another in identifying a factor: sample by factor and factor by congener (eq 1). Chemical composition of the third fingerprint (Figure 2C) closely resembled the congener profiles of PCDD/Fs arising from waste incineration processes in Kanto Region of Japan (23) (Figure S1C in the SI) and contributed strongly to sediment layers dated 1971 and younger (Figure 2G). In Japan, emission of PCDD/Fs from municipal and industrial waste VOL. 41, NO. 8, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2705
FIGURE 3. Stacked bars show the contribution of each PMF-derived fingerprint on the dated sediment layers from Lake Shinji (A-D) and Tokyo Bay (E-H). Circles show the congener profiles of the original data sets. Vertical lines separate PCDDs and PCDFs. incineration plants continued to rise since 1960s, reaching maximum by 1997 (24). From its congener profile and time trends, third factor (Figure 2C and G) was attributed to combustion. Fourth factor (Figure 2D) showed 1,2,3,4,6,7,9-HpCDD, 1,2,3,4,6,7,8-HpCDD, 1,2,4,6,7,9/1,2,4,6,8,9-HxCDD, and 1,2,3,6,7,9/1,2,3,6,8,9-HxCDD in the decreasing order of abundance. This congener profile closely resembled the oldest sediment layer in Lake Shinji dated 1947 (circles in 2706
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 8, 2007
Figure 3A). While the absolute contribution (in pg(g-dry sediment)-1) of the fourth factor did not show a clear time trend (Figure 2H), sample-specific, % factor contribution to total PCDD/F concentrations steadily increased with the sediment depth and is negligible for newer layers (Figure S2H). To capture the time trends of the fourth factor on molar basis, concentrations of the main congeners in Figure 2D were converted to % contribution to total PCDD/F concentrations in moles (mole%) and were plotted as a function of
TABLE 1. Slope, Intercept, and r 2 for the Linear Regression of the Relative Concentration (% Contribution to Total PCDD/F Concentrations in moles; mol%) versus Sediment Age (y) for the Main Congeners of the Dechlorination Factor for Lake Shinji (A; n ) 4) and Tokyo Bay (B and C; n ) 5)a homologue
isomer
HpCDD HxCDD
1,2,3,4,6,7,9 1,2,4,6,7,9/1,2,4,6,8,9 1,2,3,6,7,9/1,2,3,6,8,9 1,2,4,6,9 1,2,4,6/1,2,4,9/1,2,6,8/1,4,7,8
PeCDDb TeCDDb HpCDD HxCDD PeCDD TeCDDb HxCDF PeCDF TeCDF
1,2,3,4,6,7,9 1,2,4,6,7,9/1,2,4,6,8,9 1,2,3,4,6,8 1,2,3,6,7,9/1,2,3,6,8,9 1,2,4,6,8/1,2,4,7,9 1,3,7,9 1,2,3,4,7,8 1,2,3,7,8 1,2,3,6,8/1,2,4,7,8/ 1,3,4,6,7/1,3,4,7,8/1,2,4,6,7 2,3,7,8
a Values in parentheses show standard deviation. is given for reference.
intercept
r2
A. Lake Shinji PCDDs (-3.591 × 10-2 ((2.310 × 10-2) (-3.022 × 10-2 ((5.943 × 10-3) (-9.801 × 10-3 ((4.187 × 10-3) (-3.044 × 10-3 ((4.771 × 10-4) (-1.932 × 10-3 ((4.917 × 10-4)
76.25 ((45.21) 60.19 ((11.64) 19.84 ((8.198) 6.014((0.9341) 3.814((0.9627)
0.5473 0.9282 0.7325 0.9532 0.8854
B. Tokyo Bay PCDDs (-2.132 × 10-1 ((2.439 × 10-2) (-1.135 × 10-1 ((8.529 × 10-3) (-4.485 × 10-2 ((6.122 × 10-3) (-4.144 × 10-2 ((5.681 × 10-3) (-4.172 × 10-2 ((4.248 × 10-3) (-2.658 × 10-2 ((5.800 × 10-3)
421.7 ((47.51) 222.8 ((16.62) 88.01 ((11.93) 81.65 ((11.07) 81.90 ((8.276) 55.24 ((11.30)
0.9622 0.9833 0.9471 0.9466 0.9698 0.8750
C. Tokyo Bay PCDFs (-3.488 × 10-2 ((5.833 × 10-3) (-2.595 × 10-2 ((2.511 × 10-3)
68.62 ((11.36) 50.95 ((4.892)
0.9226 0.9727
(-3.289 × 10-2 ((5.826 × 10-3) (-3.975 × 10-2 ((2.751 × 10-3)
64.74 ((11.35) 78.00 ((5.359)
0.9139 0.9858
slope (mole%/year)
b
Isomer (not the main congener of PMF dechlorination factor) having the most negative slope
sediment age for the time period. The fourth factor showed strong contribution but the other factors did not (19471965). Of the main congeners in Figure 2D, 1,2,3,4,6,7,9HpCDD, 1,2,4,6,7,9/1,2,4,6,8,9-HxCDD, and 1,2,3,6,7,9/ 1,2,3,6,8,9-HxCDD (but not 1,2,3,4,6,7,8-HpCDD) afforded negative slopes (Table 1A), despite an increase in ΣPCDD/Fs with time (Figure S4 in the SI). For PeCDDs and TeCDDs, 10 out of 12 chromatographic peaks afforded negative slopes and an isomer (or a coeluting peak) having the most negative slope, 1,2,4,6,9-PeCDD and 1,2,4,6/1,2,4,9/1,2,6,8/1,4,7,8TeCDD, is given in Table 1A. While the slopes of 1,2,3,4,6,7,9HpCDD and 1,2,4,6,7,9/1,2,4,6,8,9-HxCDD are within the error range, the most negative slopes for PeCDDs and TeCDDs are an order of magnitude lower (Table 1A). Tokyo Bay. In contrast to Lake Shinji, Tokyo Bay (980 km2 surface area) is an inner bay in the Tokyo Metropolitan area surrounded by various industrial plants as well as agricultural fields that comprise a considerable percentage of the catchment area (approximately 20%) (16). Indeed, time trends of the total PCDD/F concentrations (pg/g) in Tokyo Bay are similar to those of Lake Shinji but are approximately 4-fold greater in magnitude (Figure S4). Figure 4 shows four factors obtained from the PMF analyses of the PCDD/Fs in a core and surface sediments from Tokyo Bay (Figure 1). By increasing the number of factors from three to four, COD of all congeners improved to 0.6 and above. An additional (fifth) factor was similar to the fourth factor in Figure 4 and was excluded from further analysis. The COD for the Tokyo Bay data are provided in Table S2 of the SI. For each factor in Figure 4, % factor contribution to total PCDD/F concentrations is provided in Figure S3 of the SI. Following the same procedure described above for the Lake Shinji sample, first three factors were attributed to PCP, CNP, and combustion (Figure 4A-C and E-G). Dechlorination Pathways. Similarly to the Lake Shinji sample, PMF analysis of PCDD/Fs in Tokyo Bay sediment core afforded a fourth factor showing an increasing % factor contribution to total PCDD/F concentrations toward older layers (1937-1959; Figure S3H). For the time period 19371959 the fourth factor contributed significantly (Figure S3H) and other factors did not. Mole% of following congeners representative of the fourth factor exhibited negative slopes as a function of sediment age (Table 1B), despite an increase
in ΣPCDD/Fs (Figure S4): 1,2,3,4,6,7,9-HpCDD, HxCDD (1,2,4,6,7,9/1,2,4,6,8,9; 1,2,3,4,6,8; 1,2,3,6,7,9/1,2,3,6,8,9), 1,2,4,6,8/1,2,4,7,9-PeCDD, 1,2,3,4,7,8-HxCDF, PeCDF (1,2,3,7,8; 1,2,3,6,8/1,2,4,7,8/1,3,4,6,7/1,3,4,7,8/1,2,4,6,7), and 2,3,7,8TeCDF. All TeCDDs exhibited negative slopes and 1,3,7,9TeCDD, an isomer having the most negative slope, is shown in Table 1B for reference. As in the Lake Shinji case, while the slopes of 1,2,3,4,6,7,9-HpCDD and 1,2,4,6,7,9/1,2,4,6,8,9HxCDD are within the error range, slopes for PeCDDs and TeCDDs are an order of magnitude lower (Table 1B). In addition, nearly an order of magnitude greater slope was obtained for the same congeners (1,2,3,4,6,7,9-HpCDD, 1,2,4,6,7,9/1,2,4,6,8,9-HxCDD, and 1,2,3,6,7,9/1,2,3,6,8,9-HxCDD) in Tokyo Bay than in Lake Shinji (Table 1A and B). Figure S5 in the SI provides individual plots for the results summarized in Table 1. The trends observed for PCDDs in Table 1 suggest in situ lateral dechlorination of OcCDD and other highly chlorinated PCDDs. Removal of chlorines substituted at 8- and 3-positions in sequence results in the main HpCDD and HxCDD of the fourth factor in Lake Shinji and Tokyo Bay: 1,2,3,4,6,7,9HpCDD and 1,2,4,6,7,9-HxCDD (note 1,2,4,6,8,9-HxCDD coelutes with this congener). Both in Tokyo Bay and Lake Shinji, while mole% of most TeCDDs and PeCDDs increased with depth, only HxCDDs and HpCDDs fully chlorinated at 1,4,6,9-positions showed the same trend. In addition, an order of magnitude greater slope was obtained for HpCDDs and HxCDDs than PeCDDs and TeCDDs (Table 1A and B). The observations above suggest that the preference for lateral mechanism (over peri) becomes less significant for lower chlorinated (particularly penta and below) PCDDs. That is, dechlorination becomes less selective and less favorable for PCDDs possessing five or less chlorines per molecule, resulting in the accumulation of HpCDDs and HxCDDs fully chlorinated at 1,4,6,9-carbons (circles in Figure 3A and E). For PCDFs, the trend observed in Table 1C indicates preferred peri dechlorination. Unlike PCDDs, slopes of tetra- to hexachlorinated congeners were within the error range for PCDFs (Table 1C). For both PCDDs and PCDFs, linear increase in the concentrations of putative dechlorination products (Table 1) suggests that decades of reductive dechlorination are yet to reach equilibrium. Inverse of slopes in Table 1A suggest that at least 27.8 ( 17.9 year(mole %)-1 (for 1,2,3,4,6,7,9VOL. 41, NO. 8, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2707
FIGURE 4. Chemical fingerprints (A-D) and total PCDD/Fs arising from each fingerprint (circles in E-H; left y-axis) in Tokyo Bay. Bars (right y-axes) represent the annual PCP (E) and CNP (F) consumption in Japan. S1-7 in G-H denote surface sediments. In A-D, vertical lines separate PCDDs and PCDFs. HpCDD) is required for the formation of lateral dechlorination products in Lake Shinji. In Tokyo Bay, at least 4.7 ( 0.5 year(mole %)-1 (for 1,2,3,4,6,7,9-HpCDD) and 25.2 ( 1.7 year(mole %)-1 (for 2,3,7,8-TeCDF) are required for the formation of lateral and peri dechlorination products, respectively. As Figure S5B indicates, fastest reaction identified in the present study (formation of 1,2,3,4,6,7,9-HpCDD in Tokyo Bay) takes 20 years to double the product concentration. Experimentally observed biotic (2) and abiotic 2708
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 8, 2007
(3) PCDD dechlorination proceeds much faster, achieving the same extent of dechlorination within a few months. Molecular orbital calculations suggested that highly chlorinated hexa- to octa- dibenzo-p-dioxins have lower HOMO-LUMO gap and are more susceptible toward dechlorination, relative to tetra- and penta- homologues (5). In addition, dechlorination is thermodynamically favorable at the most electron-deficient 2,3,7,8-chlorinated carbons (Figure S6 in SI). As a result, preference for lateral over peri
mechanism increases with the degree of chlorination (5). Laboratory studies demonstrated thermodynamic control on abiotic reduction of 1,2,3,4,6,7,9-HpCDD by Aldrich humic acid (3). Both the rate and extent of product formation increased with the degree of chlorination, reaching a maximum rate of approximately 0.17 nM(day)-1 for HxCDDs (compared to 0.04 nM(day)-1 for TeCDDs) in the presence of bulk reductant titanium(III) citrate (3). Similar rate dependence on the level of chlorination was observed in peri dechlorination of OcCDD in organic acid-amended incubation of historically PCDD/F-contaminated estuarine sediments (4). The rate of 1,2,3,4,6,7,8-HpCDD formation was more than an order of magnitude greater than that of 2,3,7,8-TeCDD (4). However, there are additional parameters that strongly influence the rate and mechanism of PCDD dechlorination: microbial community, organic carbon content, salinity, sulfate concentration, and hydrogen level (4, 25). In our present study, thermodynamic control on the dechlorination rates suggests that both biotic and abiotic dechlorination pathways are operative in Lake Shinji and Tokyo Bay sediments. Gaus et al. (10) suggested the importance of lateral dechlorination based on the shared characteristics of PCDD/F profiles for sediment core samples dated as old as 350 years collected around the globe: an increase in the relative concentrations of lower chlorinated, 1,4,6,9-substituted congeners with depth. The authors hypothesized that a PCDD/F precursor (e.g., PCP) penetrated to the deeper layers, from which OcCDD (and to lesser extend HpCDDs) was formed in situ (10). Subsequent, preferred lateral dechlorination of OcCDD likely resulted in the dominance of congeners fully chlorinated in the 1,4,6,9- positions (10). It must be noted that there are additional sources of highly chlorinated PCDD/Fs in deep sediment layers that will impact dechlorination pathways: microbially mediated coupling of naturally occurring chlorophenols (26), deposition of PCP and its photochemical products in rain (27), and forest fires (28). In a separate report, PCA of surface sediment and soilderived PCDD/Fs in Kanto Region, Japan afforded an unknown factor dominated by penta- to hexa-homologues fully chlorinated at 1,2,6,9-carbons (8). The 1,2,6,9-pattern (rather than 1,4,6,9-pattern suggested by Gaus et al. (10)) was more prominent in an older sediment collected at Lake Kasumigaura (40 years of sedimentation) than a newer sample from Tokyo Bay (10 years of sedimentation; both sampled between 1993 and 1995) (8). Indeed, the two main lateral dechlorination products 1,2,3,4,6,7,9-HpCDD and 1,2,4,6,7,9/ 1,2,4,6,8,9-HxCDD (Figures 2D and 4D) are fully chlorinated at 1,2,6,9- and 1,4,6,9- positions. Dechlorination of 1,2,4,6,7,9HxCDD (the most unstable HxCDD) is equally favorable at 2- and 7-positions (Figure S6). While dechlorination at 7-position will retain both 1,2,6,9- and 1,4,6,9- patterns, dechlorination at 2-position will retain only the 1,4,6,9pattern. Based on the time trends of dechlorination fingerprints obtained in the present study, we suggest that both 1,4,6,9- and 1,2,6,9- patterns result from preferred lateral dechlorination of highly chlorinated dibenzo-p-dioxins. In our present study, PMF-derived fingerprints reasonably well reproduced the original data sets in both Lake Shinji and Tokyo Bay samples (Figure 3). Figure 3 indicates that Lake Shinji and Tokyo Bay share common sources and dechlorination patterns possessing similar time trends. Time trends of major sources, PCP, CNP, and combustion are in agreement with previous reports based on PCA and MRA (6, 7), and inclusion of dechlorination fingerprint significantly improved reproducibility of the original data set.
Acknowledgments This work was supported by the 21st Century COE Program “Environmental Risk Management for Bio/Eco-systems” of the Ministry of Education, Culture, Sports, Science and Technology of Japan.
Supporting Information Available Diagnostic tools for the PMF analysis, reference congener profiles, percent factor contributions of PMF-derived fingerprints, total concentration of each homologue in Lake Shinji and Tokyo Bay, kinetics of dechlorination pathways, and driving force for the lateral dechlorination of OcCDD. This material is available free of charge via the Internet at http://pubs.acs.org.
Literature Cited (1) Bunge, M.; Adrian, L.; Kraus, A.; Opel, M.; Lorenz, W. G.; Andreesen, J. R.; Gorisch, H.; Lechner, U. Reductive dehalogenation of chlorinated dioxins by an anaerobic bacterium. Nature 2003, 421, 357-360. (2) Ballerstedt, H.; Kraus, A.; Lechner, U. Reductive dechlorination of 1,2,3,4-tetrachlorodibenzo-p-dioxin and its products by anaerobic mixed cultures from Saale River sediment. Environ. Sci. Technol. 1997, 31, 1749-1753. (3) Fu, Q. S.; Barkovskii, A. L.; Adriaens, P. Reductive transformation of dioxins: An assessment of the contribution of dissolved organic matter to dechlorination reactions. Environ. Sci. Technol. 1999, 33, 3837-3842. (4) Albrecht, I. D.; Barkovskii, A. L.; Adriaens, P. Production and dechlorination of 2,3,7,8-tetrachlorodibenzo-p-dioxin in historically-contaminated estuarine sediments. Environ. Sci. Technol. 1999, 33, 737-744. (5) Lynam, M. M.; Kuty, M.; Damborsky, J.; Koca, J.; Adriaens, P. Molecular orbital calculations to describe microbial reductive dechlorination of polychlorinated dioxins. Environ. Toxicol. Chem. 1998, 17, 988-997. (6) Masunaga, S.; Yao, Y.; Ogura, I.; Nakai, S.; Kanai, Y.; Yamamuro, M.; Nakanishi, J. Identifying sources and mass balance of dioxin pollution in Lake Shinji Basin, Japan. Environ. Sci. Technol. 2001, 35, 1967-1973. (7) Masunaga, S.; Yao, Y.; Ogura, I.; Sakurai, T.; Nakanishi, J. Source and behavior analyses of dioxins based on congener-specific information and their application to Tokyo Bay basin. Chemosphere 2003, 53, 315-324. (8) Sakurai, T. Dioxins in aquatic sediment and soil in the Kanto region of Japan: major sources and their contributions. Environ. Sci. Technol. 2003, 37, 3133-3140. (9) Johnson, G. W.; Ehrlich, R.; Full, W. Principal components analysis and receptor models in environmental forensics. In An Introduction to Environmental Forensics; Murphy, B. L., Morrison, R. D., Eds.; Academic Press: San Diego, CA, 2002; pp 461-515. (10) Gaus, C.; Brunskill, G. J.; Connell, D. W.; Prange, J.; Muller, J. F.; Papke, O.; Weber, R. Transformation processes, pathways, and possible sources of distinctive polychlorinated dibenzop-dioxin signatures in sink environments. Environ. Sci. Technol. 2002, 36, 3542-3549. (11) Barabas, N.; Goovaerts, P.; Adriaens, P. Modified polytopic vector analysis to identify and quantify a dioxin dechlorination signature in sediments. 2. Application to the Passaic River. Environ. Sci. Technol. 2004, 38, 1821-1827. (12) Bzdusek, P. A.; Lu, J. H.; Christensen, E. R. PCB congeners and dechlorination in sediments of Sheboygan River, Wisconsin, determined by matrix factorization. Environ. Sci. Technol. 2006, 40, 120-129. (13) Bzdusek, P. A.; Christensen, E. R.; Lee, C. M.; Pakdeesusuk, U.; Freedman, D. L. PCB congeners and dechlorination in sediments of Lake Hartwell, South Carolina, determined from cores collected in 1987 and 1998. Environ. Sci. Technol. 2006, 40, 109119. (14) Pakdeesusuk, U.; Lee, C. M.; Coates, J. T.; Freedman, D. L. Assessment of natural attenuation via in situ reductive dechlorination of polychlorinated biphenyls in sediments of the Twelve Mile Creek Arm of Lake Hartwell, SC. Environ. Sci. Technol. 2005, 39, 945-952. (15) Yao, Y.; Masunaga, S.; Takada, H.; Nakanishi, J. Identification of polychlorinated dibenzo-p-dioxin, dibenzofuran, and coVOL. 41, NO. 8, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2709
(16)
(17) (18) (19) (20) (21) (22)
planar polychlorinated biphenyl sources in Tokyo Bay, Japan. Environ. Toxicol. Chem. 2002, 21, 991-998. Sakurai, T.; Kim, J. G.; Suzuki, N.; Matsuo, T.; Li, D. Q.; Yao, Y. A.; Masunaga, S.; Nakanishi, J. Polychlorinated dibenzo-pdioxins and dibenzofurans in sediment, soil, fish, shellfish and crab samples from Tokyo Bay area, Japan. Chemosphere 2000, 40, 627-640. Kanai, Y.; Inouchi, Y.; Yamamuro, M.; Tokuoka, Y. Sedimentation rate and environment in Lake Shinji, Shimane Prefecture. Chikyukagaku (Geochemistry) 1998, 32, 71-85 (in Japanese). Johnson, G. W.; Ehrlich, R. State of the art report on multivariate chemometric methods in environmental forensics. Environ. Forensics 2002, 3, 59-79. Paatero, P.; Tapper, U. Positive matrix factorization: A nonnegative factor model with optimal utilization of error estimates of data values. Environmetrics 1994, 5, 111-126. EPA PMF 1.1 for Windows; U.S. EPA: Research Triangle Park, NC, 2005. Paatero, P. Least squares formulation of robust non-negative factor analysis. Chemometr. Intell. Lab. 1997, 37, 23-35. Masunaga, S.; Takasuga, T.; Nakanishi, J. Dioxin and dioxinlike PCB impurities in some Japanese agrochemical formulations. Chemosphere 2001, 44, 873-885.
2710
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 8, 2007
(23) Ogura, I.; Masunaga, S.; Nakanishi, J. Congener-specific characterization of PCDDs/PCDFs in atmospheric deposition: comparison of profiles between deposition, source and environmental sink. Chemosphere 2001, 45, 173-183. (24) Dioxins Emission Inventory; Environmental Agency of Japan, 2000 (in Japanese). (25) Fu, Q. S.; Barkovskii, A. L.; Adriaens, P. Dioxin cycling in aquatic sediments: the Passaic River Estuary. Chemosphere 2001, 43, 643-648. (26) Green, N. J. L.; Hassanin, A.; Johnston, A. E.; Jones, K. C. Observations on historical, contemporary, and natural PCDD/ Fs. Environ. Sci. Technol. 2004, 38, 715-723. (27) Baker, J. I.; Hites, R. A. Is combustion the major source of polychlorinated dibenzo-p-dioxins and dibenzofurans to the environment? A mass balance investigation. Environ. Sci. Technol. 2000, 34, 2879-2886. (28) Gullett, B. K.; Touati, A. PCDD/F emissions from forest fire simulations. Atmos. Environ. 2003, 37, 803-813.
Received for review November 17, 2006. Revised manuscript received February 5, 2007. Accepted February 6, 2007. ES0627444
