Identification of polychlorinated dibenzo-p-dioxin, dibenzofuran, and coplanar polychlorinated biphenyl sources in Tokyo Bay, Japan
Abstract
A dated sediment core collected from Tokyo Bay, Japan, was used to assess the historical inputs of polychlorinated dibenzo-p-dioxins (PCDDs), polychlorinated dibenzofurans (PCDFs), and coplanar polychlorinated biphenyls (Co-PCBs) from multiple sources. The levels, patterns, and profiles of these compounds in the core were congener-specifically investigated, and the results show that the PCDD and PCDF (PCDD/F) and the Co-PCB inputs increased abruptly from the late 1950s and peaked during the period 1967 to 1972. From 1972 to 1981, the inputs decreased continuously and then generally leveled off. Using principal component analysis, two herbicides widely used in the past, pentachlorophenol (PCP) and chloronitrofen (CNP), as well as combustion processes were identified as the major dioxin sources in Tokyo Bay. The PCB formulations and combustion processes were estimated to be the major sources of Co-PCBs. Furthermore, multiple regression analysis was performed for dioxin-source apportioning, and it was found that the herbicides PCP and CNP have mainly contributed to the PCDD/F burdens since the late 1950s. This study suggests that herbicide-derived PCDD/Fs remaining in agricultural land will continue to run off and pollute the aquatic environment in Japan for a long time.
INTRODUCTION
Polychlorinated dibenzo-p-dioxins (PCDDs), dibenzofurans (PCDFs), and biphenyls (PCBs) constitute a group of persistent, bioaccumulative, and toxic contaminants in the environment. Several PCDDs and PCDFs (PCDD/Fs) and coplanar PCBs (co-PCBs) have been shown to cause toxic responses similar to those caused by 2,3,7,8-tetrachlorodibenzo-p-dioxin (2,3,7,8-TCDD), the most potent congener within these groups of compounds. These toxic responses include dermal toxicity, immunotoxicity, carcinogenicity, and adverse effects on reproduction, development, and endocrine functions [1]. To conduct a comprehensive dioxin risk assessment for humans and the ecosystem, the toxic equivalent (TEQ) approach has been developed and is now used worldwide. In Japan, the tolerable daily intake was revised in June 1999 to 4 pg TEQ/kg/d for the sum of PCDD/Fs and Co-PCBs to control the exposure to these compounds. For this purpose, a full understanding of the key sources of PCDD/Fs and Co-PCBs and the relative contributions of these sources is required.
Olie et al. [2] first found PCDD/Fs in fly ash and flue gas of some municipal incinerators in The Netherlands in 1977. Since then, many PCDD/F sources have been identified, and these can generally be divided into three categories [3]: Industrial processes [4-6], thermal processes [7-9], and secondary sources or reservoirs. The contribution of dioxins from different sources is a topic of intense discussion. In Europe, incineration is generally thought to be the primary source, and atmospheric emissions of PCDD/Fs are reported to have decreased following the strong regulatory control of such processes [10, 11]. On the other hand, a Canadian study found that pentachlorophenol (PCP), which was used extensively as a wood preservative, was one of the most significant sources of dioxins in that country [12]. In addition, Baker and Hites [13] recently noted that the photochemical synthesis of octachlorodibenzo-p-dioxins (OCDD) from PCP in atmospheric condensed water might be the most significant source of OCDD to the environment.
In Japan, incineration has been deemed by many to contribute the greatest portion of PCDD/Fs to the environment, and dioxin-control measures have been focused only on incineration until now. Recently, however, Masunaga et al. [14] reported that a significant portion of dioxins in Japanese surface sediments originated from agrochemicals, especially PCP and chloronitrofen (CNP), which were widely used as paddyfield herbicides in the past. Furthermore, Masunaga and Nakanishi [15] investigated the dioxin impurities in agrochemicals used in the past in Japan, and they proposed that the annual emission of dioxins from agrochemicals was far greater than that from combustion sources during the 1960s and 1970s. For Co-PCBs, commercial PCB mixtures such as Aroclor, Kanechlor, and Clophen are known to contain Co-PCB congeners [16, 17]. Several recent studies have indicated that municipal waste incineration may result in Co-PCB formation [8, 9, 18].
The time course of dioxins and PCBs in the environment provides an important clue regarding the sources of these compounds [19, 20]. The objective of this study was to reconstruct the historical trends and to further elucidate the key sources of PCDD/Fs and Co-PCBs in the Japanese aquatic system to provide useful information for the establishment of comprehensive dioxin and PCB control measures in Japan.
Location of sampling site in Tokyo Bay, Japan.
MATERIALS AND METHODS
Study area
Tokyo Bay is a typical closed inner bay located southeast of Tokyo, Japan. It has a surface area of 980 km2; maximum and mean depths of 50 and 15 m, respectively; and a hydraulic residence time of 1.6 months. The catchment area is 7,600 km2, in which 25.6 million people reside, corresponding to nearly one-fifth the total population of Japan. Tokyo Bay was chosen because it is considered to be one of the water systems most affected by human activities in Japan. For example, many industrial plants are located along the shoreline. Municipal solid-waste incinerators in the area burn more than 6 million tons/year. Furthermore, herbicides have been used in paddy fields and other agricultural fields, which comprise approximately 20% of the catchment area [21]. Environmental pollution of Tokyo Bay proceeded intensively from the late 1950s and peaked in the 1970s, but contamination by organic compounds and heavy metals continues to date [22].
Sampling and dating
Details of the field sampling program and sediment dating methods are given by Sanada et al. [22]. Brief descriptions are provided here.
A sediment core of approximately 60 cm in length was collected from Tokyo Bay with a Matsumoto gravity corer (Rigo, Tokyo, Japan) in September 1993. The coring site (F2) is in the middle of Tokyo Bay at 35°33′ N latitude and 139°55′ E longitude (Fig. 1), with a water depth of 18 m. The core was sliced into 1-cm disks on board, and the sediment of each disk was immediately transferred to polypropylene containers. All disks were then freeze-dried and kept frozen until required.
To estimate the sedimentation rate, vertical distributions of excess 210Pb and 137Cs activity concentrations in the core were investigated. The total 210Pb and 137Cs activity concentrations in each disk were determined by γ-spectrometry using a high-purity germanium detector (Ortec, Oak Ridge, TN). Supported 210Pb was obtained by indirectly determining the activity concentration of the supporting parent 226Ra. The 226Ra was measured by analyzing its decay product, 214Pb, on the assumption that the two were in equilibrium. The excess 210Pb activity concentration was plotted on a logarithmic scale against the cumulative mass per unit area. A regression line was fitted to the data, and the slope of the regression line gave a 210Pb-derived average sedimentation rate. The distribution of 137Cs activity concentration in the core was related to fallout from nuclear weapons testing. The 137Cs-derived average sedimentation rate was obtained by assigning the peak in 137Cs deposition to 1963. Although some distraction was observed from the obtained excess 210Pb and 137Cs activity profiles of the sediment core, trends of molecular markers such as PCBs, linear alkylbenzenes, and tetrapropylene-based alkylbenzenes matched the use of these chemicals in the area, indicating that the sediment core was suitable for our historical study [22].
The average sedimentation rate was estimated to be 0.27 g/cm2/year by the 210Pb method, with a range of 0.20 to 0.40 g/cm2/year. On the basis of the 137Cs method, the average sedimentation rate was estimated to be 0.26 g/cm2/year (range, 0.18–0.29 g/cm2/year). In addition, a molecular stratigraphy approach was also performed using PCBs as a molecular marker, and the average sedimentation rate was then estimated to be 0.27 g/cm2/year (0.26–0.29 g/cm2/year) by assigning the peak in PCB deposition to 1970 [22]. All these estimates are similar, and the 210Pb-derived average sedimentation rate was used for dating in this study [23].
Extraction and cleanup
A related study on the historical trends of endocrine disrupters in Tokyo Bay was previously reported by Okuda et al. [24], using the same sediment core, in which they composited the individual 1-cm core disks into 16 subsamples at 3- or 6-cm intervals. In the present study, the disks were subsampled to make up 13 samples at various depths (0–3, 3–6, 6–9, 9–15, 15–20, 20–25, 25–30, 30–35, 35–40, 40–45, 45–50, 50–55, and 55–58 cm) for our PCDD/F and Co-PCB analysis.
After the addition of 16 13C-labeled PCDD/F and 14 13C-labeled Co-PCB internal standards (Wellington Laboratories, Guelph, ON, Canada), each sample (5–10 g) was Soxhlet (Millville, NJ, USA) extracted with toluene for 20 h. The extracts were hydrolyzed with aqueous potassium hydroxide solution at room temperature, treated with concentrated sulfuric acid, washed with n-hexane-extracted water, and then treated with activated copper. Sample cleanup included chromatography on silica gel, alumina, and activated carbon columns [25]. The silica gel column was packed with 2 g of silica gel (Wakogel S-1; Wako Pure Chemical Industries, Osaka, Japan) heated at 130°C for 3.5 h, and 120 ml of n-hexane was utilized for eluting PCDD/Fs and PCBs. The alumina column was packed with 5 g of basic alumina (Aluminium oxide 60, Activity I; Merck, Darmstadt, Germany) heated at 190°C for 3 h. This column was first eluted with 30 ml of n-hexane containing 2% (v/v) dichloromethane and then with 30 ml of n-hexane containing 50% (v/v) dichloromethane. The last fraction was loaded onto the activated carbon column packed with 0.5 g of activated carbon-impregnated silica gel (Wako Pure Chemical Industries). This column was first eluted with 20 ml of n-hexane containing 25% (v/v) dichloromethane, and the eluate was added to the initial fraction of the alumina column chromatography for collecting mono-/di-ortho and normal PCBs. Then, 200 ml of toluene was used for eluting PCDD/Fs and nonortho PCBs. The final PCDD/F and nonortho PCB fractions were concentrated to 25 μl, whereas the mono-/diortho and normal PCB fractions were concentrated to 1 ml. All these fractions were spiked with two 13C12-labeled recovery standards (Cambridge Isotope Laboratories, Andover, MA, USA) for high-resolution gas chromatography/high-resolution mass spectrometry (HRGC-HRMS) analysis.
HRGC-HRMS analysis
Both PCDD/Fs and Co-PCBs were analyzed by HRGC (HP6890; Hewlett-Packard, Wilmington, DE, USA)-HRMS (AutoSpec; Micromass, Manchester, UK). Both DB-5 and DB-17 columns (length, 60 m; inner diameter, 0.25 mm; film thickness, 0.25 μm; J&W Scientific, Folsom, CA, USA) were used to separate seventeen 2,3,7,8-substituted PCDD/F (2,3,7,8-PCDD/F) congeners, in which the congeners being interfered on DB-5 were quantified with DB-17. The DB-5 column was utilized for the analysis of other PCDD/F and Co-PCB congeners. An autosampler (gas chromatography system injector; Hewlett-Packard) was employed for injection (2 μl, splitless).
For PCDD/F determination, the following temperature programs were used: for DB-5, 160°C for 3 min, 40°C/min to 200°C, hold for 2 min, and 2°C/min to 310°C; and for DB-17, 160°C for 3 min, 40°C/min to 220°C, hold for 2 min, 2°C/min to 280°C, and hold for 33.5 min. In the case of Co-PCBs, the temperature programs used were as follows: for nonortho PCBs, 120°C for 1 min, 40°C/min to 200°C, hold for 2 min, 6°C/min to 320°C, and hold for 5 min; and for mono-/di-ortho PCBs, 70°C for 1 min, 40°C/min to 190°C, l°C/min to 240°C, 10°C/min to 310°C, and hold for 9 min. The temperatures of the injector and the ion source were 280°C and 250°C, respectively. The interface temperature was set at the maximum value of each temperature program. The carrier gas was helium, and the electron-impact ionization energy was 40 eV.
The mass spectrometer was operated at a resolution of 10,000–13,000 (5% valley) and in a selected ion monitoring mode. Tetra- to octachlorinated PCDD/Fs and 14 Co-PCBs International Union of Pure and Applied Chemistry 77, 81, 126, 169, 105, 114, 118, 123, 156, 157, 167, 189, 170, and 180) were analyzed by congener-specific analysis [25]. Toxic equivalent concentrations were calculated based on the toxic equivalency factors (TEFs) for humans and mammals established by the World Health Organization, Paris, France, in 1998 [1]. The accuracy and reliability of the analytical method used in the present study were previously approved by analyzing test samples including harbor sediment, industrial sludge, and fortified industrial soil extract in the fourth round of the International Intercalibration Study on PCDDs, PCDFs, and mono-ortho and planar PCBs [26]. Method blanks generally contained OCDD, but in concentrations no more than 2.5 pg/g. Some other congeners detected were present in much smaller concentrations. The average recoveries for 2,3,7,8-PCDDs, 2,3,7,8-PCDFs, and Co-PCBs were 88 ± 26%, 77 ± 21%, and 84 ± 20%, respectively.
Data analysis
In this study, Statistica software (Statistica 2000, Ver 5.5; StatSoft®, Tulsa, OK, USA) was used for statistical analysis. Source identification was performed using principal component analysis (PCA). The PCA is a multivariate technique that can be used to reduce the dimensionality of complex data. It has been applied to PCDD/F and PCB data by many researchers [12, 19, 27, 28]. For our PCDD/F congener-specific data set, each congener or congener cluster was considered as a variable, whereas each sediment sample was treated as a case. Because the number of cases was smaller than that of variables, congener-specific data were transformed into a correlation matrix, and the matrix was utilized as input data for PCA. Principal components (PCs) were obtained by varimax rotation, and the major PCs were determined based on the cumulative proportion (>95%). The characteristic congeners of each major PC were then extracted based on their factor loadings, and they were compared with those of known sources for source identification. Furthermore, multiple regression analysis (MRA) was applied for dioxin source apportioning. The congener-specific data of PCDD/F concentrations in PCP, CNP, and atmospheric deposition were used to estimate the historical contributions of different sources to dioxin pollution in Tokyo Bay. Both PCA and MRA were not carried out for Co-PCB source identification and apportioning, because our Co-PCB congener-specific data set was too small and Co-PCB congener-specific information on the estimated sources was lacking.
RESULTS AND DISCUSSION
Trends in PCDD/Fs
Because congener profiles (including the non-2,3,7,8-sub-stituted constituents) reflect the characteristics of dioxin sources, we conducted congener-specific analysis for our sediment core samples. Consequently, nearly all the tetra- to octachlorinated PCDD/Fs were detected in the samples examined. Part of the results are given in Table 1. The ΣPCDD/F level increased during the period 1935 to 1972, in which a drastic increase occurred from the late 1950s that peaked around 1970 (ΣPCDD/Fs, 45,000 pg/g). During 1972 to 1981, the total dioxin concentration decreased continuously to 22,000 pg/g and then generally leveled off. The ΣPCDD/F concentration was found to be 620 pg/g in the deepest sediment layer, dated at approximately 1937, which provides the information on the PCDD/F background level before World War II in Tokyo Bay. Combustion of coal and wood as well as metal production might be responsible for the pollution level at that time. For instance, on average, 35 million tons of coal were produced each year in Japan from 1930 to 1937. Using the concentration of ΣPCDD/Fs in the surface sediment layer and the average sedimentation rate estimated by the 210Pb method, the present flux of PCDD/Fs to Tokyo Bay sediment was estimated to be 5,100 pg/cm2/year [29].
A distinct homologue profile dominated by OCDD and some interesting trends were found from the bottom to the top of the core. Figure 2 shows the PCDD/F homologue trends (without those of hexachlorodibenzo-p-dioxins [HxCDDs] and hexachlorodibenzofurans [HxCDFs] for clear illustration). The OCDD, heptachlorodibenzo-p-dioxins (HpCDDs), octachlorodibenzofuran (OCDF), heptachlorodibenzofurans (HpCDFs), HxCDDs, and HxCDFs showed similar trends: They increased dramatically during 1956 to 1972, decreased rapidly during 1972 to 1981, and then leveled off. On the other hand, TCDDs, pentachlorodizenzo-p-dioxins (PeCDDs), and tetra-chlorodizenzofurans (TCDFs), particularly 1,3,6,8-TCDD and 1,3,7,9-TCDD, increased during 1962 to 1977, decreased rapidly during 1977 to 1981, and subsequently leveled off. Moreover, pentachlorodibenzofurans (PeCDFs) showed a different trend from those of other homologues mentioned above, increasing slowly to date. These characteristics indicate the existence of different dioxin sources in the Tokyo Bay area.
Sources of PCDD/Fs
To identify PCDD/F sources, PCA was performed using a correlation matrix calculated from the PCDD/F data set (83 variables and 13 cases). Three major PCs with proportions of approximately 35%, 33%, and 28% were obtained. Because the cumulative proportion of these PCs was more than 95%, it was considered that the dioxin pollution in Tokyo Bay mainly came from three sources (PC-1, PC-2, and PC-3). The factor loadings (>0.8) of the three major PCs are plotted in Figure 3 [30].
| Core depth (cm) | 55–58 | 50–55 | 45–50 | 40–45 | 35–40 | 30–35 | 26–30 | 20–25 | 15–20 | 9–15 | 6–9 | 3–6 | 0–3 |
| Average cumulative weight (g/cm2) | 15.0 | 14.1 | 12.4 | 10.8 | 9.2 | 7.7 | 6.3 | 5.1 | 3.9 | 2.7 | 1.8 | 1.0 | 0.3 |
| Age range (year) | 1935–1938 | 1938–1945 | 1945–1951 | 1951–1956 | 1956–1962 | 1962–1967 | 1967–1972 | 1972–1977 | 1977–1981 | 1981–1986 | 1986–1989 | 1989–1991 | 1991–1993 |
| Average age (year) | 1937 | 1942 | 1948 | 1954 | 1959 | 1965 | 1970 | 1975 | 1979 | 1984 | 1988 | 1990 | 1992 |
| 2378-TCDD | 0.23 | 0.23 | 0.23 | 0.23 | 0.44 | 0.90 | 1.9 | 2.0 | 1.7 | 1.8 | 1.8 | 1.8 | 1.5 |
| TCDDs | 13 | 18 | 17 | 18 | 31 | 110 | 3200 | 8500 | 3800 | 3100 | 3900 | 2800 | 2500 |
| 12378-PeCDD | 0.79 | 1.0 | 0.90 | 1.0 | 1.5 | 3.3 | 6.8 | 5.1 | 6.9 | 5.7 | 6.3 | 6.1 | 5.3 |
| PeCDDs | 16 | 20 | 17 | 20 | 26 | 59 | 500 | 1300 | 660 | 600 | 630 | 530 | 480 |
| 123478-HxCDD | 0.76 | 0.93 | 1.0 | 1.3 | 2.6 | 8.9 | 14 | 9.1 | 9.0 | 9.2 | 10 | 8.8 | 8.3 |
| 123678-HxCDD | 1.7 | 2.1 | 2.1 | 4.1 | 9.8 | 43 | 73 | 46 | 38 | 37 | 36 | 33 | 31 |
| 123789-HxCDD | 2.7 | 3.0 | 3.0 | 3.7 | 5.9 | 18 | 28 | 20 | 18 | 17 | 18 | 16 | 15 |
| HxCDDs | 34 | 40 | 37 | 45 | 82 | 270 | 410 | 360 | 270 | 340 | 340 | 300 | 270 |
| 1234678-HpCDD | 21 | 27 | 36 | 110 | 340 | 1800 | 2700 | 1500 | 1100 | 1100 | 1100 | 1000 | 940 |
| HpCDDs | 75 | 82 | 100 | 210 | 620 | 3100 | 4400 | 2600 | 1900 | 2000 | 2000 | 1800 | 1700 |
| 12346789-OCDD | 380 | 440 | 580 | 1500 | 5000 | 24000 | 30000 | 17000 | 12000 | 13000 | 14000 | 11000 | 11000 |
| 2378-TCDF | 4.5 | 4.6 | 4.7 | 5.4 | 7.4 | 12 | 17 | 11 | 10 | 9.0 | 10 | 9.5 | 8.7 |
| TCDFs | 25 | 27 | 31 | 36 | 59 | 140 | 310 | 380 | 270 | 270 | 300 | 270 | 240 |
| 12378-PeCDF | 3.2 | 3.5 | 3.5 | 4.1 | 5.1 | 7.5 | 11 | 8.7 | 8.7 | 8.8 | 9.7 | 9.2 | 8.3 |
| 23478-PeCDF | 1.6 | 1.7 | 1.8 | 2.3 | 3.2 | 6.0 | 9.7 | 8.3 | 9.6 | 9.5 | 11 | 11 | 9.9 |
| PeCDFs | 17 | 20 | 22 | 30 | 47 | 120 | 210 | 190 | 200 | 210 | 260 | 280 | 240 |
| 123478-HxCDF | 5.3 | 5.9 | 6.9 | 8.5 | 13 | 33 | 53 | 32 | 30 | 30 | 32 | 30 | 28 |
| 123678-HxCDF | 1.8 | 2.3 | 2.7 | 3.6 | 5.1 | 12 | 20 | 15 | 19 | 17 | 20 | 21 | 19 |
| 234678-HxCDF | 1.2 | 1.7 | 1.9 | 2.5 | 4.1 | 10 | 21 | 21 | 28 | 29 | 31 | 39 | 34 |
| 123789-HxCDF | 0.33 | 0.40 | 0.61 | 0.56 | 0.69 | 1.4 | 1.9 | 1.7 | 2.2 | 1.8 | 2.4 | 2.7 | 2.1 |
| HxCDFs | 20 | 26 | 33 | 60 | 140 | 610 | 930 | 560 | 490 | 530 | 610 | 620 | 570 |
| 1234678-HpCDF | 11 | 17 | 28 | 55 | 140 | 510 | 740 | 420 | 360 | 330 | 370 | 330 | 310 |
| 1234789-HpCDF | 0.9 | 1.6 | 2.3 | 3.3 | 7.8 | 32 | 56 | 29 | 26 | 28 | 29 | 28 | 28 |
| HpCDFs | 17 | 29 | 51 | 130 | 400 | 1900 | 2800 | 1500 | 1000 | 1200 | 1300 | 1100 | 1000 |
| 12346789-OCDF | 24 | 43 | 86 | 180 | 520 | 2300 | 2800 | 1700 | 1200 | 1300 | 1300 | 1200 | 1100 |
| Σ PCDDs | 520 | 600 | 750 | 1800 | 5800 | 28000 | 38000 | 30000 | 19000 | 19000 | 20000 | 17000 | 16000 |
| Σ PCDFs | 100 | 150 | 220 | 440 | 1200 | 5100 | 7000 | 4400 | 3100 | 3500 | 3800 | 3500 | 3200 |
| Σ PCDD/Fs | 620 | 740 | 970 | 2200 | 7000 | 33000 | 45000 | 34000 | 22000 | 22000 | 24000 | 20000 | 19000 |
| PCB81 | 0.30 | 0.50 | 0.83 | 1.9 | 5.6 | 25 | 47 | 25 | 22 | 19 | 20 | 18 | 17 |
| PCB77 | 9.7 | 27 | 58 | 150 | 430 | 1100 | 2100 | 1100 | 940 | 730 | 780 | 740 | 690 |
| PCB126 | 0.75 | 1.0 | 2.0 | 3.9 | 9.4 | 29 | 49 | 30 | 25 | 22 | 24 | 24 | 22 |
| PCB169 | 0.25 | 0.40 | 0.46 | 0.60 | 1.1 | 2.6 | 4.8 | 4.0 | 3.8 | 3.7 | 4.1 | 4.3 | 4.1 |
| PCB123 | NDa | 4.2 | 7.9 | 20 | 54 | 200 | 350 | 210 | 150 | 130 | 140 | 120 | 100 |
| PCB118 | 55 | 170 | 430 | 760 | 1900 | 6700 | 11000 | 6400 | 4800 | 4100 | 4300 | 3900 | 3600 |
| PCB114 | ND | ND | 3.1 | 11 | 28 | 120 | 180 | 96 | 71 | 60 | 63 | 56 | 56 |
| PCB105 | 13 | 30 | 90 | 210 | 710 | 3000 | 5200 | 2800 | 2000 | 1700 | 1800 | 1600 | 1500 |
| PCB167 | ND | 5.2 | 15 | 33 | 75 | 260 | 390 | 250 | 190 | 160 | 170 | 150 | 150 |
| PCB156 | ND | 9.4 | 28 | 56 | 150 | 530 | 820 | 490 | 390 | 330 | 360 | 320 | 300 |
| PCB157 | ND | ND | 8.8 | 17 | 42 | 140 | 210 | 130 | 100 | 87 | 93 | 86 | 80 |
| PCB189 | ND | ND | 1.9 | 4.9 | 10 | 33 | 61 | 38 | 33 | 27 | 32 | 29 | 27 |
| PCB180 | 8.3 | 32 | 50 | 140 | 440 | 1700 | 3900 | 2400 | 2100 | 1600 | 2100 | 1500 | 1400 |
| PCB170 | 5.8 | 17 | 31 | 81 | 230 | 890 | 1800 | 1100 | 1000 | 770 | 960 | 740 | 660 |
| Σ Co-PCBs | 94 | 290 | 730 | 1500 | 4100 | 15000 | 26000 | 15000 | 12000 | 9700 | 11000 | 9300 | 8600 |
| Σ TEQ (PCDD/Fs)b | 4.2 | 4.8 | 5.2 | 7.4 | 14 | 47 | 76 | 49 | 46 | 44 | 46 | 45 | 41 |
| Σ TEQ (Co-PCBs) | 0.09 | 0.13 | 0.28 | 0.55 | 1.4 | 4.4 | 7.4 | 4.5 | 3.7 | 3.2 | 3.4 | 3.3 | 3.1 |
| Σ TEQ (PCDD/Fs + Co-PCBs) | 4.2 | 5.0 | 5.5 | 7.9 | 15 | 52 | 83 | 53 | 49 | 47 | 50 | 48 | 44 |
- a ND = not detected.
- b TEQ = toxic equivalent.
- c See text for additional definitions of abbreviations.
Historical trends of polychlorinated dibenzo-p-dioxin/dibenzofuran (PCDD/F) homologues in the sediment core. Arrows in parentheses indicate the corresponding axes (left or right). TCDFs = tetrachlorodibenzofurans; PeCDFs = pentachlorodibenzofurans; HpCDFs = heptachlorodibenzofurans; OCDF = octachlorodibenzofuran; TCDDs = tetrachlorodibenzo-p-dioxins; PeCDDs = pentachlorodibenzo-p-dioxins; HpCDDs = heptachlorodibenzo-p-dioxins; OCDD = and octachlorodibenzo-p-dioxin.
The PC-1 includes OCDD, HpCDDs, OCDF, most HpCDFs, some HxCDDs and some HxCDFs as its characteristic congeners. These higher chlorinated PCDD/Fs correspond well with the impurities of PCP [3, 15], and their trends noted above are consistent with the history of PCP use in Japan as shown in Figure 4. The annual PCP usage increased rapidly from 1957 and reached a maximum of approximately 16,000 tons in 1967. It decreased by a factor of 55 during 1967 to 1975 and then generally leveled out until 1986. Based on these comparisons, PC-1 was judged to be the dioxin impurity of PCP.
Principal component loadings for the polychlorinated dibenzo-p-dioxin/dibenzofuran (PCDD/F) congener-specific data. PC = principal component; TCDD = tetrachlorodibenzo-p-dioxin; TCDF = tetrachlorodibenzofuran; PeCDD = pentachlorodibenzo-p-dioxin; PeCDF = pentachlorodibenzofuran; HxCDD = hexachlorodibenzo-p-dioxin; HxCDF = hexachlorodibenzofuran; HpCDD = heptachlorodibenzo-p-dioxin; HpCDF = heptachlorodibenzofuran; OCDD = octachlorodibenzo-p-dioxin; OCDF = octachlorodibenzofuran.
History of pentachlorophenol (PCP), chloronitrofen (CNP) and polychlorinated biphenyl (PCB) use in Japan. Arrows in parentheses indicate the corresponding axes (left or right).
In the case of PC-2, many PeCDFs and HxCDFs are the characteristic congeners. Based on the homologue pattern of a typical combustion process reported by Hutzinger and Fiedler [3], in which PeCDFs and HxCDFs are the main components, it was assumed that PC-2 corresponds to combustion processes in which incineration is the main factor.
The characteristic congeners of PC-3 include some TCDD, PeCDD, and TCDF congeners, particularly 1,3,6,8-TCDD, 1,3,7,9-TCDD, 1,3,6,9-TCDD, 1,2,3,6,8-PeCDD, 1,2,3,7,9-PeCDD, and 2,4,6,8-TCDF These lower chlorinated PCDD/Fs correspond with the primary impurities of CNP, which was used as the replacement of PCP [15]. Their trends mentioned above are consistent with the history of CNP use in Japan; that is, the annual CNP usage increased steadily from 1965, reached a maximum of approximately 5,500 tons in 1974, and stopped in 1995 (Fig. 4). Based on these comparisons, we identified PC-3 to be the dioxin impurity of CNP.
Considering the separate periods of time during which PCP and CNP were used in Japan, the observed trends among the higher and lower chlorinated homologue classes in different time frames seem to reflect the difference in period of PCP and CNP inputs. On the other hand, it should be noted that our source identification approach mentioned above is based on the assumption that variance in the congener space of the sediment profile is forced only by source variation. Other possible causes of variance such as environmental breakdown were neglected, because PCDD/Fs are reported to be resistant to microbial attack in the environment [31] and because their photodegradation in sediment is not considered to be significant.
Contributions of different sources to dioxin pollution
After dioxin source identification, MRA was carried out for source apportioning based on the congener-specific data of PCDD/F concentrations in such sources. Because of the large variability in dioxin congener profiles from different samples of each source, we used average congener profiles for these sources, and this may cause uncertainty in our estimation. The dioxin congener profile used for PCP was the average profile of four Japanese PCP formulations [15] with slight modification to adjust its homologue pattern. That is, the congener-specific concentrations of HpCDDs, OCDF, and HpCDFs in the average profile were modified based on the typical homologue ratios of HpCDDs, OCDF, and HpCDFs to that of OCDD in PCP reported by Hutzinger and Fiedler [3] to reduce calculation error. The congener profile of CNP was determined based on the average of five CNP formulations [15]. For combustion processes, the average congener profile of atmospheric deposition obtained in the Kanto area (including Tokyo Bay) [32] was used.
Historical contributions of different sources to total polychlor-inated dibenzo-p-dioxin/dibenzofuran (ΣPCDD/F; left bar) and total PCDD/F-derived toxic equivalent (ΣTEQ [PCDD/Fs]; right bar) concentrations in Tokyo Bay, Japan. Arrows indicate the corresponding axes (left [ΣPCDD/Fs] or right [ΣTEQ (PCDD/Fs)]). CNP = chloronitrofen; PCP = pentachlorophenol.
In the MRA, the three sources and the 13 sediment samples were regarded as independent and dependent variables, respectively, whereas congeners or congener clusters in these variables were treated as cases. For each sediment sample, the regression coefficients of the three sources were obtained separately on the basis of the least squares. Then, source apportioning was performed using the obtained regression coefficients. Based on the calculated apportionment of each source, the historical contributions of PCP, CNP, and atmospheric deposition to ΣPCDD/F concentration in Tokyo Bay were estimated, and the results are expressed as the left bars (corresponding to the left axis) in Figure 5. Furthermore, these results were transformed into TEQ using the TEQ:PCDD/F ratio in each source. The historical contributions of different sources to ΣTEQ (PCDD/Fs) concentration in Tokyo Bay are also shown in Figure 5 (right bar and axis).
It can be seen that the herbicides PCP and CNP have mainly contributed to the PCDD/F burdens since the late 1950s (Fig. 5). Pentachlorophenol has been the greatest contributor to both ΣPCDD/F and ΣTEQ (PCDD/Fs) concentrations in Tokyo Bay. The contribution of PCP peaked around 1970, decreased during 1972 to 1981, and then leveled off. Chloronitrofen has played a minor role in ΣPCDD/F concentration since 1967. The contribution of CNP reached its maximum around 1975, decreased during the period 1977 to 1986, and subsequently leveled off. Combustion processes are the secondary contributor to ΣTEQ (PCDD/Fs), but their proportion has been generally increasing to date. The contributions of PCP, CNP, and combustion processes to the ΣPCDD/F concentration in the surface sediment layer were estimated to be 76%, 15%, and 9%, respectively. In the case of ΣTEQ (PCDD/Fs), their contributions were 62%, 4%, and 34%, respectively. The inputs originating from PCP and CNP did not significantly decrease even after the decline in their use. This suggests that herbicide-derived PCDD/Fs remaining in agricultural land will continue to run off and to pollute the aquatic environment in Japan for a long time.
Historical trends of coplanar polychlorinated biphenyl (Co-PCB) congeners in the sediment core. Arrows in parentheses indicate the corresponding axes (left or right).
Trends in Co-PCBs
For Co-PCBs, the 14 congeners were detected in nearly all the sediment layers examined. The obtained data are presented in Table 1. The ΣCo-PCB concentration in the oldest sediment layer, corresponding to approximately 1937, was found to be 94 pg/g. It increased drastically from 1956 to 1972, reaching a peak of 26,000 pg/g in approximately 1970. Thereafter, a decrease occurred by a factor of 2.7 during 1972 to 1986, after which the total concentration generally leveled off. The PCB-118 had been the dominant congener, contributing more than 39% (up to 59%) of the ΣCo-PCB concentration throughout the core, followed by PCB-105, -180, -77, and -170. The contributions of PCB-118, -105, -180, -77, and -170 to the ΣCo-PCB concentration in the surface sediment layer were 42%, 17%, 16%, 8.0%, and 7.7%, respectively. Based on the concentration of ΣCo-PCBs in the surface sediment layer and the average sedimentation rate of 0.27 g/cm2/year, we calculated the present flux of Co-PCBs to Tokyo Bay sediment to be 2,300 pg/cm2/year [29]. On the other hand, the ΣTEQ (Co-PCBs) was dominated by PCB-126, contributing more than 66% (up to 88%) of the ΣTEQ (Co-PCBs) throughout the core. This can be explained by its having the largest TEF within the Co-PCB group. The other important TEQ contributors were PCB-118, -105, -156, and -77. The contributions of PCB-126, -118, -105, -156, and -77 to the ΣTEQ (Co-PCBs) concentration in the surface sediment layer were 71%, 12%, 4.8%, 4.8%, and 2.2%, respectively.
Also, the ΣTEQ (PCDD/Fs and Co-PCBs) level increased drastically from the late 1950s and reached its maximum of 83 pg/g in approximately 1970. It then declined to 49 pg/g over the next 9 years, after which it generally leveled off. The PCDD/Fs contributed more than 90% of the ΣTEQ (PCDD/Fs and Co-PCBs) throughout the sediment core, thus indicating that PCDD/Fs have played a major role in the toxic burden in Tokyo Bay since the late 1930s (Table 1).
From the historical trends of individual Co-PCB congeners as shown in Figure 6 (without those of PCB-114, -123, -156, -157, -167, -189, -170, and -180 for clear illustration), it was found that all the compounds, except PCB-169, showed very similar trends. They increased drastically during 1956 to 1972, then decreased quickly during 1972 to 1977. On the other hand, PCB-169 showed a different trend from those of the other compounds; the PCB-169 level has generally been increasing slowly to date. These characteristics suggest that different Co-PCB sources are present in the Tokyo Bay area.
Sources of Co-PCBs
Because the Co-PCB data set obtained from the sediment core is too small, it was difficult to apply PCA for source identification. However, from the historical trends of Co-PCB congeners described above, we found something interesting. The trends of all the congeners, except PCB-169, in the core are consistent with the history of PCB production and use in Japan (Fig. 4). Commercial PCB production began under the trademark of Kanechlor (KC) in 1954. The annual PCB usage increased significantly by a factor of more than 50 from 1954 (200 tons/year) to 1970 (10,120 tons/year), and the production was stopped in 1972 [33]. These trends are also in agreement with that of ΣPCBs in the same sediment core previously reported by Okuda et al. [24]. In addition, the observed Co-PCB pattern with a predominance of PCB-118, -105,-180, -77, and -170 mentioned above can be explained by the characteristics of KC preparations. According to Takasuga et al. [34], PCB-118 and -105 are the major Co-PCB components in KC-500, -400, and -300. Both PCB-180 and -170 exist at high levels in KC-600, whereas PCB-77 is an important component in KC-400 and -300. Kannan et al. [16] also reported that KC preparations contain PCB-77 as a characteristic non-ortho Co-PCB congener. Based on these comparisons, we believe that PCB formulations were the greatest contributing Co-PCB source in Tokyo Bay. Thus, the phenomenon of the ΣCo-PCB concentration generally leveling off since 1986 is largely attributed to the supply of formulation-derived Co-PCBs remaining in the catchment area.
In the case of PCB-169, it was found to exist in trace or undetectable levels in PCB formulations. Its time trend mentioned above is consistent with that of PeCDFs recorded in the sediment core. As described previously, PeCDFs are considered to be the main combustion-related dioxin components. Furthermore, combustion processes were reported by Ballschmiter et al. [35] to be the source of PCBs, particularly many highly chlorinated congeners, including PCB-169. Brown et al. [36] studied the Co-PCB concentration profiles in Aroclor formulations, and they noted that the environmental burden of PCB-169 was derived largely from non-Aroclor sources. Accordingly, we infer that combustion processes were another significant source of Co-PCB pollution in Tokyo Bay.
Our inference is consistent with the view of Ohsaki et al. [37]. They indicated that municipal waste incineration might be a minor Co-PCB source compared with commercial PCB preparations. On the other hand, the presence of Co-PCBs in the sediment layers corresponding to the period 1935 to 1951, the pre-PCB production era in Japan, might be attributed to the PCB importation and use at that time [33]. Long-distance atmospheric transport is another possible explanation [33]. Further investigation is required to obtain more detailed PCB congener-specific information for improved future source identification.
CONCLUSIONS
Based on the obtained congener-specific data of PCDD/Fs and Co-PCBs in a dated sediment core, the historical trends of these compounds in Tokyo Bay were reconstructed. Using a statistical analysis approach, two herbicides, PCP and CNP, as well as combustion processes were identified to be the major dioxin sources, and contamination from the use of PCP and CNP followed by agricultural runoff was further shown to be the primary factor for dioxin inputs in this watershed. For Co-PCB pollution, PCB formulations and combustion processes were estimated to be the major sources, in which the former was considered to be the primary factor. These findings will assist remediation planning for and subsequent monitoring in the Tokyo Bay area. Furthermore, our results are of significance for the establishment of comprehensive PCDD/F and Co-PCB control measures in Japan and would help in understanding the global dioxin and dioxin-like PCB problems.
Acknowledgements
This work was conducted under the research project Establishment of a Scientific Framework for the Management of Toxicity of Chemicals Based on Environmental Risk-Benefit Analysis, supported by Core Research for Evolutional Science and Technology of the Japan Science and Technology Corporation.
