Seasonal variation of the distribution of PCBs in sediments and biota in a PCB-contaminated estuary

(1)

Seasonal variation of the distribution of PCBs

in sediments and biota in a PCB-contaminated estuary

Chung-Te Fu

a,b,*

, Shian-Chee Wu

a,1

a_{Graduate Institute of Environmental Engineering, National Taiwan University, 71, Zhou Shan St., Taipei City 106, Taiwan} b_{Department of Environmental Engineering, Vanung University, 1, Vanung Rd., Chungli City 325, Taiwan}

Received 10 December 2004; received in revised form 1 May 2005; accepted 9 July 2005 Available online 9 September 2005

Abstract

To elucidate the effects of seasonal variation of precipitation on the distribution of polychlorinated biphenyls (PCBs) in estuarine sediments and benthic feeders, PCB concentrations of river surface sediments and mullet fish (Liza macrol-epis) were investigated in the estuary of Er-Jen River near former PCB contamination sites before and after each wet season from 2002 to 2004. Analyses of grain size distribution and organic matter revealed that the pre-existing surface sediments were covered by and mixed with the soil particulates brought by surface runoff after each wet season. Obvi-ous increment of PCB content and significantly elevated fraction (p < 0.005) of light PCBs of the river mouthÕs sedi-ments after each wet season indicated that the invading particles were rich in unweathered PCBs. PCBs previously buried in the surface soil of heavily contaminated sites were flushed into this estuary through surface runoff. The pre-cipitation altered the PCB patterns in sediment organic matter, the dietary source of mullet, and consequently changed that of mullets accordingly, which all possessed significant greater fraction of light PCBs. In this study, it was demon-strated that seasonal summer precipitation affected the distribution of PCBs on surface sediments and the mullets of this estuary. PCB residuals retained in this region still pose potential threats to biota resided here.

Keywords: Estuary; Mullet ﬁsh; Sediments; Precipitation

1. Introduction

Seasonal variations of patterns or concentrations of polychlorinated biphenyls (PCBs) have been observed in various environmental media. The dominating factors of seasonal variation include temperature diﬀerence,

precipitation, congener-speciﬁc properties, physiological behaviors of biota, etc. Temperature-dependent varia-tions were observed in water (Bremle and Larsson, 1997; Bruhn and McLachlan, 2002), atmosphere ( Man-chester-Neesvig and Andren, 1989; Haugen et al., 1999; Kiss et al., 2001), soil (Wilcke and Amelung, 2000) and atmospheric deposition (Carrera et al., 2002). In which higher PCBs level is commonly observed in warmer sea-son than in colder seasea-son was attributed to the higher water temperature, air/gas exchange rate and desorbing rate of PCBs from sediment to water. Meanwhile, higher PCB body burden in aquatic biota was found in spring than in autumn. This was mainly attributed to the

* _{Corresponding author. Address: Department of}

Environ-mental Engineering, Vanung University, 1, Vanung Rd., Chungli City 325, Taiwan. Tel.: +886 2 23629435/886 3 4515811x705; fax: +886 2 23629435/886 3 4622232.

E-mail address:tedfu@msa.vnu.edu.tw(C.-T. Fu).

1

Tel./fax: +886 2 23629435.

(2)

consumption of lipid of biota in winter, which resulted in a concentration effect seen in spring (Huhnerfuss et al., 1995; Harding et al., 1997; Stapleton et al., 2002; Wiklund et al., 2003). Additionally, precipitations, either seasonal or episodic, are important factors on the distribution of PCBs. Seasonal precipitation not only scavenged the airborne PCBs down to ground water bodies, but also changed the concentration and patterns of PCBs in the river water (Bremle and Larsson, 1997). On the other hands, short-term episodic precipitation brought intense surface runoff to the watershed. Original bed sediments were diluted or buried by sediments with different organic compound compositions flushed in from the river catchment (Barber and Writer, 1998). However, aforementioned investigations on the varia-tions were mostly concentrated on the temperate or around the arctic/sub arctic regions. Only few attempts have been made to investigate the seasonal variations of PCBs in previously contaminated sites of tropical areas where terrestrial sources of PCBs exist, and tem-perature difference varies slightly and intense precipita-tion occurs just in limited seasons.

Temperature difference in tropical areas is not as great as in temperate regions, insomuch it seems not to be a dominating factor for the distribution of PCBs. In-stead, surface runoff brought in by intense summer pre-cipitation might change the composition and distribution of PCBs of original sediments. An increased PCB level was observed in the river water, sediments, (Barber and Writer, 1998; Bergqvist et al., 1998; Kow-alewska et al., 2003) and in the fish that resided in a lake (Stewart et al., 2003) after a flooding event. Therefore, the seasonal precipitation happened in mid-summer, rather than the temperature difference, was more likely the governing role for the seasonal variation of PCBs in tropical area. Additionally, surface sediments are the primary food sources for benthic feeders so that the change in sediment might have also altered the body burden and pattern of PCBs in their predators through food chains. ThoughStewart and coworkers (2003)did observe the accumulation of PCBs in fish after an epi-sodic flooding event, little attention has been given to the precipitation induced seasonal variation of PCBs found in sediments and fish of tropical areas.

Er-Jen River, situated in southwestern part of Tai-wan, is the most polluted river of the entire island due to its long-term receiving of untreated sewage and indus-trial wastewater (ROC EPA, 2004). In the 1980s, open burning and acid washing operations held along the riv-erside from the metal recovery industry led to large quantities of PCBs being discharged into the estuary and causing serious pollution. A PCB concentration as high as 77 lg g 1was found in the soil samples collected at the riverside in early 1990s (Huang et al., 1992). In addition, PCBs and dioxins were found in the ﬁsh resid-ing in the estuary (Ling et al., 1995) and nearby ﬁsh

farms (Lu et al., 1995). A PCB body burden of 3400 ± 470 ng g 1was detected in the estuarine mullets (Chen et al., 1994). The contaminated sites were all found near the riverside where storm runoff has long been carrying the contaminated topsoil into the river and letting it settle at the bottom of the estuary water. Storm waters also conveyed large amounts of upstream sediments downward, which might dilute or cover the pre-existing sediments. The mid-summer storms that bring more than 90% of the yearly precipitation (ROC, CWB, 2005) might have great influences on the distribu-tion of river sediment and probably to the sediment feeders. Therefore, to elucidate the precipitation effects on the seasonal variation of PCBs in sediment and bot-tom feeders in tropical regions, this study investigated the contents and patterns of PCBs in surface sediments and bottom feeders of Er-Jen River before and after the rainy seasons in year 2002–2004.

2. Materials and methods 2.1. Sample collection

Eleven sampling sites to collect surface sediments were situated along the estuary and were named EJ1– EJ11 in sequence from upstream to river mouth (Fig. 1). To avoid site deviations in each seasonal sam-pling, sampling sites were chosen based on their distin-guishable locations and were positioned by a global positioning system (GPS) and terrestrial surrounding. EJ1 was located most upstream as the background refer-ence. EJ2 was at the intersection of the mainstream and the heavily polluted main branch Sanyekon Creek, which contributes more than 60% of the total pollutant load in Er-Jen River (Wen, 2002). EJ4 was near the for-mer metal recovery operation site, Nan-Din Bridge; EJ6 was in front of the sewer outlet that holds both the sew-age and storm water, and EJ9, EJ10 and EJ11 were

Fig. 1. The schematic diagram of the sampling sites in the Er-Jen Estuary.

(3)

situated at the river mouth. Among the eleven sites, EJ2, EJ4 and EJ6 were suspected contaminated sites which have either been near the former operation site of metal recovery (EJ4) or in front of the pollution plume (EJ2 and EJ6). Sediment samples were collected by a sedi-ment grabber and each sample was a composite of three grabs to average the heterogeneity of sediments. Sedi-ment samples were collected before and after the rainy season in May 2002, September 2002, September 2003 and May 2004, respectively. A total of sixty mullet fish, chosen as the target fish due to their abundance (Kuo and Shao, 1999) and high accumulation potential (Chen et al., 1994; Wang et al., 2000), were caught near site EJ6 in September 2002 (n = 7), May 2003 (n = 35) and Sep-tember 2003 (n = 18), then froze and transported to the laboratory immediately. Water samples were collected simultaneously with the river surface sediment. Grain size distribution of sediment, total organic content of water, organic/lipid content and PCB content in sedi-ment and fish were analyzed.

One might wonder why samples were not taken dur-ing the dry season in May 2003, but waited until May 2004, which interrupted the annual Ôpre and postÕ rainy season samplings. In fact, this unforeseen change of date and plan was attributed to the outbreak of severe acute respiratory symptom (SARS) in spring 2003. All unnec-essary meetings and long-distance traveling, including our scheduled sediment sampling in May 2003, were ad-vised to cancel or postpone by authoritative regulations. The sampling of ﬁsh, unlike the sampling of sediment delayed by SARS until May 2004, was implemented in May 2003 by our cooperative ﬁsherman alone. Fortu-nately, the defect of the experimental design was alleviated by the intrinsic characteristics of this study, the seasonal variation of sediment and biota. If the pos-tulated periodic change did exist, then the similar sea-sonal variation before and after the rainy season should happen not only in the year 2002, but also in the succeeding years (2003 and 2004). Therefore, it was assumed that the variations due to precipitation might recur periodically and the distribution of PCBs in May 2003 was similar to May 2004. Thus, the original dataset before wet season in the second studied year (May 2003) was replaced by the dataset collected in May 2004.

2.2. Sample preparation and analyses

Sediment samples were freeze-dried for 100 h and ground into ﬁne powders using a stainless steel mortar and pestle. About 10 g of sediments were transferred into a soxhlet apparatus and extracted with 300 ml of a mixture of residue grade acetone and n-hexane (1:1, v/v) in a 70C water bath for 36 h. The sediment ex-tracts were then concentrated to 1 ml in a rotary evapo-rator (Bu¨chi) and cleaned with 5 ml of concentrated

sulfuric acid (Merck). Afterwards, it was back extracted with hexane three times and concentrated to 1 ml by nitrogen stream. The 1 ml concentrate was then passed through a pipette (10 mm ID) packed with sulfuric-acid (Merck) treated silica gel (70–230 mesh, for column chromatography, Macherey-Nagel), silver-nitrate (J.T. Baker) coated silica gel and florisil (Merck). This extract was finally eluted with 70 ml n-hexane and rotary-evap-orated to 1 ml for GC analysis. Each fish was sliced to fillets and was treated as an individual sample. The fish fillets were prepared with the same procedure as the sedi-ment. Lipid content analyses were performed on the fish fillet by weighing them after rotary evaporation, but prior to the purification process with sulfuric acid.

PCB contents were analyzed with a high resolution gas chromatograph (Agilent 6890N) equipped with a 63

Ni electron capture detector and a 30 m· 0.25 mm · 0.25 lm, 5% phenyl phase fused silica capillary column (RT-5mx, J&W). The column flow was set at 1 ml/min with a linear velocity of 27 cm/min. The oven tempera-ture was programmed as follows: 80C for 4 min, in-creases to 160C at an increment of 20 C per min, to 265C by increasing 2 C per min, to 295 C by increas-ing 25C per min and holding it at 295 C for 5 min. The injector was set at 250C in splitless mode, and the detector was set at 295C. Twelve environmentally significant congeners, IUPAC Nos. 18, 28/31, 52, 44, 101, 149, 118, 153, 138, 180, and 194, provided by Supe-lco were introduced as a quantification standard for specific congeners and summarized as PCB12. Decachlo-robiphenyl (Supelco) was spiked into samples before extraction and served as a surrogate of target com-pounds. The internal standard of GC performance was 2,4,6-Trichlorobiphenyl (Supelco). Only the peaks within proper range (2%) of the retention time when compared to the standards were counted for the quanti-fication of PCBs congeners. A standard reference mate-rial (NIST 1939a) was applied to verify the feasibility of the measuring method. The average recovery of the congeners in the SRM is 82 ± 6% with the range of 73% (CB118) to 90% (CB138/163). Every batch of eight samples was accompanied with one duplicate sample, one QC sample and one blank sample. Method detection levels (MDLs) were in the range of 0.032 (CB44) to 0.088 (CB18) ng g 1 with recovery of 96 ± 23% from spiked PCB12 standards. The recovery of surrogate CB209 was 70 ± 16% of sediment samples and 73 ± 7% of fish samples. All data was presented without cor-rection to the recovery.

Organic content of sediment was determined by sul-furic-dichromate digestion and back titration with stan-dard ferrous ammonium sulfate solution. Grain size distribution analysis was performed through wet sieving and grain size analyzer (CILAS, M175) with 12 sieves ranged from less than 2 lm to larger than 710 lm. Total organic matter of river water was determined by the

(4)

persulfate oxidation method and analyzed by an O.I. Analytical Type 1010 TOC Analyzer.

3. Results and discussion

3.1. Seasonal change of the water quality in the Er-Jen Estuary

Unlike temperate estuarine regions where high river flow is brought about by snow melting in spring, the high flow of river water in this studied region was resulted from the intense precipitation brought in by typhoons from the Pacific Ocean in a short period, normally from June to August. The precipitation is so intensive that a huge amount of surface runoff and serious erosion of sur-face soils in the river catchment are commonly observed. Solid content as high as 5000 mg l 1and resultant solid loading of 5· 105

ton d 1were observed in the upstream of the river during a summer storm in 2002, which con-tributed more than 85% of the annual solid loading. (ROC, WRA, 2005).

The characteristics of climate and water quality of this estuary, including atmosphere temperature, precipi-tation, river flow rate, and aqueous concentrations of ammonia nitrogen, biochemical oxygen demand and suspended solid, are expressed as the average of the ratio of the monthly average to the annual average in the per-iod from 1991 to 2000 (Fig. 2). The monthly averaged values of water quality parameters, except water pH, deviated significantly from the annual mean during rainy seasons. Ammonium nitrogen (r = 0.91) and BOD (r = 0.81) were negatively correlated with the flow rate, while the variation of SS (r = 0.63) was synchronized with the variation of flow rate. Meanwhile, river flow rate was positively correlated to the precipitation

(r = 0.93). Therefore, precipitation was considered the main factor on the fluctuation of river water quality. The average TOC found in river water for May and Sep-tember were 6.1 ± 0.56 mg/l and 1.0 ± 0.16 mg/l, respec-tively. This significant reduction of TOC (p < 0.0001) in September revealed the dilution effect of runoff water on the water constituents within the studied period.

Consequently, precipitation is simultaneously resp-onsible for both the dilution of water concentrations of BOD and ammonium nitrogen through the storm runoﬀ, and the thickening of water solid content by up-stream erosion.

3.2. Seasonal variations of organic content and grain size of sediment

The organic matter content of sediments collected in alternative seasons was shown inFig. 3. Higher organic content of surface sediment was observed in mid-estua-rine sector where industrial eﬄuents were discharged into this region from Sanyekon creek. In comparison, organic contents were found lower in the upstream and river mouth sectors. After summer storms, the average organic content declined along the river (from 1.52% in May to 1.18% in September of 2002 and 2.10–1.66% during 2003–2004). An exception was found in the river mouth sites where organic contents increased signiﬁcantly after summer and even exceeded the values of dry season.

A deposition of eroded material from the catchment was discovered in the river, especially at the river mouth site EJ9, EJ10 and EJ11. Precipitation might have been the cause in diluting or moving the organic-rich sedi-ments down the river. An increase of organic content in the sediments around the river mouth was attributed to the conveying and re-deposition of materials from

Fig. 2. The seasonal variations of some physical and chemical parameters (water temperature, pH, SS, BOD, ammonia, precipitation and ﬂow rate) of Er-Jen River. The variations were expressed as the average of the ratios of monthly average to annual average in the year 1991–2000.

Fig. 3. The effects of precipitation on the fine grain fraction and organic matter content of the river sediments in year 2002 (A) and 2003–2004 (B). GSF denotes the fine grain size fraction of sediment particles, while M and S denote the sampling periods in May and September, respectively.

(5)

surface runoﬀ and original sediments in the mid-sector of the estuary.

Grain size distribution of the river sediment, before and after the wet season of 2002 and during 2003– 2004, is also shown inFig. 3. By grouping sediment par-ticles into fine (0–38 lm), medium (38–250 lm) and coarse (250–710 lm) fractions, the grain size of sedi-ments increased gradually along the river from fine (up-stream site EJ1) to medium (EJ2–EJ8) and eventually to coarse (river mouth sites EJ9–EJ11) in May. The weight percentage of fine particles decreased from more than 90% at EJ1 to 0% at EJ10, while that of coarse particu-lates increased from 1% to more than 90%. However, grain sizes of SeptemberÕs sediments were uniformly dis-tributed along the whole estuary, which indicated that new materials had been deposited evenly after precipita-tions. In the dry seasons, from September to May of next year, tidal current would selectively remove fine particles in the sediment at the river mouth sites EJ9– EJ11, which resulted in the declining of fine particle frac-tion along the river.

Shift of organic matter content and grain size distri-bution along the river indicated that fine and organic-rich particulates had been transported from upstream sector and river catchment to the river mouth through surface runoff and strong river flow. Original surface sediments in the estuarine sector might have been mixed or masked by the particulates from external sources after precipitation. However, the original sediments in the estuarine sector and river mouth were exposed again in dry season owing to the tidal current. Thus, a period-ical change in the organic content and grain size distri-bution occurred seasonally.

3.3. PCB concentrations of sediments

Dry solid based and organic matter adjusted PCB content (PCB12), found in sediment samples that were collected along the estuary is shown in Fig. 4. Before the wet season, the spatial distribution of PCB concen-tration indicated high PCB contamination in the estua-rine sector at points between EJ2 and EJ8, especially at EJ2, EJ4, and EJ6, which were near three suspected contaminated sites, Sanyekon Creek, Nan-Din Bridge and the sewage outlet, respectively. On the other hand, PCB12 of upstream site EJ1 and river mouth site EJ9, EJ10, and EJ11 were lower than those found on other estuarine sites were. However, after the wet season, high concentrations of PCB12were observed at EJ4, EJ7 and the river mouth and not at EJ2 and EJ6.

For dry-solid-based PCB content (Fig. 4A), a dilu-tion effect similar to what has found in the organic con-tent was observed. Surface runoff brought in by precipitation demonstrated its masking and re-deposi-tion effects on the PCB content of sediments in the estu-ary, where PCBs were reduced in the mid-estuarine

sector but increased in the river mouth. Before the wet season, higher PCB12 were found near the suspected contaminated sites, but they were shifted to the lower estuarine sector after summer storms. Particles with low PCB content from upstream erosion were deposited evenly over the estuarine sector and resulted in dilution and masking eﬀects on the original distribution pattern in sediments.

Contrary to the varied dry-solid-based PCB contents, organic-adjusted PCB content of the sediments along the estuary all increased after summer precipitation (Fig. 4B), which indicated that sediment enriched in PCBs were flushed into the estuary through surface run-off. The consequence of the re-deposition of high PCB content particles resulted in the average PCB12of river mouth to increase slightly in 2002 and significantly in 2003. An increase in sedimentary PCB content was ob-served at the river mouth in September 2003, to a level higher than the suspected hot spots, suggests that exter-nal contamination source other than upstream erosion dominated the distribution of the river mouthÕs surface sediment. Higher OM-normalized concentration denotes higher PCB activity and less aging material exported from PCB contaminated source during storm runoffs.

Similar to the periodic changes of grain size distribu-tion, these new-deposited particles in the sediment were mostly ﬁne particles that might be gradually lost to the sea by tidal current in the dry seasons. Therefore, PCB content found in the dry season following a wet one was similar to that of the previous dry seasons. 3.4. PCB pattern deviation in sediment

The change of PCB patterns on each site before and after the rainy season is expressed as the relative

Fig. 4. The dry solid based (A) and organic adjusted (B) PCB content of the surface sediments in studied seasons and locations.

(6)

deviation, i.e. ((CBn/PCB12)Sep (CBn/PCB12)May)/(CBn/ PCB12)Sep, where CBnis the organic adjusted concentra-tion of nth congener. Principal component analysis (PCA) of the relative deviation of 12 selected congeners indicates that the main components of the first factor are CB138 and 149, and those of the second factor are CB52 and 44. These two factors explain more than 80% of the total variation. Since CB52 and CB44 are lightly rinated PCBs, and CB138 and CB149 are heavily chlo-rinated ones, degree of chlorination seems to be an appropriate indicator to characterize the relative devia-tions. Fraction of light PCBs to total PCB12was applied to interpret the PCB distribution along this estuary. The fraction of light PCBs, including tri- and tetra-chlori-nated biphenyls such as CB18, CB28&31, CB52 and CB44, found at eleven sites for the months of May and September are shown inFig. 5. The fraction of light PCB of the surface sediments collected in May within estuarine sector (EJ2–EJ8) were in the range of 30– 40% and remained unchanged or varied insignificantly after the wet season (p = 0.8 of year 2002 and p = 0.54 of year 2003–2004). Contrarily, those collected from sur-face sediment of the river mouth zone (EJ9, EJ10 and EJ11) increased significantly from 30% to 50% (p < 0.005 of year 2002 and p < 0.001 of year 2003– 2004). Based on the phenomenon of varying light PCB fractions, these sampling sites can be roughly grouped into three main clusters, the upstream background EJ1, the estuarine sector EJ2–EJ8, and the river mouth zone EJ9, EJ10 and EJ11. Based on their relative devia-tions of PCBs in 2002, a two-dimensional PCA plot of these sampling sites is displayed inFig. 6. The plot also displayed three discrete clusters, which was similar to the findings from the light fractions inFig. 5.

The seasonal variation of light PCB fraction in the river mouth sediments revealed that external sources had been introduced into this region. The most probable sources of invading mass were non-point sources flushed into the estuary during precipitation, including particles from atmospheric wet deposition, upstream erosion and surface runoff of river catchment. In addition, relatively high PCB content of the invading particles observed in the river mouth, as compared with upstream sectors, suggests that there was another major contributor of external contamination other than the re-suspended up-stream sediments. Original surface sediments in the river mouth zone with coarser size, lower organic and lower PCB content were more susceptible to fine and organ-ic-rich invading particles than those in the other estua-rine sectors. Thus, a distinct PCB fingerprint indicating the terrestrial contribution of PCBs was observed clearly in the river mouth after the wet season. Furthermore, with the knowledge that sediments do not increase their light fraction spontaneously during precipitation, these elevated contents and light fractions found in the down-stream of storm water outlet (EJ6) were resulted from the surface runoff of nearby catchment.

Previous studies (Wania and Mackay, 1993; Cousins and Jones, 1998; Ashley and Baker, 1999; Meijer et al., 2002) suggested that a heavy-PCB-dominated pattern is an indicator of near source emission of PCBs, while a light-PCB-dominated pattern reﬂects a more weath-ered pattern or one that is from air–water exchange. In other words, light PCBs were believed to possess higher potential than heavy ones to partition or evapo-rate into water column and atmosphere. Thus, PCB residuals retained near a contamination source exhibited a more heavily chlorinated fraction than remote ones did. Such eﬀects are regarded as aging or weathering.

In this study, although sediments in the river mouth exhibited elevated light PCB fraction after each wet

Fig. 5. The light PCB fractions of the eleven sampling sites in year 2002 and 2003–2004.

Fig. 6. The two-dimensional PCA plot of the sampling sites based on the relative deviation of PCB patterns before and after rainy season in 2002.

(7)

season in the succeeding years, signiﬁcant increment of PCB content in this region suggested that its sources were not from remote weathered background contami-nation, but rather from a fresh local contamination source.Ling et al. (1995)reported that the major type of contamination in this estuary resembles light and medium chlorinated rich mixture Aroclor 1242 and 1254. Meanwhile, PCB residues in the estuarine sector and nearby villages seemed not weathered enough to show an enhancement of heavy fraction within only one decade. Therefore, newly deposited river mouth sed-iments still exhibited an unweathered PCB pattern that resembled the original AroclorÕs composition with prominent light PCB fraction. A similar pattern was ob-served byAshley et al. (2000)in the Hudson River where light homologs dominated the near source emission.

Light PCB fraction is believed to be higher in the top-soils surrounding the estuary than in the river sediment due to the lower aging rate of PCBs in the topsoils. PCB residuals in the river sediments, especially the light ones, are presumed to partition easily into the aqueous phase by their intrinsically higher water solubility and by the external mixing effects from water current and sediment scouring. On the contrary, there is no strong disturbance to the topsoils as what the river sediment had experi-enced, most light PCBs were preserved in the topsoils in spite of minor losses mainly through evaporation. Meanwhile, tidal current also conveyed new-deposited particles to the sea. In accordance, as the consequence of continual partitioning, evaporation and tidal convey-ing, light PCB fractions of surface sediments declined from wet seasonÕs peak to dry seasonÕs valley, and in-creased again by fresh light PCB source from storm runoff. The alternation of PCB patterns observed in the river mouth was attributed to the periodic settlement of fresh PCBs, the aging of them and the tidal flushing. Because settlement of eroded soil particles was minor within the estuary sector (EJ2–EJ8), as compared with what is at the river mouth, due to strong river flow. Invading sediments only comprised an insignificant por-tion of the upper part of sample cores. Therefore, the light PCBs fractions of the samples from the sites were nearly unchanged. Particles that lay beneath the newly deposited particles were aged particles with higher or-ganic content and higher heavy PCB fraction that might buffer the influence of the invading particles and obscure the influence. However, the above consideration is not essential when applied to the river mouth, of which ori-ginal PCBs and organic contents are much lower than those of the estuarine sector, and where the newly depos-ited particles are comprised of the majority of the incre-ment of PCB content in the river mouth after precipitation. Therefore, surface sediment of the river mouth could better reflect the real characteristics of newly deposited surface sediments than the estuarine sediment.

3.5. PCBs in ﬁsh

The average size, weight and lipid content of mullet fish caught in May and September of year 2002 and 2003 are shown inFig. 7. The average PCB burden of mullet varied insignificantly (p = 0.79) from MayÕs 9012 to SeptemberÕs 8143 ng g 1_{(lipid adjusted).} Inter-estingly it did not increase simultaneously with an in-crease in their presumed dietary source, the surface sediment (Horinouchi and Sano, 2000), which increased significantly (p < 0.05) from 769 to 1225 ng g 1(organic adjusted mean of all locations). One possible explana-tion is the larger partiexplana-tion capacity of fish caught in au-tumn, which were significantly larger in size (p < 0.005), heavier in weight (p < 0.00001) and richer in lipid (p < 0.05) than those caught in spring. The enrichment rate of fish lipid might be faster than the accumulation rate of PCBs and resulted in the dilution of PCBs. Therefore, the significant increase of lipid content in fish lowered their PCB concentration. This effect of dilution is similar to the findings in the temperate regions (Stapleton et al., 2002; Wiklund et al., 2003) that the change of lipid content of fish deviate their body burden. However, besides the lipid content (Wiklund et al., 2003), dietary preference (Loizeau et al., 2001), sediment burial rate (Berglund et al., 2001), habitat eutrophic sta-tus (Teil et al., 1996) and embryo development (Wiklund et al., 2003) were other possible governing factors. The mechanism behind the reduction of PCB level in fish is not fully understood. To better understand it, more re-searches are needed to focus on the enrichment of lipid content in fish.

The PCB patterns found in the mullet ﬁsh, unlike the content, exhibited a clear seasonal variation before

Fig. 7. The box plots of PCB content, length (denoted as L), weight (wt.) and lipid content (Lp.) of ﬁsh in May (M) and September (S). Bottom and top boundaries of box represent 25th and 75th percentiles. The line within the box marks the medians. Whiskers above and below the boxes indicate the 90th and 10th percentiles. Dots outside the whiskers represent the 95th and 5th percentiles.

(8)

and after each rainy season. Light PCB fraction in-creased significantly from dry season to wet season (p < 0.00005), while no significant change (p = 0.75) was observed for fish caught in September of the suc-ceeding years 2002 and 2003, which denotes a periodic change of patterns in fish.

With congener patterns changing seasonally, the influence of surface sediments on the patterns of these mullets could be identified.Fig. 8demonstrates the syn-chronous change of light PCB fraction of fish and sedi-ment in May and September. It should note that the light fractions of mullets and river mouthÕs sediments lifted up significantly after precipitation. The simulta-neous increase of the sediment and the fish revealed that PCB patterns of the fish body rapidly reflected those found in the sediments.

In actuality, the concept of the exposure route of hydrophobic chemicals to fish has been widely discussed in numerous studies and remains doubtful to many. The greater fraction of light PCB of mullets caught in Sep-tember was owing to the elevated light PCB fraction in the surface sediment. Unfortunately, lack of information on the aqueous PCB content in this estuary weakens the aforementioned postulation, thus one may argue that the contribution of PCB body burden to these mullets, rather than the dietary source, was through gill exchange or so-called bioconcentration. Gray (2002) reviewed more than forty articles related to bioaccumulation and indicated that some of them, though more than half to two-thirds of these articles claimed that biomag-nification was the dominant mechanism, lacked direct or appropriate evidences. Nevertheless, a commonly acceptable mechanism is that biomagnification occurs in highly hydrophobic chemicals, while bioconcentration dominates less hydrophobic chemicals, and the divide line is approximately in the log Kowrange of 5–6 (Clark

et al., 1990; Thomann et al., 1992; Russell et al., 1999). The log Kowof the light PCB congeners in current study is 5.24–5.84 (Kowvalue adopted fromHawker and

Con-nell, 1988), which are just in the border region to sup-port that both dietary uptake and gill exchange are possible mechanisms. However, a significant increase of PCB content in the surface sediment of the river mouth should not be attributed to the river water, be-cause water might not carry such high PCB content to result in the lift of PCB in the sediment during the period of a summer storm. Conversely, it is more likely that intrinsically higher water solubility of light PCBs may partition from polluted sediment to water and increase aqueous PCB content according to an equilibrium con-cept. Therefore, successive lifts of light PCBs in the sedi-ments and water may increase the light PCBs in these mullets due to simultaneous exposure from food and water. Although bioconcentration and biomagnification are all possible accumulation mechanisms to increase the light PCBs in fish, the exact PCB source was more likely to be from the newly deposited sediments rather than the water.

In this study, through analyses of the precipitation induced seasonal variation of PCB distribution in the surface sediment and mullet ﬁsh, an exposure pathway from sediment to biota has been established. Congener pattern is a useful indicator to identify the origins and path of PCBs. PCB residues from early contamination retained in nearby areas could be brought into the estu-arine ecosystem and concentrated in the bottom feeders, mullets.

References

Ashley, J.T.F., Baker, J.E., 1999. Hydrophobic organic contaminants in surﬁcial sediment of Baltimore harbor: Inventories and sources. Environ. Toxicol. Chem. 18, 838– 849.

Ashley, J.T.F., Secor, D.H., Zlokovitz, E., Wales, S.Q., Baker, J.E., 2000. Linking habitat use of Hudson River striped bass to accumulation of polychlorinated biphenyl congeners. Environ. Sci. Technol. 34, 1023–1029.

Barber, L.B., Writer, J.H., 1998. Impact of the 1993 ﬂood on the distribution of organic contaminants in bed sediments of the upper Mississippi River. Environ. Sci. Technol. 32, 2077–2083.

Berglund, O., Larsson, P., Edwald, G., Okla, L., 2001. Inﬂuence of trophic status on PCB distribution in lake sediments and biota. Environ. Pollut. 113, 199–210. Bergqvist, P.A., Strandberg, B., Ekelund, R., Rappe, C.,

Granmo, A., 1998. Temporal monitoring of organochlorine compounds in seawater by semipermeable membranes following a ﬂooding episode in Western Europe. Environ. Sci. Technol. 32, 3887–3892.

Bremle, G., Larsson, P., 1997. Long term variations of PCB in the water of a river in relation to precipitation and internal sources. Environ. Sci. Technol. 31, 3232–3237.

Fig. 8. The box plots of contents and light fractions of PCBs in the sediments and ﬁsh in May and September, in which S denotes sediments, F denotes ﬁsh, and LFrac means light PCB fraction.

(9)

Bruhn, R., McLachlan, M.S., 2002. Seasonal variation of polychlorinated biphenyl concentrations in the southern part of the Baltic Sea. Mar. Pollut. Bull. 44, 156–163. Carrera, G., Fernandez, P., Grimalt, J.O., Ventura, M.,

Camarero, L., Catalan, J., Nickus, U., Thies, H., Psenner, R., 2002. Atmospheric deposition of organochlorine com-pounds to remote high mountain lakes of Europe Environ. Sci. Technol. 36, 2581–2588.

Chen, W.F., Wang, H.C., Liu, C.Z., 1994. Fish species and PCBs concentration in the Er-Jen River. Ann. Rept. of Nat. Inst. Environ. Anal. Taiwan, ROC 2, 305–318 (in Chinese). Clark, K.E., Gobas, F.A.P.C., Mackay, D., 1990. Model of organic chemical uptake and clearance by ﬁsh from food and water. Environ. Sci. Technol. 24, 1203–1213.

Cousins, I.T., Jones, K.C., 1998. Air–soil exchange of semivol-atile organic compounds (SOCs) in the UK. Environ. Pollut. 102, 105–118.

Gray, J.S., 2002. Biomagniﬁcation in marine systems: the perspective of an ecologist. Mar. Pollut. Bull. 45, 46–52. Harding, G.C., LeBlanc, R.J., Vass, W.P., Addison, R.F.,

Hargrave, B.T., Pearre, S., Dupuis, A., Brodie, P.F., 1997. Bioaccumulation of polychlorinated biphenyls (PCBs) in the marine pelagic food web, based on a seasonal study in the southern Gulf of St. Lawrence, 1976–1977. Mar. Chem. 56, 145–179.

Haugen, J.E., Wania, F., Lei, Y.D., 1999. Polychlorinated biphenyls in the atmosphere of southern Norway. Environ. Sci. Technol. 33, 2340–2345.

Hawker, D.W., Connell, D.W., 1988. Octanol–water partition coeﬃcients of polychlorinated biphenyl congeners. Environ. Sci. Technol. 22, 382–387.

Horinouchi, M., Sano, M., 2000. Food habits of ﬁshes in a Zostera marina bed at Aburatsubo. Central Japan Ichthyol. Res. 47, 163–171.

Huang, C.W., Miyata, H., Lu, J.R., 1992. Levels of PCBs, PCDDs and PCDFs in soil samples from incineration sites for metal reclamation in Taiwan. Chemosphere 24, 1669– 1676.

Huhnerfuss, H., Pfaﬀenber, B., Gehrcke, B., 1995. Stereochem-ical eﬀects of PCBs in the Marine environment—seasonal variation of coplanar and atropisometric PCBs in blue mussels of the German bight. Mar. Pollut. Bull. 30, 332– 340.

Kiss, G., Varga-Puchony, Z., Tolnai, B., 2001. The seasonal changes in the concentration of polycyclic aromatic hydro-carbons in precipitation and aerosol near Lake Balaton, Hungary. Environ. Pollut. 114, 55–61.

Kowalewska, G., Konat, S.J., Wawrzyniak, W.B., Szymczak, Z.M., 2003. Transfer of organic contaminants to the Baltic in the Odra Estuary. Mar. Pollut. Bull. 46, 703–718. Kuo, S.R., Shao, K.T., 1999. Species composition of ﬁsh in the

coastal zones of the Tsengwen Estuary, with descriptions of ﬁve new records from Taiwan. Zool. Stud. 38, 391–404. Ling, Y.C., Soong, D.K., Lee, M.K., 1995. PCDD/DFs and

coplanar PCBs in sediment and ﬁsh sample from the Er-Jen River in Taiwan. Chemosphere 31, 2863–2872.

Loizeau, V., Abarnou, A., Menesguen, A., 2001. A steady-state model of PCB bioaccumulation in the sea bass (Dicentrar-chus labrax) food web from the Seine estuary, France. Estuary 24, 1074–1087.

Lu, J.R., Miyata, H., Huang, C.W., Tsai, H.T., Sheng, V.Z., Nakao, T., Mase, Y., Aozasa, O., Ohta, S., 1995. Conta-mination levels of PCDDs, PCDFs and non-orthochlorine substituted coplanar PCBs in milkﬁsh and crab from culture pond and coastal area near open-air incineration sites for metal reclamation in Wan Li, Taiwan, Republic of China. Chemosphere 31, 2959–2970.

Manchester-Neesvig, J.B., Andren, A.W., 1989. Seasonal var-iation in the atmospheric concentration of polychlorinated biphenyl congeners. Environ. Sci. Technol. 23, 1138–1148. Meijer, S.N., Steinnes, E., Ockenden, W.A., Jones, K.C., 2002.

Inﬂuence of environmental variables on the spatial distri-bution of PCBs in Norwegian and UK soils: implications for global cycling. Environ. Sci. Technol. 36, 2146–2153. ROC. Central Weather Bureau.<http://www.cwb.gov.tw/V4e/

index.htm>(accessed March 2005).

ROC. Environmental Protection Agency, 2005. Yearbook of Environmental Protection Statistics: 2004, pp. 2–54 (in Chinese).

ROC. Water Resource Agency, <http://gweb.wra.gov.tw/ wrweb/>(accessed March 2005).

Russell, R., Gobas, F.A.P.C., Haﬀner, G.D., 1999. Role of chemical and ecological factors in trophic transfer of organic chemicals in aquatic food web. Environ. Toxicol. Chem. 18, 1250–1257.

Stapleton, H.M., Skubinna, J., Baker, J.E., 2002. Seasonal dynamics of PCB and toxaphene bioaccumulation within a Lake Michigan food web. J. Great lakes Res. 28, 52–64. Stewart, A.R., Stern, G.A., Lockhart, W.L., Kidd, K.A., Salki,

A.G., Stainton, M.P., Koczanski, K., Rosenberg, G.B., Savoie, D.A., Billeck, B.N., Wilkinson, P., Muir, D.C.G., 2003. Assessing trends in organochlorine concentrations in Lake Winnipeg ﬁsh following the 1997 red river ﬂood. J. Great lakes Res. 29, 332–354.

Teil, M.J., Blanchard, M., Carru, A.M., Chesterikoﬀ, A., Chevreuil, M., 1996. Partition of metallic and organochlo-rinated pollutants and monoorthosubstituted PCB pattern in the trophic web from diﬀerent areas of river Seine. Sci. Total Environ. 181, 111–123.

Thomann, R.V., Connolly, J.P., Parkerton, T.F., 1992. An equilibrium-model of organic chemical accumulation in aquatic food webs with sediment interaction. Environ. Toxicol. Chem. 11, 615–629.

Wania, I.T., Mackay, D., 1993. Global fractionation and cold condensation of low volatility organochlorine compounds in Polar Regions. Ambio 22, 10–18.

Wang, C.H., Huang, H.Z., Hong, W.C., 2000. Analyses of PCB content in the ﬁshes in rivers of Taiwan. Ann. Rept. Nat. Inst. Environ. Anal. Taiwan, ROC. 8, 115–130 (in Chinese). Wen, C.K., 2002. Integrated Planning of Watershed Man-agement for Chiang chun, Yensuei and Er-Jen Rivers. EPA-034910093, ROC Environmental Protection Agency, (Chapter 2) (in Chinese).

Wiklund, A.K.E., Wiklund, S.J., Axelman, J., Sundelin, B., 2003. Dynamics of lipids and polychlorinated biphenyls in a Baltic amphiod (Monoporeia aﬃnis): a ﬁeld study. Environ. Toxicol. Chem. 22, 2499–2507.

Wilcke, W., Amelung, W., 2000. Persistent organic pollutant in native grassland soils along a climosequence in North America. Soil Sci. Soc. Am. J. 64, 2140–2148.