Organochlorine contaminants in mummichog (Fundulus heteroclitus) living downstream from a bleachedkraft pulp mill in the Miramichi Estuary, New Brunswick, Canada

Study sites

Mummichogs were sampled at 4 (M1) and 39 km (M3) downstream from a BKM located near the town of Miramichi, within the Miramichi Estuary. Site M1 has a muddy and rocky bottom. In operation since 1945, the mill discharged relatively high levels of chlorinated compounds, including PCDDs and PCDFs, until May 1993, when chlorine dioxide was partially substituted for chlorine in the bleaching process. Concentrations of 2,3,7,8 tetrachlorodibenzo-p-dioxin (D41) in the final effluent were reduced from 100 ppq in 1988 to 5 ppq in 1995, and those of 2,3,7,8-tetrachlorodibenzofuran (F41) were reduced from 1,000 to 13 ppq during the same time period. Since May 1993, the bleaching sequence used at the BKM is (C50/D50)_o,pDE_pD, where C50/D50 is 50% chlorine and 50% chlorine dioxide, E_o,p indicates caustic extraction with oxygen and peroxide, D signifies chlorine dioxide, and E_p indicates caustic extraction with peroxide; in addition, the effluent undergoes secondary treatment. In 1994, based on the annual production, the daily production was 1,306 air-dry metric tons of paper, and the mean flow discharged into the estuary was 72,923 m³/d [10]. The BKM is the source of more than 95% of the organic contamination in the Miramichi Estuary, whereas the groundwood mill at Nelson-Miramichi and municipal effluents are smaller contributors. A municipal dump and a former woodpreservation plant (which was closed in 1986) are located near M1. This wood-preservation plant operated from 1924 to 1986 and had used creosote (1924–1986), pentachlorophenol (1957- 1986), and copper/arsenic (1957–1986) [7]. Levels of pentachlorophenol (PCP) in water at M1 were still elevated in October 1988 (2.27 μg/L) but were not detectable in May 1993 and 1994. Site M3 is an intertidal marsh and has a sandy bottom. It is located downstream, at the mouth of the Miramichi Estuary, and it has no local source of anthropic pollution.

Mummichogs were also sampled at two sites upstream and downstream in the Bouctouche Estuary, which is located 100 km south of the Miramichi Estuary. The Bouctouche Estuary has no direct source of industrial contamination. Sources of organic wastes are municipal effluents and agricultural activities. Upstream site B1 is an intertidal marsh with a sandy bottom; site B2, located 20 km downstream, near the mouth of the estuary, has a sandy bottom covered with brown algae. The Bouctouche Estuary (33 km2 and 25 km long; 10,000 habitants) is 10 times smaller than the Miramichi Estuary (300 km2 and 80 km long; 50,000 habitants) [10]. Sited M1 and B1 are both located at the upstream limit of distribution of mummichog in the Miramichi and Bouctouche Estuaries, respectively. Forest spraying with pesticides has been more intense in the Miramichi River basin than in the Bouctouche River basin [8].

Sampling procedures

At each site, fish were captured from May 22 to June 3, 1995, using minnow traps baited with capelin. Fish were stunned by a blow to the head, and total length (to the nearest millimeter), total weight (to the nearest gram), and carcass weight (to the nearest gram) were measured. One separate kit of dissection tools, washed with acetone and then with hexane, was used per site. Gutted carcasses were stored in the freezer at −20°C. Carcasses were partially thawed, and otoliths were removed for age determination [12]. Three pools of 11 carcasses per sex and per site were prepared. Fish lengths were paired among pools from the same sex. Frozen carcasses were homogenized with a glass mortar and pestle, which were washed with soap, rinsed with water, and then rinsed with ultrapure hexane between pools. Homogenates were stored at −20°C in glass bottles that were covered with aluminum foil that had been rinsed with hexane as well as a plastic lid.

Chemical analyses

Dioxins, furans, and PCBs. Fish tissue samples were homogenized unfrozen, and 5-g aliquots were removed for routine moisture and lipid determinations. PCDD/Fs non-ortho and mono-ortho-substituted PCBs in males and females were analyzed in the laboratory of M.G. Ikonomou (Institute of Ocean Sciences, Department of Fisheries and Oceans, Sidney, British Columbia, Canada). Details of the analytical method are described elsewhere [13]. Briefly, 10 g (wet weight) of the fish homogenates was spiked with a mixture of 9 ¹³C₁₂-labeled PCDD/F and of 4 ¹³C₁₂-labeled PCB internal standards, as supplied by Cambridge Isotope Laboratories (Andover, MA, USA). The samples were dried with sodium sulfate and extracted with dichloromethane (DCM) from a glass column using gravity flow. Sample cleanup took place in four stages. Aliquots were first passed through a gel permeation chromatography column packed with Biobeads SX-3 resin (ATS Scientific, Oakville, ON, Canada) and eluted with 1:1 DCM/cyclohexane. The gel permeation chromatography column extract was then passed through a multilayer silica gel column and eluted with DCM/hexane (1:1). The third cleanup step was conducted via a neutral alumina-activated column capped with anhydrous sodium sulfate, washed with hexane, and further eluted with 1:1 DCM/hexane. Fractionation of the eluant was accomplished with an automated high-performance liquid chromatography system utilizing a carbon fiber packed with a 1:12 mixture of activated carbon/filter paper homogenate. Compounds of interest were eluted successively with (I) 20 ml of 5% DCM/hexane; (II) 44 ml of 50% DCM/hexane; (III) 50 ml of 50% ethyl acetate/benzene, and (IV) 60 ml of toluene in a reverse flow direction. Fractions II (containing the mono-ortho-PCBs), III (the non-ortho-PCBs), and IV (the PCDDs and PCDFs) were concentrated to less than 10 μl and spiked with the corresponding ¹³C₁₂-labeled method performance standards prior to gas chromatography-high-resolution mass spectrometry analysis [13]. ¹³C₁₂-labeled PCB 101 was added to fractions II and III, and ¹³C₁₂-1,2,3,4-tetrachlorodibenzo-p-dioxin and ¹³C₁₂-1,2,3,7,8,9-D63 were added to fraction IV. PCDD/Fs and non-ortho- and mono-ortho-PCBs were analyzed with a VG-Autospec high-resolution mass spectrometer (Micromass, Manchester, UK; E1, 10,000) equipped with a Hewlett Packard Model 5890 Series II gas chromatograph (Hewlett Packard, Avondale, PA, USA) and a CTC A200S autosampler (LEAP Technologies, Chapel Hill, NC, USA).

CPEs and CPs. Chlorinated pesticides in male fish were analyzed by Axys Analytic Services Ltd. (Sidney, British Columbia, Canada). Approximately 10 g (wet weight) of the fish homogenates was spiked with ¹³C₁₂-labeled hexachlorobenzene, γ-hexachlorocyclohexane, p,p′-1,1-dichloro-2,2-bis(pchlorophenyl) ethylene) (DDE), p,p′-DDT, and mirex and with deuterium-labeled endosulphan. The samples were dried with sodium sulfate and extracted with DCM from a glass column by gravity flow. In order to remove lipids, the extract was loaded onto a Biobeads SX-3 column (ATS Scientific) and eluted with DCM. Then, each extract was separated, by column chromatography on Florisil, into fractions Fr1 + Fr2 and Fr3. The Fr1 + Fr2 fraction was analyzed for nonpolar and moderately polar pesticides by gas chromatography with a highresolution mass spectrometer. Analyses were carried out on a VG Ultima AutoSpec high-resolution mass spectrometer (Micromass) equipped with a Hewlett Packard 5890 gas chromatograph, a CTC autosampler, and a VAX work station (Compaq, Houston, TX, USA). Data were acquired in the multiple ion detection mode in order to enhance sensitivity. The most polar pesticides, those in fraction Fr3, were analyzed using a Hewlett Packard 5890 gas chromatograph equipped with a 60 m × 0.25 mm, 0.25-μm film DB5 Durabond-fused silica capillary column (Scarborough, ON, Canada) and a ⁶³Ni gas chromatograph electron-capture detector.

Axys Analytical Services also analyzed chlorophenols and chloroguaiacols in homogenates from male fish. Samples were spiked with ¹³C₁₂-labeled surrogate standards (monochloro to pentachlorophenol and mono-, tri-, and tetrachloroguaiacol). The samples were dried with sodium sulfate and extracted with DCM from a glass column by gravity flow. The extracts were reacted with acetic anhydride in order to convert the CPs into their acetate derivatives. The derivatized extracts were cleaned up using column chromatography on silica gel prior to analysis for derivatized chlorophenolic CP. Analyses were carried out by gas chromatography with low resolution (quadrupole) mass spectrometry using a Finnigan INCOS 50 mass spectrometer (Finnigan, Bremen, Germany) equipped with a Varian 3400 GC, a CTC autosampler, and a DG10 data system (Varian, Walnut Creek, CA, USA).

Quality control and calculations. For PCDD/F and PCB analyses, the quality control measures undertaken for the gas chromatography with high-resolution mass spectrometer analysis were based on procedures established by Environment Canada for PCDD/F analyses [14]. Samples were processed in batches of 11; each batch consisted of nine samples, a procedural blank, and a blind duplicate (split sample). Pesticides and CP analyses were carried out in batches of nine; each batch consisted of six samples plus one reference sample (a spiked tissue sample), a procedural blank, and a blind duplicate. The procedural blanks demonstrated mainly nondetectable or low background levels of the target compounds. Octachlorodibenzo-p-dioxins (D81) and PCB 77 and 15 were detected in blanks at variable levels that were sometimes comparable to those measured in the samples. Thus, data for these compounds were withdrawn from the statistical analyses. The precision of the analytical measurements was evaluated by comparisons of data for blind duplicates. For most contaminants, the variance of the concentrations measured in split duplicates of the same homogenate was below 25% for values higher than two times the detection limit and below 75% for values close to the detection limit. Polychlorinated biphenyl 11, 13, and 127 demonstrated variances of more than 75% for duplicates and were withdrawn from the statistical analyses. The recoveries of the ¹³C₁₂-labeled surrogate standards ranged from 53 to 102% for PCDD/Fs, from 52 to 113% for PCBs, from 62 to 120% for CPEs, from 44 to 120% for chlorophenols, and from 34 to 120% for chloroguaiacols.

The concentrations of identified compounds were calculated with the internal standard method using mean relative response factors determined from calibration standard runs that were made before and after each batch of samples was run. Concentrations were corrected for recovery. Contaminants that were not detectable in all samples from all sites were withdrawn. Other undetected values were replaced by one-half of the limit of detection. Values with incorrect isotope ratio were used only if the concentrations were comparable to similar samples from the same site. Because only 2,3,7,8 PCDD and 2,3,7,8 PCDF congeners were measured, a non-2,3,7,8 congener value was calculated by subtracting the 2,3,7,8-substituted congeners from the homologue group; if the resulting value was zero, one-half of the detection limit of the homologue group was substituted.

Dioxins, furans, and coplanar PCBs may bind to the aryl hydrocarbon cellular receptor. The toxic potency of a mixture of these compounds is estimated by calculating and adding the toxic contributions of each using toxic equivalent factors. Toxic equivalent factors express the potency of each congener relative to D41 to induce specific biological responses, such as EROD induction, and they have been determined from a variety of in vivo and in vitro studies in mammals and in fish. Two sets of toxic equivalent factors were used to assess the toxicity of PCDD/Fs and coplanar PCBs: one is based on fish embryotoxicity [15, 16] and the other, which is more complete, is based on induction of EROD induction in mammalian data [11, 17](Table 2). The concentrations of individual congeners were multiplied by the appropriate toxic equivalent factors, and the sum of these products for one particular sample is the D41-equivalent concentration (TEQ).

Univariate statistical analyses

Length, carcass weight, and condition factor (K = 100 weight/length₃) data were analyzed by a two-way factorial analysis of variance (ANOVA) design that used site and sex as factors and that assessed the interaction between these factors. The variables were transformed to logarithms prior to analyses in order to meet the assumptions of normality and homoscedasticity of the residuals. Age data were analyzed with a two-way ANOVA design for ranked data. For both types of ANOVA, an a posteriori multiple comparisons test that used the least-squares means was applied when a factor was found to be significant (p < 0.05). Because a large number of comparisons were made and in order to reduce the likelihood of type I errors, these comparisons were performed at a corrected level of significance, 0.05/c, where c is the total number of pairwise comparisons [18].

Because of the small number of samples per site and sex (n = 3), a two-way ANOVA for ranked data followed by an a posteriori multiple comparisons test (least-squares means) were used to detect possible effects of site or sex on percentages of water and lipids of the homogenates, on PCDD/F and PCB concentrations, and on the TEQs. The same test was used to compare the percentages of PCDD/Fs and PCBs with a different number of chlorine atoms among sites and between sexes. Because no data were available for female homogenates, concentrations of CPs and CPEs were compared among sites with a one-way ANOVA, applied to ranks, followed by multiple comparison tests, when significant. The same test was used to compare percentages of DDE ([o,p′-DDE + p,p′-DDE]/[total DDT]*100) among sites. For each site, linear relationships between concentrations of PCBs and PCDD/Fs (log transformed) in males and concentrations in females were assessed by least-squares regression. Analysis of covariance was used to compare slopes and intercepts of the lines among sites.

Multivariate analyses

We have used two approaches of multivariate analysis in order to identify and locate various sources of contamination: the traditional approach involving comparison of congener profiles (nature of the source) and a novel approach involving comparison of spatial gradients of contamination (location of the source). The traditional approach compares the relative proportions of different congeners (percent concentrations among congeners for each site) among sites and reveals different patterns of composition of chemical contaminants at different sampling sites; these patterns will vary as a function of the nature of the source and of the pathways of dissemination of the various congeners [19, 20]. The novel approach discloses the different patterns of spatial gradients of contamination of different groups of contaminants (percent concentrations of contaminants among sites for each group of contaminants); these patterns will vary as a function of the location of the respective sources and of the pathways of dissemination. With such an analysis, it is possible to compare the pattern of distribution of a chemical with a known source (for example mirex, an airborne contaminant originating from a distant source) with those of other compounds and, in turn, to infer from that comparison the location of their source.

Multivariate analysis of contaminant composition (nature of the source). For each sample and each group of contaminants, the relative contributions (%) of each contaminant or congener were calculated, and the data were then centered and scaled [19, 20]. A principal component analysis (PCA) was performed for each group of contaminants; in this analysis, contaminant concentrations were used as variables and sites were used as objects. In order to simplify interpretation and for a better separation of variables among principal component factors, an orthogonal VARIMAX axis rotation was applied to PCA [20, 21]. In order to limit the number of factors to be extracted, only the factors having latent roots (eigen values) of greater than 1 [21] were used. When some grouping trends of samples (by site) were visualized by plotting the factor scores, a canonical discriminant analysis was used to find linear combinations of the contaminants that best summarized the differences among sites.

Multivariate analysis of spatial distribution of contaminants (location of the source). The concentrations at sites were expressed in percentages, and the data were then centered and scaled. Only male fish were used in this analysis. A canonical discriminant analysis was performed, in which sites were used as variables and contaminant concentrations were used as objects [21]. Major groups of contaminants (PCDD/Fs, PCBs, CPs, DDT and metabolites, and CPEs other than DDT) were used as categorical variables.

Univariate analyses

The length and weight of fish used to compose the homogenates did not differ among sites for each sex. Females were longer and heavier than males but were of the same age (Table 3). The condition factor was higher for fish from site B1 compared with other sites and in males compared with females for all sites. Fish from site M3 were 1 year older than the fish from other sites. For all sites, percentages of water were slightly higher in homogenates from males compared with females (Table 3). Percentages of water varied little among sites. Lipids were higher at the upstream sites compared with the downstream sites in both estuaries (Table 3).

Higher levels of most contaminants were found at M1, the upstream site of the Miramichi Estuary, except for some CPEs. Compared with the Bouctouche Estuary, D41, D51, and D71 were 25 to 50 times more concentrated in fish from M1 than in those from B1 and B2 (Table 4). Tissue concentrations of PCPs were 35 to 60 times higher, and those of TCP2, TeCP, and TeCG were 10 to 30 times higher (Table 5) in fish from M1 than in those from B1 and B2. Nine congeners of PCBs were 5 to 10 times more concentrated in fish sampled at M1, as compared with sites in the Bouctouche Estuary (Table 6). Concentrations of total DDT were about three times higher at M1, compared with other sites, and the percentage of DDE was lowest at site M1 (Table 7). The downstream site in the Miramichi Estuary (M3) had the lowest levels of CPEs, compared with other sites (Table 7). Concentrations of D41 and F41 were 2 to 3 times higher at M3, compared with those at B1 and B2, and concentrations of TCP2 were 2 to 5 times higher (Tables 4 and 5).

Within the Bouctouche Estuary, concentrations of most PCBs, of F41, and of cis-nonachlor were up to two times higher at downstream site B2 (Tables 4, 6, and 7). On the other hand, concentrations of PCP, TCP2, and DDT were up three times lower at B2 compared with those at B1 (Tables 5 and 7).

Data for both sexes were only available for PCDD/Fs and PCBs. For seven congeners of PCDD/Fs, concentrations were higher in males compared with females. For 5 out of these 7 congeners, there was a significant interaction between sexes and sites (Table 4). However, these interactions were explained by the fact that concentrations of these congeners were not detectable in males and females at most sites (except M1). Higher concentration of PCBs in males, compared with females, were detected for eight congeners (Table 6). Concentrations of PCDD/Fs and of PCBs in males were strongly correlated with concentrations in females at all sites, with concentrations that were about 1.4 to 1.5 times higher in males compared with females (log concentration in females = 0.965(log concentration in males) − 0.014; r² = 0.99; p = 0.0001). Analysis of covariance revealed no differences among sites in terms of the linear relationships between concentrations observed in males and in females (comparison of slopes, F = 0.26; p = 0.85; comparisons of intercepts, F = 0.54; p = 0.65).

Higher proportions of tetrachlorinated PCDD/Fs (F-9.1, p = 0.009) and PCBs (F = 13.1; p = 0.003) and lower proportions of pentachlorinated PCBs (F = 18.0; p = 0.0008) were found in females compared with males at all sites (Fig. 2). Higher proportions of tetrachlorinated PCBs (F = 3.24; p = 0.05) and lower proportions of pentachlorinated PCBs (F = 4.16; p = 0.03) were found in the Miramichi Estuary compared with the Bouctouche Estuary (Fig. 2).

Multivariate analysis of contaminant composition among sites

Principal component and discriminant canonical analyses gave similar results. Plots are presented only for canonical analyses because, generally, they led to a better discrimination among samples. The variances explained by the first two principal components were, respectively, 53.7 and 15.0% for PCDD/Fs, 46.1 and 30.2% for CPs, 61.2 and 20.1% for CPEs, and 54.8 and 22.0% for PCBs.

Dioxins and furans. The discriminant score plot (Fig. 3A) indicated a separation of the samples into three groups according to site: M1, M3, and B1 and B2. Higher percentages of D41, D51, D71, and, to a lesser extent, F71 were found at M1. The distribution of the loadings along the second canonical variate (Can2) revealed that, compared with other sites, M3 had higher percentages of D41 and F41. Higher percentages of all these congeners were found in the Miramichi compared with the Bouctouche Estuary.

Chlorophenols. Again, the discriminant score plot divided samples into three groups according to site: M1, M3, and B1 and B2 (Fig. 3B). The loading score plot demonstrated that PCP was present in higher proportion relative to other chlorophenolics at M1 compared with other sites. It also indicated that higher percentages of TCP2, TCG, TeCP, and TECG were observed at site M3. The distribution of loadings along the second canonical variate indicated that higher percentages of all these compounds were found in the Miramichi Estuary compared with the Bouctouche Estuary.

Chlorinated pesticides. The discriminant score plot (not shown) separated samples into four groups according to site: M1, M3, B1, and B2. The loading plot demonstrated that there were higher percentages of (1,1-dichloro-2,2-bis (p-chlorophenyl) ethane) (DDD) and DDT at the upstream sites, M1 and B1, compared with the downstream sites. Higher percentages of chlordane derivatives tended to be associated with site B2.

Polychlorinated biphenyls: The discriminant score plot (not shown) separated samples into four groups according to site: M1, M3, B1, and B2. The samples from the Miramichi Estuary were located on the left side of the score plot, whereas the samples from the Bouctouche Estuary were on the right side. The loading plot indicated that higher percentages of congeners 61, 66, 70, 114, 123, 156, and 189 were found at site M1.

Multivariate analysis of spatial distribution of contaminants

The discriminant scoring plot showed that the patterns of percent concentrations of contaminants among sites differed among different groups of contaminants (Fig. 4). Most PCDD/ Fs and CPs were located at the right side of the discriminant scoring plot, indicating that higher percentages of these compounds were found at the upstream site of the Miramichi Estuary (site M1), located on the right side of the loading plot. Most PCBs and DDT were also associated with M1 but were located more centrally. In contrast, dieldrin, mirex, chlordanes, and related compounds (nonachlors) were located to the left side, because higher percentages of these compounds were found in the Bouctouche Estuary.

Spatial gradients of contamination

Concentrations of contaminants in fish sampled in estuaries may vary among sites in terms of the function of different factors, including sources of contamination, drainage basin area, suspended particle loads and water residence times, sedimentation rates, and local food webs. The patterns of distribution of persistent contaminants originating from a distant location may be used as an indicator of the extent of among site variation that is attributable to natural ecological gradients. Mirex is a persistent lipophilic chlorinated compound used as a pesticide as well as a flame retardant. The production and use of this pesticide were restricted to the Great Lakes area from 1959 to 1976. In the Gulf of St. Lawrence, the major source of mirex is probably atmospheric deposition in precipitation and dryfall, with a limited contribution attributable to transport of contaminated particles in water from the St. Lawrence Estuary [22]. In order to investigate the location of sources of contamination, we have used multivariate analysis to compare the spatial distribution of mirex with that of other groups of contaminants. This analysis (Fig. 4) clearly indicates the presence of local sources of PCDD/Fs, CPs, DDT, and PCBs at the upstream site of the Miramichi Estuary. These results have complemented the findings of the traditional analysis of congener composition (see below), which disclosed profiles of PCDD/Fs and CPs that are typical of pulp and paper and wood-preservation industries at M1, and, in addition, this analysis revealed the presence of sources of DDT and PCBs near M1.

Concentrations of mirex in sampled fish were greatest at the upstream sites of the Bouctouche and Miramichi Estuaries and were smallest (12 times) at the downstream site of the Miramichi Estuary (site M3). This pattern of distribution is probably a consequence of particle entrapment at the upstream sites of the estuaries and may as well be due to the sandy nature of the sediment at site M3. The larger drainage basin of the Miramichi Estuary (as compared with the Bouctouche Estuary) does not seem to affect the loading of mirex, as observed previously for atmospherically borne organochlorines deposited in lakes with different sizes and drainage areas [23]. Most other pesticides, except DDT, and metabolites were found at similar concentrations in fish from M1 (compared with B1), whereas most PCDD/Fs (up to 50 times for D71), CPs (up to 35 times for PCP), PCBs (up to 10 times for PCB70), and DDT (2.6 times) were more concentrated at M1 compared with B1, which indicates local sources of contamination. Despite the relatively low capacity of retention of contaminants at M3, concentrations of TCP2 were five times higher and those of D41 were three times higher at M3, compared with B2.

Table Table 5.. Concentrations of chlorophenols and chloroguaiacols in homogenates (n = 12) of male mummichog collected in the Miramichi and the Bouctouche estuaries^a

Table Table 6.. Concentrations of PCB congeners in homogenates (n = 23) of male (M) and female (F) mummichog collected in the Miramichi and the Bouctouche estuaries^a

Comparisons between sexes

Males accumulated 1.5 times more PCDD/Fs and PCBs than did females. Higher concentrations of organochlorine contaminants in males have repeatedly been reported in fish sampled close to BKMs [24]. Female mummichog are usually larger than males of the same age [3], as was observed in this study, and a higher growth rate may lead to a greater dilution of the contaminants in their female mummichog tissues [25]. Lower levels of lipophilic contaminants in female fish have frequently been attributed to a greater loss of contaminants in eggs, rather than in sperm, at the time of spawning. However, the loss of contaminants in eggs does not appear to be significant, since the relationships between levels of contaminants in males and in females are similar among sites, whereas the number of eggs produced by the females was two times higher at M1 than at other sites in the Miramichi or Bouctouche Estuaries [10]. Higher hepatic microsomal cytochrome P450 in males, compared with females, may be responsible for the smaller proportions of tetrachlorinated PCB and PCDD/F congeners in males, since these congeners are eliminated more readily than are congeners with a higher number of chlorine atoms [26].

Multivariate analysis of contaminant composition

As expected, larger proportions of PCDD/Fs, characteristic of pulp mills with chlorine bleaching, were found in mummichog at site M1: congeners associated with chlorine bleaching including D41 and F41, and those produced by the condensation of polychlorinated phenols during pulp digestion include D41 and D51 [19]. Following partial conversion to chlorine dioxide in bleaching in 1993, PCDD/F levels were markedly reduced but were still detectable in the effluent in 1995. Thus, the observed body burdens are probably due to a combination of low levels of current exposure and sediment exposure from previous releases. Chlorophenols typical of BKME, mostly TCP2, TCG, and TeCP, were also detected in the tissues of fish from this site [27]. Higher relative levels of D41, F41, and tri- and tetrachlorinated CPs were found in mummichog from M3, indicating that these compounds, and probably other compounds contained in BKME, were transported 40 km downstream from the Miramichi Estuary. Longrange transport of CPs (up to 250 km downstream from a source) and of F41 have been observed in different rivers [27].

The PCDD present in highest proportion (25–30%) at M1 was D71. Fish tissues are usually depleted in the highly chlorinated homologues compared with sediments from the same site [28]. Thus, the presence of relatively high concentrations of heptachlorinated congeners in mummichog indicates that these compounds are found at high concentrations in sediments at M1. Indeed, up to 3.40 ng/g dry weight D71 and 0.032 ng/ g D41 were measured in grab samples of sediments collected near M1 in 1990 (Thus, D71:D41 = 106:1 in sediments, compared to 4:1 in mummichog)[29]. Relatively high concentrations of PCP (7.95 ng/g) were detected in mummichog sampled at M1. Previous studies also revealed relatively high levels of PCP metabolites (6,858 ng/g at M1 and 200 ng/g at M3) in the bile of mummichog sampled in the Miramichi Estuary in 1992 [30]. Typically, concentrations in fish fillets are at least 100 to 1,000 times lower than in bile [27]. Pentachlorophenyl, D71, and F71, which were also more concentrated at M1, may originate from the former wood-preservation plant. High levels of D71 (350–1,100 ng/g) were measured in sediments collected in 1984 in the receiving environment of aqueous effluent discharges from the Domtar wood-preserving plants in Newcastle [31].

Table Table 7.. Concentrations of chlorinated pesticides in homogenates (n = 12) of male mummichog collected in the Miramichi and the Bouctouche estuaries^a

In the industrialized Miramichi basin, possible sources of PCBs include direct industrial discharges, combustion processes, and municipal dumps. Polychlorinated biphenyls may be produced in situ by the pulp mill or may enter into the effluent as contaminants [32]. Thus, the highest proportion of less-chlorinated PCB congeners in the Miramichi compared to the Bouctouche Estuary might reflect a different source of contamination. Alternatively, it might reflect a more recent contamination or different processes of degradation. Anaerobic bacterial degradation tends to reduce concentrations of highly chlorinated congeners, whereas aerobic degradation decreases concentrations of less-chlorinated congeners [33]. Thus, both the profiles of PCBs and the persistence of PCP indicate that the degradation of chlorinated compounds is altered at site M1. This could be related to toxic and/or anoxic conditions in the sediments at this site. Enrichment of the sediment with organic matter, with black discoloration of the surface (suggestive of anoxic conditions), has been observed downstream from the BKM, in the upper Miramichi Estuary, in the sector of M1 [34].

Larger concentrations of DDT and lower proportions of DDE at the upstream site of the Miramichi Estuary reflect the spruce budworm aerial spraying carried out from 1954 to 1962 in the Miramichi watershed [8]. DDE is a degradation product from DDT and is more persistent than the parent compound. DDE has a higher volatility than DDT and is more easily transported through the atmosphere to areas where application has not taken place. Thus, it is usually found in lower proportions closer to the source [35]. The degradation of DDT may also be altered because of the toxic/anoxic conditions of the sediments.

Toxic levels

Levels of TEQs were higher in mummichog collected at site M1 than in mummichog collected from other sites. In this study, up to 3 pg/g were found in fish tissues, whereas in rainbow trout (Oncorhynchus mykiss), the threshold for EROD induction is 20 pg/g D41 in the liver [36]. In mummichog, the carcass and the liver contain approximately 1 and 10% lipids, respectively. Thus, levels of TEQs in the target tissue (the liver) might exceed (30 pg/g) the threshold for EROD induction, since at equilibrium, it is expected that persistent chlorinated compounds will be equally distributed in the tissues on a lipid-normalized basis [24]. A threshold whole-body TEQ of 0.3 to 1.0 pg/g (3% lipid) for induction was predicted in fingerling chinook salmon (Oncorhynchus tshawytsha) exposed to BKME in the Fraser River, but the study did not prove that PCDD/Fs were the only inducing compounds in the effluent [37]. Thus, PCBs and dioxins may contribute to the 2.5-fold induction of EROD observed in mummichog captured at M1. However, several studies have shown that nonpersistent compounds released in the pulp-mill effluents, including wood extractives, play a predominant role as inducers in fish exposed to BKME and that persistent chlorinated compounds may be used as tracers for the presence of these compounds [24]. A gradient of cytochrome P4501A mRNA has been observed in Atlantic tomcod (Microgadus tomcod) exposed in cages located downstream from the BKM in Newcastle, suggesting that the effluent from the mill or from another nearby point source of inducers was responsible for the observed induction [38].

Concentrations of PCP were one order of magnitude below the known thresholds of toxicity. The no-observed adverse effect level for growth in rainbow trout is 11 μg/L [39], and this concentration in water would lead to accumulation of PCP (μg/g) in fish tissues [40]. Toxicity of other less-toxic CP originating from the BKME may be additive to that of PCP, but total concentration of CPs was in the nanogram-per-gram range (11.1 μg/g). Concentrations of DDT (ng/g) were also below those associated with detrimental effects in wildlife (μg/ g) [41].

Sedentary character of the mummichog

Differences in concentrations and patterns of contamination between M1 and M3 and, to a lesser extent, between B1 and B2 indicated that mummichog from these locations did not mix during their growing period. Repeated observations of different prevalences of chronic pathological lesions, such as fin erosions and vertebral lesions, and of different meristic counts in fish sampled at M1 compared with fish sampled at M3 also support reduced movements of fish among sites [9, 42]. The effects of pulp-mill effluents on fish reproduction have been associated with nonpersistent compounds and are more likely detected with sedentary species exposed continuously to the effluent. Thus, these findings support the use of mummichog for adult fish surveys.