BIOINDICATORS OF CONTAMINANT EXPOSURE, LIVER PATHOLOGY, AND REPRODUCTIVE DEVELOPMENT IN PRESPAWNING FEMALE WINTER FLOUNDER (Pseudopleuronectes americanus) FROM URBAN AND NONURBAN ESTUARIES ON THE NORTHEAST ATLANTIC COAST
Lyndal L. Johnson, John E. Stein, Tracy K. Collier, Edmundo Casillas, Bruce McCain, and Usha VaranasiNational Marine Fisheries Service
Northwest Fisheries Science Center
Environmental Conservation Division
2725 Montlake Blvd. E.
Seattle WA 98112
U.S. DEPARTMENT OF COMMERCE
Barbara Hackman Franklin, Secretary
National Oceanic and Atmospheric Administration
John A. Knauss, Administrator
National Marine Fisheries Service
Relationships between liver pathology and ovarian development, and exposure to xenobiotic compounds were evaluated in prespawning female winter flounder (Pleuronectes americanus, formerly Pseudopleuronectes americanus) sampled from 11 sites on the Northeast coast of the United States during the 1988 and 1989 spawning seasons. Three sites were located in Boston Harbor, Massachusetts, four sites were in Raritan Bay, New Jersey, and four sites were in nearby, less urbanized embayments. Sediments from these sites exhibited a wide range in concentrations of xenobiotic compounds (e.g. concentrations of polycyclic aromatic hydrocarbons (PAHs) ranged from 20 to 50,000 ng/g dry weight and concentrations of polychlorinated biphenyls (PCBs) ranged from 2 to 1,400 ng/g dry weight), with the sites in Boston Harbor and Raritan Bay the most heavily contaminated. The following parameters associated with ovarian development were measured: ovarian developmental stage, ovarian atresia, gonadosomatic index, plasma estradiol, fecundity, and egg weight. Contaminant exposure was assessed by measuring concentrations of fluorescent aromatic compounds (FACs) in the bile; hepatic aryl hydrocarbon hydroxylase (AHH) activity; concentrations of polychlorinated biphenyls (PCBs) in liver, ovary, and brain; and levels of xenobiotic-DNA adducts in liver tissue. Additionally, liver tissue was examined histologically for the presence of suspected toxicopathic lesions. In general, indicators of contaminant exposure were elevated and prevalences of suspected toxicopathic lesions were highest in fish from sites within Boston Harbor and Raritan Bay. Moreover, prevalences of two categories of lesions--hydropic vacuolation and biliary or hepatocellular proliferation--were positively correlated with concentrations of PCBs in tissue and FACs in bile. Hepatic AHH activity, however, was significantly depressed in reproductively active fish and showed little correlation with other indicators of contaminant exposure. Evidence of decreased egg weight and increased atresia in fish exposed to high levels of PCBs or PAHs was observed. However, contaminant exposure had no clear negative impact on gonadal recrudescence, gonadosomatic index, plasma estradiol concentrations, or fecundity in female winter flounder. These results are in contrast to results with another pleuronectid species, English sole (Pleuronectes vetulus, formerly Parophrys vetulus), which shows inhibited gonadal development and lower plasma estradiol concentrations at contaminated sites in Puget Sound, Washington. The apparent difference between English sole and winter flounder in susceptibility to contaminant-induced reproductive dysfunction could be related to a number of factors, including possible interspecific differences in the activation and detoxication of contaminants. Additionally, English sole reside in contaminated estuaries throughout vitellogenesis and move offshore to spawn, while winter flounder often remain offshore for extended periods during early vitellogenesis and move into contaminated estuaries prior to spawning. Because of these contrasting migration patterns, both the duration and timing of exposure to contaminants during gonadal recrudescence may differ substantially in these two species.
Winter flounder (Pleuronectes americanus, formerly Pseudopleuronectes americanus) is an important commercial and recreational species on the Northeast coast. In recent years, however, there has been evidence of a decline in landings of winter flounder in the nearshore portion of its range (Witherell et al. 1990, Childs 1987, Smith 1989). Several factors may contribute to the apparent drop in winter flounder populations, including fishing pressure, variations in water temperature that could influence larval survival (Buckley et al. 1990, Jeffries and Terceiro 1985, Rogers 1976), changes in food supply (Laurence 1977), predation on larvae (Pearcy 1962), and destruction of larval habitat through dredging or other forms of disturbance (Monooch 1988, NOAA 1990). Additionally, because winter flounder often reside in highly contaminated urban estuaries, there has been concern that exposure to anthropogenic compounds, which may cause diseases or reproductive impairment, could be at least partially responsible for the decline in landings of this fish.
Studies conducted to date on the effects of contaminants on reproductive success in winter flounder have focused primarily on the later stages of the reproductive cycle, such as fertilization success, larval development or larval size, fecundity, and egg weight (e.g., Topp 1967, Smith and Cole 1979, Klein-MacPhee et al. 1984, Black et al. 1988, NOAA 1990). While no clear link has been established between exposure to environmental contaminants and egg viability in winter flounder, several studies present evidence of reduced egg or larval size in flounder from contaminated sites within Boston Harbor, Massachusetts and Narragansett Bay, Rhode Island (Black et al. 1988, NOAA 1990). In contrast, little is known about the impact of contaminant exposure on steroid metabolism or vitellogenesis in free-living winter flounder, in spite of the fact that recent field and laboratory studies (Singh and Singh 1987; Sivarajah et al. 1978a,b; Saxena and Garg 1978; Payne et al. 1978; Thomas 1988, 1989; Stein et al. 1991; Johnson et al. in press) show that this phase of the reproductive cycle may be disrupted by exposure to xenobiotic compounds.
In the present study, gonadal development and plasma estradiol concentrations, as well as egg weight and fecundity were assessed in female winter flounder from sites along the northeast coast with a wide range of concentrations of contaminants in sediment. The objective was to determine whether flounder showing evidence of exposure to contaminants, particularly exposure to polychlorinated biphenyls (PCBs) or polycyclic aromatic hydrocarbons (PAHs), exhibited any signs of impairment in these reproductive parameters. Although a number of chemicals may alter reproductive function in fish (Lam 1983), in this study we chose to measure concentrations of PAHs and PCBs in flounder because
Several additional bioindicators of contaminant exposure were measured in conjunction with concentrations of FACs in bile and PCBs in tissues of winter flounder. Levels of xenobiotic-DNA adducts in the liver were determined using the 32P-postlabelling assay which measures the level of hydrophobic aromatic compounds bound to DNA (Varanasi et al. 1989a). Also, liver tissue was examined for toxicopathic lesions, so their relationship with other bioindicators of contaminant exposure and with reproductive indicators could be assessed. The relationship between hepatic lesions and reproductive development was of interest for two reasons: first, hepatic lesions serve as bioindicators of exposure to environmental contaminants; and second, it is possible that lesions could affect the reproductive process directly by interfering with the production of vitellogenin or other functions associated with gonadal development that are carried out by the liver. Relationships between liver lesions and other bioindicators of contaminant exposure were also examined, because, although the occurrence of neoplastic and nonneoplastic liver disease in winter flounder from contaminated areas on the northeast coast of the United States is well-documented (Murchelano and Wolke 1985, 1991; Moore 1991; Gronlund et al. 1991), there is little information on relationships between levels of specific contaminants in winter flounder tissues or other biological indicators of contaminant exposure (e.g., hepatic aryl hydrocarbon hydroxylase (AHH) activity or xenobiotic-DNA adducts) and the risk of disease occurrence.
Multivariate techniques were used to correlate indicators of exposure with indicators of reproductive development. Because biological factors such as fish age or condition may have a strong influence on reproductive development, and the timing of the reproductive cycle can vary clinally, variables accounting for these parameters were also included in analyses.
In addition to the indicators of exposure mentioned above, hepatic
AHH activity was determined as a measurement of cytochrome P4501A
(CYP1A). Increases in CYP1A are generally indicative of contaminant
exposure in marine species (Collier and Varanasi 1991, Payne et
al. 1987, Stegeman et al. 1988), and increased CYP1A has been
positively correlated with reproductive dysfunction in some species
(Spies and Rice 1988, Johnson et al. 1988). However, in certain
teleosts, hepatic CYP1A is suppressed by estradiol (Snowberger
and Stegeman 1987, Snowberger Gray et al. 1991, Forlin and Haux
1990, Pajor et al. 1990), diminishing its effectiveness as an
indicator of contaminant exposure in reproductively active female
fish. Hepatic AHH activity was measured in this study so its
utility as an indicator of contaminant exposure in reproductively
active female winter flounder could be assessed.
Female winter flounder were collected by otter trawl from the sites shown in Figure 1. Latitudes and longitudes for each site are given in Table 1. Sampling was conducted in December 1988 and in October and December 1989, during the season when vitellogenesis normally occurs in this species (Burton and Idler 1984). Approximately 30 adult females (length > 280 mm) were randomly selected for necropsy from each site at each sampling time. Females smaller than 280 mm were not included in the study to insure that sampled females were large enough and old enough to have reached sexual maturity, which in female winter flounder occurs at approximately 270-290 mm or 3 years of age (Klein-Macphee 1978).
Flounder were individually weighed and measured, otoliths were collected for age determination, and a one mL blood sample was taken with a heparinized syringe. Blood samples were subsequently centrifuged at 800 x g, and the plasma was stored at -20°C for later measurement of plasma estradiol concentrations. Ovaries were weighed and samples were taken for histological examination. Ovarian tissues samples were also collected and stored at -20°C for later determination of PCB concentrations. In selected vitellogenic females, one entire ovary was removed, slit lengthwise, and preserved in modified Gilson's fluid for determination of fecundity (Bagenal and Braum 1971). Livers were preserved in Dietrichs' fixative (Gray 1954) while ovaries were preserved in Davidsons' fixative (Mahoney 1973). A portion of the liver was also frozen immediately in liquid nitrogen, and stored at -80°C for later determination of AHH activity. Additionally, tissue samples from the liver and brain were collected and stored at -20°C for later determination of PCB concentrations. Bile was collected and stored at -20°C for measurement of FACs. Gonadosomatic index (GSI) was calculated according to the following formula:
Because low body weight may be associated with suppressed ovarian
development in adult female fish (Burton and Idler 1987), a condition
factor was determined for all sampled animals so the influence
of emaciation on ovarian development could be distinguished from
any potential effects of contaminant exposure. Condition factor
was calculated according to the following formula:
Tissues collected for histology were embedded in paraffin, sectioned, stained with hematoxylin and eosin (Luna 1968) and examined microscopically. Hepatic lesions were classified according to the criteria outlined in Myers et al. (1987, in press), then grouped into the following categories:
A complete description of the types of individual lesions included in these categories is given in Table 2.
Ovaries were classified into the following developmental stages (Table 3), using histological criteria modified from Wallace and Selman (1981): regressed (oocytes in the perinucleolar stage), previtellogenic (oocytes with cortical alveoli), and vitellogenic (oocytes with exogenous yolk deposition in the cytoplasm). Spawning or spawned out females were not observed because sampling was conducted before the onset of spawning.
Ovaries were also examined histologically for follicular atresia and inflammatory lesions associated with oocyte resorption, including lymphoid or macrophage infiltrates. Atretic follicles were classified according to the scheme described in Hunter and Macewicz (1985) and Johnson et al. (1991) as alpha yolked atretic follicles, alpha nonyolked atretic follicles, or beta, gamma, or delta follicles (Table 4). Ovarian lesions were then divided into the following groups for statistical analysis: yolked atretic follicles (alpha stage); nonyolked atretic follicles (alpha stage); late stage atretic follicles (beta, delta, or gamma stage); and idiopathic ovarian inflammatory lesions (i.e., lymphoid or macrophage infiltrates not associated with parasitic infection). In addition, the actual proportion of yolked oocytes undergoing atresia in individual fish was determined by morphometric analysis for a subsample of approximately 10 females per site.
Concentrations of PCBs in liver and ovary were determined from approximately two gram samples, and in brain from approximately 0.2 g samples, according to the method described by MacLeod et al. (1985), with modifications as described in Stein et al. (1987). Briefly, tissue samples were ground with 10 g of silica, added to a column (270 x 23 mm) containing activated silica gel (Amicon Corp., Danvers, MA), and eluted with 50 mL pentane:methylene chloride (90:10, V/V). The eluant was concentrated and exchanged with 1 mL hexane before analysis by gas chromatography with electron capture detection. Biliary concentrations of FACs which fluoresce at benzo[a]pyrene (BaP) wavelengths were measured according to the method of Krahn et al. (1986). Hepatic microsomes were prepared and hepatic AHH activity was measured in duplicate using 14C-BaP (80 µm) as a substrate as described by Collier et al. (1986). Levels of xenobiotic-DNA adducts in liver tissue were measured using the 32P- postlabelling method as described in Varanasi et al. (1989a). Plasma estradiol levels were determined by radioimmunoassay as described by Sower and Schreck (1982).
Fecundity was determined gravimetrically using procedures described by Bagenal and Braum (1971). Preserved ovaries were allowed to remain in Gilson's fluid for at least 3 months to allow eggs to harden and ovarian connective tissue to disintegrate. Preserved eggs were then washed with water, separated from ovarian connective tissue, filtered, and dried at 60o C for 24 hours. All eggs collected were weighed, then three random samples of 200 eggs each were weighed. Fecundity was determined according to the formula:
Fecundity = 2 [(total weight of eggs)(number of
eggs in subsample)/mean weight of eggs in subsample].
Hepatic and ovarian lesion prevalences were calculated for each sampling site and the G-statistic (Sokal and Rohlf 1981) was used to determine whether prevalences in winter from other sampling sites differed significantly from lesion prevalences in winter flounder from Niantic Bay, the site with the lowest levels of PAHs and PCBs in sediments. The significance level for these tests was set at p < 0.05.
Analysis of variance (Sokal and Rohlf 1981) was used to compare mean concentrations of estradiol in plasma and GSI at the sampling sites. Analysis of variance was also used to determine the effect of ovarian maturation stage on hepatic AHH activity, so any variation in AHH activity associated with the reproductive cycle could be separated from changes in AHH activity associated with contaminant exposure.
Regression analysis (Sokal and Rohlf 1981) was used to examine relationships between variables associated with contaminant exposure (i.e., concentrations of FACs in bile, hepatic DNA adducts, concentrations of PCBs in tissues, site of capture, and the presence of hepatic lesions) and variables associated with reproductive activity (i.e., GSI, plasma estradiol, fecundity, and egg weight). The effects of fish age, length, condition factor, year of sampling (1988 vs. 1989), season of sampling (October vs. December), and ovarian developmental stage were taken into account in these analyses. For analyses involving the effect of site of capture, Niantic Bay was chosen as the reference site, and other sites were evaluated relative to Niantic Bay. Regression models were fitted using stepwise linear regression, with the significance level for entry of variables into the equation set at p < 0.05. Analyses were conducted using the Statview II (Mention of trade names is for information only and does not constitute endorsement by the Department of Commerce) statistical package. Regression analysis was also used to assess the influence of fish age and sampling season on bioindicators of contaminant exposure. Spearman rank correlation, a nonparametric correlation technique (Sokal and Rohlf 1981), was used to examine relationships between liver lesion prevalences and other bioindicators of contaminants exposure.
Logistic regression (Breslow and Day 1980) was used to assess the relationships between ovarian recrudescence and ovarian atresia and indicators of contaminant exposure (i.e., site of capture, bile FACs, hepatic DNA adducts, tissue PCBs, and the presence of hepatic lesions). This statistical method is similar to multiple regression but is suitable for use with binomially distributed outcome data (e.g., data denoting whether or not ovarian development is occurring, or whether ovarian lesions are present or absent). Fish age, length, and condition factor were included in all analyses so the influence of these factors could be accounted for. As in the multiple regression analyses, when the effect of site of capture was examined, Niantic Bay was chosen as the reference site and the effects of other sites were evaluated relative to Niantic Bay. The EGRET statistical package was employed to fit the logistic regression models using a procedure similar to stepwise regression (Draper and Smith 1980). Variables were discarded from the models if their entry failed to produce a change in scaled deviance that was statistically significant (p < 0.05).
Fish Size, Age, and Condition
Mean length, age, and condition factor of fish captured at the sampling sites are shown in Figure 2; mean values of age, length, and condition factor broken down by site, year, and sampling season are given in Appendix table A-1. While the range in these parameters (e.g. 1-6 cm, 1-2 years) was not great, significant differences were found between the sampling sites, and between fish sampled from different geographical areas and during different years (Fig. 2, Table 5). In general, fish from north of Cape Cod were larger and had higher condition indices than fish from south of Cape Cod. In addition, there were size and age differences among fish collected from various sites. Fish from Deer Island, Mystic River, and Quincy Bay were significantly older and larger than those at the Niantic reference site, while fish from Gravesend were younger and smaller. There also appeared to be some year-to-year variation in fish condition; condition indices were significantly lower in the 1989 than in the 1988 sampling (Appendix table A-1).
Biliary Fluorescent Aromatic Compounds
Of all the fish sampled, those from Mystic River had the highest overall levels of FACs in bile (Fig. 3, Appendix table A-2). Over the 2 years of sampling, mean concentrations of FACs-BaP in bile ranged from 1,500 to 14,000 ng/g. In both October and December of 1989, FACs-BaP concentrations in flounder from Mystic River were roughly the same, with FAC-BaP concentrations ranging from approximately 1,500 to 12,000 ng/g, and means of 4,000 to 5,000 ng/g. However, in December 1988, bile FAC concentrations were much higher, exceeding 10,000 ng/g in 58% of fish sampled. The reason for the exceptionally high FAC-BaP concentrations observed in 1988 is unknown. In addition to Mystic River, FAC concentrations (mean ± SD) were also elevated at the Massachusetts Bay outside Boston Harbor (2,900 ± 3,000 ng/g) and Old Orchard Shoals within Raritan Bay (4,100 ± 2,000 ng/g). However, at other sites within Boston Harbor and Raritan Bay FAC concentrations were generally less than 2,000 ng/g and were not statistically different from concentrations at the Niantic Bay or Duxbury reference sites. Apart from the unusually high FAC concentrations in flounder sampled from Mystic River in 1988, sampling season appeared to have little impact on FACs concentrations (Table 6); the grand mean concentrations were 1,400 ± 2,000 ng/g in December (n = 145), and 1,400 ± 1,300 ng/g in October (n = 65).
Mean concentrations of PCBs in winter flounder ovaries ranged from approximately 300 ng/g wet weight at Duxbury to approximately 500 ng/g wet weight at the Boston Harbor and Raritan Bay sites (Figure 4a). Concentrations of PCBs in ovary tissue were significantly elevated in comparison to Niantic Bay levels in flounder from four sites: Mystic River and Deer Island in Boston Harbor, and Sandy Hook Channel and Shrewsbury River in Raritan Bay. Mean concentrations of PCBs in liver of fish from Boston Harbor were about 1,000 ng/g wet weight or less at all three sampling sites (i.e., Mystic River, Deer Island, and Quincy Bay), and were not significantly different from liver PCB concentrations in fish from Niantic Bay (Figure 4b). In Raritan Bay, on the other hand, liver PCB concentrations were significantly elevated at both Sandy Hook Channel and Shrewsbury River (Figure 4b), the same sites with elevated concentrations of PCBs in ovaries. In brain tissue, mean concentrations of PCBs ranged from a low of 530 ng/g at Deer Island to a high of 1,000 ng/g at Gravesend Bay (Figure 4c), but intersite differences were not statistically significant. Concentrations of PCBs in brains of fish from Duxbury and Niantic Bay (720 and 740 ng/g, respectively) were comparable to those in fish from heavily contaminated sites in Boston Harbor and Raritan Bay. Concentrations of PCBs in winter flounder tissues listed by site, year, and sampling season are given in Appendix table A-2. Wet weight concentrations of PCBs in tissues are approximately 1/5 of concentrations expressed in dry weight.
Concentrations of PCBs in liver, ovary, and brain were positively correlated: for liver vs. ovary, r2 = 0.107, n = 70, p = 0.006; for brain vs. liver, r2 = 0.068, n = 70, p = 0.03; for brain vs. ovary, r2 = 0.14, n = 79, p = 0.007. However, relative concentrations of PCBs in various tissue compartments appeared to be affected by the sampling season and reproductive stage of winter flounder (Table 6, Appendix table A-2). Ovarian PCB concentrations were significantly higher (p = 0.0018) in fish sampled in December than in fish sampled from the same areas in October (360 ± 200 ng/g wet weight (n = 54) and 250 ± 140 ng/g wet weight (n = 52), respectively) and were positively correlated with gonadosomatic index (p = 0.001, r2 = 0.231, n = 105). Liver PCB concentrations showed the opposite pattern; they were significantly (p = 0.0002) lower in fish sampled in December than in fish sampled in October (1,300 ± 750 ng/g wet weight, n = 66, and 550 ± 400 ng/g wet weight, n = 19, respectively), and tended to decline as GSI increased (for log[liver PCB] vs. GSI, r2 = 0.09, p = 0.012, n = 70). Brain PCB concentrations, in contrast, did not change significantly during the sampling season. The mean PCB concentration in October was 760 ± 700 ng/g wet weight (n = 82), while in December it was 860 ± 540 ng/g wet weight (n = 82). Brain PCB concentrations were not significantly correlated with GSI.
Hepatic Xenobiotic-DNA Adducts
Like concentrations of PCBs in tissues and FACs in bile, levels of xenobiotic-DNA adducts in liver were generally elevated in winter flounder from within Boston Harbor and Raritan Bay (Figure 5, Appendix table A-2). Xenobiotic-adduct levels were significantly higher at Deer Island and Mystic River in Boston Harbor, and at Sandy Hook Channel in Raritan Bay than at the Niantic Bay reference site. For flounder from Deer Island and Mystic River, the mean level of DNA adducts in liver tissue was approximately 50 pmol/mol bases, and levels of DNA adducts in liver tissue of fish from Sandy Hook Channel were also relatively high (38 pmol/mol bases). In contrast, DNA adduct levels in flounder from other sampling sites were generally less than 20 pmol/mol bases. Concentrations of xenobiotic-DNA adducts in liver were also positively correlated with concentrations of PCBs in tissues and FACs in bile. For FAC-BaP vs. DNA adducts, r2 = 0.27, n = 37; and for ovarian PCBs vs. DNA adducts r2 = 0.22, n = 22, and for liver PCBs vs. DNA adducts, r2 = 0.34, n = 19. Levels of xenobiotic-DNA adducts in liver remained relatively stable throughout the sampling season (Table 6). In October, the mean concentrations was 34 ± 7.0 pmol/mol bases (n = 13), while in December it was 29 ± 21 (n = 40) pmol/mol bases.
Hepatic AHH Activity
Hepatic AHH activity was not significantly elevated in flounder from sites within Boston Harbor or Raritan Bay in comparison to flounder from the Niantic Bay reference site, in spite of the relatively high concentrations of sediment-associated contaminants in Boston Harbor and Raritan Bay (Figure 6). In fact, in some cases AHH activity was significantly lower (analysis of variance (ANOVA), p < 0.05) in flounder from heavily contaminated sites. Activity levels varied considerably, ranging from 61 to 780 pmol/mg/min in fish from Mystic River, 210 to 730 pmol/mg/min in fish from Deer Island, and from 100 to 750 pg/mg/min in fish from the Raritan Bay sampling sites. In fish from Niantic Bay, one of the least contaminated of the sampling sites, AHH activity ranged from 460 pmol/mg/min to 1,100 pmol/mg/min. Moreover, hepatic AHH activity was not positively correlated with other bioindicators of contaminant exposure. However, hepatic AHH activity was closely related to the stage of ovarian developmental stage in sampled animals (Table 6, Figure 7). In flounder from Boston Harbor and Raritan Bay as well as in flounder from the less contaminated embayments, hepatic AHH activity was significantly lower (ANOVA, p = 0.001) in vitellogenic flounder (220 ± 330, n = 250) than in flounder that were nonvitellogenic (980 ± 620, n = 29) (Figure 5). Even in October, when animals were in early stages of vitellogenesis, differences in AHH activity were marked. For vitellogenic fish sampled at this time, mean AHH activity was 480 ± 400 pmol/mg/min (n = 55), while in nonvitellogenic fish it was 1,500 ± 800 pmol/mg/min (n = 13). Moreover, AHH activity was negatively correlated with both concentrations of estradiol in plasma (r2 = 0.312, p = 0.0001, n = 312, log-transformed AHH) and GSI (r2 = 0.437, p = 0.0001, n = 342, log-transformed AHH), declining as estradiol levels and ovary weight increased. Because of the effect of the reproductive cycle on hepatic CYP1A in female winter flounder, it was not used as an exposure indicator in subsequent analyses of relationships between contaminant exposure and ovarian development.
Neoplastic lesions were relatively rare in the winter flounder sampled in this study (Figure 8, Appendix table A-3). Of the 586 fish examined, only one fish from the Mystic River had a neoplasm. Foci of cellular alteration were more prevalent, affecting 3% of all fish examined. Of the 16 fish affected, 63% were from Boston Harbor and Raritan Bay, but 37% were from the moderately to minimally contaminated sites such as Niantic Bay, Connecticut; Duxbury Bay, Massachusetts; and Narragansett Bay, Rhode Island. The two other categories of hepatic lesions, hydropic vacuolation and proliferative lesions (predominantly biliary proliferation), were more common, affecting 22 and 5% of fish, respectively. Prevalences of proliferative lesions were significantly elevated at the Deer Island, Mystic River, and Quincy Bay sites in Boston Harbor, and at the Shrewsbury River, Sandy Hook Channel, and Old Orchard Shoals sites in Raritan Bay (Figure 6). Prevalences of hydropic vacuolation were significantly elevated at all sampling sites within Boston Harbor and Raritan Bay, and at Narragansett Bay. Highest prevalences (>40%) were found at Mystic River, Old Orchard Shoals, and Sandy Hook Channel.
Neoplastic and putatively preneoplastic lesions showed no relationship with tissue contaminant concentration, partly because they were observed in so few animals (Table 7). However, concentrations of FACs-BaP in bile were positively correlated (Spearman-Rank correlation coefficient, p < 0.05) with prevalences of both hydropic vacuolation (rho = 0.497, n = 21) and proliferative liver lesions (rho = 0.565, n = 21). Concentrations of PCBs in liver were also positively correlated with both of these lesions (for hydropic vacuolation, rho = 0.870, n = 21; for proliferative lesions, rho = 0.660, n = 15). Proliferative liver lesion prevalences were positively correlated with ovarian PCB concentrations (rho = 0.647, n = 15) and levels of DNA adducts in liver (rho = 0.717, n = 9).
In general, prevalences of hepatic lesions were somewhat higher
in October than in December (Appendix table A-3). Prevalences
of FCA, proliferative lesions, and hydropic vacuolation were 4.5,
8 and 40%, respectively, in October (n = 224), while in December,
prevalences were 1.6, 3, and 20% (n = 362).
Assessment of ovarian development in winter flounder showed that
although gonadal recrudescence was well under way in most females
sampled, some adult females had not entered vitellogenesis (Figure 9,
Appendix table A-4).
The majority of these animals were from
sites north of Raritan Bay. In Raritan Bay over 99% of sampled
females were maturing, while only 77% were maturing from sites
north of Raritan Bay. Furthermore, the prevalence of inhibited
ovarian development showed no clear relationship to contaminant
levels at the sampling sites; females from the Raritan Bay sites,
which had relatively high contaminant levels, showed no impairment
of ovarian development, and fish from Mystic River, which was
the most contaminated of all sites sampled, showed prevalences
of ovarian development comparable to those at minimally contaminated
sites north of Cape Cod. Logistic regression analysis (Figure 10)
revealed that the most important factor influencing ovarian development
was fish age. In addition, fish captured north of Cape Cod were
less likely to be vitellogenic than animals captured farther south.
However, none of the measured indicators of contaminant exposure
were related to a decreased probability of ovarian development.
Neither sampling year nor sampling season (i.e., October vs.
December) significantly influenced the stage of ovarian development.
Apparently, by October most females that were going to undergo
gonadal development that season had entered vitellogenesis.
Plasma Estradiol and GSI
Significant intersite differences in GSI and plasma estradiol
concentrations in female winter flounder were observed (Figure 11 a,b);
however, these reproductive indicators were not depressed
in fish from contaminated sites. Moreover, no negative relationships
could be found between either GSI or estradiol and any indicator
of contaminant exposure that was measured. Similarly, the presence
of hepatic lesions had no significant effect on either plasma
estradiol concentrations or GSI (Table 8). Both plasma estradiol
concentrations and GSI were significantly higher in fish sampled
in December than in fish sampled in October (Table 8).
Fecundity and Egg Weight
No significant intersite differences in fecundity were found in
winter flounder (Figure 12a, Appendix table A-4), and fecundity
showed no relationship with concentrations of FACs in bile (Table 8).
Fecundity tended to be lower in flounder with the highest
concentrations of PCB in ovary and brain, but these relationships
were not statistically significant after the influence of fish
length on fecundity had been taken into account (Table 8). Mean
individual egg weight was somewhat lower in fish from Deer Island
and Mystic River than in fish from other sites, but the differences
were not significant (Figure 12b, Appendix table A-4). However,
a significant negative correlation was found between egg weight
and biliary FACs, which remained significant even after biological
factors had been taken into account (Table 8). The effects of
sampling year and sampling season could not be evaluated because
data were collected only in December 1989.
Ovarian lesions were found in substantial proportions of both vitellogenic and nonvitellogenic winter flounder (Figure 13 a,b). Prevalences of ovarian lesions were similar for flounder at all sites sampled, except for fish from Mystic River, which showed significantly higher prevalences of atresia in both vitellogenic and nonvitellogenic females. However, the morphometrically-determined percentage of yolked oocytes undergoing atresia was approximately the same at all sites sampled, ranging from 5.4 to 6.6%. Logistic regression analysis (Table 9) indicated that high concentrations of PCBs in the liver were associated with a significantly increased probability of atresia of yolked oocytes in vitellogenic females (p = 0.026, 6% of variation explained), but the actual percentage of atretic oocytes within ovaries of individual females was not correlated with any of the indicators of contaminant exposure. In nonvitellogenic females, atresia of nonyolked oocytes was significantly more common (logistic regression, p = 0.042, 4% of variation explained) in fish with elevated biliary FAC concentrations. No correlations were found between ovarian or brain PCB concentrations, hepatic lesions, or xenobiotic-DNA adducts and the probability of ovarian atresia. In vitellogenic fish, GSI was found to be negatively associated with atresia of yolked oocytes (p = 0.016, 1% of variation explained), nonyolked oocytes (p = 0.001, 8% of variation explained), and ovarian inflammatory lesions (p = 0.011, 2% of variation explained). The negative correlation between GSI and the presence of ovarian lesions may be an indication that lesions are more likely to develop during early stages of ovarian development.