The Hudson River is contaminated with polychlorinated biphenyls (PCBs) from Fort Edward, New York, USA, to New York City. Capacitor manufacturing facilities at Fort Edward and Hudson Falls, New York, USA, are considered to be the major sources of PCBs in the upper Hudson River, with discharges beginning in 1947. Between 1966 and 1974, the Fort Edward and Hudson Falls facilities purchased 35,000 metric tons of PCBs or 15% of domestic sales in the United States. It is estimated that approximately 600 metric tons of PCBs were released into the Hudson River between the 1940s and 1977 1, 2.
Concentrations of PCBs measured in the livers of mink (Mustela vison) collected in the vicinity of the Hudson River are above thresholds associated with effects and have not decreased over time. Mink collected over 25 years ago had hepatic concentrations of PCBs 3 that were equivalent to concentrations associated with reproductive impairment in ranch mink experimentally exposed to PCBs 4, 5. More recently, mink collected in the vicinity of the Hudson River 16 years after the initial study 3 had no apparent decrease in hepatic PCB concentrations 6, which were above the criteria for impairment of mink health and reproduction 7, 8.
Mink are among the most sensitive species to PCBs 9–18. Because mink are fish-eating mammals that satisfy criteria, including chemical sensitivity, for a sentinel wildlife species 19, regulatory agencies such as the U.S. Environmental Protection Agency (U.S. EPA), U.S. Fish and Wildlife Service, Environment Canada, and the Swedish Environmental Protection Agency recognize the mink as such, making it one of the most commonly selected receptors in ecological risk assessments for sites involving aquatic habitats with elevated concentrations of PCBs and related compounds 20.
Thus, the objective of the present study was to evaluate the health effects of feeding farm-raised mink diets containing PCB-contaminated fish from the Hudson River. In the present report, the effects on adult reproductive performance (percentage of females whelping, gestation length, percentage of stillbirths, and live kits whelped) and offspring growth and mortality through 31 weeks of age are discussed in terms of dietary and hepatic concentrations of ∑PCBs and World Health Organization (WHO) 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) toxic equivalents (TEQsWHO 2005). A companion paper presents the effects on organ mass and pathology, including the mandible and maxilla, of adult mink and their offspring [21; this issue].
Common carp (Cyprinus carpio) and Atlantic herring (Clupea harengus) were used as fish components of the experimental diets. Carp were collected from the upper Hudson River from Northumberland Pool (114 kg), from the vicinity of Lock 2 (879 kg), and from the first 610 m of Moses Kill (528 kg; Fig. 1) by the New York Department of Environmental Conservation and transported frozen to the Michigan State University Experimental Fur Farm. Fish were ground, blended, sampled for contaminant and nutrient analyses, and frozen for subsequent incorporation into mink feed. Whole Atlantic herring, purchased from Finicky Pet Food, was chosen as the control fish because they, like carp, contain the enzyme thiaminase 22. The herring was ground, blended, and sampled as described above.
Treatment diets were based on the Michigan State University Experimental Fur Farm ranch diet formulated to meet the nutrient requirements of mink 22. Diets contained 25% whole ground chicken, 24% water, 20% fish, 15% wheat middlings (Akey), 4% spray-dried eggs (Van Eldren), 3.5% spray-dried liver (Van Eldren), 3.5% soybean oil (North American Nutrition), 3% spray-dried blood (California Spray Dry), 1% phosphoric acid (food grade, 75%; Astaris), 0.48% vitamin premix (Akey), 0.48% mineral premix (Akey), and 0.04% biotin (Archer Daniels Midland). The proportion of fish incorporated into the feed (20%) is in the range of fish consumed by mink in the wild 12. The control diet contained 20% ocean herring, and the remaining five treatment diets contained a mixture of the herring and Hudson River fish in the following proportions: 2.5% Hudson River fish/17.5% herring; 5% Hudson River fish/15.0% herring; 10% Hudson River fish/10% herring; 15% Hudson River fish/5% herring; 20% Hudson River fish. The decision to incorporate no more than 20% Hudson River fish into the diet was based on the high mean ∑PCB concentration in the collected fish (36 µg ∑PCBs/g wet wt) and the desire to not cause complete reproductive failure and/or adult mortality. Targeted dietary concentrations of ∑PCBs were 0, 0.90, 1.8, 3.6, 5.4, and 7.2 µg/g feed based on 36 µg ∑PCBs/g wet weight in the Hudson River fish.
Dietary and tissue samples were analyzed for a number of chemicals. Samples of each treatment diet were shipped to Alpha Woods Hole Laboratory for analysis of ∑PCBs, PCB homolog groups, and potentially toxic and bioaccumulative metals and minerals. For the PCB analysis 23, samples were homogenized in diatomaceous earth containing 1:1 dichloromethane:acetone and exchanged to hexane. The extracts were cleaned with sulfuric acid. The extract was then analyzed for PCB congeners and homolog groups by low-resolution gas chromatography–mass spectrometry (GC–MS) in selective ion monitoring (SIM) mode. Blanks, duplicates, and spikes were analyzed at a frequency of one per analytical batch of up to 20 samples. In addition, to evaluate accurate and consistent extraction and analysis, a standard reference material (SRM), SRM 1946 (National Institute of Standards and Technology, USA, Lake Superior Fish Tissue), was extracted and analyzed with each analytical batch. This SRM has certified values for 30 PCB congeners. Results for the congener analyses were to be within 20% of the 95% confidence interval (CI) for certified concentrations. Overall, greater than 93% of the reference material results were within the control limits for the certified concentrations.
Individual metals (Supplemental Data, Table S1) were analyzed by U.S. EPA method 6020A 24 using inductively coupled plasma-mass spectrometry. Mercury was analyzed by U.S. EPA method 7471A 25 using cold vapor atomic absorption spectroscopy.
Dietary and tissue PCB congeners
Diet samples were also sent to Axys Analytical Services for analysis of PCB congeners, polychlorinated dibenzo-p-dioxins (PCDDs), polychlorinated dibenzofurans (PCDFs), chlorinated pesticides, and polybrominated diphenyl ethers (PBDEs). In addition, liver samples from adults, six-week-old kits, and 31-week-old juveniles were submitted for analysis of PCB congeners. For chlorinated pesticides (Supplemental Data, Table S1) and PCB congeners, the sample was homogenized with anhydrous powdered sodium sulfate and extracted with dichloromethane. The extracts were cleaned with a Biobeads SX-3 (Bio-Rad) gel permeation column, followed by a Florisil (US Silica) column cleanup 26. The extract was then split, and one split was analyzed for pesticides 27 and most PCB congeners 26 using low-resolution GC–MS. The other split was analyzed for coplanar PCBs 26 with high-resolution GC–MS. The split for coplanar PCBs was further cleaned using an acid/base-layered silica column, a 4.5% carbon AX-21 (Anderson Development)/celite 545 (Sigma-Aldrich) mixture, and finally an alumina column.
Dietary and tissue PCDD, PCDF, and PBDE congeners
For PCDD/PCDF and PBDE analyses, the procedure was the same through the Biobeads SX-3 gel permeation column. After that, a fluid-management system was used with the following order of columns: jumbo acid silica, layered acid/base silica, alumina, and a carbon column. Individual PCDD/PCDF congeners 28 and individual PBDE congeners 29 were analyzed using high-resolution GC–MS. The percentage of lipids was determined gravimetrically for all samples by the weight of the total residue contained in an aliquot of the sample extract. Dietary and hepatic concentrations of analytes are presented on a wet-weight basis unless otherwise noted.
Data validation was based on the quality-assurance/quality-control criteria documented in the Analytical Quality Assurance Plan: Hudson River Natural Resource Damage Assessment (Version 2.0, September 1, 2005), U.S. EPA National Functional Guidelines for Organic Data Review (1999), U.S. EPA National Functional Guidelines for Inorganic Data Review (1994), and the individual laboratory standard operating procedures cited above.
The Michigan State University Institutional Animal Care and Use Committee approved the use of animals in the present study. Seventy-five first-year (virgin), natural dark, female mink and 30 first-year, natural dark, male mink from the Michigan State University Experimental Fur Farm herd were assigned to the six treatment groups on December 26, 2006. Littermates were not placed in the same treatment group, to minimize genetic predisposition to PCB toxicity. The control and targeted 5.4- and 7.2 µg ∑PCBs/g feed groups had 15 females and five males each and the targeted 0.90-, 1.8-, and 3.6 µg ∑PCBs/g feed groups had 10 females and five males each. The number of mink placed on trial balanced a reasonable expectation of detecting biologically meaningful events subject to the limitations of available time and resources to conduct the study. The control and the two highest treatment groups were assigned 15 females rather than 10 to increase sample sizes at dietary concentrations where severe effects were anticipated.
Female mink were housed individually in suspended wire cages (76 cm L × 61 cm W × 46 cm H) on the outside aisles of an open-sided mink shed. Five animals per treatment group were assigned to a bank of five cages. Assignment of treatments to banks of cages was done randomly. A wooden nest box (38 cm L × 28 cm W × 27 cm H) bedded with aspen shavings and excelsior (wood wool) was attached to the outside of each cage. Males were assigned to banks of five cages (61 cm L × 30 cm W × 38 cm H) on an inside aisle of the same shed. Feed and water were available ad libitum. The “Standard Guidelines for the Operation of Mink Farms in the United States” 30 were followed for housing and maintenance of animals.
Mink were started on their respective treatment diets on January 3, 2007, after a one-week acclimation period. Fresh feed was provided daily, and water was available ad libitum. One hour prior to feeding the treatment diets each day, animals were given 10 g of control feed to which thiamine had been added (0.25 mg thiamine hydrochloride/animal) to prevent Chastek's paralysis that could result from the incorporation of thiaminase-containing fish into the feed 22. Feed consumption was determined for the same 2-d period each week, and body mass was recorded every two weeks. On February 14, 2007, in response to a decrease in feed consumption in the groups fed the diets that contained predominantly herring, the decision was made to incorporate a flavor enhancer (Proliant B3301 Spray-dried beef flavor) and additional vitamin E into all treatment diets at rates of 66 and 4.4 g/kg feed, respectively. Animals were provided the diets containing the flavor enhancer beginning February 15, 2007, through the end of the trial. Measurement of feed consumption and determination of body mass were discontinued at the initiation of breeding, to reduce stress to the females. Measurement of feed consumption was not resumed after whelping because the females and their litters consumed the same feed until weaning. After weaning, kits were group-housed until they were approximately 10 weeks old, making accurate determination of individual feed consumption difficult. Because adult feed consumption cannot be measured from initiation of breeding through weaning of kits, the value for average daily feed consumption determined for the first seven weeks of the trial was applied for the total duration of the trial.
Females were mated to males within their respective treatment groups between March 1 and March 21, 2007. Each female was given an opportunity to mate every fourth day until a presumed successful mating occurred as determined by posture during mating and postcoital appearance of the female's vaginal area. A mated female was given an opportunity to breed with a different male the day following a successful mating and on the eighth and ninth days after a successful mating (a common commercial mink breeding practice).
The whelping period began on April 22 and ended on May 9, 2007. Nest boxes were checked on a daily basis for the presence of mink kits. The gender of the kits was determined at birth, and the live and stillborn young were counted. Body mass of kits was recorded at birth and at three and six weeks of age. Body mass of adult females was recorded at the time their litters were weighed.
All surviving adult mink were killed with CO2 and necropsied between June 6 and June 19, 2007. Adult males were necropsied on June 6, 2007, adult females that did not whelp or lost their kits before weaning were necropsied on June 7, 2007, and adult females that whelped and successfully weaned their kits were necropsied on June 12 and June 19, 2007, with the exception of one female in the 7.2 µg ∑PCBs/g feed group that was still nursing a kit. The necropsy dates were chosen to allow assessment of placental scarring in females that did not whelp and in females that lost their kits before the uterine horns regressed and to allow those females that had kits to continue nursing until kits were six weeks of age. Altogether, 14 control females, nine 0.90 µg ∑PCBs/g feed females, nine 1.8 µg ∑PCBs/g feed females, two 3.6 µg ∑PCBs/g feed females, and three 5.4 µg ∑PCBs/g feed females were killed and necropsied during the last phase. The single surviving female in the 7.2 µg ∑PCBs/g feed group was necropsied on June 28, 2007.
On June 12 and June 19, 2007, a sample of weanling kits (approximately six weeks old) was killed and necropsied as previously described. This sample included 15 control kits (seven females, eight males), nine 0.90 µg ∑PCBs/g feed kits (four females, five males), and 13 1.8 µg ∑PCBs/g feed kits (seven females, six males). There were not sufficient numbers of kits in the 3.6-, 5.4-, and 7.2 µg ∑PCBs/g feed groups to justify necropsy of weanlings. Sampling was done to insure that as many litters as possible were represented.
The remaining kits were maintained on their respective treatment diets. On July 17, 2007, littermates were separated, with animals being assigned to individual cages (61 cm L × 30 cm W × 38 cm H) in the two interior rows of the open-sided shed. Body mass was determined on a monthly basis until termination of the trial during the first week in December. On December 4 and 5, 2007, seven-month-old animals in the two remaining treatment groups (16 females and 14 males in the control group and 11 females and 12 males in the 0.90 µg ∑PCBs/g feed treatment group) were killed and necropsied as previously described.
Toxic equivalents were calculated by summing the products of individual PCDD, PCDF, and PCB congener concentrations and their respective toxic equivalency factors (TEF) 31. Congeners that had concentrations less than the detection limit were assigned a concentration equal to one-half the detection limit. The choice of assigned value (0, one-half detection limit, or detection limit) had no substantive influence on ∑PCB congener or TEQ concentrations based on a quantitative assessment.
Measurement end points of interest were classified into one of three data types and statistically analyzed according to data type. Possible data types were as follows: continuous measurements, such as ΣPCBs in livers; counts, such as kits per litter; or binary outcomes, such as whether or not an individual kit survived to three weeks. Statistical analyses of measurement end points were conducted using a generalized linear model framework 32 where the most appropriate class of linear models was selected based on classification of data type and correlation structure (e.g., repeated measures, kits clustered within litters). A summary of endpoints classified by data type and analysis method is provided in Table 1.
Table 1. Summary of study end points, data types, and statistical methods
Stillbirth and kit mortality at three and six weeks of age
Change in body mass of seven-month-old juveniles during growth trial
Linear GEE regression
∑PCBs and ∑TEQs in adult livers
∑PCBs and ∑TEQs in six-week-old kit livers
Linear GEE regression
∑PCBs and ∑TEQs in seven-month-old juvenile livers
Linear GEE regression
Continuous endpoints were analyzed by analysis of variance (ANOVA) 33 when the endpoint was measured at the experimental unit level and the experimental units within a treatment group were expected to be independent (e.g., adult mink). For example, gestation lengths for pregnant females in the same treatment group were expected to be independent. Continuous endpoints having repeated measures on an individual animal or kits clustered within litters were analyzed with linear generalized estimating equation (GEE) regression. The GEE models are a contemporary extension of generalized linear models for clustered data that adjust for within-cluster correlation 34. Examples of clustered data in the present study were repeated measures of adult female body mass and kit liver ΣPCB concentrations within litters. The adult female body mass model for the prebreeding period was adjusted for baseline body mass and days on treatment. Likewise, the adult female feed consumption model was adjusted for days on treatment.
Count endpoints measured on adults, including number of kits whelped and kits whelped alive, were analyzed using negative binomial regression models. Treatment effects on the rate of kits whelped alive were estimated as differences (%) in rate of kits whelped alive per litter from control. Refitting the model with ΣPCBs in feed as a continuous variable, the relative rate of live kits per litter for a given increase in micrograms ∑PCBs/g feed (Δ) was estimated.
The minimum dietary concentrations necessary to induce 20 and 50% kit mortality (LC20 and LC50) were estimated based on the maximum likelihood estimates provided by the beta-binomial regression. Dose–response relationships were estimated for ΣPCBs as well as TEQs. The LC20 and LC50 values expressed as ΣPCBs and TEQs were estimated (with 95% CIs) for stillbirth, mortality at three weeks, and mortality at six weeks of age. Control-adjusted LC20 and LC50 values were also estimated based on the same maximum likelihood estimates of the beta-binomial regression and summarized in Supplemental Data, Table S2.
An LC50 expressed as dietary ΣPCB and TEQ concentrations for mortality by 31 weeks of age was estimated as the concentration corresponding to 1 minus the joint probability of survival from birth to the six-week necropsy period and from the six-week necropsy period to the end of the juvenile growth trial (31 weeks of age). The product of these survival probabilities is the joint probability of surviving to 31 weeks of age, and the complement (1 minus the joint probability) is the probability of death before 31 weeks of age. This method accounted for the kits necropsied postweaning. The survival models were beta-binomial models previously described for modeling mortality at three and six weeks of age. The LC50 was determined by iteratively solving for ΣPCBs.
All statistical analyses were conducted using R statistical software (http://www.r-project.org/) including the additional R packages MASS for negative binomial models, aod for beta-binomial models, geepack for GEE models, and doBy for linear functions of estimated GEE regression parameters.
Dietary contaminant concentrations
The analyzed concentrations of ∑PCBs in feed were similar to the nominal doses, and PCDDs and PCDFs contributed less than 2% of the TEQs (Table 2). The analyzed concentrations of ∑PCBs (± standard deviation) were 0.0074 (0.0016), 0.72 (0.12), 1.5 (0.21), 2.8 (0.38), 4.5 (0.49), and 6.1 (0.51) µg/g feed. These values will be used subsequently rather than the targeted concentrations. The corresponding concentrations of TEQsWHO 2005 (± standard deviation) were 0.41 (0.022), 4.8 (0.15), 10 (0.074), 18 (0.13), 28 (0.12), and 38 (0.38) pg/g feed, respectively. The percentage of contribution by PCDDs, PCDFs, non-ortho PCBs, and mono-ortho PCBs to the TEQsWHO 2005 in the treatment diets averaged 1.5, 1.4, 75, and 22%, respectively. Thus, 97% of the TEQsWHO 2005 in the diet was from PCBs with 3,3′,4,4′,5-pentachlorobiphenyl (PCB 126 using International Union of Pure and Applied Chemistry nomenclature) contributing 74% of the TEQsWHO 2005. The predominant organochlorine pesticide present in the feed was total toxaphene at a maximum concentration of 0.24 µg/g feed, followed by ∑dichlorodiphenyltrichloroethane (DDT) at a maximum concentration of 0.019 µg/g feed (data not shown). Mercury was present at a maximum concentration of 0.031 µg/g feed. The maximum ∑PBDE concentration was 0.032 µg/g feed, with BDE 47 accounting for 73% of the ∑PBDE concentration (Supplemental Data, Table S1).
Table 2. Mean concentrations of total polychlorinated biphenyls (PCBs), polychlorinated dibenzo-p-dioxins (PCDDs), polychlorinated dibenzofurans (PCDFs), non-ortho PCBs, and mono-ortho PCBs as well as toxic equivalents (TEQs)a in experimental mink diets
Hepatic ∑PCB and TEQWHO 2005 concentrations of PCB-exposed adult female mink were significantly greater (p < 0.025 and p < 0.004, respectively) than control concentrations and increased monotonically with feed ∑PCB concentrations (Table 3). Concentrations in livers of adult males (not shown) were not significantly different from those in females. The percentage of contribution by PCDDs, PCDFs, non-ortho PCBs, and mono-ortho PCBs to the TEQsWHO 2005 in the maternal livers averaged 0.80, 1.4, 82, and 16%, respectively. Thus, 98% of hepatic TEQsWHO 2005 was from PCBs, with 82% being contributed by PCB 126.
Table 3. Mean toxic equivalents (TEQs) contributed by polychlorinated dibenzo-p-dioxins (PCDDs), polychlorinated dibenzofurans (PCDFs), non-ortho polychlorinated biphenyls (PCBs), and mono-ortho PCBs as well as ∑ PCBs in livers of adult female mink fed diets containing contaminated fish from the Hudson River
Feed group based on dietary concentration (µg ∑PCBs/g feed)
Hepatic ∑PCB and TEQWHO 2005 concentrations (based only on non-ortho and mono-ortho PCBs) in livers of six-week-old kits (nine animals) and 31-week-old juveniles (23 animals) were significantly greater (p < 0.001) in the 0.72 µg ∑PCBs/g feed group compared to controls (15 kits and 30 juveniles; Table 4). The percentage of contribution by non-ortho and mono-ortho PCBs to the TEQsWHO 2005 averaged 77 and 23% in the kit livers and 81 and 19% in the juvenile livers, with PCB 126 contributing all of the non-ortho TEQsWHO 2005.
Table 4. Mean concentrations of total polychlorinated biphenyls (PCBs), non-ortho PCBs, and mono-ortho PCBs as well as toxic equivalents (TEQs) in livers of six-week-old kits and 31-week-old juvenile mink fed diets containing contaminated fish from the Hudson River
Feed group based on dietary concentration (µg ∑PCBs/g feed)
Feed consumption of adult female mink did not differ significantly (p = 0.084) across treatment groups during the seven-week period prior to breeding. Average daily feed consumption for adult female mink was 115 g feed/d.
Reproductive performance and offspring survivability
Although all 75 adult females on the trial were successfully mated during the three-week breeding period, the mean number of kits whelped alive per litter was less at higher dietary PCB concentrations compared to lower dietary PCB concentrations (p < 0.004). The percentage of females whelping at least one kit (dead or alive) decreased slightly with increasing dietary PCB concentration, although this decrease was not statistically significant (p = 0.110). The percentage of females whelping at least one kit was 100% for females in the control and 0.72- and 1.5 µg ∑PCBs/g feed groups and 90, 87, and 73% for females in the 2.8-, 4.5-, and 6.1 µg ∑PCBs/g feed groups, respectively. One female each in the 2.8- and 4.5 µg ∑PCBs/g feed groups and three females in the 6.1 µg ∑PCBs/g feed group did not whelp. Of these, one female each in the 2.8- and 6.1 µg ∑PCBs/g feed groups had uterine implantation sites, indicating conception, while the female in the 4.5 µg ∑PCBs/g feed group and two females in the 6.1 µg ∑PCBs/g feed group did not. Average gestation lengths, which ranged from 46.3 to 48.7 d, were similar for all treatment groups (p = 0.74). Compared to controls, the mean number of kits whelped alive per litter was 44% less (CI 17–63%, p = 0.0039) in the 4.5 µg and 49% less (CI 21–67%, p = 0.0023) in the 6.1 µg ∑PCBs/g feed groups (Table 5). The estimated relative rate of live kits per litter for a given increase in micrograms ∑PCBs/g feed (Δ) was e−0.17Δ (p < 0.001). Increases of 1 and 2 µg ∑PCBs/g feed were associated with 15.9 and 29.3% reductions in mean live kits whelped per litter.
Table 5. Mean number of kits whelped alive per litter from mink fed diets containing polychlorinated biphenyl (PCB)–contaminated fish collected from the Hudson Rivera
Kit mortality increased over time, ultimately resulting in no kits surviving to the end of the trial with the exception of those in the control and 0.72 µg ∑PCBs/g feed groups. The odds of kit mortality at six weeks of age were 22 (CI 3.7–129, p < 0.001) times greater in the 2.8 µg, 9.4 (CI 2.1–42, p = 0.0035) times greater in the 4.5 µg, and 40 (CI 4.4–369, p = 0.0011) times greater in the 6.1 µg ∑PCBs/g feed groups than the control group (Table 6). The numbers of kits alive at weaning, taking into consideration animals that were necropsied at six weeks of age, were 51, 30, 38, 9, 15, and 2 in the control and 0.72-, 1.5-, 2.8-, 4.5-, and 6.1 µg ∑PCBs/g feed groups, respectively. Between six and 10 weeks of age (when young mink were housed individually), 4 (7.8%), 6 (20%), 33 (87%), 8 (89%), 15 (100%), and 2 (100%) animals in the control and 0.72-, 1.5-, 2.8-, 4.5-, and 6.1 µg ∑PCBs/g feed groups died, respectively. The seven juvenile deaths that occurred from July 12, 2007, to the termination of the trial were one juvenile in the 0.72 µg ∑PCBs/g feed group, the remaining five juveniles in the 1.5 µg ∑PCBs/g feed group, and the remaining juvenile in the 2.8 µg ∑PCBs/g feed group. Thus, at the end of the trial, the only animals remaining were 47 juveniles in the control group and 23 juveniles in the 0.72 µg ∑PCBs/g feed group. Figure 2 depicts offspring mortality between 6 and 31 weeks of age, and Supplemental Data, Table S3 presents the number of offspring alive at key time points throughout the study.
Table 6. Mortality of kits whelped and nursed by mink fed diets containing polychlorinated biphenyl (PCB)–contaminated fish collected from the Hudson Rivera
At the start of the experiment, adult female body mass did not differ significantly across treatment groups (p = 0.916) but generally decreased in all treatment groups over the first eight weeks of the trial. Body mass was significantly greater in the 1.5 µg (1,166 g, CI 1,119–1,213, p = 0.014), 4.5 µg (1,190 g, CI 1,156–1,224, p < 0.001), and 6.1 µg (1,206 g, CI 1,179–1,232, p < 0.001) ∑PCBs/g feed groups compared to controls (1,103 g, CI 1,072–1,134).
Body mass of individual kits was not significantly different at whelping (p = 0.21), but average kit masses in the 1.5-, 2.8-, and 4.5 µg ∑PCBs/g feed groups (10, 19, and 28 TEQsWHO 2005/g feed, respectively) were, respectively, 27 g (CI 8.3–45.0, p = 0.0044) or 25%, 29 g (CI 3.7–54.6, p = 0.025) or 27%, and 50 g (CI 29.1–71.3, p < 0.001) or 46% less than controls at three weeks of age and 47 g (CI 8.5–86.0, p = 0.017) or 21%, 47 g (CI 2.0–92.5, p = 0.041) or 21%, and 73 g (CI 23.8–123.1, p = 0.0037) or 33% less than controls at six weeks of age. The 6.1 µg ∑PCBs/g feed group was omitted from analysis at three and six weeks of age because the regression models were not estimable with only two live kits remaining in this treatment group (Supplemental Data, Table S4).
The only significant treatment effect on juvenile growth was that males in the 0.72 µg ∑PCBs/g feed group gained 253 g (CI 82–401, p = 0.003) or 34% more than males in the control group. At the end of the trial, mean change in body mass in the control group was 476 g for females and 750 g for males, and in the 0.72 µg ∑PCBs/g feed group mean change in body mass was 488 g for females and 1,003 g for males (data not presented).
Lethal concentration estimates
Estimated dietary and maternal hepatic LC20 and LC50 values based on kit stillbirths and kit mortality at three and six weeks of age were derived in terms of ∑PCBs/g and TEQsWHO 2005/g (Supplemental Data, Table S5). Control adjusted LC20 and LC50 values were also estimated (Supplemental Data, Table S2). Plots illustrating dose–response curves used to derive LC20 and LC50 values for kit mortality at six weeks of age based on dietary ∑PCB and TEQWHO 2005 concentrations and maternal hepatic ∑PCB and TEQWHO 2005 concentrations are presented in Figures 3 and 4, respectively. The estimated LC50 for mortality by 31 weeks of age was 0.78 µg ∑PCBs/g feed.
Dietary contaminant concentrations
The analyzed concentrations of ∑PCBs (0.72–6.1 ∑PCBs/g feed) in the treatment diets and the derived TEQsWHO 2005 (4.8–38 pg TEQsWHO 2005/g feed) were similar to those used in other mink feeding studies with a similar design 12–17. In addition, the congener profile in most cases was similar across studies. Non-ortho PCBs (75%), and PCB 126 specifically (74%), contributed the majority of dietary TEQsWHO 2005 derived from HR fish. A mink feeding study of a similar design was conducted using PCB-contaminated fish collected from the Housatonic River (Massachusetts, USA) 16, 17. Housatonic River diets contained ∑PCB concentrations ranging from 0.34 to 3.7 µg/g feed and TEQWHO 2005 concentrations ranging from 2.5 to 51 pg/g feed (recalculated using TEF values presented in Van den Berg et al. 31), with 87% of the TEQsWHO 2005 being contributed by non-ortho PCBs and 81% by PCB 126 specifically. Two additional mink feeding studies of a comparable design used fish containing PCBs collected from Saginaw Bay (Lake Huron, Michigan, USA) or the mouth of the Saginaw River (Michigan, USA), which empties into Saginaw Bay. In the Saginaw Bay study 12–14, dietary concentrations ranged from 0.72 to 2.6 µg ∑PCBs/g feed (17–66 pg TEQsWHO 2005/g feed, recalculated using TEF values presented in Van den Berg et al. 31), with 64% of the TEQsWHO 2005 contributed by non-ortho PCBs (62% by PCB 126). In the Saginaw River study 15, which was conducted 14 years after the Saginaw Bay study, dietary ∑PCB concentrations ranged from 0.83 to 1.7 µg ∑PCBs/g feed and 22 to 57 pg TEQsWHO 2005/g feed (recalculated using TEF values presented in Van den Berg et al. 31). In this study, non-ortho PCBs and PCDDs each contributed 35% of the TEQsWHO 2005 and PCDFs contributed 25%. The single largest contributor of TEQsWHO 2005 was PCB 126 (33%). It should be pointed out that there are some limitations with the TEQ approach. Toxic equivalency factors are consensus values of the relative potencies of the various PCB/PCDD/PCDF congeners rather than precise values. They may vary depending on species, measurement end points, and the relative proportions of individual congeners in complex mixtures. Some TEFs are based on in vitro studies that do not account for the potential differences between animal species in absorption, distribution, metabolism, and elimination. Other TEFs are based on quantitative structure–activity relationships because of a lack of toxicity data for some congeners, which can introduce a degree of uncertainty into the determination of TEQs. As such, TEFs may under- or overestimate the relative potencies of congeners and, thus, should be considered as protective estimates as opposed to predictors of effect thresholds 20.
The dietary concentrations of ∑PCBs used in the present study bracket dietary concentrations of ∑PCBs that are relevant for wild mink residing along the upper Hudson River. The National Oceanic and Atmospheric Administration (NOAA) database of tissue residue data for specific watershed projects was searched for whole-body analyses of fish sampled in the Hudson River between latitudes 42.87804 and 43.199263, which encompasses the three carp collection sites for the present study and fish sizes of 7 to 20 cm, which is the expected size of typical prey fish consumed by mink 17. The data used were from sampling studies conducted by NOAA in 1993, 1995, and 1999 and by the General Electric Company from 2004 to 2009. Fish species sampled in this region included the largemouth bass (Micropterus salmoides), pumpkinseed sunfish (Lepomis gibbosus), redbreast sunfish (Lepomis auritus), yellow perch (Perca flavescens), golden shiner (Notemigonus crysoleucas), bluntnose minnow (Pimephales notatus), spottail shiner (Notropis hudsonius), common shiner (Luxilus cornutus), spotfin shiner (Notropis spilopterus), and mimic shiner (Notropis volucellus). The average ∑PCB concentration for all fish sampled between these two latitudes was 4.0 µg/g tissue (n = 430, range 0.32–25). Fish collected at sites closest to the carp collection sites in the present study had average ∑PCB concentrations of 4.6 µg/g tissue for the Moses Kill area (n = 55, range 0.73–16), 4.5 µg/g tissue for the Northumberland Pool area (n = 36, range 0.81–15), and 2.4 µg/g tissue for the above Lock 2 site (n = 70, range 0.48–5.5). The average ∑PCB concentration for fish sampled at these three sites on the Hudson River was 3.8 µg/g tissue (Supplemental Data, Table S6).
Daily doses (based on dietary concentrations and an average daily feed consumption of 97 g/kg body mass that was calculated by dividing average feed consumption per animal [115 g] by average body wt of adult females across the six treatment groups [1,186 g]) of the various organochlorine pesticides as well as the potentially toxic metals (arsenic, cadmium, chromium, copper, mercury, nickel, lead, selenium, and zinc) present in the treatment diets were less than the toxicity reference values for mink reported by Hink et al. 35. Similarly, dietary concentrations of PBDEs were less than those reported to cause adverse effects in mink 36.
Hepatic ∑PCB and TEQWHO 2005 concentrations
Hepatic concentrations of ∑PCBs and TEQWHO 2005 concentrations increased with dietary concentration and were similar between adult male and female mink. The percentage of hepatic TEQsWHO 2005 compared to dietary TEQsWHO 2005 contributed by non-ortho PCBs increased slightly (from 75 to 82%), while the contribution by mono-ortho PCBs decreased by 5%. The PCB 126 contribution increased from 74 to 82% when comparing dietary to hepatic TEQsWHO 2005, respectively. Bioaccumulation factors (BAFs) were derived by dividing the hepatic concentration (wet wt) of ∑PCBs and TEQsWHO 2005 by dietary ∑PCB and TEQ concentrations (wet wt), respectively. The BAF for ∑PCBs was 1.2 and the BAF for TEQsWHO 2005 was 5.9. These BAFs are similar to those based on dietary and hepatic ∑PCB and TEQWHO 2005 concentrations reported for the Housatonic River study (1.2 and 4.0, respectively) 16, 17. The BAFs for individual PCB congeners that contributed the majority of TEQsWHO 2005 resembled the BAF for ∑TEQsWHO 2005 (PCB 126 = 6.5, PCB 105 [2,3,3′,4,4′-pentachlorobiphenyl] = 4.1, PCB 118 [2,3′4,4′,5-pentachlorobiphenyl] = 4.3, and PCB 156 [2,3,3′,4,4′,5-hexachlorobiphenyl] = 7.7). Hepatic concentrations of ∑PCBs and TEQsWHO 2005 in six-week-old kits and 31-week-old juveniles were similar to adult concentrations, as was reported for mink exposed to PCBs derived from fish collected from the Housatonic River 17.
Adult feed consumption
Adult feed consumption did not differ significantly over the first seven weeks of the trial, although a flavor enhancer was added to the feed beginning in week 6 following the observation during week 5 that feed consumption decreased by approximately 50% in those groups fed diets containing primarily herring (control and 0.72- and 1.5 µg ∑PCBs/g feed groups). The inclusion of a flavor enhancer in all of the treatment diets resulted in an immediate increase in feed intake such that feed consumption was again comparable across all groups by week 6. The average daily feed consumption of 115 g reported here is identical to the value reported for adult female ranch mink 37 and similar to the 127 g/d rate reported for mink fed diets containing fish collected from the Housatonic River 16.
Reproductive performance and offspring mortality
The effects of consumption of diets containing PCB-contaminated fish collected from the Hudson River on reproductive performance were similar to those reported for mink fed diets containing PCB/PCDD/PCDF-contaminated fish collected from Saginaw Bay 12, 14. There was no significant effect on the percentage of females whelping at least one kit, but there was significant reduction in the mean number of kits whelped alive per litter in both studies. The no observed adverse effect levels (diet-based)/concentrations (tissue-based; NOAELs/NOAECs) and lowest observed adverse effect levels/concentrations (LOAELs/LOAECs) are presented in Table 7.
Table 7. Comparison of threshold concentrations of ∑polychlorinated biphenyls (PCBs) and toxic equivalents (TEQs) relating to offspring mortality and growth
∑PCBs (µg/g feed)
TEQs (pg/g feed)
∑PCBs (µg/g liver)
TEQs (pg/g liver)
Concentration given is for 3,3',4,4',5-pentachlorobiphenyl (PCB 126).
No observed adverse effect level/concentration (NOAEL/NOAEC)
An increase in kit mortality is the effect most often reported in mink reproduction studies using PCB/PCDD/PCDF-contaminated fish. In addition to the present study, kit mortality was significantly increased in the Saginaw Bay study 12, 14 and the Housatonic River study 16. In the latter study, the contribution by non-ortho PCBs and PCB 126 to ∑TEQs was similar to the present study. The NOAEL/NOAEC and LOAEL/LOAEC for each study are presented in Table 7. It is interesting to note that an equivalent dietary concentration of TEQsWHO 2005 provided only by PCB 126 had no effect on kit mortality through six weeks of age 18. The differences in NOAEL(C)s ranged from 3.6-fold (hepatic TEQs) to 106-fold (dietary ∑PCBs). The differences in LOAEL(C)s for increased kit mortality were considerably less, ranging from 1.5-fold (hepatic ∑PCBs) to fivefold (dietary ∑PCBs).
A treatment-related increase in juvenile mortality between weaning at six weeks of age and the end of the growth period at approximately 30 weeks of age was reported only in the present study. Between 6 and 10 weeks of age, siblings were still group-housed and the larger, stronger animals were more effective at competing for feed compared to their smaller, weaker littermates. Smaller animals that continue to grow poorly may die, even after being housed singly at 10 to 12 weeks of age. Because dietary concentrations of 1.5 µg ∑PCBs/g feed and greater resulted in decreased kit body mass, it was not unexpected that animals in these treatment groups experienced excessive mortality after weaning. In the Saginaw Bay study 12, 14, kits were not maintained beyond weaning. In the Housatonic River study 16, offspring maintained through 27 weeks had similar growth rates.
The significantly greater body masses of adult females at the higher dietary concentrations of PCBs compared to controls could be a reflection of the initial decrease in feed intake of control animals fed diets containing 20% ocean herring. While feed consumption was equivalent across groups after beef flavoring was added to all treatment diets, it is apparent that control females continued to have a lower body mass.
Body masses of treated kits were comparable at whelping but significantly less compared to controls at six weeks of age in both the Saginaw Bay study 12, 14 and the present study. The NOAELs/NOAECs and LOAELs/LOAECs for the respective studies are presented in Table 7. In contrast, body mass of kits exposed to an equivalent or greater concentration of TEQsWHO 2005 provided only by PCB 126 was not adversely affected 18.
It is not clear why male juveniles in the 0.72 µg ∑PCBs/g feed group gained 253 g more mass compared to controls by the end of the trial. In the Housatonic River study 16, exposure to PCBs had no effect on juvenile growth through 27 weeks of age.
Lethal concentration estimates
Because there are acknowledged limitations associated with use of NOAELs/NOAECs and LOAELs/LOAECs 37, 38 in risk assessment, LC20 and LC50 values were derived for kit mortality in the present study. Regression estimates of lethal concentrations (LC20 and LC50) are less sensitive to the idiosyncrasies of experimental design, and estimates are not limited to the concentrations under study. The only mink feeding study of a similar design as the present study that reported lethal dietary concentrations was the Housatonic River study 16. There is a 2.9-fold difference between the Hudson River and Housatonic River LC20 values based on ∑PCBs with overlapping confidence intervals. The LC20 values based on TEQsWHO 2005 differ by 5.5-fold with no overlap in confidence intervals (Table 7). Assuming an average daily feed intake of 97 g/kg body mass (based on average feed consumption and body weight of adult females across the six treatment groups), the dose lethal to 20% of the population (LD20) was 33 µg ∑PCBs/kg body mass daily or 281 pg TEQsWHO 2005/kg body mass daily. The corresponding LD20 values for the Housatonic River were 105 µg ∑PCBs/kg body mass daily or 1,730 pg TEQsWHO 2005/kg body mass daily 16.
Comparing results among studies in which mink are fed site-specific fish is complicated by potential differences in the presence or absence of toxic contaminants, the ratios among them, and the presence of other dietary constituents that could alter the apparent overall toxicity of the measured contaminants. As mentioned previously, dietary concentrations of PBDEs, organochlorine pesticides, and potentially toxic metals were less than those reported to cause adverse effects in mink 35, 36. Difference in dietary, and thus hepatic profiles of TCDD-like congeners between the sites could also influence toxicity. While the TEF system theoretically normalizes the contribution of individual TCDD-like chemicals to overall TCDD toxicity, the system may not adequately address agonistic and antagonistic interactions between components of a complex mixture 14, 20. The addition of thiamine to the feed to counteract the effects of thiaminase present in the fish could potentially ameliorate the effects induced by PCBs in that there is evidence that dietary exposure to PCBs and other organochlorine chemicals interferes with thiamine metabolism, resulting in reduced thiamine concentrations in tissues 40, 41.
The results of the present study indicated that mink reproductive performance as well as offspring survivability and growth were adversely affected by consumption of feed containing PCB-contaminated fish collected from the Hudson River from two months prior to breeding through the growth period. All offspring exposed to dietary concentrations of 1.5 µg ∑PCBs/g feed and above died by 31 weeks of age. A dietary concentration of 0.34 µg ∑PCBs/g feed (2.6 pg TEQsWHO 2005/g feed) was predicted to result in 20% kit mortality by six weeks of age. Assuming that mink residing along the upper Hudson River consume fish containing an average of 4.0 µg ∑PCBs/g and that fish is the only component of the mink's diet that contains PCBs or other aryl hydrocarbon receptor–active compounds, a diet comprised of less than 10% fish with these environmentally relevant PCB concentrations could be expected to result in kit mortality.
Tables S1–S6. (48 KB XLS).
This work was supported by the Hudson River Natural Resource Trustees as part of the ongoing Hudson River Natural Resource Damage Assessment. The conclusions and opinions presented here are those of the authors and do not represent the official position of any of the funding agencies, the Hudson River Trustees, or the United States of America.