Research
Comparison of Risk Assessment
Methodologies for Exposure of Mink
to PCBs on the Kalamazoo River,
Michigan
S T E P H A N I E D . M I L L S A P , * ,†,§,⊥
A L A N L . B L A N K E N S H I P , §,|,⊥
PATRICK W. BRADLEY,§
P A U L D . J O N E S , ‡,§,⊥ D E N I S E K A Y , ‡
ARIANNE NEIGH,§ CYRUS PARK,§
K A R L D . S T R A U S E , §,|
M A T T H E W J . Z W I E R N I K , §,⊥ A N D
JOHN P. GIESY§
Department of Fisheries and Wildlife, Zoology Department,
National Food Safety and Toxicology Center, and Center for
Integrative Toxicology, Michigan State University, East
Lansing, Michigan 48824, and ENTRIX, Inc. 4295 Okemos
Road, Okemos, Michigan 48864
Risk assessments are generally based on exposures
predicted by use of simple models of accumulation from
abiotic compartments or the diet. The use of tissue-specific
measurements of residue concentrations in wildlife
tissues is more accurate and subject to less uncertainty,
but these data are often not available. This report compares
the results of two different site-specific approaches for
assessing the risk of PCBs to mink residing along the
Kalamazoo River, MI. The first approach was based on
hepatic concentrations of PCBs and 2,3,7,8-tetrachlorodibenzo-p-dioxin equivalents (TEQs) of mink. The second
approach was based on measured concentrations of both
PCBs and TEQs in the diets of mink. For each of these
methodologies, assessments were based on no observed
adverse effect levels (NOAELs) or concentrations
(NOAECs) and lowest observed adverse effect levels
(LOAELs) or concentrations (LOAECs). Samples of mink
(Mustela vison) and its diet were collected from within the
Kalamazoo River Area of Concern (KRAOC) and an
upstream reference area in the Fort Custer Recreation
Area (FC). Hazard quotient (HQ) values were calculated
based on congener-specific concentrations of PCBs or TEQs,
several toxicity reference values (TRVs), and several
assumed dietary compositions. Mean total hepatic
concentrations of PCBs were 2.7 and 2.3 mg PCBs/kg,
ww, in mink from the KRAOC and FC, respectively. HQs
based on the LOAEC and mean hepatic PCB concentrations
ranged from 0.37 to 0.87 at KRAOC and 0.31-0.73 at FC.
HQs based on PCBs in the diet ranged from 0.20 to 1.8 at
* Corresponding author phone: (734)692-7628; fax: (734)692-7603;
e-mail: stephanie_millsap@fws.gov. Present address: U.S. Fish and
Wildlife Service, 9311 Groh Rd., Grosse Ile, MI 48138.
† Department of Fisheries and Wildlife, Michigan State University.
‡ Zoology Department, Michigan State University.
§ National Food Safety and Toxicology Center, Michigan State
University.
| Center for Integrative Toxicology, Michigan State University.
⊥ ENTRIX, Inc.
10.1021/es049600e CCC: $27.50
Published on Web 10/27/2004
 2004 American Chemical Society
KRAOC and from 0.04 to 0.35 at FC. Dietary HQs were less
than 10-fold different than tissue-based HQs.
Introduction
There are two primary approaches for assessing risk to wildlife
for persistent, bioaccumulative, and chronically toxic compounds such as polychlorinated biphenyls (PCBs) (1). The
method used most often is the dietary-based approach, which
predicts potential concentrations of the residue of concern
by use of simple bioaccumulation models and concentrations
of residues measured at lower levels of the food chain or in
water, soil, or sediment (2). Assumptions regarding dietary
composition are used to predict an average daily dose
consumed by the species of interest. This estimated dose is
then compared to dietary toxicity reference values (TRVs)
(2). This technique, which is used when information on the
tissue concentrations in receptors of concern cannot be
collected, is limited by the accuracy of the assumptions used
to predict accumulation. The uncertainty can be particularly
great for complex mixtures such as PCBs, as individual PCB
congeners exhibit differential accumulation, potential for
metabolism, and toxicity. A major factor in applying dietary
exposure models is the dietary composition used. The dietary
composition is especially difficult to characterize for opportunistic predators such as the mink (Mustella vison), since
the diet can vary seasonally and among habitats. Although
mink are a semiaquatic predator, they can occupy a wide
range of habitat types including estuaries, rivers, streams,
and wetlands (3).
The alternative and presumably more accurate method,
the tissue-based approach measures the actual concentrations of the residues of interest in the tissues of the receptors
of concern (1). The site-specific, tissue-specific residue
concentrations are then compared to tissue-based TRVs
derived from studies in which concentrations of residues
were measured synoptically with responses. The tissue-based
approach can relate the measured tissue concentrations to
diet or sediment concentrations in order to determine
appropriate remedial options. However, it might not always
be possible to collect tissues of the receptors of concern,
especially for animals, such as mink, that are difficult to
sample because they are solitary, nocturnal, and have large
territories (3). In other cases, the sensitive species to be
protected may not be present at the location (4). While a
lines of evidence approach, which considers both approaches,
has been advocated by risk assessors, both methods are
seldom applied at the same location. For this reason, the two
methods were compared to determine how similar the
predictions would be.
In this study, the ranges of possible risk estimates were
expressed as hazard quotients (HQs) based either on dietarybased (predicted) or tissue-based (measured) estimates of
exposure. Estimated dietary concentrations and measured
hepatic concentrations were compared to TRVs based on no
observable adverse effect levels (NOAELs) for dietary comparisons and no observable adverse effect concentrations
(NOAECs) for tissue-based comparisons. The HQ values
based on the lowest observable adverse effect level (LOAEL)
and lowest observable adverse effect concentration (LOAEC)
were compared to NOAEL(C)-based HQs. HQ values predicted by use of both total PCBs and TEQs and a range of
possible HQ values estimated by use of several different TRVs
available in the literature were compared. Finally, potential
VOL. 38, NO. 24, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6451
aryl hydrocarbon receptor (AhR) (17, 18). A congener-specific
approach allows for the calculation of 2,3,7,8-TCDD equivalents (TEQs) by integrating the total concentrations and
relative potencies of individual congeners identified in a
mixture (19, 20). The use of TEQs has been advocated, but
congener-specific information is not always available so it
is not always possible to use this method. Therefore, the
risks based on concentrations of PCBs and TEQs from the
same data set were compared.
Studies of mink diets, including those conducted on
Michigan rivers and streams, indicate that mink are opportunistic feeders and eat a variety of food items (aquatic
and terrestrial animals) (21-24). Additionally, their food
habits vary by habitat type and season. Therefore, using only
one standard dietary composition may not be appropriate
for every site. The results of this field study were used to test
the accuracy of the assumptions commonly made in ecological risk assessments based on minimal information and
allows for the calibration of the more commonly used dietarybased, predictive approach to the tissue-based method based
on actual tissue concentrations of PCBs.
Methodology
FIGURE 1. Map of the Kalamazoo River Area of Concern (KRAOC).
The inset map of Michigan indicates the locations of the two counties
in which the study was conducted. The KRAOC extends from Morrow
Lake Dam to Lake Michigan. The mink collection area (MCA) in
KRAOC is indicated by a box surrounding the river. The former
Trowbridge (TB) impoundment is also located within the KRAOC.
Fort Custer (FC) is the upstream reference area and is also indicated
by a box surrounding the river.
ranges of HQ values calculated from several assumptions
about diet composition were compared. The range of risk
values can be used to describe both the range and magnitude
of possible risks, giving assessors an estimate of the uncertainty involved in these types of risk assessments.
The mink was chosen as a receptor of concern because
it is sensitive to PCBs, especially in terms of kit weight, litter
size, and kit mortality (5-11). In addition, as a piscivorous
top predator with a high rate of metabolism, it is likely to be
the receptor most likely to receive the greatest exposure.
This semiaquatic member of the family Mustelidae is
commonly found along waterways throughout North America,
including Michigan. For these reasons, mink have been
selected as an ecological receptor for risk assessments at
other sites, such as the Fox River (WI) (12), Clinch River (TN)
(13), and Housatonic River (MA) (14), which are large riverine
systems with PCBs as the contaminant of concern (COC). In
addition, experimental data from mink PCB feeding studies
have been used to assess risk to other mustelid species (15).
The study was conducted on the main stem of the
Kalamazoo River, which is located in southwestern Michigan
(Figure 1). The Kalamazoo River is approximately 195 km
long and flows northwest, entering Lake Michigan near
Saugatuck, MI. In 1990, approximately 125 km of the
Kalamazoo River was designated a Superfund site, referred
to as the Kalamazoo River Area of Concern (KRAOC). The
site extends from Morrow Dam in Kalamazoo County to Lake
Michigan (Figure 1). The release of PCBs, the COC, resulted
primarily from PCB-contaminated waste discharged from
the recycling and processing of carbonless copy paper (16).
Exposures and risks measured in the KRAOC were compared
to those from an upstream, less PCB-contaminated site in
the Fort Custer Recreational Area (FC).
A congener-specific approach, including analysis of nonortho-substituted congeners, was utilized in this study since
the most sensitive biological effects of PCBs have been
attributed to planar congeners that resemble 2,3,7,8-tetrachlorodibenzo-p-dioxin (2,3,7,8-TCDD) and act through the
6452
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 24, 2004
Sample Collection. Samples of wild mink and their prey
were collected throughout the KRAOC and FC during the
winters of 2000 through 2002. Mink and muskrats (Ondatra
zibethicus) were trapped in the same areas. Terrestrial small
mammals were sampled in June/July 2000 and again in
August/September 2000 at four locations within the Trowbridge (TB) area of KRAOC and three locations at FC. The TB
area (Figure 1) is the largest of the four former impoundments
(approximately 132 hectares of former sediments and approximately 70 hectares of existing impounded water), has
the greatest estimated mass of PCBs, and has the greatest
mean surficial concentration of PCBs in soils (∼11 mg/kg,
dw) (16). Sampling occurred in a 35 × 35 m area in which
alternating Sherman traps (Forestry Suppliers, Jackson, MS)
and pitfall traps (#10 canning tins) were placed approximately
5 m apart. Concentrations of PCBs were measured in the
following dietary items: eastern chipmunk (Tamias striatus),
red squirrel (Tamiasciurus husdonicus), meadow vole (Microtus pennsylvanicus), deer mouse (Peromyscus maniculatus), white-footed mouse (Peromyscus leucopus), and
meadow jumping mouse (Zapas hudsonius). Fish were
collected by electro-fishing (Smith Root, Inc., Vancouver,
WA) in early winter 2000 (11/9/99-12/3/00) at TB and FC.
The following fish species were analyzed for PCB concentrations to use in the dietary assessment: carp (Cyprinus carpio)
ranging in length of 46-56 cm, forage fish (collected based
only on size, predominately Cyprinids, immature Catastomids
and Centrarchids) all less than 23 cm long, smallmouth bass
(Micropterus dolomieu) ranging in length from 31 to 51 cm,
and suckers (Catostomus commersoni, Hypentelium nigricans,
and Moxostoma erythrurum) ranging in length from 31 to 51
cm. Crayfish (Cambarus and Oronectes species) were collected
by use of wire minnow traps between June and September,
2000 from areas adjacent to where small mammals were
collected within TB and FC.
Weights, lengths, visual health, sex, and species of
potential prey items were recorded. For mink, the GI tract
was removed and contents preserved for prey item identification. The baculum (os-penis) of male mink was removed,
cleaned, and weighed. Liver tissue was removed for PCB
analysis and histology, and a tooth was removed for dental
cementum analysis to determine age. The jaws were also
removed for histological analysis. Gross necropsies and liver
histology were performed by board-certified pathologists at
the Animal Health Diagnostics Laboratory (Michigan State
University, East Lansing, MI), dental cementum analysis was
performed by Matson’s Laboratory (Milltown, MT), and jaw
histology was performed by Kerrie Beckett (Department of
Animal Science, Michigan State University). Gut contents
were removed from small mammals, and weights, lengths,
visual health, and sex were determined. These and all
remaining samples were wrapped whole in aluminum foil
and then frozen until homogenization for chemical analysis.
Samples were thawed and weighed, and the whole body was
homogenized in a grinder (model 4732 SS; Hobart Corp.,
Troy, OH) which resulted in a coarse grind. The fur was
removed from muskrats but not other mammals. This
coarsely ground material was then further homogenized in
a Waring Commercial Blender (model 36BL29, Waring
Commercial Inc., New Hartford, CT), placed in cleaned glass
jars (I-Chem, New Castle, DE), and stored at -20 °C until
PCB analysis.
Chemical Analysis. The extraction method used was based
on EPA method 3540 (Soxhlet extraction). Briefly, a known
quantity (approximately 5 g) of tissue was homogenized,
extracted, and passed through a neutral/acidic silica gel
column, and extract volume was reduced to 5 mL. All PCB
congeners are numbered using IUPAC nomenclature. Surrogate standards, PCB 204 and PCB 30 (AccuStandard, New
Haven CT), were added to all samples, blanks, and matrix
spikes before extraction. The extract volume was reduced to
1.0 mL under a stream of nitrogen. An aliquot of 0.5 mL was
transferred to a gas chromatograph (GC) vial for total PCB
analysis, while the remaining 0.5 mL was subjected to carbon
column chromatography for the separation of non-orthosubstituted (coplanar) PCB congeners.
PCB congeners were quantified by use of a GC (PerkinElmer AutoSystem) equipped with a 63Ni electron capture
detector (GC-ECD) according to previously published methods (25). A solution containing 100 individual PCB congeners
was used as a standard. PCB congeners were identified by
relative retention time and quantified by comparing peak
areas to those of the appropriate peaks in the standard
mixture. TurboChrom software (Perkin-Elmer, Boston, MA)
was used to integrate the peaks. Concentrations of all resolved
PCB congeners were summed to obtain total PCB concentrations.
Non-ortho-substituted PCB congeners 77, 81, 126, and
169 were separated from coeluting congeners and interferences by carbon column chromatography (26). A mixture of
radiolabeled 13C coplanar PCB congeners (77, 81, 126, and
169) (Cambridge Isotope Laboratories, Andover MA) was
added to each extract as internal quantification standards.
The volume of the final extract containing non-ortho coplanar
PCB congeners was then reduced to 20 µL under a stream
of nitrogen.
The coplanar PCB fraction was analyzed by GC-mass
spectrophotometer (Hewlett-Packard 5890 series II gas
chromatograph equipped with a HP 5972 series mass selective
detector) (Agilent Technologies, Foster City, CA). Non-orthosubstituted PCB congeners were identified by use of selected
ion monitoring of the two ions of the molecular cluster.
Congener concentrations were calculated based on ion ratios
for the native and 13C internal standard congener.
TEQ Computation. Concentrations of TEQs in mink livers
were calculated by multiplying the concentration of individual PCB congeners by its respective World Health
Organization toxic equivalency factors (TEF) (27). Total TEQ
concentrations were determined by summing the concentrations of TEQs of PCB congeners 77, 81, 105, 118, 126, 156,
157, 167, and 169 (IUPAC nomenclature). When a congener
was not detected, a surrogate value of one-half of the limit
of quantification was multiplied by the TEF to calculate the
congener-specific TEQs. PCB congeners 156 and 157 frequently coeluted with PCB congeners 171 and 200, respectively. In those instances, it was assumed that the entire
concentration was due to the non-ortho-substituted (co-
TABLE 1. Toxicity Reference Values (TRVs) Used To Calculate
Hazard Quotients for Total PCBs and Total TEQs
tissue-based TRVs
(mg PCBs/kg)
or (ng TEQs/kg)
total PCBs
total TEQs
dietary-based TRVs
(mg PCBs/kg bw/d)
or (ng TEQs/kg bw/d)
NOAEC
LOAEC
NOAEL
LOAEL
reference
6.0
3.1
56
71
7.3
3.1
220
270
0.12
0.17
1.7
1.1
0.35
0.23
0.41
7.7
4.5
2.4
(29)
(28)
(28)
( 8)
( 7)
planar) congeners 156 and 157, providing the maximum
estimate of TEQ concentration. By making this assumption
PCB congener 157 contributed < 5% of the total concentration
of TEQs, whereas PCB congener 156 contributed between <
0.01% and 50% of the total TEQ concentrations with a mean
TEQ contribution of 10%. The relatively greater contribution
of PCB congener 156 to the total concentration of TEQs in
some samples occurred when absolute concentrations of
total PCBs and total TEQs were both small.
Toxicity Reference Values. TRVs were derived from
multiple feeding studies to capture the range, variability,
and associated uncertainties. TRVs were selected based on
the following criteria: (1) close relatedness of the test species
to the wildlife receptor of concern (in this case all studies are
based on mink); (2) chronic duration of exposure which
included sensitive life stages to evaluate potential developmental and reproductive effects; (3) measurement of ecologically relevant endpoints; and (4) minimal impact of cocontaminants. This last criterion is especially relevant to
selection of appropriate TRVs. Presence of co-contaminants
confounds one’s ability to assign causality to one particular
contaminant. Similarly, in the case of mink feeding studies,
many have been conducted with technical Aroclors, while
others have been conducted with environmentally weathered
mixtures. Selection of an Aroclor-based study may not reflect
the actual toxicity of an environmentally weathered mixture
and may thus contribute to uncertainty in the results. On the
other hand, if TRVs are derived from an environmentally
weathered PCB mixture, careful evaluation of the potential
effects of co-contaminants is necessary.
Dietary-Based PCB TRVs. The PCB-based dietary TRVs
were derived from the results of two studies in which mink
were fed PCBs in the diet under laboratory conditions (28,
29). First, Halbrook et al. (29) conducted a well-designed,
chronic laboratory study, in which mink were fed fish
containing environmentally weathered PCBs, at concentrations that are relevant to the KRAOC. From this study, TRVs
based on the NOAEL and LOAEL were determined to be 1.0
and 1.9 mg/kg, ww (diet). The measurement endpoint used
was a statistically significant (17%) decrease in male kit body
weight at 6 weeks of age. When normalized to ingestion rate
and body weight, the TRV values based on the NOAEL and
LOAEL were 0.12 and 0.23 mg/kg BW/d, respectively (Table
1). The TRVs derived from the study conducted by Halbrook
et al. (29) are strengthened by the fact that the dose intervals
were very close together, there were three doses at which
there were no observable adverse effects, and the effects of
co-contaminants were minimal.
A well-designed, chronic laboratory study in which mink
were fed fish with environmentally weathered PCBs, at
concentrations that are relevant to the KRAOC, was conducted by Bursian et al. (28). From this study, TRVs based
on the NOAEL and LOAEL were determined to be 3.1 and 3.7
mg/kg, ww (diet), respectively. The measurement endpoints
on which these TRVs were based were sensitive and ecologically relevant reproductive and developmental endpoints,
VOL. 38, NO. 24, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6453
including reduced kit body weight after 3 wk and decreased
kit survival after 6 wk. When normalized to ingestion rate
and body weight, the NOAEL and LOAEL TRVs were 0.17 and
0.41 mg/kg bw/d, respectively (Table 1). The study, by Bursian
et al. (28), is strengthened by the fact that the dose intervals
were very close together, there were four doses at which
there were no observable adverse effects, and the effect of
cocontaminants were minimal.
Tissue Residue-Based PCB TRVs. TRVs based on tissue
residues of PCBs were calculated based on two studies (28,
29). A hepatic-based mink LOAEC of 7.3 mg PCBs/kg, ww
was reported by Halbrook et al. (29). However, a hepaticbased NOAEC was not reported by the authors. The hepaticbased NOAEC was estimated from the NOAEC-based concentration in adipose tissue and the relationship between
the PCB concentrations in adipose and liver at the LOAEC.
The NOAEC was estimated to be 6.0 mg PCBs/kg, ww. In the
study by Bursian et al. (28), the LOAEC and NOAEC were
both reported to be 3.1 mg PCBs/kg, ww.
Dietary-Based TEQ TRVs. TRVs based on TEQs were
calculated from the results of three laboratory-based mink
feeding studies (7, 8, 28). In the study by Bursian et al. (28),
the NOAEL and LOAEL were determined to be 16.1 and 68.5
ng/kg, ww (diet), respectively. When normalized to ingestion
rate and body weight, the TRVs based on the NOAEL and
LOAEL were 1.7 and 7.7 ng TEQs/kg bw/d, respectively (Table
1). In a second study, Heaton et al. (8) reported TRVs based
on decreased kit survivability and decreased weights of kits
at 3 and 6 wk after whelping. There is some uncertainty
concerning this study since co-contaminants (dioxins and
furans) are known to be present. When the TRVs based on
the NOAEL and LOAEL were recalculated based on the more
recent WHO TEFs, they were determined to be 1.1 and 4.5
ng TEQs/kg bw/d, respectively (Table 1). The TRVs from
Brunstrom et al. (7) were calculated using a study-specific
normalized ingestion rate (NIR) to yield LOAEL and NOAEL
TRVs of 2.4 and 0.35 ng TEQs/kg bw/d, respectively. However,
the TRV based on the NOAEL corresponds to a 7-fold lesser
concentration of TEQs than did the LOAEL and also has the
greatest difference in dose range of the studies used.
Tissue Residue-Based TEQ TRVs. The studies by Heaton
et al. (8) and Bursian et al. (28) were utilized for determining
tissue residue-based TRVs for TEQs. TRVs based on the
NOAEC and LOAEC hepatic concentrations were determined
to be 56 and 220 ng TEQs/kg, ww respectively (28). In a study
by Heaton et al. (8), the TRVs based on concentrations of
TEQs in the liver associated with the NOAEC and LOAEC
were reported to be 71 and 270 ng TEQs/kg, ww respectively.
Site-Specific Dietary Analysis. Stomachs and large intestines of mink were dissected from mink carcasses and
stored frozen until analysis. The stomach was cut open and
the contents were removed by squeezing through a stacking
sieve (mesh numbers 5-230; Hubbard Scientific, Fort Collins,
CO). After rinsing, the contents were dried at 90 °C for 24 h
and identifiable contents separated and classified as bone,
feathers, exoskeleton, hair, teeth, or scales (30). Dietary
composition was expressed as percent frequency of mammal,
fish, bird, and crayfish (31). In instances in which no
identifiable components were found, it was recorded as
unknown, and these individuals were not used to calculate
the dietary composition for use with estimating dietary doses.
Average Potential Daily Dose Calculations. Average
potential daily dose (ADDpot) was used to estimate the total
dose of PCBs and TEQs present in food ingested (3) (eq 1)
ADDpot ) Σ(Ck × FRk × NIRk)
(1)
where Ck ) contaminant concentration in each prey item,
FRk ) fraction of time spent on site, and NIRk ) normalized
ingestion rate ) mean daily ingestion rate/mean body weight.
6454
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 24, 2004
Dietary exposure was estimated using three different
assumptions of dietary composition. Results from the sitespecific dietary analysis represented what mink from the
Kalamazoo River ate immediately prior to their capture. Due
to the opportunistic nature of mink, two other dietary models
were also chosen to estimate ADDpot. One assumed different
proportions of dietary items based on literature values (32).
The other assumed an equally opportunistic consumption
of specified prey items. The proportion of prey in the
literature-based diet represented only one study because it
was a year-round study on a Michigan river and because it
provided a maximum estimate of both fish and PCBs in diet
compared to other studies (32). This literature-based diet
consisted of 85% fish, 9% crayfish, and 6% small mammals.
The equally opportunistic model gave equal weight (33.3%)
to all three prey components. That model was chosen because
it represents a median exposure of mink diet that can vary
greatly, depending on prey availability. However, regardless
of the model chosen, the portion of the diet that consisted
of mammals was assumed to consist of 50% small mammals
and 50% muskrat, while the fish portion of the diet was
assumed to consist of 25% carp, 25% forage fish, 25%
smallmouth bass, and 25% sucker. An additional assumption
in all models was that the site use factor (FRk) was 100%.
Therefore, it was assumed that mink captured all prey items
from within the contaminated area.
Concentrations of total PCBs and TEQs in site-specific
prey items were then used to calculate the daily dose of PCBs
and TEQs to mink from KRAOC and FC. Potential risks were
estimated by use of an HQ approach (eq 2), in which the
ADDpot for each dietary model was compared to appropriate
TRVs for PCBs and TEQs from laboratory-based mink feeding
studies (7, 8, 28, 29).
HQ )
ADDpot (mg PCBs/kg/d or ng TEQs/kg/g)
(2)
dietary toxicity reference value
Statistical Analysis. Normality was determined using the
Kolmogorov-Smirnov one sample test using the Normal
distribution with a mean of zero (0) and standard deviation
of one (1). The test statistic was Lillifors Probability. If the
value of Lillifors test was g 0.05, the data were considered
to be normally distributed. Data that were normally distributed were compared using a t-test. If data did not exhibit
a normal distribution, then a nonparametric version of the
t-test was used. Therefore, if only two means were being
compared, a Mann-Whitney U test was performed, and if
there were more than 2 means being compared a KruskalWallis test would have been used. Correlations were performed using a Pearson Product Correlation.
Results
Mink. Nine male and one female mink were collected within
the KRAOC. Three mink (two male and one female) were
collected from FC, approximately 25 river kilometers upstream of the PCB-contaminated area.
Male mink from the KRAOC had a mean (95% CI) weight
of 1048 ( 150 g and a mean length of 40 cm. Male mink from
FC had a mean weight of 990 ( 65 g and a mean length of
39 cm, respectively (Table 2). The female mink from FC
weighed 390 g and was 34 cm long. All of the mink collected
appeared to be in good condition and did not exhibit external
or internal gross abnormalities. Baculum mass was not
correlated to PCB concentrations (r2 ) 0.09, p ) 0.36), even
after correction for mink body weight. Based on a histological
examination, the livers of the mink were normal. Mink from
KRAOC ranged in age from 1 to 3 y (average 1.6 y), and mink
from FC ranged in age from juvenile (<1 y) to 2 y (average
1.2 y) (Table 2). There were five individuals that exhibited
lesions in the squamous epithelium tissue in the jaw bones
TABLE 2. Hepatic Total PCB Concentrations, Hepatic Total TEQ Concentrations, and Biological Data for Mink Collected from
KRAOCa and FCb
mink ID
gender
total PCBs
(mg/kg ww)
total TEQs
(pg/g ww)
lipid
(%)
age
(years)
baculum
mass (g)
body
weight (g)
jaw
lesion
KRAOC1
KRAOC2
KRAOC3
KRAOC4
KRAOC5
KRAOC6
KRAOC7
KRAOC8
KRAOC9
FC1
FC2
FC3_F
male
male
male
male
male
male
male
male
male
male
male
female
3.4
5.0
1.0
6.0
0.05
1.7
2.9
3.3
1.1
1.6
3.7
1.6
250
580
32
1300
31
25
330
210
21
52
200
76
4.5
4.2
12
7.7
5.2
4.0
3.1
3.6
3.0
4.0
4.2
1.9
2
1
1
3
1
1
1
2
2
0.5
1
2
0.35
0.28
0.39
0.33
0.37
0.23
0.33
0.47
0.44
0.23
0.22
N/A
1100
970
1000
1100
1000
790
990
1200
1300
1000
940
390
absent
present
absent
present
absent
absent
present
present
absent
present
absent
absent
a
Kalamazoo River Area of Concern.
b
Fort Custer Recreation Area (Reference area).
FIGURE 2. Mean total PCBs (mg/kg, ww with 95% confidence
intervals) and total TEQs (ng/kg, ww) in mink liver tissue from
Kalamazoo River Area of Concern (KRAOC) and Fort Custer (FC), an
upstream reference area. Sample sizes and p-values (Student t-test)
comparing mean PCB and TEQ concentrations between sites are
also presented.
(33), and three of these individuals also contained the greatest
liver PCB concentrations (Table 2).
Total PCB and TEQ Concentrations. Total concentrations
of PCBs in livers of mink from the KRAOC ranged from 0.05
to 6.0 mg PCBs/kg, ww (mean 2.7 ( 1.3 mg PCBs/kg, ww).
Total PCB concentrations in livers of mink from FC ranged
from 1.6 to 3.7 mg PCBs/kg, ww (mean 2.3 ( 1.4 mg PCBs/kg,
ww) (Figure 2). There were no significant differences in PCB
concentrations between the sites (Student’s t-test unequal
variance, p ) 0.33).
Concentrations of total TEQs in mink liver from KRAOC
ranged from 21 to 1300 ng TEQs/kg, ww (mean 300 ( 270
ng TEQs/kg, ww) (Figure 2), while total TEQs in mink liver
from FC ranged from 52 to 200 ng TEQs/kg, ww (mean 110
( 87 ng TEQs/kg, ww). There were no statistically significant
differences in TEQ concentrations between sites (Student’s
t-test unequal variance, p ) 0.10). Total concentrations of
PCBs were significantly correlated with total concentrations
of TEQs (r2 ) 0.76, p < 0.01). PCB 126 contributed the greatest
amount of the TEQs at both locations with average contributions of 77% at KRAOC and 69% at FC (Figure 3). PCBs 118
and 156 were the next greatest contributors, but their relative
rankings were different for the two sites.
Concentrations of total PCBs at KRAOC ranged from 0.01
mg PCBs/kg, ww in muskrat as well as small mammals to 10
mg PCBs/kg, ww in smallmouth bass (Table 3). At FC, PCB
concentrations in prey items ranged from 0.0014 mg PCBs/
kg, ww in small mammals to 2.4 mg PCBs/kg, ww in carp.
PCB concentrations in prey items from KRAOC were significantly greater (Student t-test p < 0.05) than those from
the upstream reference area at FC.
FIGURE 3. Relative contribution of individual PCB congeners to
total TEQ concentrations in livers of mink. The last two bars to the
right of the figure represent the arithmetic mean contribution of
individual congeners to total TEQs in livers of mink from the KRAOC
(n ) 9) and FC (n ) 3), respectively.
Concentrations of TEQs in prey items from KRAOC ranged
from a minimum of 0.22 ng TEQs/kg, ww in small mammals
to 111 ng TEQs/kg, ww in smallmouth bass, while those at
FC ranged from 0.08 ng TEQs/kg, ww in small mammals to
47 ng TEQs/kg, ww in carp (Table 3). Prey items from TB
contained significantly greater concentrations of TEQs than
those from FC (Student t-test p < 0.05), with the exception
of muskrat and carp, which were not statistically different
(Student t-test p ) 0.24 and p ) 0.11, respectively).
Site-Specific Diet Composition. No individual mink had
more than one identifiable prey category in its gastrointestinal
tract. One mink contained scales, five contained fur, one
contained crayfish exoskeleton, and five contained no major
components in their GI tract. This resulted in the following
site-specific dietary composition for mink on the Kalamazoo
River: 72% mammals, 14% fish, and 14% crayfish.
Tissue-Based Risk Calculations. Hazard quotients (HQs)
were calculated based on both LOAEC and NOAEC for total
concentrations of PCBs and TEQs in livers (Table 4). Also,
since two different TRVs were used, two values are presented
for each combination of exposure measure. HQs based on
the mean PCB concentrations in mink livers were all less
than 1.0 based on both the LOAEC and NOAEC (Table 4).
HQs based on the LOAEL of mean TEQ concentrations ranged
from 1.1 to 1.4 and 0.39 to 0.49 for the KRAOC and FC,
VOL. 38, NO. 24, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6455
TABLE 3. Total PCBs (mg PCBs/kg ww) and TEQs (ng TEQs/kg ww) in Prey Items Collected from Two Sites on the Kalamazoo
Rivera
KRAOC
N
mean
sucker
smallmouth bass
carp
forage fish
crayfish
muskrat
small mammals
5
5
5
5
13
7
20
3.8 ( 2.3a
6.7 ( 2.6a
3.8 ( 2.3a
3.2 ( 1.3a
0.54 ( 0.53a
0.07 ( 0.03a
0.11 ( 0.13a
sucker
smallmouth bass
carp
forage fish
crayfish
muskrat
small mammals
5
5
5
5
13
7
20
40 ( 27a
80 ( 25a
48 ( 20
38 ( 15a
6.7 ( 5.5a
1.4 ( 0.78
9.4 ( 20a
FC
range
lipid (%)
N
mean
range
lipid (%)
PCB Concentrations
0.42 -5.9
5.2
6
3.5-10
1.9
5
1.4 -7.1
3.7
5
1.9 -4.7
4.0
5
0.08 -1.9
0.84
4
0.01 -0.11
2.4
4
0.01 -0.55
4.8
17
0.79 ( 0.07
0.89 ( 0.30
1.1 ( 0.74
0.64 ( 0.21
0.06 ( 0.03
0.01 ( 0.01
0.02 ( 0.04
0.67-0.85
0.71-1.4
0.46-2.4
0.37-0.88
0.03-0.09
0.01-0.03
1.4 × 10-3 - 0.18
2.8
2.7
10
3.8
5.8
3.7
4.1
TEQ Concentrations
8.0 -80
5.2
6
40-111
1.9
5
29-77
3.7
5
15-54
4.0
5
1.8 -22
0.84
4
0.22 -2.3
2.4
4
0.3 -93
4.8
17
12 ( 1.3
17 ( 4.2
33 ( 16
9.9 ( 3.0
2.0 ( 1.3
5.9 ( 11
0.60 ( 0.57
10-13
14-22
8.8-47
6.0-13
0.36-3.4
0.17-23
0.08-1.9
2.8
2.7
10
3.8
5.8
3.7
4.1
a Indicates a statistical difference (p < 0.05) when comparing a prey item from KRAOC (Kalamazoo River Area of Concern) to FC (Fort Custer
Recreation Area (Reference area)).
TABLE 4. Tissue-Based Hazard Quotients (HQs) Calculated
Using Total PCBs or Total TEQs Concentrations in Mink Livers
Collected from the Kalamazoo Rivera
total PCB HQs
NOAEC
KRAOCb mean (n)9)
KRAOC upper 95% CL
FCc mean (n)3)
FC upper 95% CL
0.45-0.88
0.67-1.3
0.38-0.74
0.61-1.2
LOAEC
0.37-0.87
0.55-1.3
0.31-0.73
0.50-1.2
total TEQ HQs
NOAEC
4.3-5.4
8.0-10
1.5-1.9
2.7-3.5
TABLE 5. Range of Average Potential Daily Doses (ADDpot),
Based on Mean and the Upper 95% Confidence Limit (U95%
CL) of Prey Items for Total PCBs (mg PCBs/kg bw/d) and TEQs
(ng TEQs/kg w/d), at Both KRAOCdand FCe
KRAOC
LOAEC
dietary models
1.1-1.4
2.1-2.6
0.39-0.49
0.71-0.89
mean
U95% CL
a The TRVs used to calculate HQs based on total PCBs are from
feeding studies conducted by Halbrook et al. (29) and Bursian et al. (28).
The TRVs used to calculate total HQs based on TEQs are based upon
mink feeding studies conducted by Bursian et al. (28) and Heaton et al.
(8). b Kalamazoo River Area of Concern. c Fort Custer Recreation Area
(Reference area).
respectively. HQ values based on the NOAEC for TEQs ranged
from 4.3 to 5.4 and from 1.5 to 1.9 for the KRAOC and FC,
respectively.
Dietary-Based Risk Calculations. Dietary exposures
(average daily doses) were reported as both concentrations
of total PCBs and TEQs (Table 5). The rank order of ADDpot
for the three different dietary models were literature-based
> equally opportunistic > site-specific. The resulting dietarybased NOAEL and LOAEL HQs are presented as a range of
values (Table 6), based on several TRV values. For both NOAEL
and LOAEL-based HQs for PCBs, the TRVs derived from the
laboratory feeding study by Bursian et al. (28) resulted in
lesser HQs than those derived from the study by Halbrook
et al. (29). Of the three studies used to calculate HQs based
on TEQs, the TRV based on the study by Bursian et al. (28)
resulted in the least HQs. Whereas, the TRVs based on the
study by Brunström et al. (7) resulted in the greatest HQs,
and the TRVs based on the study by Heaton et al. (8) resulted
in intermediate HQs.
Among the three diets assumed, mean LOAEL-based HQ
values for PCBs ranged from 0.20 to 1.8 at KRAOC, while
those based on concentrations in prey from FC ranged from
0.04 to 0.35. The NOAEL-based HQs ranged from 0.50 to 3.4
at KRAOC and from 0.09 to 0.68 at FC. LOAEL-based HQ
values for TEQ concentrations ranged from 0.17 to 2.0 at
KRAOC, while those based on concentrations at FC ranged
from 0.07 to 0.71. The mean NOAEL-based HQs of TEQ
concentrations from KRAOC and FC ranged from 0.79 to 14
6456
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 24, 2004
PCBs
FC
TEQs
PCBs
TEQs
Site-Specifica
0.08
1.3
0.12
2.1
0.02
0.02
0.57
1.1
mean
U95% CL
Equally Opportunisticb
0.18
2.3
0.26
3.3
0.03
0.05
0.85
1.3
mean
U95% CL
Literature-Basedc
0.41
4.9
0.59
6.7
0.08
0.11
1.7
2.3
a Model is based on results from field collected mink GI tract contents.
Model assumes that consumption of specified prey items is equal.
c Model based upon one mink diet reported in the literature (32).
d Kalamazoo River Area of Concern. e Fort Custer Recreation Area
(Reference area).
b
and 0.34 to 4.9, respectively. Overall, there was relatively
good agreement between HQ values based on total PCBs
and those based on TEQs. Among locations, PCB-based HQs
were 4-5-fold greater at TB than FC, while TEQ-based HQs
were 2-3-fold greater at KRAOC than FC.
Discussion
Total PCB concentrations in mink liver tissue from both sites
were similar to one another. The resulting HQs were less
than 1.0 at both sites for both the LOAEC and NOAEC (Table
4). This suggests that adverse effects from PCBs were unlikely.
However, there was more of a difference among sites when
comparing TEQ concentrations in liver tissues, although this
difference was not statistically significant. This may indicate
an increased relative potency of the PCB exposure mixture
at KRAOC compared to FC. The resulting TEQ-based HQs
are greater than 1.0 at KRAOC for both the LOAEC and
NOAEC, which indicates the potential for risk to mink.
Additionally, the TEQ-based HQs for the NOAEC at FC would
also suggest potential but relatively low level of risk to mink.
There was considerable agreement among the four
different feeding studies that were used to derive tissuebased and dietary-based TRVs. With the exception of the
FIGURE 4. a. Comparison of tissue-based and dietary-based HQs at the Kalamazoo River Area of Concern (KRAOC) and Fort Custer (FC)
calculated from LOAEC and LOAEL-based TRVs for PCBs and TEQs. The range in values is a result of using more than one TRV. Exact
values for HQs are presented in Tables 4 and 6. b. Comparison of tissue-based and dietary-based HQs at the Kalamazoo River Area of
Concern (KRAOC) and Fort Custer (FC) calculated from NOAEC and NOAEL-based TRVs for PCBs and TEQs. The range in values is a result
of using more than one TRV. Exact values for HQs are presented in Tables 4 and 6.
study by Brunström et al. (7), there was less than a 2.5-fold
difference among TRVs. The dissimilarity of TRVs from the
Brunström study is likely because it was designed to answer
questions regarding effects of ortho vs non-ortho substituted
PCB fractions. That study was not designed as a typical doseresponse study since there was a 7-fold difference in the
LOAEL and NOAEL expressed as TEQs. There was also
uncertainty associated with the NOAEC-based TRV from the
study by Heaton et al. (8). That NOAEC was not directly based
on a tested dose. Instead, it was estimated by taking the
geometric mean of the control and lowest dose, which was
the LOAEC. The experimental design that allows for an
accurate estimate of the NOAEC is one that has a tight dosing
interval and results in two tested doses being less than the
LOAEC.
The use of three different sets of assumptions of dietary
composition encompasses a wide range of potential mink
feeding habits and also provides a measure of uncertainty
when estimating risk to mink. The greatest HQ values for
mink on the Kalamazoo River were obtained by using the
literature-based dietary model because of the greater proportion of the diet that was assumed to be fish and the fact
that concentrations of both PCBs and TEQs were greater in
fish than in other prey items. The Great Lakes Water Quality
Initiative assumes that fish comprise 85% of a mink’s diet for
deriving water quality criteria protective of mammals (34).
However, site-specific determination of the mink diet on the
Kalamazoo River suggests that during the winter the diet of
mink is comprised predominantly of small mammals. This
is similar to results from Sealander (23) who also reported
a high percentage of mammals in the diet of male mink during
winter.
The determination of risk to mink due to PCB concentrations in prey items is dependent upon the dietary assumptions and which TRVs were used. At the KRAOC, LOAELbased HQs for total PCBs and TEQs, based on mean estimated
VOL. 38, NO. 24, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6457
TABLE 6. Hazard Quotient (HQ) Values Based on Mean and the
Upper 95% Confidence Limit (U95% CL) of Potential Average
Daily Doses of Total PCBs and TEQs, When Assuming Three
Different Dietary Compositions for Minka
KRAOCb
LOAEL
NOAEL
FCc
LOAEL
NOAEL
PCB Based Dietary Models
mean
U95% CL
0.20-0.36
0.29-0.52
Site-Specific
0.50-0.70 0.04-0.07
0.71-1.0
0.05-0.10
0.09-0.13
0.13-0.18
mean
U95% CL
Equally Opportunistic
0.44-0.79 1.1-1.5
0.08-0.15
0.63-1.1
1.6-2.2
0.11-0.20
0.20-0.29
0.28-0.39
mean
U95% CL
1.0-1.8
1.4-2.6
Literature-Based
2.4-3.4
0.20-0.35
3.5-4.9
0.26-0.47
0.48-0.68
0.64-0.90
Acknowledgments
TEQ-Based Dietary Models
mean
U95% CL
0.17-0.56
0.27-0.86
Site-Specific
0.79-3.8
0.07-0.24
1.2-5.9
0.15-0.46
0.34-1.6
0.66-3.2
mean
U95% CL
Equally Opportunistic
0.30-0.97 1.4-6.7
0.11-0.35
0.43-1.4
2.0-9.5
0.17-0.54
0.50-2.4
0.77-3.7
mean
U95% CL
0.64-2.0
0.88-2.8
Literature-Based
2.9-14
0.22-0.71
4.0-19
0.29-0.94
1.0-4.9
1.3-6.5
a The TRVs used to calculate HQs were based on feeding studies
(7, 8, 28, 29). The range of calculated HQs is a result of using more than
one TRV. b Kalamazoo River Area of Concern. c Fort Custer Recreation
Area (Reference area).
dietary exposure concentrations, are less than 1.0 for all
dietary models except the literature-based model in which
fish comprise 85% of the mink’s diet (Table 6). Thus, if fish
comprise 85% or more of the mink’s diet, then exposure is
potentially great enough to cause adverse reproductive effects
in mink. If the most conservatively modeled diet (85% fish)
is combined with the most conservative NOAEL-based dietary
TRV, the mean PCB and TEQ-based HQs were 3.4 and 14,
respectively. However, the most likely threshold for effects
lay somewhere between the LOAEL and NOAEL. At FC, total
PCBs pose minimal risk to mink regardless of the percentage
of fish in the diet or which TRV was utilized. However, when
the NOAEL TRVs for TEQs are utilized, there is potential for
risk to mink at FC, based on the equally opportunistic and
literature-based dietary models.
The results of the two basic approaches to determining
exposure in risk assessments were compared to determine
the accuracy of predicted versus measured exposures. This
comparison was done by comparing the range of resulting
mean HQs from the two methods (Figure 4). Based on mean
exposure, all dietary models resulted in HQs that were less
10-fold different than tissue-based HQs, and half of the dietary
models were less than twice as different than their tissuebased HQ counterparts. There did not appear to be a large
difference among the LOAEL(C)-based HQs for PCBs and
TEQs (based on mean exposure) between the tissue-based
and dietary-based approaches (Figure 4). For example, the
LOAEL(C)-based maximum HQs for PCBs and TEQs were
1.8 for the dietary model and 1.4 for the tissue-based
approach. The range of NOAEL(C)-based HQs was greater
than the range of LOAEL(C)-based HQs for both tissue-based
and dietary-based approaches, reflecting a greater uncertainty in the NOAEL(C)-based TRVs.
Results from this study suggest that it would be appropriate to estimate risk based on either tissue-based or dietarybased methodologies. In the future, if researchers use a
6458
9
dietary-based approach to estimate risk to mink, site-specific
data should be collected from a variety of potential prey
items and incorporated into the assessment. Researchers
could also attempt to reduce uncertainty in the dietary-based
assessment by measuring availability of prey species, collecting only fish sizes that mink would likely capture and
adding a seasonal component. At either site on the Kalamazoo
River, there was no single dietary model that most closely
approximated the exposure measured in liver tissue. When
dietary-based methods are used for risk assessment, a variety
of feeding assumptions should be used to evaluate the
sensitivity of the dietary model. Risk assessors should clearly
communicate how dietary assumptions impact risk calculations. Both approaches to assessing risk were in agreement
that the degree of exposure to PCBs and TEQs was near the
threshold for effects on reproduction in mink on the
Kalamazoo River.
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 24, 2004
Funding was provided through a grant from the Kalamazoo
River Study Group to Michigan State University. Fish
sampling was conducted with the assistance of Blasland,
Bouck and Lee Inc. (Syracuse, NY), who provided electrofishing personnel and equipment. This research involved
the assistance of the following people: Scott Fitzgerald necropsies; Steve Bursian - necropsies and equipment; Kevin
Allen, Jim Burns, Dan Keith, Daniel Villeneuve, and Alan
Zwiernik - trapping; Dr. K. Kannan, Dong-Hoon Khim,
George Klemolin, and Jamie Kober - laboratory support and
analysis; Ryan Holem - organizational support. Steve
Bursian, Tom Burton, and Scott Winterstien reviewed the
manuscript and provided valuable suggestions. Additional
thanks to the MSU Kellogg Biological Station for housing
and laboratory space during sample collection.
Literature Cited
(1) Fairbrother, A. Hum. Ecol. Risk Asses. 2003, 9, 1475-1491.
(2) McDonald, B. G.; Wilcockson, J. B. Hum. Ecol. Risk Assess. 2003,
9, 1585-1594.
(3) Environmental Factors Handbook; U.S. Environmental Protection Agency: Washington, DC, 1993.
(4) Guidelines for ecological risk assessment; EPA/630/R-95/002F;
U.S. Environmental Protection Agency: Washington, DC, 1998.
(5) Aulerich, R. J.; Ringer, R. K.; Seagran, H. L.; Youatt, W. G. Can.
J. Zool. 1971, 49 (5), 611-616.
(6) Aulerich, R. J.; Ringer, R. K.; Iwamato, S. J. Reprod. Fertil. 1973,
Suppl 19, 365-376.
(7) Brunström, B.; Lund, B. O.; Bergman, A.; Asplund, L.; Athanassiadis, I.; Athanasiadou, M.; Jensen, S.; Orberg, J. Environ.
Toxicol. Chem. 2001, 20 (10), 2318-2327.
(8) Heaton, S. N.; Bursian, S. J.; Giesy, J. P.; Tillitt, D. E.; Render,
J. A.; Jones, P. D.; Verbrugge, D. A.; Kubiak, T. J.; Aulerich, R. J.
Arch. Environ. Contam. Toxicol. 1995, 28, 334-343.
(9) Hochstein, J. R.; Bursian, S. J.; Aulerich, R. J. Arch. Environ.
Contam. Toxicol. 1998, 35, 348-353.
(10) Kihlstrom, J. E.; Olsson, M.; Jensen, S.; Johansson, A.; Ahlbom,
J.; Bergman, A. Ambio. 1992, 21 (8), 563-569.
(11) Restum, J. C.; Bursian, S. J.; Giesy, J. P.; Render, J. A.; Helferich,
W. G.; Shipp, E. B.; Verbrugge, D. A. J. Toxicol. Environ. Heal.
A. 1998, 54, 343-375.
(12) Thompson, T.; Morgan, D.; Mortensen, l.; Pearson, A.; Morris,
D.; Traynor, C. Final baseline human health and ecological risk
assessment: Lower Fox River and Green Bay, Wisconsin remedial
investigation and feasibility study; Wisconsin Department of
Natural Resources: Madison, WI, 2002.
(13) Sample, B. E.; Suter, G. W. I. Environ. Toxicol. Chem. 1999, 18
(4), 610-620.
(14) Ecological risk assessment for the General Electric (GE)/Housatonic
River site, rest of river: Volume 6; U.S. Army Corps of Engineers
and U.S. Environmental Protection Agency: Concord, MA, 2003.
(15) Leonards, P. E. G.; van Hattum, B.; Cofino, W. P.; Brinkman, U.
A. T. Environ. Toxicol. Chem. 1994, 13, 129-142.
(16) Camp, Dresser, McKee. Final Baseline Ecological Risk Assessment,
Allied Paper, Inc./Portage Creek/Kalamazoo River Superfund Site;
Michigan Department of Environmental Quality: Lansing, MI,
1999.
(17) Giesy, J. P.; Kannan, K. Crit. Rev. Toxicol. 1998, 28 (6), 511-569.
(18) Kannan, K.; Blankenship, A. L.; Jones, P. D.; Giesy, J. P. Hum.
Ecol. Risk Assess. 2000, 6 (1), 181-201.
(19) Blankenship, A. L.; Giesy, J. P. In Environmental analysis of
contaminated sites: toxicological methods and approaches;
Sunahra, G. I., Renoux, A. Y., Thellen, C., Gaudet, C. L., Pilon,
A., Eds.; John Wiley and Sons: New York, U.S.A., 1999; Chapter
9.
(20) Framework for Application of the Toxicity Equivalence Methodology for Polychlorinated Dioxins, Furans and Biphenyls in
Ecological Risk Assessment (External Review Draft); EPA 630/
P-03/002A; U.S. Environmental Protection Agency, Risk Assessment Forum: Washington, DC, 2003.
(21) Casson, J. E.; Klimstra, W. D. Trans. Ill. State Acad. Sci. 1983,
76, 281-292.
(22) Hamilton, W. J., Jr. J. Mammal. 1936, 17 (2), 169.
(23) Sealander, J. A. J. Wildl. Manage. 1943, 7 (4), 411-417.
(24) Whitman, J. S. M.S., University of Idaho, 1981.
(25) Nakata, H.; Kannan, K.; Jing, L.; Thomas, N.; Tanabe, S.; Giesy,
J. P. Environ. Poll. Marine Res. 1998, 103, 45-53.
(26) Senthilkumar, K.; Kannan, K.; Sinha, K.; Tanabe, S.; Giesy, J. P.
Environ. Toxicol. Chem. 1999, 18 (7), 1511-1520.
(27) Van den Berg, M.; Birnbaum, L.; Bosveld, A. T. C.; Brunstrom,
B.; Cook, P.; Feeley, M.; Giesy, J. P.; Hanberg, A.; Hasegawa, R.;
(28)
(29)
(30)
(31)
(32)
(33)
(34)
Kennedy, S. W.; Kubiak, T. J.; Larsen, J. C.; van Leeuwen, F. X.
R.; Liem, A. K. D.; Nolt, C.; Peterson, R. E.; Poellinger, L.; Safe,
S.;, D. S.; Tillitt, D.; Tysklind, M.; Younes, M.; Waern, F.;
Zacharewski, T. Environ. Health Perspect. 1998, 106 (12), 775792.
Bursian, S. J.; Aulerich, R. J.; Yamini, B.; Tillitt, D. E. Dietary
exposure of mink to fish from the Housatonic River: Effects on
reproduction and survival; Final report submitted to Weston
Solutions, West Chester, PA, 2003; 106 pp. Available at http://
www.epa.gov/NE/ge/thesite/restofriver-reports2.html.
Halbrook, R. S.; Aulerich, R. J.; Bursian, S. J.; Lewis, L. Environ.
Toxicol. Chem. 1999, 18 (4), 649-654.
Litvaitis, J. A.; Titus, K.; Anderson, E. M. In Research and
management techniques for wildlife and habitats, Fifth, rev, ed.;
Bookhout, T. A., Ed.; The Wildlife Society: Bethesda, MD, 1996.
Lockie, J. D. J. Wildl. Manage. 1959, 23 (2), 224-227.
Alexander, G. Michigan Academician 1977, 10, 181-195.
Beckett, K. Michigan State University. Personal communication,
2003.
Great Lakes water quality initiative criteria documents for the
protection of wildlife; EPA-820-B-95-008; U.S. Environmental
Protection Agency: Washington, DC, 1995.
Received for review March 15, 2004. Revised manuscript
received May 27, 2004. Accepted June 22, 2004.
ES049600E
VOL. 38, NO. 24, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6459