SERA
United States
Environmental Protection
Agency
Great Lakes
National Program Office
77 West Jackson Boulevard
Chicago, Illinois 60604
EPA 905-R94-024
October 1994
Assessment and
Remediation of
Contaminated Sediments
(ARCS) Program
HAZARD RANKING OF CONTAMINATED SEDI-
MENTS BASED ON CHEMICAL ANALYSIS,
LABORATORY TOXICITY TESTS, AND BENTHIC
COMMUNITY STRUCTURE: METHOD OF PRIORI-
TIZING SITES FOR REMEDIAL ACTION
United States Areas of Concern
ARCS Priority Areas of Concern
-------
Hazard Ranking of Contaminated Sediments Based on
Chemical Analysis, Laboratory Toxicity Tests, and
Benthic Community Structure: Method of Prioritizing
Sites for Remedial Action
M. L. Wildhaber and C. J. Schmitt
National Biological Survey
Midwest Science Center
4200 New Haven Road
Columbia, Missouri 65201
September 22, 1994
Prepared for:
Assessment and Remediation of Contaminated Sediments Program
Great Lakes National Program Office
U.S. Environmental Protection Agency
77 West Jackson
Chicago, Illinois 60604
U..S. Fish and Wildlife Service/National Biological Survey Contract Number
DW14933874-1
-------
DISCLAIMER
Mention of trade names or commercial products does not constitute
endorsement or recommendation for use.
-------
TABLE OF CONTENTS
Page
INTRODUCTION 1
OVERALL HAZARD RANKING 2
Ranking Sites Based on Toxicity Estimated from Chemistry 3
-Toxic units model 5
-Estimated pore-water concentrations 6
-Water quality standards 9
-Other information 12
Ranking Sites Based on Toxicity as Measured by
Laboratory Toxicity Tests 12
Ranking Sites Based on Toxicity as Measured by
Benthic Community Structure 14
Final Ranking 15
Objectives 16
MATERIALS AND METHODS 16
Sediment Sampling Locations 16
Sampling and Biological Measures 16
Chemical Analyses 17
Detection Limits 18
Toxic Units Model Evaluation 19
Ranking Evaluation 20
-------
TABLE OF CONTENTS (continued)
Page
RESULTS 21
Toxic Units Model 21
-Laboratory toxicity 21
-Benthic community structure 23
Final Ranking 25
DISCUSSION AND CONCLUSIONS 26
Effectiveness of the Toxic Unit Model 26
-Laboratory toxicity 26
-Benthic community structure 30
-Data quality 32
-Potential for model enhancements 33
Ranking Approach 35
ACKNOWLEDGMENTS 38
REFERENCES 39
TABLES 50
FIGURES 74
-------
INTRODUCTION
Contaminated sediments have become a major focus of environmental concern
and research, particularly in the Great Lakes (USEPA 1990). The concern is that
even with elimination of current and future sources of contamination, those
contaminants already present in sediments are a threat to aquatic life and human
health. The Assessment and Remediation of Contaminated Sediments (ARCS)
program was managed by the Great Lakes National Program Office of the United
States Environmental Protection Agency (USEPA) specifically to address
contaminated sediment issues in the Great Lakes and to examine new and
innovative ways to both assess and treat contaminated sediments (USEPA 1990).
One goal of the ARCS program was to develop a method by which the relative
risks associated with contaminated sediment from different sites can be evaluated.
A general method of ranking contaminated sediments based upon direct sediment
analyses and tests proposed by Kreis (1989) and was used to proportionally scale
sediment variables so that they could be compared and combined. The ranking
method developed for ARCS was modified and enhanced from the version
proposed by Kreis (1989) by incorporating bioavailability, control-adjusted
laboratory toxicity tests, and mean tolerance to pollution of the benthic community
for the sediments of concern.
The numerical ranking system developed by Kreis (1989) was intended as a
guide, to be used in evaluating regulatory and remediation alternatives for
contaminated Great Lakes sediments. Kreis (1988) previously showed that the
1
-------
ranking process can be an effective tool for determining which sites, of a set of
contaminated sites, have the highest concentrations of total contaminants and
associated parameters. The results of the ranking process can then be used to
prioritize sites for remediation; this prioritization is necessary due to the high cost
of sediment remediation. As resources and technologies become available, the
sediments needing remediation could each be "cleaned-up" in the order of their
ranking. The remediation procedure or combination of procedures chosen is site-
specific and would depend on ecological, chemical, economic, and engineering
considerations.
OVERALL HAZARD RANKING
In this report, we used the process of proportional scaling from 1 to 100 (Kreis
1989) to scale different types of information (i.e., sediment chemistry, laboratory
toxicity, and benthic community structure) in a way that they can be evaluated
equivalently. As suggested by Kreis (1989), once all the information was on the
same scale, the data were combined by averaging (i.e., arithmetic mean) the
information from all three categories. The site mean of the proportionally scaled
values was then considered to be the best estimate of relative hazard, among
sites, of sediment contaminants to aquatic life.
The primary difference between this relative ranking process and the earlier
process (Kreis 1989) is that this one incorporates more information, including: 1)
adjustment of the bulk sediment chemistry concentrations for estimated
-------
bioavailability and cumulative chronic toxicity through the use of "toxic units"
(defined in the next section); 2) incorporating of control-adjusted laboratory toxicity
test results; and 3) incorporating of benthic community information, in the form of
mean tolerance to pollution per organism for each site.
Ranking Sites Based on Toxicity Estimated from Chemistry Data
In the ranking system developed by Kreis (1989), each chemical or group of
chemicals (e.g., metals, dioxins, etc.) analyzed was ranked independently. For
each analyte or group of analytes, the measured values or totals, respectively,
representing each site under consideration were scaled from 1 to 100, relative to
each other (i.e., the minimum measured value became 1, the maximum, 100). The
equation used to calculate the ranks for each chemical or group of chemicals was:
Rank = 1 S/te value - Minimum value
Maximum value - Minimum value
The independent ranks calculated for each chemical or group of chemicals was
then combined by averaging (arithmetic mean) for each site, yielding a mean rank
for each site based on chemical concentrations. In this ranking process the
chemicals analyzed are scaled relative to each other based only on the
concentrations present; the process does not scale those chemicals based on the
true measure of concern, which is hazard. In addition to analyte concentration.
-------
hazard also includes a toxicity component.
The approach we present for evaluating the hazard associated with chemicals
in sediments includes toxicologies!, ecological, and bioavailability information. In
contrast to the earlier method (Kreis 1989), in our approach the "toxic units"
(Sprague and Ramsay 1965) for each contaminant measured are calculated.
Sprague and Ramsay (1965) defined a toxic unit as the ratio of observed water
concentration of a contaminant to the incipient lethal concentration of that
contaminant. Our definition of a toxic unit is similar, but based on the USEPA
Ambient Water Quality Criteria (AWQC); we define a toxic unit as the ratio of the
estimated bioavailable component of the contaminant to the chronic toxicity water
quality criteria (e.g. USEPA 1986a) for that contaminant. We estimate
contaminant bioavailability by assuming that concentrations of the water-soluble,
bioavailable fractions of organic chemicals are controlled by equilibrium partitioning
(DiToro et al. 1991), and that acid volatile sulfides control the solubility and
bioavailability of metals (DiToro et al. 1990). Our estimates of chronic toxicity are
based heavily on the AWQC for chronic toxicity (e.g., USEPA 1986a), which
incorporate laboratory-measured chronic toxicity—which indirectly incorporates
bioaccumulation and measured bioaccumulation.
In our approach, total potential toxicity is calculated by summing toxic units
over all contaminants measured for each site. This total potential toxicity is then
ranked among sites using the ranking equation described earlier (Kreis 1989). The
result is a relative ranking of the sites under investigation based on the cumulative
-------
knowledge of what is known about the potential bioavailability and toxicity of the
contaminants found in the sediments at each site. In our approach the analytes
present in sediments are scaled for toxicity and bioavailability and combined
(summed). These sums are then ranked.
Toxic units model
To estimate toxicity of a complex mixture of chemicals, we used an additive
model based on toxic units. A toxic unit is defined here as the ratio of the
estimated concentration of a contaminant in the pore water of a test sediment to
an estimate of chronic toxicity of that contaminant in water. The equation for
toxic units is:
Q
Toxic unit = —^~
r = Estimated pore-water concentration
^wp
C = Water quality standard
The toxic units for the contaminants in a sediment are then summed to produce a
total toxicity estimate for that sediment.
-------
Estimated pore-water concentrations
Pore-water concentrations of the contaminants are estimated on the basis of
equilibrium partitioning and organic carbon control for organic compounds (DiToro
et al. 1991) and AVS and sulfide control for metals (DiToro et al. 1990). The
estimated pore-water concentrations are considered estimates of the bioavailable
portions of the total concentrations of each contaminant measured in the
sediments (i.e., the concentration to which sediment-dwelling organisms would be
exposed at equilibrium).
Pore-water concentrations of organic compounds were estimated from the
total sediment concentration in two ways: 1). by assuming complete bioavailability
of bulk sediment concentrations; and 2). by correcting for organic carbon using
equilibrium partitioning (DiToro et al. 1991). The intent of using both methods
was to evaluate how well the toxic units model predicted toxicity of sediments as
measured by control-adjusted laboratory toxicity tests and mean tolerance to
pollution of the benthic community and how the ranking of hazard among sites
compared with and without accounting for bioavailability through equilibrium
partitioning. In the bulk-sediment approach, organic analytes are considered
completely bioavailable; that is, the pore-water concentration was estimated by
multiplying the dry weight concentration of each analyte by the dry weight to
moisture content ratio of the sediment. In the equilibrium partitioning method, the
pore-water concentration for organic contaminants and, hence, the bioavailability,
of each analyte is assumed to be controlled by organic carbon content of the
-------
sediment, and is consequently a function of the organic carbon partition coefficient
(K^) of the analyte. We used equilibrium partitioning modeling to estimate the
pore-water concentration of each analyte on the basis of the bulk (i.e., dry-weight)
sediment concentration, the arithmetic mean log octanol/water partition coefficient
(Table 1), and the proportion of organic carbon in the sediment sample associated
with that analyte, as follows:
oc
Where:
CWP = Pore-water analyte concentration,
Cs = Bulk-sediment analyte concentration;
LogtfJ = 0.983 x Log(Kov) + 0.00028 (Ditoro et al. 1991)
KQW = Octanoti water partition coefficient,
^ = Sediment organic carbon partition coefficient, and
POC = Proportion of organic carbon in the sediment
The main source of Kow values was USEPA (1987a); however, this document
did not contain Kow values for all analytes. For those organic analytes lacking a
Kow, the following substitutions were made: Chlorodioxins and chlorodibenzofurans
lacking Kow values were assumed to have the same Kow as a chlorodioxin with the
same number of chlorines for which a value was available; for endrin ketone (a
-------
polar metabolite of endrin) the Kow of endrin aldehyde (another polar metabolite)
was substituted; and for cis- and trans-chlordane. the Kow for technical chlordane,
a complex mixture containing these and other constituents, was used.
Analogous to the methods used for organic contaminants, pore-water
concentrations of metals were also estimated from the bulk sediment concentration
in two ways: 1). by assuming complete bioavailability of bulk sediment
concentrations; and 2). by correcting for AVS by using the concentrations of
metals simultaneously extracted with AVS (i.e., simultaneously extracted metals-
SEMs) adjusted for AVS (DiToro 1990). In both methods, the estimated pore-
water concentration was the estimated bioavailable concentration (entire sediment
concentrations in the first method) multiplied by the dry weight to moisture
content ratio of the sediment. In the first method, the full concentration of each
metal present in the sediment is considered bioavailable. In the second, only those
metals extractable with a weak acid (1-IM HCI) and adjusted for the potential sulfide
salts that could be formed by the extracted metals are considered bioavailable. As
described for organic contaminants and organic carbon, the intent of using these
two methods was to evaluate how well the toxic units model predicted the toxicity
of sediments based on control-adjusted laboratory toxicity tests and mean
tolerance to pollution of the benthic community and how the ranking of hazard
among sites compared with and without AVS limitation.
The AVS model is based on the assumption that under the reducing conditions
present within sediments, sulfides control the pore-water concentrations and,
8
-------
hence, bioavailability of divalent metals. The sulfide salts of the metals are
relatively insoluble; the formation of these salts thereby renders the divalent metals
unavailable. AVS modeling of pore-water concentrations therefore adjusts the
maximum potential pore-water concentrations of metals downward based on the
amount of AVS present.
In the ARCS sediments, the total molar concentrations metals simultaneously
extracted with AVS frequently exceeded the molar sulfide concentration.
Consequently, it was necessary to apportion the sulfide among the divalent metals
and arsenic present. We allocated AVS to metals and arsenic based on the
solubility product constants, K,p, of their sulfide salts (Weast et al. 1988).
Accordingly, AVS was allotted to metals in the following order: mercury, silver,
copper, cadmium, lead, zinc, nickel, arsenic, iron, manganese, and chromium (i.e.,
mercuric sulfide is the least soluble and chromium sulfide the most soluble sulfide).
An equimolar amount of AVS was allotted to each metal in the order of their K.p
until either all AVS was allotted or all metals were sulfide-bound. For selenium,
which is not controlled by AVS, we used the bulk-sediment concentration and the
sediment quality standard proposed by Lemly et al. (1993).
Water quality standards
Relative potential toxicity was estimated from chronic toxicity information for
each analyte (Table 2), the primary source of which was the series of AWQC
documents. The calculation of chronic toxicity AWQC incorporates information
-------
both on chronic toxicity of a contaminant to various forms of aquatic life and the
bioconcentration factor of the contaminant in aquatic ecosystems. Thus, this
estimate of chronic toxicity is also more indicative of the severity of long-term
exposure to a contaminant for aquatic organisms than would be a direct measure
of toxicity. In addition, many of the so-called "short-term" tests performed under
the ARCS Program, and against which we compare the results of the toxic units
model, are in fact longer in duration than 96 h and include sublethal endpoints
such as growth and behavior. Hence, they are more appropriately compared to the
chronic criteria.
For those analytes without AWQC, chronic toxicity values were obtained from
other sources (Table 2). We used the chronic toxicity values of the Michigan
Department of Natural Resources 1993 Water Quality Standards (unpublished data)
for the dichlorobenzenes, naphthalene, silver, and tributyltin. For dibenzofuran,
bis(2-ethylhexyl)phthalate, and dimethyl phthalate, we estimated a value by
proportionally adjusting the Michigan DNR chronic toxicity standard for 1,2-
dichlorobenzene on the basis of the ratio of 1,2-dichlorobenzene toxicity to that of
the other compounds (LeBlanc 1980). For the chlorodioxins, chlorodibenzofurans,
and polychorinated biphenyls (PCBs), we estimated a value by proportionally
adjusting the Aroclor8 1254 AWQC (USEPA 1980a) in two steps: First, we
determined the ratio of the aryl hydrocarbon hydroxylase (AHH) activity of Aroclor8
1254 to that of TCDD and the ratio of the AHH activity of Aroclor8 1254 to that of
each compound (Smith et al. 1990). For some isomers and congeners, the ratio
10
-------
of AHH activity of Aroclor8 1254 to that of TEF of each compound {Safe 1990)
was used. We then used the ratio of these two values as a proportionation factor
to adjusted Aroclor8 1254 values. For most polycyclic aromatic hydrocarbons
(PAHs), a chronic toxicity estimate was back-calculated using equilibrium
partitioning from a sediment threshold (USEPA 1985a) by assuming 4% organic
carbon (the value used to generate the threshold concentrations) and the K,,,. values
used in the model presented here; for benzo(
-------
values used in the calculations for individual sediments were those measured in the
14-d flow-through sediment toxicity test with Chironomous riparius. one of the
ARCS laboratory toxicity tests, for that sediment (Nelson et al. 1993).
Lack of available information precluded the use of several of the compounds
measured in the example data set in the ranking process. These included: the a, 13,
and 6 isomers of hexachlorocyclohexane; dimethyl butyl phthalate; di-n-
octalphthalate; 2-methyl naphthalene; 4-methyl naphthalene; and
benzo(b)flouranthene, for which toxicity information was lacking; and monobutyl-
and dibutyl-tin, for which Kow values were not available.
Other information
The potential toxicity of ammonia in the ARCS sediments prompted the
inclusion of the unionized ammonia levels associated with the 14-d Chironomous
riparius sediment toxicity test for each sediment (Nelson et al 1993). The toxicity
of ammonia and, hence, the AWQC for ammonia (USEPA 1985b) are pH-
dependent. Consequently, we used the pH values measured during the same
toxicity tests to estimate the AWQC for unionized ammonia.
Ranking Sites Based on Toxicity as Measured by Laboratory Toxicity Tests
When the results of multiple laboratory sediment toxicity tests and multiple
endpoints measured within some tests are to be used together, the data from each
toxicity test and multiple endpoints measured within some tests had to be scaled
12
-------
equivalently (analogous to the toxic units scaling performed on the suite of
chemistry measures). This was accomplished by normalizing the measured
response associated with an endpoint to the control sediment response for that
endpoint:
Control-adjusted laboratory = .. Endpoint value for test sediment
toxicity response Endpoint value for control sediment
Adjusting a test response to its control scales that measure as a proportion of its
control, and it also adjusts the responses for the conditions present at the time of
the test. The latter accounts for variation attributable to the tests being run at
different times, in different locations, by different investigators, or combinations of
these factors. So that the control-adjust laboratory toxicity response ordered the
sites in the same fashion with respect to toxicity as the toxic units model (i.e.,
lowest value = least toxic and highest value = most toxic), the ratio of the
endpoint value for the test sedimentiendpoint value for the control sediment was
subtracted from one, as shown in the last equation. The estimates of hazard—the
control-adjusted laboratory toxicity responses for each endpoint—were then
averaged (i.e., arithmetic mean) over all measured endpoints for a site to estimate
the mean hazard for each site based on laboratory toxicity. This mean estimated
hazard was then ranked among sites.
13
-------
In some tests (e.g., 48-hour Daphnia magna survival-Table 3), multiple
concentrations of the test medium were used to estimate LCSOs, which yield little
information compared to the actual response at individual concentrations, and
which are not scalable to the control. For such tests, a response was first
multiplied by the proportion of the full concentration of the medium at which the
test was run (i.e., 50% elutriate response was multiplied by 0.5) before the
proportional laboratory toxicity response was calculated. This "proportionation
factor" gave the observed response for each concentration an equivalency to
100% (i.e., the concentration at which the other tests were run) and enabled
scaling of the responses relative to controls.
Ranking Sites Based on Toxicity as Measured by Benthic Community Structure
As we described for laboratory toxicity tests, when different measures of
benthic community structure and well-being are measured, they must be uniformly
scaled for evaluation. The measure we used was Lenat's (1993) biotic index (i.e.
mean tolerance-to-pollution per organism). Lenat (1993) presents a extensive list
of tolerance to pollution values for all the organisms considered in his estimates of
mean tolerance per organism. The mean tolerance is calculated by first assigning
each species a relative tolerance value to pollution; the set of tolerance to pollution
values for each species presented by Lenat (1993) was based on correlations
between water quality and abundance data for each species. Once the tolerance
value for each species is obtained, the mean tolerance (T) is calculated by
14
-------
summing the product of the abundance (IM,) and tolerance value for each species at
a site (Tj) and then dividing that total by the total number of organisms at a site:
E N,
Calculating the mean tolerance to pollution of the benthic community assures that
the presence of less tolerant orders influences the ranking of a site (i.e., the site is
considered less toxic). Here, it is this mean tolerance to pollution per organism at
a site that is ranked among sites. Because Lenat's values were derived for flowing
waters in the southeast, tolerance values for some taxa had to be obtained from
other sources. Table 4 gives the list of organisms observed in the ARCS
sediments, their associated tolerance values, and the source of the value; a few of
the tolerance values had to be obtained from another source.
Final Ranking
The rankings that result from the different types of information discussed (i.e.
chemistry, laboratory toxicity tests, and benthic community structure) can be
combined to produce a final ranking for each site. At this point each type of
information has been scaled from 1 to 100. The estimate of relative hazard for the
sites under investigation, based on all three types of information, is the arithmetic
15
-------
mean of the three ranks.
Objectives
In this report, we focus on two main aspects of the just mentioned ranking
procedure. The first goal of this report was to evaluate the effectiveness of the
toxic units approach at predicting biological endpoints (i.e., laboratory toxicity and
benthic community structure). The second goal was to evaluate how site ranking
changed as the level of information collected at a site declined.
MATERIALS AND METHODS
Sediment Sampling Locations
Sediment samples were collected at 19 different sites for the ARCS program
(Ingersoll et al. 1993). Samples were obtained from the lower reaches of two
rivers and one harbor complex: in the Buffalo River, samples from five sites were
collected in October, 1989; in the Saginaw River, three sites were sampled in
December, 1989, and seven were sampled in June, 1990 (only one of which had
been sampled in December); and four Indiana Harbor sites were sampled in August,
1989.
Sampling and Biological Measures
Grab samples of sediment were collected with a 23- x 23-cm Ponar sampler
and brought into the laboratory for analysis. At each site, one sample was
16
-------
collected for chemical analysis and laboratory toxicity tests and five samples were
collected for benthic community analyses. Those samples designated for both
chemical analysis and laboratory toxicity testing were split into two portions; one
portion of each sample was analyzed for contaminants and the other portion was
used in laboratory toxicity tests. Those samples designated for benthic community
analysis were sieved with a 500-um screen and organisms were identified to the
lowest possible taxonomic level. Table 3 gives a list of laboratory toxicity tests
carried out on the set of sediment samples modeled in this paper and the endpoints
measured. Table 4 gives a list of the taxa observed in the set of sediment samples
modeled in this paper, the tolerance value used for each taxon, and the reference
from which the tolerance value was obtained. Further details of sample collection,
handling, preparation, and testing are presented elsewhere (Ingersoll et at. 1993).
Chemical Analyses
A variety of methods were used to measure contaminant concentrations in the
ARCS sediment samples. A subsample of the analytical portion of each sediment
was freeze-dried to gravimetrically estimate percent solids. A second subsample
was analyzed with a carbon determinator to estimate total organic carbon. A third
subsample was used to estimate acid volatile sulfides (AVS) by leaching with 1-N
HCL; simultaneously extracted metals (SEM) were determined by analyzing the
resulting HCL solutions by atomic absorption spectroscopy. A fourth subsample
was completely digested with acids and analyzed for total metals by atomic
17
-------
absorption. Crustal elements were estimated from a fifth, freeze-dried subsample
of sediment analyzed by x-ray fluorescence. Methylmercury was estimated from a
sixth subsample of sediment digested in potassium hydroxide and analyzed by
atomic fluorescence. Organotins were quantified by gas chromatography from a
seventh subsample after extraction with 0.2% tropolone. Organic chemical
residues (i.e., PAHs, PCBs, chlorinated pesticides, PCDDs, and PCDFs) were
extracted with methylene chloride (all but PCDDs and PCDFs) or benzene (PCDDs
and PCDFs) from subsamples eight and nine, respectively, and analyzed by gas
chromatography and mass spectrometry. Further details of these chemical
analyses can be found elsewhere (Ingersoll et al. 1993).
Detection Limits
To complete this study, we had to account for differences in analytical
sensitivity among samples by defining for each analyte one censoring level to be
applied to all samples. We needed to ensure that an analyte measured with low
sensitivity (i.e., high detection limit) did not control the toxic units estimate for any
sample only because of that analyte's high detection limit. To eliminate this
possibility, we censored at the highest detection limit among samples for each
analyte; i.e., any concentrations, detected or not, that were at or below this
censoring level were assumed to be zero. The censoring process we used for the
ARCS chemistry data was the most appropriate alternative because the data had
been collected prior to the development of the model; preliminary analyses revealed
18
-------
that many of the toxic units values derived from the ARCS sediment chemistry
data were defined by the analytical sensitivity (i.e., detection limits) of the
methods used to measure the various analytes. The detection limits for one set of
Saginaw River sediments were so high that they completely dominated the hazard
assessment. Consequently, data from the seven Saginaw River sites sampled in
June of 1990 were not included in the analyses presented here; our analyses are
based on samples from the other 12 ARCS sites, all of which had much lower and
more consistent detection limits.
Toxic Units Model Evaluation
Model predictions were analyzed by simple regression and correlation. We
assessed the ability of various forms of the toxic units model (e.g., with and
without certain variables) to predict mean control-adjusted laboratory toxicity
and/or mean tolerance to pollution of the benthic community on the basis of
regression results, which were evaluated in terms of R2- and P- values. A model
was considered a better predictor of toxicity than other models if it accounted for a
larger percentage of the variability observed in the measured response (i.e., mean
control-adjusted laboratory toxicity or mean tolerance to pollution of the benthic
community).
19
-------
Ranking Evaluation
In the ARCS program, there were three categories of collection sites for
sediments: Priority Master Stations, Master Stations, and Reconnaissance
Stations. The difference between the categories of stations was in the number of
parameters measured. At Priority Master Stations, all chemistry (Table 2),
laboratory toxicity (Table 3), and benthic community (Table 4) measurements were
made. At Master Stations, all chemistry, a shortened list of laboratory toxicity ('*'
tests in Table 3), and all benthic community measurements were made. At
Reconnaissance Stations, only a short list of contaminants (i.e. cadmium,
chromium, copper, iron, nickel, lead, zinc, unionized ammonia, and total AroclorRs)
and Microtox" were measured. The intent of this gradient of information collected
at each site was designed to determine how much information was actually needed
to effectively evaluate the relative contaminant hazard among a group of sites.
As directed under the ARCS program, we evaluated the effect of different
amounts of information on the resultant final ranking of the sites. We compared
and contrasted the final ranking among sites when Priority Master Station
indicators were used as opposed to when: 1). Master Station indicators were used;
2). Reconnaissance indicators were used; and, 3). only the contaminants
considered bioaccumulative were used in the toxic units model with the full bulk
sediment concentration of each contaminant considered bioavailable. The latter
question was added on our part based upon our review of the draft Great Lakes
Water Quality Initiative (USEPA 1993) (Table 5). We wanted to know the extent
20
-------
to which the relative ranking of sites changed when the set of information used in
the process was reduced. In other words, what is the minimal, most cost-effective
suite of measurements required to accurately rank the sites? The assessment of
the extent to which the ranking changed was measured by the change in the
amount of variation explained by the full model rankings when reduced sets of data
were used to rank the sites (i.e., change in R2). The fact that comparisons were
limited to pairs of ranks allowed us to use simple correlation analysis. We used the
Statistical Analysis System (SAS Institute 1990) for all data management,
computations, and statistical analyses.
RESULTS
Toxic Units Model
Laboratory toxicity
The best predictor of laboratory toxicity was the full model of bioavailable
contaminants, which included both organic carbon control for organic
contaminants and AVS control for inorganics (Figure 1). This model accounted for
more than 89% of the variability present in mean laboratory toxicity. When bulk
contaminant concentrations were substituted for bioavailable fractions, the model
accounted for only 68% of the variability (Figure 2). Among the ARCS sites,
predicted toxicity of Indiana Harbor site 3 was the most greatly affected when
equilibrium partitioning and AVS were excluded from the model; toxicity at this site
was accurately estimated by the full model, but greatly underestimated by the bulk
21
-------
chemistry model (cf. Figures 1 and 2).
Tables 6-8 give the list of individual contaminants that were estimated to be
present at at least one toxic unit at one or more sites within an area of concern
(e.g., Saginaw River), based on bioavailability chronic toxicity. These lists are
shorter than the full list of contaminants measured because not all PCB mixtures,
dibenzodioxins, dibenzofurans, etc. were present at toxic concentrations at all
sites. The contaminants that contributed the most to the toxic units estimate for a
sediment, including iron and ammonia, varied among sites; no single contaminant
dominated at all sites (Tables 6-8). In general, only a few contaminants accounted
for the greatest percentage of estimated toxicity at each site. For example,
endosulfan, endrin, methylmercury, and iron contributed most in the Buffalo River;
chromium, dieldrin, endosulfan, endrin, heptachlor, iron, methylmercury, PCBs, and
tributyl tin the most in Indiana Harbor; and chlordane, heptachlor, iron, and PCBs
the most in the Saginaw River.
The full model indicated that unionized ammonia, chromium, and iron were
present at potentially toxic concentrations and accounted for much observed
variability at most sites (Tables 6-8). Although it probably influenced many
laboratory toxicity test results, unionized ammonia alone did not explain all the
toxic effects of the sediments observed in the laboratory; alone it accounted for
more than 60% of the variability (Figure 3), substantially less than the full model.
The relative toxicity of Saginaw River site 6 was much higher than predicted when
ammonia alone was used in the model (cf. Figures 1 and 3). Although iron is only
22
-------
toxic at relatively high concentrations—1 mg/l (USEPA 1976)—it was also important
in the prediction of laboratory toxicity from sediment chemistry. Without iron, the
toxic units model (including ammonia) accounted for just over 52% of the
variability of the laboratory toxicity test (Figure 4) as compared to 89% with both
iron and ammonia included (Figure 1). Without iron, the site with the highest
relative toxicity (i.e., Indiana Harbor site 7) was predicted to have about half its
measured toxicity (cf. Figures 1 and 4). And finally, replacing the bioavailable
fractions with whole-sediment concentrations of those contaminants that are
considered Bioaccumulative Contaminants of Concern by USEPA (1993-Table 5),
and assuming them to be completely bioavailable, also reduced the model's ability
to predict laboratory toxicity. Using only the bioaccumulatives in this fashion, the
model accounted for less than 70% of the variability in the laboratory toxicity tests
(Figure 5).
Benthic community structure
As described for laboratory toxicity, the best predictor of the mean tolerance
value for the benthic community was the full model of bioavailable contaminants,
which included both equilibrium partitioning for organics and AVS for metals . This
model accounted for 40% of the variability present in the mean tolerance value
(Figure 6). When bulk contaminant concentrations were substituted for
bioavailable fractions, the model accounted for only 34% of the variability (Figure
7). If only contaminants considered Bioaccumulative Contaminants of Concern by
23
-------
USEPA (Table 5) are used in the toxic units model, the model accounted for 40%
of the variability in the mean tolerance value for the benthic community (Figure 8).
The variability of the mean tolerance value of the benthic community had a
different set of relationships with ammonia and iron than did the laboratory toxicity
results. In contrast to what was observed for the laboratory toxicity tests,
unionized ammonia did not explain a significant amount of the variability in the
mean tolerance value of the benthic organisms (R2 = 0.04, P =0.56). Moreover,
without iron, the full model no longer accounted for a significant percentage of the
variability in the tolerance value (R2 = 0.11, P =0.31); for laboratory toxicity
tests, the level of significance declined but the full model could still predict the
response without iron (Figure 4). In fact, when iron is the only contaminant
considered, the model accounted for 40% of the variability in benthic community
structure, as it did when all contaminants were considered (Figure 9).
Two other important observations must be noted. First, when only the
taxonomic order of each organism, instead of the lowest level of taxonomic
identification possible (i.e. mostly genera and species), were used to calculate the
mean tolerance value for the benthic community, there were no significant
relationships with toxic units in any form. Second, even though we can predict
laboratory toxicity and benthic community mean tolerance using the toxic units
model, there was no correlation between laboratory toxicity and benthic
community mean tolerance (R2 = 0.21, P = 0.13).
24
-------
Final Rankings
The shortened list of laboratory toxicity indicators (i.e., from Priority Master
Station to Master Station indicators), in general, produced the same results as the
full set of data (R2 = 0.99, P = 0.0001; Figure 10). Further exclusion of
laboratory toxicity (except MicrotoxR) and benthic community data, and using only
the short list of sediment contaminant measures (i.e., reducing the Priority Master
Stations to Reconnaissance stations) resulted in both increases and decreases in
relative toxicity estimates within the set of sites (R2 = 0.75, P =0.0003; Figure
11). Eight sites, as opposed to only five for the Priority Master Station indicators,
had relative toxicity rankings greater than 40 when only Reconnaissance data were
used. Also, some sites switched position relative to other sites; for example,
Saginaw River site 3 was least toxic with only the Reconnaissance indicators
whereas it was sixth most toxic based on all Priority Master Station data.
Using only the data on bioaccumulative contaminants (i.e., ignoring laboratory
toxicity and benthic community measures) and assuming the complete
bioavailability of all bioaccumulative contaminants (by use of measured bulk
sediment concentrations found in the sediment) produced the most dramatic
change in relative toxicity of the sites (R2 = 0.75, P = 0.0003; Figure 12).
Relative to the most toxic site (i.e., Indiana Harbor site 7), the estimated toxicity of
all other sites decreased greatly. All but three sites that had relative hazard values
between 40 and 60 in the case of Priority Master Station data (including Indiana
Harbor site 3) had relative toxicities less than 12 when only bioaccumulatives were
25
-------
used. As a final note, even though we tried different measures to rank the relative
toxicity of sites, Indiana Harbor site 7 was always the most toxic, followed by
Saginaw River site 6.
DISCUSSION AND CONCLUSIONS
Effectiveness of the Toxic Units Model
Laboratory toxicity
Based upon our comparisons of different model permutations, the best estimate
of toxicity was produced by the model that incorporated bioavailability calculations
and included the fullest list of potential contaminants. Without estimates of
bioavailability, comparisons with toxicity data inherently assumed that all
contaminants are completely bioavailable, which greatly reduced model fit.
Conversely, without consideration of the toxic effects of contaminants that are
only toxic at relatively high concentrations, such as iron, it is inherently assumed
that such contaminants are completely nontoxic. The chronic AWQC for iron, for
example, is 1 mg/L (USEPA 1976); nevertheless, when iron was excluded the
estimated toxicity was also less effective than the full model at predicting observed
laboratory effects. Collectively, these findings corroborate previous studies
reporting that sediment toxicity, as defined by short-term tests such as those
incorporated in the ARCS program, is mediated by pore-water concentrations
(Hamelink et al. 1971, Ditoro et al. 1991), and that this toxicity is additive (Safe
1990).
26
-------
One possible reason for a lack of concordance between the bioaccumulatives
and toxicity test results may lie with the PCB component of the ARCS chemistry
data set, which is inherently weak. Analyses based on Aroclor8 mixtures, as
performed for the ARCS program, may not accurately represent the distribution or
abundance of toxic PCB congeners in weathered environmental mixtures (Tillitt et
al. 1992). Consequently, the TEF approach we used to estimate the toxicity of the
PCBs in ARCS sediments, which was based on the dioxin equivalency of
unweathered mixtures, may not have been accurate. Better estimates would be
obtained through instrumental analyses of individual AHH-active congeners (Kuehl
et al. 1991) or, for reconnaissance activities such as those conducted under the
ARCS program, through the use of biologically based measurements of dioxin-like
activity, as measured by the H4IIE in vitro assay (Tillitt et al. 1991).
Kinetic factors may also greatly affect the sensitivity of sediment toxicity tests
to bioaccumulative contaminants. Implicit in the comparison of toxicity test results
with equilibrium-based toxic units estimates is the assumption that the equilibrium
{or, at least, steady-state) conditions of the site from which the sediment was
collected are re-established at the time of the test. No measure of the degree to
which this assumption was satisfied was included in the ARCS data set, however,
and the rates of such processes may be relatively slow. For this and other
reasons, the uptake of sediment-bound, hydrophobic contaminants such as PCBs
and dioxins is also slow (Pruell et al. 1993). Moreover, delayed mortality may
occur (Mehrle et al. 1988). Collectively, these kinetic factors would tend to make
27
-------
short-term tests less responsive to contaminants controlled by partitioning and
other relatively slow processes than to others. Nevertheless, our modeling results
indicate a contribution by RGBs and other hydrophobic contaminants to laboratory-
measured toxicity.
Many of the environmental chemistry variables measured in the ARCS
sediments were highly intercorrelated, as were the toxicity test results. Such
intercorrelation impedes the assessment of causality. This was particularly true for
assessing the effects of iron in sediments. Because iron is soluble only in the
absence of oxygen {Drever 1982), it could be argued that toxicity attributable to
iron is actually caused by a lack of oxygen. In the ARCS study, however, and
contrary to our original suspicions, toxicity attributable to the high concentrations
of iron in many sediments appeared to be more than just an artifact or surrogate
measure of low dissolved oxygen concentrations. Most of the laboratory toxicity
tests were performed with reconstituted water, and the dissolved oxygen levels of
waters overlying the test sediments were monitored regularly and found to be
acceptable in most tests {Nelson et al. 1993). Thus, the improved relationship
between laboratory toxicity and toxic units estimates with the inclusion of iron
seemed to be directly related to the actual toxicity of the iron which, at the high
concentrations present in some ARCS sediments (Tables 6-8), is not surprising.
Although ammonia was present at potentially toxic concentrations at most
sites, the contribution to total toxicity of ammonia in the toxic units model was
assessed on the basis of unionized ammonia concentrations and pH measured
28
-------
regularly during the conduct of laboratory toxicity tests. In these tests, the
overlying water was reconstituted water and not ambient water. We are not sure
how accurately the measurements of ammonia and pH incorporated into the toxic
units model represent the ammonia actually present in the natural system prior to
sediment collection. To make the estimates of toxicity attributable to ammonia
representative of the natural system, the ammonia and pH concentrations in the
water overlying the sediment within the natural system should be measured. The
same is true for dissolved oxygen, which should also be measured under ambient
conditions. Despite the presence of ammonia at potentially toxic levels and its
potential to influence the results of the laboratory toxicity tests, ammonia alone
was not able to effectively predict the toxicity of the sediments, as noted earlier.
Our findings support the hypothesis that benthic toxicity, as measured by
short-term toxicity tests, is mediated by pore-water contaminant concentrations, as
has been suggested by others for some time (e.g., Hamelink et al. 1971; DiToro et
al. 1990, DiToro et al. 1991). Ecosystem risk, however, has much broader
implications, especially for bioaccumulative contaminants, where trophic transfers
predominate. With our findings, we have shown that the short-term laboratory
toxicity tests are sensitive to iron and ammonia, responses to which may mask the
effects of slow-acting, bioaccumulative contaminants. The short-term laboratory
toxicity tests, by themselves, appear to reflect toxicity of the sediments in the
immediate vicinity from which they are collected, but not necessarily the
ecosystem risk represented by the transfer of the contaminants they contain
29
-------
through the food chain.
Benthic community structure
In the ARCS data, the benthic community structure of contaminated sediments
was related to the toxicity of the contaminants present as defined by the toxic
units approach. As the estimated toxic units of a sediment increased the mean
tolerance to pollution of the benthos community increased. Like the relationship
between toxic units and laboratory toxicity test results, the strength of the
relationship between toxic units estimates and mean tolerance to pollution declined
when, instead of estimated bioavailable concentrations, bulk sediment
concentrations of contaminants were used in the toxic units estimates. Unlike the
work with laboratory toxicity and toxic units, bioavailable iron toxic units alone had
as strong a relationship with mean tolerance to pollution as any other toxic units
estimates.
The differences in the observed strength of the relationships among sediment
community tolerance, laboratory toxicity test results, and toxic units estimates
seems to be an indicator of an information short fall. The much stronger
relationship observed between laboratory toxicity and toxic units than between
sediment community tolerance and toxic units, seems to indicate an information
short fall related to sediment community identification. The observed increased
strength of the relationship between toxic units and benthic community tolerance
with a finer scale of resolution in taxa identification is an indicator of one possible
30
-------
way to improve the strength of the relationship. In the ARCS data, the greatest
percentage of organisms fell into the category of unidentifiable Tubificidae (i.e.
family level identification) (Canfield et al. 1993). From Lenat's (1993) tolerance
values, it is apparent that there is a wide range of tolerances among Tubificidae
species. It is possible that a finer level of identification (i.e. genera and/or species)
would have strengthened the relationship between toxic units estimates and
benthic community tolerance.
Another possible contributing factor to the differences in the strength of the
relationships among sediment community tolerance, laboratory toxicity test results,
and toxic units estimates is that the sediment community tolerance measures a
different aspect of contamination than the other two measures. The much
stronger relationship between laboratory toxicity and toxic units than between
sediment community tolerance and toxic units may be that the sediment
community tolerance is more of a long-term measure of effect than the other two
measures. This is speculation but it requires further evaluation before it can either
be accepted or reject as a cause.
Overall, our results are an indication that the toxic units model is an effective
tool for understanding benthic community structure in relationship to complex
contaminant mixtures, and that Lenat's (1993) biotic index is a good measure of
the effects of pollution, in general, on benthic community structure.
31
-------
Data quality
Important in the application of a toxic units approach is the consistency and
quality of the data. To make accurate environmental-chemistry based comparisons
of sediment samples, based on environmental chemistry, the concentrations of any
given analyte must be measured with the same sensitivity among the samples to
be compared; the censoring level (i.e., detection limit) for any given analyte must
be consistent (preferably constant) among samples. If this is not the situation,
differences among samples or sites may be controlled by analytical sensitivity
rather than true differences among the samples (e.g., two identical samples may
appear different if represented by differing detection limit values) which
necessitated our elimination of most of the Saginaw River ARCS data from
consideration. The best way to avoid the influence of detection limits on toxic
units estimates is to define a minimum level of sensitivity for each analyte (i.e., a
performance-based criterion). Moreover, this minimum level of acceptable
sensitivity should be determined by the level of concern for the analyte. Defining a
minimum level of sensitivity for contaminant measurements is only the first step in
addressing the problem of detection limits associated with contaminant
measurements, the second being how to treat those censored values that
inevitably occur. The most environmentally conservative approach would use all
values, including censored values in the toxic units model. Because of the expense
associated with remediation, the high toxic units estimated using this approach
would encourage consistently high sensitivity.
32
-------
Potential for mode! enhancements
The methods we used to estimate the bioavailable portions of measured
contaminants assume solid-phase control by organic carbon for organic compounds
and sulfide for metals and arsenic (DiToro et al. 1990, DiToro et at. 1991). We
made no attempt to account for other mechanisms, such as sorption or
complexation by other organic or inorganic ligands that might be present in the
pore waters. We also assumed complete insolubility of sulfides, an obvious over
simplification. Consequently, the models do not predict toxicity of all
contaminants at all times {Ankley et al. 1993); nevertheless, they represent the
current best estimate of the potential toxicity of sediment contaminants. The toxic
units approach incorporated into the model is also versatile; as better methods for
estimating the bioavailability of sediment contaminants are developed and water
quality standards are revised, such refinements can be incorporated into the
modeling process.
It is important to note that if chronic toxicity standards for sediment quality
were available, they and actual measured sediment concentrations would be used
in the toxic units model instead of AWQC and estimates of pore-water
concentrations, respectively. In fact, the proposed Sediment Quality Thresholds
for organic chemicals (USEPA 1985a) are founded on estimated pore-water
concentrations, equilibrium partitioning, and the AWQC. In the absence of
published sediment criteria for all contaminants, however, AWQC and estimation of
the pore-water concentration of individual analytes remains the best available
33
-------
method for estimating toxic units.
The toxic units approach assumes that each contaminant is independent of all
other contaminants--!.e., effects are strictly additive. The model does not account
for synergistic or antagonist interactions that may occur among contaminants in
complex mixtures, even though such interactions are well documented (e.g.
Warren 1971, de March 1988, Parrott and Sprague 1993, Schmitt et al. 1993).
As information becomes available on the nature of the contradictions to this
assumption, quantifiable interactions can be incorporated into the model, along
with improvements in the bioavailability estimates. As noted earlier, the model
also presently assumes that the bioavailability of divalent metals and arsenic is
controlled by AVS. Future model improvements should also include incorporation
of other inorganic and organic ligands, perhaps through speciation modeling, to
estimate the pore-water concentrations of the divalent ions.
Yet another assumption inherent in our approach is that all endpoints in all
toxicity tests conducted for the ARCS program are equivalent. We used this
approach because it is analogous to the procedure used by USEPA (1986a) to
derive the AWQC-upon which the toxic units are based-wherein toxicity test
results from a broad suite of taxa are averaged to obtain mean acute and chronic
values for individual contaminants. Given the equally broad array of contaminants
present in the ARCS sediments, we felt that such an unbiased approach was
prudent. We recognize, however, that subsequent evaluation of the model may
suggest that certain test results should be weighted more heavily than others.
34
-------
The toxic units approach needs further validation with other data sets spanning
a broader range of conditions. The ARCS data set from which the approach was
developed contained no sediments that were not toxic to some degree to all the
organisms tested. In this current evaluation of the toxic units approach, the lack
of sediments containing only background contaminant levels and little to no
observed toxicity allows us to state only that the approach is a good measure of
relative toxicity among sediment samples. Full evaluation of the effectiveness of
the toxic units approach to predict sediment toxicity will require a sediment data
base containing sediments with a full range of toxicity (i.e., nontoxic to extremely
toxic) resulting from a wide variety of contaminants.
Ranking Approach
From our analysis, we found that it is very important how much and what type
of information is included in the hazard ranking of a site. As the amount of
information incorporated into the ranking process declined (i.e.. Priority Master
Station indicators to Master Station indicators to Reconnaissance Station
indicators) the hazard ranking of the sites changed. As the information used to
rank the sites changed so did both the relative hazard of the sites to the most
hazardous site and the order of the sites based on relative hazard. More
dramatically, when the information used to estimate the relative hazards of the
sites included only the bioaccumulative compounds, only three of the six sites
estimated to be the most hazardous sites using the full set of information (i.e.,
35
-------
relative ranks over 30) still had relative ranks above 30. The most dramatic
change occurred in the relative ranking of site 3 in Indiana Harbor; it went from a
relative hazard rank greater than 57 using the full set of information to less than
12 using only bioaccumulatives.
The minimal amount of information necessary for an appropriate assessment of
the relative hazard among a set of sites seems to be the Master Station level (i.e.,
chemistry, moderate range of laboratory toxicity data, and benthic community
structure). The information gained by the inclusion of macrophytes and toxicity
tests with more species of invertebrates (i.e.. Priority Master Stations) had little to
no effect on relative hazard ranking of the sites. If only Microtox and a short list of
contaminants (i.e.. Reconnaissance Station data) are used to produce the relative
hazard ranking for a set of sites, the relative ranking of the sites becomes
confused.
Not only the amount of information, as discussed, but also the type of
information used to construct the relative hazard ranking of a set of sites may
dramatically affect the resulting hazard ranking. If the intent of using the ranking
process is to determine overall relative toxicity, the set of information used should
include all the types of information collected at Master Stations in this study (i.e.,
chemistry, laboratory toxicity, and benthic community structure). If the intent of
using the ranking process is to determine hazard to organisms on the upper end of
the aquatic food chain, including wildlife, the set of information would be similar to
the bioaccumulatives tested in this study. Thus, it is crucial that the question of
36
-------
concern for a set of sites be correctly formulated before such a relative ranking
process is employed.
The purpose of the described ranking process is to allow different types of
data, measured on different scales, to be combined into one overall estimate of
relative hazard for the set of contaminated sites under investigation. The scaling
done for each class of data (i.e., chemistry, laboratory toxicity, benthic
community) allows for the incorporation into the estimates of relative hazard as
much information as is available in the scientific literature. The result is a current
best estimate of relative hazard for the sites under investigation. This approach
enables the comparison and combination of sediment contamination information,
measured on different scales, on one relative scale that has a foundation in
environmental chemistry, toxicology, and ecology. The process is alsp dynamic; as
more information becomes available about sediment processes, chemical fates,
toxicity, etc., new information can be incorporated into the ranking model. Thus,
the estimates of relative hazard become more accurate as the base of knowledge
increases. It can also become a planning tool by pointing out where information is
most needed.
If not already apparent, the relative site rankings generated in this study are
based on contaminant levels at one point in space. The next step would be to
adjust (i.e., weight) the toxicity of each site based on an estimate of the area (or
volume) of contaminated sediment at each site. Thus, the single point relative
hazard ranking of sites may change dramatically if less toxic sites represent a
37
-------
greater proportional area (or volume).
Finally, the ranking process need not be limited to the types of data described
here. Other classes of information (i.e., potential for resuspension, aesthetics,
recreational potential, etc.) can also be incorporated by scaling the observed values
from 1 to 100 as was done with the information presented in this report. The
model can also be extended to other quantifiable hazards, such as carcinogenicity,
by defining a toxic unit to be the level of concern for such hazards. Again, this
demonstrates the general utility of the ranking process as one way of assessing
the relative hazard among many sites when limited resources require prioritization.
ACKNOWLEDGMENTS
This study was jointly undertaken by the USEPA, Great Lakes Program Office
(GLNPO) and the U.S Department of the Interior, Fish and Wildlife Service, National
Fisheries Contaminant Research Center (NFCRC), now part of the National
Biological Survey. We thank M. Tuchman and R. Fox of GLNPO, W. Lick and J.
DePinto of the ARCS modeling workgroup, R. Kreis and J. Rathbun of the USEPA
Large Lakes Research Station, P. Landrum of the National Oceanic and
Atmospheric Administration, G. Hurlburt of the Michigan Department of Natural
Resources, M. Miah and J. Boyd of Lockheed Engineering and Sciences Company,
B. Schumacher of USEPA Environmental Monitoring Laboratory-Las Vegas, and J.
Huckins, L. Jacobs, D. Tillitt, C. Ingersoll, and T. Canfield of the NFCRC for their
assistance.
38
-------
REFERENCES
Ankley, G.T., V.R. Mattson, E.N. Leonard, C.W. West, and J.L. Bennett, 1993.
Predicting the acute toxicity of copper in freshwater sediments: evaluation
of the role of acid-volatile sulfide. Environ. Toxicol. Chem. 12:315-320.
Canfield, T.J., La Point, T.W., M.C. Swift, G.A. Burton, J.A. Fairchild, and N. E.
Kemble. 1993. Benthic community structure evaluations. In: NFCRC-Battelle
final report for the USEPA GLNPO Assessment and Remediation of
Contaminated Sediments (ARCS) project: biological and chemical
assessment of contaminated Great Lakes sediment. C. G. Ingersoll, D. R.
Buckler, E. A. Crecelius, and T. W. La Point (eds). U.S. Fish and Wildlife
Service, Columbia, Missouri.
de March, B.G.E. 1988. Acute toxicity of binary mixtures of five cations (Cu2 + ,
Cd2+, Zn2 + , Mg2+, and K + ) to the freshwater amphipod Gammarus
lacustris (Sars): alternative descriptive models. Can. J. Fish. Aquat. Sci.
45:625-633.
DiToro, D. M., J. D. Mahony, D. J. Hansen, K. J. Scott, M. B. Hicks, S. M. Mayr,
and M. S. Redmond. 1990. Toxicity of cadmium in sediments: the role of
acid volatile sulfide. Environ. Toxicol. Chem. 9:1487-1502.
39
-------
DiToro, D.M., C.S. Zarba, D.J. Hansen, W.J. Berry, B.C. Swartz, C.E. Cowan, S.P.
Pavlou, H.E. Allen, IM.A. Thomas and P.R. Paquin, 1991. Technical basis for
establishing sediment quality criteria for non-ionic organic chemicals by using
equilibrium partitioning. Environ. Toxicol. Chem. 10:1541-1583.
Drever, J.I. 1982. The geochemistry of natural waters. Prentice-Hall, Inc.,
Englewood Cliffs, N.J. 07632.
Hamelink, J.L., R.C. Waybrandt, and R.C. Ball. 1971. A proposal: exchange
equilibria control the degree to which chlorinated hydrocarbons are
biologically magnified in lentic environments. Trans. Am. Fish. Soc.
100:207-214.
Hilsenhoff, W. L. 1988. Rapid field assessment of organic pollution with a family
level biotic index. J. IM. Am. Benthol. Soc. 7(1):65-68.
Ingersoll, C. G., D. R. Buckler, E. A. Crecelius, and T. W. La Point. 1993. IMFCRC-
Battelle final report for the USEPA GLIMPO Assessment and Remediation of
Contaminated Sediments (ARCS) project: biological and chemical
assessment of contaminated Great Lakes sediment. U.S. Environmental
Protection Agency, Great Lakes National Program Office, Chicago, IL.
40
-------
Kreis, R. G. 1988. Integrated study of exposure and biological effects of in-place
sediment pollutants in the Detroit River, Michigan: an upper Great Lakes
connecting channel. Final report. U.S. Environmental Protection Agency,
Office of Research and Development, ERL-Duluth, MM, and LLRS-Grosse He,
Ml. 153p.
Kreis, R. G. 1989. Numerical ranking of hazardous sediments to prioritize sites for
remedial action. U.S. Environmental Protection Agency, Environmental
Research Laboratory-Duluth, Large Lakes Research Station, Grosse lie. Ml.
28p.
Kuehl, D.W., B.C. Butterworth, J. Libal, and P. Marquis. 1991. An isotope dilution
high resolution mass spectrometric method for the determination of coplanar
polychlorinated biphenyls: application to fish and marine mammals.
Chemosphere 22:849-858.
LeBlanc, G.A., 1980. Acute toxicity of priority pollutants to water flea (Daphnia
magna). Bull. Environ. Contam. Toxicol. 24(5):684-691.
Lemly, A.D., S.E. Finger and M.K. Nelson, 1993. Sources and Impacts of Irrigation
drainwater contaminants in arid wetlands. Environ. Toxicol. Chem. 12:2265-
2279.
41
-------
Lenat, D.R., 1993. A biotic index for the southeastern United States: derivation
and list of tolerance values, with criteria for assigning water-quality ratings.
J.N. Am. Benthol. Soc. 12(3):279-290.
Maguire, R.J., J.H. Carey and E.J. Hale, 1983. Degradation of the tri-n-butyltin
species in water. J. Agric. Food Chem. 31(5):1060-1065.
Mehrle, P.M., D.R. Buckler, E.E. Little, L.M. Smith, J.D. Petty, P.M. Peterman, D.L.
Stalling, G.M. Graeve, J.J. Coyle, and W.J. Adams. 1988. Toxicity and
bioconcentration of 2,3,7,8-tetrachloro-dibenzodioxin and 2,3,7,8-
tetrachloro-dibenzofuran in rainbow trout. Environ. Toxicol. Chem. 7:47-62.
Michigan Department of Natural Resources (DNR). 1993. Rule 57(2) guideline
levels. Surface Water Quality Division, Lansing, Michigan.
Nelson, M. K., J. J. Coyle, L. B. King, I\l. E. Kemble, E. A. Crecelius, and I. E.
Greer. 1993. Whole sediment toxicity tests. In: NFCRC-Battelle final report
for the USEPA GLNPO Assessment and Remediation of Contaminated
Sediments (ARCS) project: biological and chemical assessment of
contaminated Great Lakes sediment. C. G. Ingersoll, D. R. Buckler, E. A.
Crecelius, and T. W. La Point (eds). U.S. Fish and Wildlife Service,
Columbia, Missouri.
42
-------
Newsted, J.L. and J.P. Giesy, 1987. Predictive models for photoinduced acute
toxicity of polycyclic aromatic hydrocarbons to Daphnia magna. Strauss
(Cladocera, Crustacea). Environ. Toxicol. Chem. 6(6):445-461.
Parrott, J.L., and J.B. Sprague. 1993. Patterns of toxicity of sublethal mixtures of
metals and organic chemicals determined by Microtox8 and by DIMA, RNA,
and protein content of fathead minnows (Pimephales promelas). Can. J.
Fish. Aquat. Sci. 50:2245-2253.
Pruell, R.J., N.I. Rubinstein, B.K. Taplin, J.A. Livolsi, and R.D. Bowen. 1993.
Accumulation of polychlorinated organic contaminants from sediment by
three benthic marine species. Arch. Environ. Contam. Toxicol. 24:290-297.
Safe, S., 1990. Polychlorinated biphenyls (PCBs), dibenzo-p-dioxins (PCDDs),
dibenzofurans (PCDFs), and related compounds: environmental and
mechanistic considerations which support the development of toxic
equivalency factors (TEFs). CRC Crit. Rev. Toxicol. 21, 51-88.
43
-------
Schmitt, C. J., M. L. Wildhaber, J. B. Hunn, T. Nash, M. N. Tieger, and B. L.
Steadman. 1993. Biomonitoring of lead-contaminated Missouri streams with
an assay for erythrocyte d-aminolevulinic acid dehydratase activity in fish
blood. Archives of Environmental Contamination and Toxicology 25:464-
475.
Sijm, D.T.H.M., H. Wever and P.J. DeVries, 1989. Octan-1-ol/water partition
coefficients of polychlorinated dibenzo-p-dioxins and dibenzofurans:
experimental values determined with a stirring method. Chemosphere
19:263-266.
Smith, L.M., T.R. Schwartz and K. Feltz, 1990. Determination and occurrence of
AHH-active polychlorinated biphenyls, 2,3,7,8-tetrachloro-p-dioxin and
2,3,7,8-tetrachlorodibenzofuran in Lake Michigan sediment and biota. The
question of their relative toxicological significance. Chemosphere
21(9):1063-1085.
Sprague, J. B. and B. A. Ramsay. 1965. Lethal levels of mixed copper-zinc
solutions for juvenile salmon. J. Fish. Res. Bd. Canada 22(2):425-432.
Statistical Analysis System. 1990. Version 6.08 for OS/2. SAS Institute
Incorporated, Gary, IMC.
44
-------
Tillitt, D.E., G.T. Ankley, and J.P. Giesy. 1991. Characterization of the H4IIE rat
hepatoma cell bioassay as a tool for assessing toxic potency of planar
halogenated hydrocarbons in environmental samples. Environ. Sci. Technol.
25:87-92.
Tillitt, D.E., G.T. Ankley, J.P. Giesy, J.P. Ludwig, H. Kurita-Matsuba, D.V.
Weseloh, P.S. Ross, C. Bishop, L. Sileo, K.L. Stromberg, J. Larson, and T.J.
Kubiak. 1992. Polychlorinated biphenyl residues and egg mortality in double-
crested cormorants from the Great Lakes. Environ. Toxicol. Chem. 11:1281-
1288.
U.S. Environmental Protection Agency. 1993. ASTER data base. ERL-Duluth,
Duluth, Minnesota.
U.S. Environmental Protection Agency. 1976. Quality criteria for water. EPA
440/9-76-023. Washington DC.
U.S. Environmental Protection Agency. 1980a. Ambient water quality criteria for:
polychlorinated biphenyls (PCBs). EPA 440/5-80-068. Washington DC.
U.S. Environmental Protection Agency. 1980b. ambient water quality criteria for:
Aldrin/Dieldrin. EPA 440/5-80-019. Washington DC.
45
-------
U.S. Environmental Protection Agency. 1980c. Ambient water quality criteria for:
chlordane. EPA 440/5-80-027. Washington DC.
U.S. Environmental Protection Agency. 1980d. Ambient water quality criteria for:
DDT. EPA 440/5-80-038. Washington DC.
U.S. Environmental Protection Agency. 1980e. Ambient water quality criteria for:
endosulfan. EPA 440/5-80-046. Washington DC.
U.S. Environmental Protection Agency. 1980f. Ambient water quality criteria for:
endrin. EPA 440/5-80-047. Washington DC.
U.S. Environmental Protection Agency. 1980g. Ambient water quality criteria for:
heptachlor. EPA 440/5-80-052. Washington DC.
U.S. Environmental Protection Agency. 1980h. Ambient water quality criteria for:
hexachlorocyclohexane. EPA 440/5-80-054. Washington DC.
U.S. Environmental Protection Agency. 1980i. Ambient water quality criteria for:
phthalate esters. EPA 440/5-80-067. Washington DC.
46
-------
U.S. Environmental Protection Agency. 1980J. Ambient water quality criteria for:
copper. EPA 440/5-80-036. Washington DC.
U.S. Environmental Protection Agency. 1980k. Ambient water quality criteria for:
mercury. EPA 440/5-80-058. Washington DC.
U.S. Environmental Protection Agency. 1984a. Ambient water quality criteria for:
arsenic. EPA 440/5-84-037. Washington DC.
U.S. Environmental Protection Agency. 1984b. Ambient water quality criteria for:
cadmium. EPA 440/5-84-032. Washington DC.
U.S. Environmental Protection Agency. 1984c. Ambient water quality criteria for:
chromium. EPA 440/5-84-029. Washington DC.
U.S. Environmental Protection Agency. 1984d. Ambient water quality criteria for:
lead. EPA 440/5-84-027. Washington DC.
U.S. Environmental Protection Agency. 1984e. Ambient water quality criteria for:
mercury. EPA 440/5-80-058. Washington DC
47
-------
U.S. Environmental Protection Agency. 1985a. National prospective on sediment
quality. Office of Water, Washington D.C.
U.S. Environmental Protection Agency. 1985b. Ambient water quality criteria for
ammonia. EPA 440/5-85-001.
U.S. Environmental Protection Agency. 1986a. Ambient water quality criteria. EPA
440/5-86-001. Washington DC.
U.S. Environmental Protection Agency. 1986b. Ambient water quality criteria for:
toxaphene. EPA 440/5-86-006. Washington DC.
U.S. Environmental Protection Agency. 1986c. Ambient water quality criteria for:
nickel. EPA 440/5-86-004. Washington DC.
U.S. Environmental Protection Agency, 1987a. Processes, coefficients, and models
for simulating toxic organics and heavy metals in surface waters. EPA
600/3-87-015. Washington DC.
U.S. Environmental Protection Agency, 1987b. Ambient water quality criteria for:
zinc. EPA 440/5-87-003. Washington DC.
48
-------
U.S. Environmental Protection Agency. 1989. Rapid bioassessment protocols for
use in streams and rivers: benthic macroinvertebrates and fish. EPA 440/4-
89-001. Washington, DC.
U.S. Environmental Protection Agency. 1990. Assessment and Remediation of
Contaminated Sediments (ARCS) work plan. U.S. Environmental Protection
Agency, Great Lakes Program Office, Chicago, IL. 59 p.
U.S. Environmental Protection Agency. 1993. Proposed water quality guidance for
the Great Lakes system. U.S. Environmental Protection Agency, Chicago, IL.
Warren, C. E. 1971. Biology and water pollution control. W. B. Saunders
Company, Philadelphia, PA.
Weast, R. C., M. J. Astle, and W. H. Beyer. 1988. CRC handbook of chemistry
and physics. CRC Press, Inc., Boca Raton, FL.
49
-------
Table 1. Octanol/water partition coefficients used in the toxic units model for the
contaminants measured in ARCS sediments.
Contaminant
Octanol/water
Partition
Coefficient
(Log(K0J)
Reference
Organochlorine compounds
Chlorodioxins
2,3,7,8- 6.42
Tetrachlorodibenzodioxin
1,2,3,7,8- 6.64
Pentachlorodibenzodioxin
1,2,3,4,7,8- 7.80
Hexachlorodibenzodioxin
1,2,3,6,7,8- 7.80
Hexachlorodibenzodioxin
1,2,3,7,8,9- 7.80
Hexachlorodibenzodioxin
1,2,3,4,6,7,8- 8.00
Heptachlorodibenzodioxin
Octachlorodibenzodioxin 8.00
Chlorodibenzofurans
2,3,7,8- 6.53
Tetrachlorodibenzofuran
1,2,3,7,8- 6.79
Pentachlorodibenzofuran
2,3,4,7,8- 6.92
Pentachlorodibenzofuran
1,2,3,4,7,8- 7.80
Hexachlorodibenzofuran
Sijm et al. 1989
Sijm et al. 1989
Sijm et al. 1989
1,2,3,4,7,8-Hexachlorodi-
benzodioxin-Sijm et al.
1989
1,2,3,4,7,8-Hexachlorodi-
benzodioxin-Sijm et al.
1989
Sijm et al. 1989
Sijm et al. 1989
Sijm et al. 1989
Sijm et al. 1989
Sijm et al. 1989
1,2,3,4,7,8-Hexachlorodi-
benzodioxin—Sijm et al.
1989
50
-------
Contaminant
Octanol/water
Partition
Coefficient
(Log(K0J)
Reference
1,2,3,6,7,8-
Hexachlorodibenzofuran
2,3,4,6,7,8-
Hexachlorodibenzofuran
1,2,3,7,8,9-
Hexachlorodibenzofuran
1,2,3,4,6,7,8-
Heptachlorodibenzofuran
1,2,3,4,7,8,9-
Heptachlorodibenzofuran
Octachlorodibenzofuran
Chlorobenzenes
1,2-Dichlorobenzene
1,3-Dichlorobenzene
1,4-Dichlorobenzene
Polychlorinated biphenyls
Aroclor6 1016
Aroclor6 1221
Aroclor6 1232
Aroclor6 1242
Aroclor6 1248
Aroclor6 1254
Aroclor6 1260
Pesticides
Aldrin
xis-chlordane
7.80
7.80
7.80
7.92
7.92
7.97
3.483 (n = 6)
3.507 (n = 6)
3.455 (n = 8)
5.58 (n = 3)
4.045 (n = 2)
3.87 (n = 2)
4.845 (n = 2}
5.933 (n = 4)
6.123 (n - 3)
6.80 (n = 3)
5.30
4.13 (n = 2)
1,2,3,4,7,8-Hexachlorodi-
benzodioxin—Sijm et al.
1989
1,2,3,4,7,8-Hexachlorodi-
benzodioxin--Sijm et al.
1989
1,2,3,4,7,8-Hexachlorodi-
benzodioxin-Sijm et al.
1989
Sijm et al. 1989
Sijm et al. 1989
Sijm et al. 1989
USEPA 1987a
USEPA 1987a
USEPA 1987a
USEPA
USEPA
USEPA
USEPA
USEPA
USEPA
USEPA
1987a
1987a
1987a
1987a
1987a
1987a
1987a
USEPA 1987a
Chlordane USEPA 1987a
51
-------
Contaminant
Octanol/water
Partition
Coefficient
(Log(K0J)
Reference
Trans-chlordane
P_,p_-DDD
p.,fi-DDE
Dieldrin
a-Endosulfan
yff-Endosulfan
Endosulfan-sulfate
Endrin
Endrin aldehyde
Endrin ketone
Heptachlor
Heptachlor epoxide
K-Hexachlorocyclohexane
(Lindane)
Methyoxychlor
Toxaphene
Polycyclic aromatic
Compounds
Acenaphthylene
Anthracene
Benz(a)anthracene
Benzo(a)pyrene
Benzo(cj,h,j)perylene
Benzo(k)fluoranthene
Chrysene
4.13 (n = 2}
6.11 (n = 2)
6.013 (n = 4)
5.695 (n = 6)
3.54
-1.70
-1.70
-1.30
4.827 (n = 3)
3.15
3.15
4.41
2.65
3.573 (n = 3)
4.24 (n - 2)
3.30 (n = 2)
3.895 (n = 2)
4.477 (n = 6)
5.61 (n = 2)
5.383 (n = 3)
6.863 (n = 3)
6.45 (n = 2)
5.71 (n = 3)
Chlordane USEPA 1987a
USEPA 1987a
USEPA 1987a
USEPA 1987a
USEPA 1987a
USEPA 1987a
USEPA 1987a
USEPA 1987a
USEPA 1987a
USEPA 1987a
Endrin aldehyde USEPA
1987a
USEPA 1987a
USEPA 1987a
USEPA 1987a
USEPA 1987a
USEPA 1987a
USEPA
USEPA
USEPA
USEPA
USEPA
USEPA
USEPA
1987a
1987a
1987a
1987a
1987a
1987a
1987a
52
-------
Contaminant
Octanol/water
Partition
Coefficient
(Log(K0J)
Reference
Dibenzofuran
Fluoranthene
Fluorene
lndeno(1,2,3-c,d)pyrene
Napthalene
Phenanthrene
Pyrene
4
5
4
7
3
4
5
.09
.150
.118
.085
.341
.524
.119
(n
(n
(D
(n
(n
(n
= 3)
= 4)
= 2)
= 7)
= 5)
= 8)
USEPA
1993
USEPA
USEPA
USEPA
USEPA
USEPA
USEPA
ASTER database
1
1
1
1
1
1
987a
987a
987a
987a
987a
987a
Phthalate esters
Bis(2-ethylhexyl)phthalate
Butyl benzyl phthalate
Dimethyl phthalate
Organo-metals
Tributyltin
9.17 (n = 2)
5.556 (n = 4)
1.763 (n = 3)
3.2
USEPA 1987a
USEPA 1987a
USEPA 1987a
Maguire et al. 1983
*N equals number of values averaged (arithmetic mean) to produce the indicated
value.
53
-------
Table 2. Estimation of the chronically toxic pore-water concentration defining a toxic unit for each contaminant
used in the toxic units model. "MT" minimal or no observed toxicity.
Contaminant
Toxic Unit Calculations (ug/L)
Reference
Organochlorine Compounds
Chlorodioxins
2,3,7,8-Tetrachlorodibenzodioxin
1,2,3,7,8-Pentachlorodibenzodioxin
1,2,3,4,7,8-Hexachlorodibenzodioxin
1,2,3,6,7,8-Hexachlorodibenzodioxin
1,2,3,7,8,9-Hexachlorodibenzodioxin
1,2,3,4,6,7,8-Heptachlorodibenzodioxin
Octachlorodibenzodioxin
Chlorodibenzofurans
2,3,7,8-Tetrachlorodibenzofuran
0.0141 x 0.00000992 = 0.0000376
0.0141 x (0.0000099/0.0087)3
= 0.0000160
0.0141 x (0.0000099/0.045)3
= 0.00000308
0.0141 x (0.0000099/0.004)3
= 0.0000348
0.0141 x (0.0000099/0.0037)3
= 0.0000376
0.0141 x (0.0000099/0.0028)3
= 0.0000495
0.0141 x (0.0000099/0.1)4
= 0.00000139
0.0141 x (0.0000099/0.025)3
= 0.00000556
USEPA 1980a and Smith et
al. 1990
USEPA 1980a and Smith et
al. 1990
USEPA 1980a and Smith et
al. 1990
USEPA 1980a and Smith et
al. 1990
USEPA 1980a and Smith et
al. 1990
USEPA 1980a and Smith et
al. 1990
USEPA 1980a, Smith et al.
1990, and Safe 1990
USEPA 1980a and Smith et
al. 1990
54
-------
Contaminant
Toxic Unit Calculations (ug/L)
Reference
1,2,3,7,8-Pentachlorodibenzofuran
2,3,4,7,8-Pentachlorodibenzofuran
1,2,3,4,7,8-Hexachlorodibenzofuran
1,2,3,6,7,8-Hexachlorodibenzofuran
1,2,3,7,8,9-Hexachlorodibenzofuran
2,3,4,6,7,8-Hexachlorodibenzofuran
1,2,3,4,6,7,8-Heptachlorodibenzofuran
1,2,3,4,7,8,9-Heptachlorodibenzofuran
Octachlorodibenzofuran
Chlorobenzenes
1,2-Dichlorobenzene
1,3-Dichlorobenzene
1,4-Dichlorobenzene
0.0141 x (0.0000099/0.038)3
= 0.00000366
0.0141 x (0.0000099/0.38)3
= 0.000000366
0.0141 x (0.0000099/0.27)3
= 0.000000515
0.0141 x (0.0000099/0.065)3
= 0.00000214
0.0141 x (0.0000099/0.1)
= 0.00000139
0.0141 x (0.0000099/0.14)3
= 0.000000993
0.0141 x (0.0000099/0.1 )4
= 0.00000139
0.0141 x (0.0000099/0.1)4
= 0.00000139
0.0141 x (0.0000099/0.1)4
= 0.00000139
7.0
180.0
43.0
USEPA 1980a and Smith et
al. 1990
USEPA 1980a and Smith et
al. 1990
USEPA 1980a and Smith et
al. 1990
USEPA 1980a and Smith et
al. 1990
USEPA 1980a, Smith et al.
1990, and Safe 1990
USEPA 1980a and Smith et
al. 1990
USEPA 1980a, Smith et al.
1990, and Safe 1990
USEPA 1980a, Smith et al.
1990, and Safe 1990
USEPA 1980a, Smith et al.
1990, and Safe 1990
Michigan DNR 1993
Michigan DNR 1993
Michigan DNR 1993
55
-------
Contaminant
Toxic Unit Calculations (ug/L)
Reference
Polychlorinated byphenyls (PCBs)
Aroclor6 1016
Aroclor6 1221
Aroclor6 1232
Aroclor6 1242
Aroclor6 1248
Aroclor6 1254
Aroclor6 1260
Pesticides
Aldrin
cis-Chlordane
trans-Chlordane
p.,p/-DDE
fi,fi'-DDT
Dieldrin
a-Endosulfan
MT
MT
0.0141 x (0.0000099/ 0.001934)3
= 0.0000717
0.0141 x (0.0000099/0.0000137)3
= 0.0101
0.0141 x (0.0000099/0.0000173)3
= 0.00801
0.014
MT
0.0019
0.0043
0.0043
0.001
0.001
0.001
0.0019
0.056
Smith et al. 1990
Smith et al. 1990
USEPA 1980a and Smith et
al. 1990
USEPA 1980a and Smith et
al. 1990
USEPA 1980a and Smith et
al. 1990
USEPA 1980a
Smith et al. 1990
Dieldrin USEPA 1980b
Chlordane USEPA 1980c
Chlordane USEPA 1980c
DDT USEPA 1980d
DDT USEPA 1980d
USEPA 1980d
USEPA 1980b
Endosulfan USEPA 1980e
56
-------
Contaminant
Toxic Unit Calculations (ug/L)
Reference
/ff-Endosulfan
Endosulfan sulfate
Endrin
Endrin aldehyde
Endrin ketone
Heptachlor
Heptachlor epoxide
K-Hexachlorocyclohexane (Lindane)
Methyoxychlor
Toxaphene
Polycyclic aromatic hydrocarbons (PAHs)
Acenaphthylene
Anthracene
Benz(a)anthracene
Benzo(a)pyrene
0.056
0.056
0.0023
0.0023
0.0023
0.0038
0.0038
0.08
0.03
0.0002
24000 ug/kg5/(0.046 x Koc7}
44000 ug/kg5/(0.046 x Koc7)
220000 ug/kg5/(0.046 x Koc7)
1800000 ug/kg5/(0.046 x Koc7)
Endosulfan USEPA 1980e
Endosulfan USEPA 1980e
USEPA 1980f
Endrin USEPA 1980f
Endrin USEPA 1980f
USEPA 1980g
Heptechlor USEPA 1980g
USEPA 1980h
USEPA 1976
USEPA 1986b
USEPA 1985a and DiToro et
al. 1991
USEPA 1985a and DiToro et
al. 1991
USEPA 1985a and DiToro et
al. 1991
USEPA 1985a and DiToro et
al. 1991
57
-------
Contaminant
Toxic Unit Calculations (ug/L)
Reference
Benzo(c[,h,j)perylene
Benzo(k)fluoranthene
Chrysene
Dibenzofuran
Fluoranthene
Fluorene
Indenod ,2,3-c,d)pyrene
Napthalene
Phenanthrene
Pyrene
(1800000 ug/kg5/(0.046 x Koc7))
x 3.118
5000000 ug/kg5/(0.046 x Koc7)
460000 ug/kg5/(0.046 x Koc7)
x 0.70810 = 4.956
(1800000 ug/kg5/(0.046 x Koc7))
x 2.438
28000 ug/kg5/(0.046 x Koc7)
24000000 ug/kg5/(0.046 x Koc7)
29.0
56000 ug/kg5/(0.046 x Koc7)
198000 ug/kg5/(0.046 x Koc7)
USEPA 1985a, DiToro et al.
1991, and Newsted and
Giesy 1987
USEPA 1985a and DiToro et
al. 1991
USEPA 1985a and DiToro et
al. 1991
1,2-Dichlorobenzene
Michigan DNR 1993 and
LeBlanc 1980
USEPA 1985a, DiToro et al.
1991, and Newsted and
Giesy 1987
USEPA 1985a and DiToro et
al. 1991
USEPA 1985a and DiToro et
al. 1991
Michigan DNR 1993
USEPA 1985a and DiToro et
al. 1991
USEPA 1985a and DiToro et
al. 1991
58
-------
Contaminant
Toxic Unit Calculations (ug/L)
Reference
Phthalate esters
Bis(2-ethylhexyl)phthalate
Butyl benzyl phthalate
Dimethyl phthalate
Metals, organo-metals, and metaloids
Arsenic
Cadmium
Chromium
Copper
Iron
Lead
Manganese
Mercury
Methyl mercury
Nickel
Selenium
79x4.58310 = 32.081
3.0
79x 13.7510 = 96.25
190.0
e (0.7852 x In(hardness) - 3.49)
_(0.81 9 x In(hardness) + 1.561)
\2
5.6
1000
_(1.273 x In(hardness) - 4.705)
not toxic
0.012
0.01211/1012 = 0.0012
-(0.846 x In(hardness) + 1.1645)
1.0 (ug/g dry weight sediment)
1,2-Dichlorobenzene
Michigan DNR 1993 and
LeBlanc 1980
USEPA 1980i
1,2-Dichlorobenzene
Michigan DNR 1993 and
LeBlanc 1980
USEPA 1984a
USEPA 1984b
USEPA 1984c
USEPA 1980J
USEPA 1976
USEPA 1984d
USEPA 1976
USEPA 1984e
Mercury USEPA 1980k and
USEPA 1984e
USEPA 1986c
Lemly et al. 1993
59
-------
Contaminant
Toxic Unit Calculations (ug/L)
Reference
Silver
Tributyl tin
Zinc
Other compounds
Ammonia
29.0
0.009
s (0.8473 x In(hardness) + 0.7614)
((0.8/FT13)/FPH14)/RATI015
Michigan DNR 1993
Michigan DNR 1993
USEPA 1987b
USEPA 1985b
1Aroclore 1254 chronic AWQC (USEPA 1980a).
2AHH activity of Aroclor6 1254 relative to that of 2,3,7,8-Tetrachlorodibenzodioxin (Smith et al. 1990).
3The ratio AHH activity of Aroclor8 1254 relative to 2,3,7,8-Tetrachlorodibenzodioxin:AHH activity of listed
contaminant relative to 2,3,7,8-Tetrachlorodibenzodioxin (Smith et al. 1990).
4The ratio AHH activity of Aroclor6 1254 relative to 2,3,7,8-Tetrachlorodibenzodioxin (Smith et al. 1990):Toxic
Equivalency Factor (TEF) of listed contaminant relative to 2,3,7,8-Tetrachlorodibenzodioxin (Safe 1990).
Estimated sediment threshold concentration for listed contaminant (USEPA 1985a).
6Proportion of organic carbon in sediment used to estimate sediment threshold concentrations (USEPA 1985a).
7KOC = 10(0-00028 + osaa-iooiKow^ KOW js octano|/water partition coefficient (DiToro et al. 1991).
8Relative toxicity of listed contaminant to Benzo(a)pyrene (Newstead and Giesy 1987).
91,2-Dichlorobenzen 1993 water quality standard from Michigan Department of Natural Resources.
10Relative toxicity of listed contaminat to1,2-Dichlorobenzene (LeBlanc 1980).
11AWQC for mercury (USEPA 1984e).
12Relative toxicity of methyl mercury to mercury (USEPA 1980k).
13py _ 1Q(0.03 x (20- 15))
14FPH = 1, IF pH < = 8 THEN FPH = (1 + 10(74-pH))/1.25.
15RATIO = 16, IF pH < = 7.7 THEN RATIO = 24 x 10'7 7-pH)/(1 + 10(74-pH|).
60
-------
Table 3. Laboratory toxicity tests included in the assessment of sediment toxicity.
Laboratory toxicity tests used in evaluation of Master Station data are denoted by
Test Organism
Fishes
*Pimephales promelas
embryo larvae
*Pimephales promelas
larvae
Zooplankters
Ceriodaphnia dubia
Length Sediment
of Test Phase
7 d Whole
sediment
7 d Whole
sediment
7 d Sediment
elutriate
(6.25, 12.5,
25, 50,
100%) and
whole
sediment
Endpoints
Survival, length,
and terata
Survival and
weight
Reproduction and
survival
*Daphnia maqna
*Daphnia magna
Benthic invertebrate
Chironomous riparius
*Chironomous tentans
48 h Sediment
elutriate
(3.125, 6.25,
12.5, 25, 50,
and 100%)
and whole
sediment
7 d Whole
sediment
14 d Whole
sediment
10 d Whole
sediment
Survival
Survival and
reproduction
Survival and length
Survival and length
61
-------
Test Organism
*Hexagenia bilineata
*Hvalella azteca
*Hvalella azteca
Pontoporeia hoyi
Pontoporeia hoyi
Length
of Test
10 d
14 d,
28 d
7 d
5 d
20 d
Sediment
Phase
Sediment
elutriate and
whole
sediment
Whole
sediment
Whole
sediment
Whole
sediment
Whole
sediment
Endpoints
Survival and
molting frequency
Survival, length,
antenna length,
and sexual
maturation
Survival
Preference
Survival
Phytoplankters
*Selenastrum capricornutum
24 h
*Selenastrum capricornutum 48 h
* Selenastrum capricornutum 96 h
Macrophytes
Hvdrilla verticillata
Sediment
elutriate
(6.25, 10.2,
12.5, 25,
28.6, 50, 57,
71.4%)
Sediment
elutriate
(6.25, 12.5,
25, 50,
100%)
Sediment
elutriate
(6.25, 12.5,
25, 50,
100%)
Whole
sediment
14
C uptake
Growth
Growth
Root length, shoot
length, chlorophyll,
dehydrogenase,
and peroxidase
62
-------
Test Organism
Length
of Test
Sediment
Phase
Endpoints
4d
Whole
sediment
Biomass,
chlorophyll a, and
frond number
Microbes
*Photobacterium
phosphoreum
5 min,
15 min
Sediment
elutriate
(5.65, 6.25,
11.25, 12.5,
22.5, 25, 45,
50, 100%)
Luminescence
63
-------
Table 4. Tolerance to pollution values for taxa identified in the full group of
sediment samples used in the analyses. Also included is the associated reference
from which each tolerance value was taken. Those tolerance values with an 'n'
value were estimated because no value was available from any reference. An
estimate was either calculated from the mean of values for the species found in
the sediments (i.e. family or high taxon tolerance) or from the mean of members of
the same genera with tolerance values. The values for the hemiptera were based
on educated guess.
Scientific Name
Amphipoda
Crustacea
Gammarus lacustris
Annelida
Hirudinea
Helobdella elongata
Helobdella stagnalis
Batracobdella phalera
Placobdella ornata
Naididae
Dero digitata
Oligocheata
Tubificidae
Aulodrilus pigueti
Aulodrilus limnobius
Aulodrilus plureseta
llvodrilus templetoni
Limnodrilus cervix
Limnodrilus claparedianus
Limnodrilus hoffmeisteri
Limnodrilus maumeensis
Tolerance Value
6.0
6.9
7.9 (n = 3)
9.9
6.7
7.1
7.8 (n = 2)
10
8.2
8.55 (n = 8)
4.7
5.2
4.95 (n = 2)
9.4
10
9.78 (n = 4)
9.8
9.78 (n = 4)
Reference
Hilsenhoff 1988
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
64
-------
Scientific Name
Limnodrilus undedemianus
Limnodrilus sp.
Quistadrilus multisetosus
Tubifex tubifex
Bivalvia
Sphaeriidae
Musculium sp.
Pisidium sp.
Sphaerium sp.
Unionidae
Anodonta imbecillis
Anodonta grandis
Eliptio complanata
Coleoptera
Pshenidae
Haliplus
Decapoda
Diptera
Ceratopogonidae
Plapomyia sp.
Chaoboridae
Chaoborus sp.
Tolerance Value
9.7
9.6
8.55 (n = 8)
10
7.25 (n = 2)
7.25 (n = 2)
6.8
7.7
5.4
5.4
5.4
2.5
5.7
6.0
6.9
8.5
Reference
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
USEPA 1989
Lenat 1993
Lenat 1993
Chironomidae
5.7
Lenat 1993
65
-------
Scientific Name
Chironomous sp.
Cladopelma sp.
Coelotanypus sp.
Cricotopus sp.
Cryptochironomous sp.
Dicrotendipes sp.
Glyptotendipes sp.
Microchironomous sp.
Polypedilum sp.
Procladius sp.
Tanypus sp.
Ephemeroptera
Caenis sp.
Tolerance Value
9.8
2.5
7.7
8.12 (n - 5)
7.35 (n - 2)
7.9
8.5
7.79 (n - 6)
6.67 (n - 7)
9.3
9.6
2.7
7.6
Reference
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Lenat 1993
Gastropoda
Ancylida
Laevapex fucus
Bithyniidae
Bithynia tentaculata
Hydrobiidae
Cincinnatia cincinnatiensis
7.3
6.1
6.1
Lenat 1993
Lenat 1993
(Mullusca)
Lenat 1993
(Mullusca)
66
-------
Scientific Name
Tolerance Value
Reference
Valvatidae
Valvata lewisi
Valvata tricarinata
6.1
6.1
Lenat 1993
(Mollusca)
Lenat 1993
(Mollusca)
Hemiptera
Sigara sp.
6.0
6.0
(estimate)
(estimate)
Odanata
Gomphus sp.
6.9
6.2
Lenat 1993
Lenat 1993
Trichoptera
Oecetis sp.
5.7
Lenat 1993
67
-------
Table 5. Bioaccumulative Contaminants of Concern1 used in toxic units model
Organochlorine compounds
Chlorodioxins and furans
2,3,7,8-Tetrachlorodibenzodioxin
1,2,3,7,8-Pentachlorodibenzodioxin
1,2,3,4,7,8-
Hexachlorodibenzodioxin
1,2,3,6,7,8-
Hexachlorodibenzodioxin
1,2,3,7,8,9-
Hexachlorodibenzodioxin
1,2,3,4,6,7,8-
Heptachlorodibenzodioxin
Octachlorodibenzodioxin
2,3,7,8-Tetrachlorodibenzofuran
1,2,3,7,8-Pentachlorodibenzofuran
2,3,4,7,8-Pentachlorodibenzofuran
1,2,3,4,7,8-
Hexachlorodibenzofuran
1,2,3,6,7,8-
Hexachlorodibenzofuran
1,2,3,7,8,9-
Hexachlorodibenzofuran
2,3,4,6,7,8-
Hexachlorodibenzofuran
1,2,3,4,6,7,8-
Heptachlorodibenzofuran
1,2,3,4,7,8,9-
Heptachlorodibenzofuran
Octachlorodibenzofuran
Polychlorinated
biphenyls
Aroclor6 1016
Aroclor8 1221
Aroclor6 1232
Aroclor6 1242
Aroclor6 1248
Aroclor6 1254
Aroclor6 1260
Pesticides
Aldrin
cis-Chlordane
trans-Chlordan
p.,fi-DDD
S,fi-DDE
Dieldrin
a-Endosulfan
/ff-Endosulfan
Endosulfan
sulfate
Polycyclic aromatic
hydrocarbons (PAHs)
Benzo(a)pyrene
Benzo(cj,h,j)perylene
Benzo(k)fluoranthene
Endrin
Endrin aldehyde
Endrin ketone
Heptachlor
Heptachlor epoxide
K-Hexachlorocyclo-
hexane (Lindane)
Methyoxychlor
Toxaphene
Metals, organo-
metals, and metaloids
Arsenic
Methyl mercury
Mercury
Selenium
'Modified from USEPA (1993). Draft GLWQI.
68
-------
Table 6. Toxic units, based on the estimated bioavailable fraction, for individual
contaminants in Buffalo River sediments. Only contaminants given are those for
which toxic units were one or greater at one or more sites; all contaminants
included in total.
Contaminant
Ammonia
(unionized)
Chromium
p_,p.-DDT
yff-Endosulfan
Endrin aldehyde
Iron
Methyl mercury
Naphthalene
Octachlorodibenz
o-dioxin
Aroclor6 1248
Aroclor6 1254
Selenium
Tributyl tin
Total
BR 1
5.31
NT1
2.06
NT
NT
NT
581.59
4.01
1.33
1.05
NT
3.80
23.12
623.97
BR 3
1.56
11.13
8.24
NT
1033.58
984.36
NT
NT
NT
NT
11.50
NT
NT
2051.75
Site
BR 7
2.25
3.11
14.74
1128.38
NT
701.8
NT
NT
NT
NT
NT
NT
NT
1851.03
BR 8
2.32
1.77
NT
NT
NT
514.56
NT
NT
NT
NT
NT
NT
NT
518.94
BR9
4.12
2.39
NT
NT
NT
449.76
NT
NT
NT
NT
NT
NT
NT
456.44
1NT stands for not toxic (i.e., the estimated pore-water concentration for the
individual contaminant was less than the level of concern).
69
-------
Table 7. Toxic units, based on the estimated bioavailable fraction, for individual
contaminants in Indiana Harbor sediments. Only contaminants given are those for
which toxic units were one or greater at one or more sites; all contaminants
included in total.
Contaminant
Aldrin
Ammonia
(unionized)
Anthracene
trans-Chlordane
Chromium
BL,EL-DDE
fi,fi-DDT
Dibenzofuran
Dieldrin
/?-Endosulfan
Endrin aldehyde
Heptachlor
Heptachlor
epoxide
Iron
Methyl mercury
Naphthalene
Octachlorodibenz
o-dioxin
Octachlorodibenz
o-furan
Aroclor6 1242
IH 3
3.56
7.93
NT
17.47
97.97
NT
NT
NT
637.74
675.76
NT
NT
NT
4719.95
NT
1.71
NT
NT
223.04
Site
IH 4
NT1
8.70
NT
NT
63.91
NT
NT
NT
NT
NT
NT
NT
NT
3681.06
1350.47
1.14
NT
NT
90.76
70
IH 6
5.32
12.75
NT
26.23
85.06
1.06
2.18
NT
496.54
NT
195.32
10.50
1804.79
1335.52
NT
1.31
3.66
NT
353.63
IH 7
12.21
16.06
1.35
39.26
779.21
2.94
NT
11.63
NT
NT
NT
44.37
2010.72
10712.79
364.63
1.02
5.17
3.85
836.59
-------
Contaminant
Aroclor6 1254
Selenium
Tributyl tin
Total
IH 3
NT
2.60
249.13
6645.32
Site
IH4
11.50
2.30
154.88
5354.88
IH 6
NT
3.80
1028.63
5369.41
IH
NT
3.10
NT
14848.
7
66
1NT stands for not toxic (i.e., the estimated pore-water concentration for the
individual contaminant was less than the level of concern).
71
-------
Table 8. Toxic units, based on the estimated bioavailable fraction, for individual
contaminants in Saginaw River sediments. Only contaminants given are those for
which toxic units were one or greater at one or more sites; all contaminants
included in total.
Contaminant
Aldrin
Ammonia
(unionized)
trans-Chlordane
Chromium
B..B.-DDE
Heptachlor
Heptachlor
epoxide
1234678-
Heptachlorodi-
benzofuran
Iron
Octachlorodibenz
o-dioxin
Octachlorodibenz
o-furan
Aroclor8 1242
Aroclor8 1254
12378-
Pentachlorodi-
benzofuran
SR 3
NT1
1.67
NT
2.98
NT
NT
NT
NT
648.96
1.35
NT
223.04
NT
NT
Site
SR 6
56.14
1.93
136.31
90.73
8.26
81.84
5965.91
1.08
27.36
4.41
1.37
4921.87
25.96
1.76
SR 10
NT
NT
NT
2.31
NT
NT
NT
NT
661.99
1.32
NT
NT
16.20
NT
23478- 2.17 11.21 1.29
Pentachlorodi-
benzofuran
72
-------
Site
Contaminant SR 3 SR 6 SR 10
Tetrachlorodi- NT 6.94 NT
benzofuran
Tributyl tin 39.84 NT NT
Total 699.97 11344.15 685.84
1NT stands for not toxic (i.e., the estimated pore-water concentration for the
individual contaminant was less than the level of concern).
73
-------
Relative Laboratory Toxicity
-< s: 3
n 8?
O § CD
81"
• i o
O
01
p
bo
§
8 a i
8 § CL
o —r
N)
8
o
o
o
o
-r 0)
CO =
-Q o
c o
QJ _
CD oT
tt 1
II 5"
0 §
8 ?
3 H
II CD
N CD
^"^1 ^^
u CD
II ~£|
OCD
rn
V?
O CD
0 D
2 ET
>^B*^ ^^
IT
CD
CD
^
O
^-
CD
CO
CD
Q.
CD
D
^^-
O
X
o
<
[ft
w
1
CO
r^>
O
£
o
X
o'
C
CO
cr
0)
CO
CD
Q.
o
D
O
•— h
9L
•
o
X
o'
D
«-*•"
ty)
o
O)
o
o
o
00
8-
O
O
o —
o
o
_t,
o —
o
o
_fc
§
o
o
.-- *
8-
• 0 D • >
• 0 D • >
• O D • >
C/5C/55-5-OOJ3
Q} Q) o^Q C CD
fQfft ^"^-^CQ
^Kt«nv*Q3 Q) Q* ^^
^j ^3 ^j "^ ^™™CD
^T^T JP^~ *
DDDDo: Q) <'2
< < S-3-5 2-.
^_ __ [i (j ^j
s- ^^ ro
® Q'
Oi - .
CO
0
o
-------
g
"x
o
o
2
o
n
OJ
CD
1
0
rr
D
A A A
ODD
O O O
Degression line
Buffalo River
ndjana Harbor site 3
ndiana Harbor
Saginaw Rjver
Saqinaw River site 6
0
20000000 40000000
Bulk Sediment Chemistry Toxic Units
60000000
Figure 2. Observed relative sediment toxicity versus total toxic units based on
bulk sediment concentrations of all contaminants. The line represents the
relation Y = 0.32 - 0.0000000089X (R-squared = 0.68, n = 12, P =0.0009).
-------
33
CO
c
03
0
Q.
II
O
Pp
-
3
II
JVD
"
^
II
O
b
o
CO
c
o'
IJ
N
0
Q.
03
O
U
07
03
o
0
H
rr
0
5'
0
0
•^•w
~n
CD'
c
0
CO
O
C7
CO
0
0
Q.
0
£3-
0
CO
0
Q.
3'
0
D
^
Relative Laboratory Toxicity
X o
^* ^^
O
GO
0
CJl
o
0)
p
b
GO —
C .&>_
D
§' 01-
8' O)-
QL
0
CO
0
CO
0
•n
0
03.
6'
ZJ
II
o
ro
i
o
b
N>
00
00
X
X
o'
0
c!
c
CO
o
2L
^^
o
X
o'
c
S-'
CO
CT
03
CO
0
Q.
O
33
3
o
D
0)'
H
O
X
o"
c
2,
r™F
C/)
OD —
CD —
_».
O ~~
.
IV) ~~
GO ""
.
^ —
_JL
Ol ~~
_t
0) ~~
_L
^J "~"
0 O
• O D
• 0 D
Q) 0) Q.Q.C CD
-
CD
05
D
-------
Relative Laboratory Toxicity
CD
11 a
p •§
CD
0)
CD
r-+
CO
o
X
o'
c
D
CD
Q_
-------
0.8 H
0.2-1
D
nan
ooo
legressipn line
luffalo R
nd|aha Harbor site 3
ndiana H
agjnaw
aqinaw
irbor
Ijver
River site 6
0 20000000 40000000 60000000
Bulk Chemistry Toxic Units for Bioaccumulatives
Figure 5. Observed relative sediment toxicity versus total toxic units based on
bulk sediment concentrations of bioaccumulative contaminants. Line represents
the relation Y = 0.32 - 0.0000000089X (R-squared - 0.68, n = 12, P = 0.001).
-------
0
i i i i i i i I
2,000 4,000 6,000 8,000 10,000 12,000 14,000 16,000
Total Toxic Units
Figure 6. Benthic community mean tolerance to pollution versus total toxic units
based on bioavailable fractions of all contaminants. The line represents the
relation Y = 8.71 - 0.000018X (R-squared - 0.40, n - 12, P = 0.0278).
-------
0
10 20 30 40 50
Bulk Sediment Chemistry Toxic Units (millions)
60
Figure 7. Benthic community mean tolerance to pollution versus total toxic units
based on bulk sediment concentrations of all contaminants. The line represents
the relation Y = 8.74 - 0.000000005X (R-squared = 0.34, n - 12, P - 0.0463).
-------
0
60
10 20 30 40 50
Toxic Units of Bioaccumulatives (millions)
Figure 8. Benthic community mean tolerance to pollution versus total toxic units
based on bulk sediment concentrations of only bioaccumulative contaminants. The
line represents the relation Y = 8.74 - 0.000000005X (R-squared = 0.34,
n = 12, P = 0.0466).
-------
0
2,000
4,000 6,000
Iron Toxic Units
8,000 10,000
Figure 9. Benthic community mean tolerance to pollution versus total toxic units
based on bioavailable fractions of iron only. The line represents the relation
Y = 8.73 - 0.000027X (R-squared = 0.40, n = 12, P = 0.0261).
-------
120
0
T I I I
BUFFALO RIVER
INDIANA HARBOR
SAGINAW RIVER
AREA OF CONCERN
PRIORITY MASTER VARIABLES
MASTER VARIABLES
Figure 10. Comparison of the overall ranking of sites based on Priority Master
Stations (shaded bars) versus Master Stations (crosshatched bars). Bars
represent the average of the relative ranking among sites of toxic units,
control-adjusted laboratory toxicity, and mean tolerance to pollution of the
benthic community for both Priority Master Stations and Master Stations.
-------
120
BUFFALO RIVER
INDIANA HARBOR
I I
SAGINAW RIVER
AREA OF CONCERN
PRIORITY MASTER VARIABLES
RECONNAISANCE VARIABLES
Figure 11. Comparison of the overall ranking of sites based on Priority Master
Stations (shaded bars) versus Reconnaissance Stations (crosshatched bars). Bars
represent the average of the relative ranking among sites of toxic units,
control-adjusted laboratory toxicity, and mean tolerance to pollution of the
benthic community for Priority Master Stations. For Reconnaissance Stations,
bars represent the average of the relative ranking among sites of toxic units,
which were base only a short list of contaminants, and Microtox.
-------
LJU
LU
CC
120
100
80
60
40
20
BUFFALO RIVER
INDIANA HARBOR
SAGINAW RIVER
AREA OF CONCERN
PRIORITY MASTER VARIABLES
BIOACCUMULATIVE CONTAMINANTS
Figure 12. Comparison of the overall ranking of sites based on Priority Master
Stations (shaded bars) versus bioaccumulative contaminants (crosshatched bars).
Bars represent the average of the relative ranking among sites of toxic units,
control-adjusted laboratory toxicity, and mean tolerance to pollution of the
benthic community for Priority Master Stations. For bioaccumulative
contaminants, bars represent the toxic units based on bulk concentrations of
those contaminants listed as bioaccumulative contaminants of concern (Table 5).
------- |