-
GEI Consultants, Inc.TECHNICAL COMMENTS
ON
FEASIBILITY STUDY AND PROPOSED PLAN FOR MIDDLE MARSH
SULLIVAN'S LEDGE SITE SECOND OPERABLE UNIT
Submitted to:
United States Environmental Protection AgencyAttention: Jane
Downing, RPM
Region IJ.F. Kennedy Federal BuildingBoston, Massachusetts
02003
On behalf of:
Acushnet CompanyAmtel, Inc. and United Dominion Industries
AVX CorporationBerkshire Hathaway, Inc.
Bridgestone/Firestone, Inc.Chamberlain Manufacturing
Corporation
Commonwealth Electric Company and Commonwealth Gas CompanyEmhart
Industries, Inc.
The Goodyear Tire and Rubber CompanyParamount Communications,
Inc. and Universal American Corporation
Teledyne Rodney Metals
Prepared by:
GEI Consultants, Inc.53 Regional Drive
Concord, New Hampshire 03301
AUGUSTS, 1991
-
EXECUTIVE SUMMARY
These Comments address the Supplemental Remedial Investigation
(SRI), Feasibility Study (FS)and Proposed Cleanup Plan (Plan) for
the Middle Marsh Study Area (Middle Marsh OperableUnit) at the
Sullivan's Ledge Superfund Site. The SRI, FS and Plan were issued
in April 1991,May 1991 and May 1991, respectively, by the US
Environmental Protection Agency (EPA). Thebasic conclusions
expressed in these Comments are that the proposed remedy was
notappropriately selected, that the proposed remedy will cause more
harm to the Marsh than isjustifiable, and that EPA should return to
its original "no action" decision. Specific technicalanalyses
demonstrate flaws in the supporting data, assumptions and
studies.
The Comments consist of seven volumes. The first volume contains
the Technical Commentsand associated tables, figures and
calculations. Appendix B consists of two volumes andcontains all
references cited in the Technical Comments and the resumes of all
technicalpersonnel who contributed to the development of the
Technical Comments or whose work isrelied upon. Appendix C consists
of four volumes and contains the associated documentationand
references that support the analysis in the Technical Comments.
In the SRI and FS, EPA itself has determined that the Marsh
poses no significant risk to humanhealth. EPA has also determined
that no contaminants, other than PCBs, pose unacceptablerisks to
the environment in the Marsh. In 1989, EPA concluded that remedial
action was notrequired at the Middle Marsh. Nothing in the SRI or
FS justifies a departure from EPA's previous"no action"
determination.
EPA rests its change in position solely on the contention that
the "preferred alternative" isnecessary to reduce the risk to
wildlife (specifically mink) from exposure to PCBs in the
Marsh.This reasoning is flawed. The concentrations of PCBs have
decreased significantly since EPAfirst determined that no action
should be taken in the Marsh. The purported risk to wildlife
inreality is nothing more than one chance in 10,000 that one female
mink could sustainreproductive impairment. In fact, it is unlikely
that even one mink lives in the Marsh. None havebeen observed. The
proposal to spend as much as $8 million, or more, to prevent a
non-existent risk is irrational.
There are also a number of serious technical problems with the
data reported and methodologiesapplied in the SRI and the FS. Those
include the improper use of the sediment quality criterion,a
failure to investigate other sources of contamination of the Marsh,
problems with computermodeling, and discrepancies in analytical
data.
While the technical basis for the SRI and FS is flawed and the
benefits of the preferred alternativeare negligible, the dangers to
the Marsh and its wildlife from implementation of the
preferredalternative are substantial. There is a strong likelihood
that the proposed remedy could not beperformed without irreversible
harm to the Marsh, including destruction of the habitat of
thespotted turtle, a species of special concern in Massachusetts.
The preferred remedy alsocontains unnecessary components, and the
FS is unrealistic about both the implementationschedule and the
expected results of the remedial work.
EPA's original 'no action" decision remains the appropriate
alternative by which to preserve theenvironment of the Middle
Marsh.
-
TABLE OF CONTENTS
TABLE OF CONTENTSEXECUTIVE SUMMARYTABLESFIGURESAPPENDIX A
PageNo.
I. THERE IS NO UNACCEPTABLE RISK TO HUMAN HEALTH DUE TOEXPOSURE
TO CONTAMINANTS IN MIDDLE MARSH 1
A. EPA's May 1991 Proposed Plan correctly concluded that there
are nosignificant excess risks to humans at Middle Marsh 1
B. The risk assessment used in calculating human exposure to
PCBs isbased upon several unrealistic assumptions• Maximum exposure
estimates are inappropriate and are inconsistent
with Superfund guidance 1• The health risk assessment
inappropriately relies on PCB data from
both the 1989 and 1991 Remedial Investigations for Middle
Marsh,and therefore, overestimates the maximum risks 2
• Use of arithmetic averages of sampling data for several
substancesare unreliable; these averages likely do not reflect
actual conditions 2
• Exposure frequencies for contaminants in the Middle Marsh
areoverestimated, and future land-use assumptions are inconsistent
withthe ecological risk assessment 3
• The assumed levels of exposure to contaminated soils and
surfacewaters overestimate realistic human exposures 4
C. The human health risk assessment for PAHs in the SRI is based
onunrealistic and inaccurate factors 4
SUMMARY 5
II. THERE IS NO UNACCEPTABLE RISK TO THE ENVIRONMENT FROM PCBOR
OTHER CONTAMINATION IN MIDDLE MARSH 6
A. The concentrations of PCBs measured in Middle Marsh are not
highand have significantly decreased since the measurements
reported inthe 1989 Remedial Investigation report for Sullivan's
Ledge site 6
B. Not all elevated concentrations of contaminants appear to be
originatingfrom the Sullivan's Ledge site 8
C. EPA's withdrawal of the original "no action" remedy appears
to havebeen based on an erroneous comment, the contents of which
were noteven addressed in the SRI 8• The Carr memo incorrectly
infers that statistical probabilities relating
to the Sediment Quality Criterion (SQC) correspond to
probabilitiesof harm to benthic organisms 9
• The SQC for PCBs is designed to protect mink, not
benthicorganisms 9
D. The SRI and FS studies for Middle Marsh assume that the
PCBconcentrations pose an unacceptable risk to female mink in
theMarsh. However, there is no evidence that any mink, female
ormale, inhabit the Marsh 10
-
TABLE OF CONTENTS(Continued)
PageNo.
E. Ranges and territorial habits of the mink would indicate
that, at most,only one mink would feed in the Marsh 10
F. The habitat ranges for mink and other animals which EPA
considers tobe potentially present are applied inappropriately and
inconsistentlyacross the Marsh in the computation of the ecological
risk. Therefore,the exposure predictions are excessive and biased
11
G. The mink's average dietary concentration of PCBs, if obtained
solelyfrom food sources in Middle Marsh, will be lower than the
dietaryconcentration used in the derivation of the sediment
qualitycriterion 12
H. The food chain presented in the SRI risk assessment is based
upon theerroneous presumption that mink eat contaminated trout in
the Marsh.The SRI does not provide any evidence of trout (and any
other fish)being present in Middle Marsh 13
I. The SQC methodology is applicable only if contaminated
sediments aresubmerged for sufficient periods of time to establish
an equilibriumbetween the sediments and the overlying water column.
This conditionis not true for most of Middle Marsh where there is
NO truly aquaticenvironment 14
J. The sediment quality criteria methodology, a highly
controversial andcomplex method for evaluating risk, produces
extremely unpredictableand inaccurate results 16
K. The SQC methodology improperly assumes that aquatic
organismssuch as trout will be exposed to pore water concentrations
ofcontaminants 17
L The SRI correctly concludes that volatile and semi-volatile
organics,heavy metals, and pesticides in the Middle Marsh
sediments, porewaters, and surface waters pose no threat to the
environment 18
SUMMARY 19
THE FEASIBILITY STUDY CONTAINS UNREALISTIC ASSUMPTIONSABOUT THE
LIKELIHOOD OF SUCCESS OF THE PREFERRED REMEDY,INCLUDES UNNECESSARY
STEPS IN THE PREFERRED REMEDY, ANDDEMONSTRATES NO REASONABLE
JUSTIFICATION FOR THECONTINGENCY REMEDIAL ALTERNATIVE 20
A. The remedial schedule proposed in the FS is unrealistic.
Remediationof the Marsh will be seriously jeopardized by the
remediation of theUnnamed Stream (during the remedial action for
the First OperableUnit) if the actions are not performed
simultaneously 20
B. There is no reasonable justification for EPA's proposed
contingencyalternative 20
C. Backfilling the disturbed wetlands, as required in the
preferred remedy,is not absolutely necessary and requires
destruction of wetlandselsewhere in order to obtain the most
suitable materials 21
-
TABLE OF CONTENTS(Continued)
PageNo.
D. There is no justifiable reason to manage the cleared
vegetation as ahazardous waste 22
E. Production rates and treatment technologies stated in the FS
areunrealistic and likely unattainable 22
SUMMARY 23
IV. THE PREFERRED ALTERNATIVE WILL SIGNIFICANTLY HARM THEWETLAN
DS, WITHOUT ANY ASSURANCE THAT TH E WETLANDS CAN BEADEQUATELY
RESTORED 24
A. The proposed plan would require excavation of more wetland
acreagethan necessary, even accepting the SQC in the SRI 24• Areas
2, 3 and 4 24• Area 1 24
B. The proposed plan will disturb and/or destroy the wildlife it
is supposedto protect 25
C. The proposed plan poses a substantial and unreasonable danger
ofdestroying the habitat of the spotted turtle, a species of
special concernin Massachusetts 26
D. The proposed wetlands restoration plan is inadequate and
notconsistent with existing wetland species 26
SUMMARY 27
V. THE FEASIBILITY STUDY, BASED UPON QUESTIONABLE DATA
ANDMODELING STUDIES, FAILED TO ADEQUATELY INVESTIGATE
ALLPOTENTIALLY APPROPRIATE REMEDIAL ALTERNATIVES, AND FAILED
TOINCLUDE SIGNIFICANT FACTORS IN EVALUATING THE PREFERRED
ANDCONTINGENCY ALTERNATIVES 28
A. The data which is relied upon in the FS regarding PCB
concentrationsin surface water and pore water samples contain
severaldiscrepancies 28
B. The TOC Analytical Method produces inconsistent, unreliable
resultswhich are not reproducible to even an order of magnitude in
thelaboratory 29
C. The basic assumptions of the hydrologic computer models are
notincluded in the SRI or available for review, either by EPA or
the public.The limited information on the hydrologic computer
modeling which isavailable indicates that the models may not have
been calibratedcorrectly 29
D. The Feasibility Study for Middle Marsh did not adequately
evaluatepotentially appropriate remedial alternatives for Middle
Marsh and failedto address critical aspects of the preferred and
contingencyremedies 30
SUMMARY 32
-
TABLE OF CONTENTS(Continued)
PageNo.
REFERENCES
TABLES
FIGURES
APPENDIX A - Area Averages of PCB Concentrations
APPENDIX B - Volume I of IIAPPENDIX B - Volume II of II
APPENDIX C - Volume I of IVAPPENDIX C - Volume II of IVAPPENDIX
C - Volume III of IVAPPENDIX C - Volume IV of IV
-
UST OF TABLES
TABLE 1 - Carcinogenic Potencies of Various PAHs Relative to
that ofBenzo(A)pyrene
TABLE 2 - Comparison of Estimated Excess Carcinogenic Risks in
the SRIwith Estimates of Adjusted Values That Incorporate
RelativePotency Factors for PAHs
TABLE 3 - Comparison of 1988 and 1990 PCB ConcentrationsTABLE 4
- Summary of Sediment/Soil Criteria Developed Within the SRITABLE 5
- Comparison of 1988 and 1990 PCB ConcentrationsTABLE 6 - Estimate
of the Average Concentration of PCBs in the Mink's
Diet Assuming All of Its Food is Obtained in Middle Marsh
LIST OF FIGURES
FIGURE 1 - Analysis of Equilibrium Partitioning
-
August 5, 1991Page 1
MAJOR TECHNICAL ISSUES
I. THERE IS NO UNACCEPTABLE RISK TO HUMAN HEALTH DUE TOEXPOSURE
TO CONTAMINANTS IN MIDDLE MARSH
A. EPA's May 1991 Proposed Plan correctly concluded that there
are nosignificant excess risks to humans at Middle Marsh
The Summary of Study Risks in the May 1991 Proposed Plan states,
"EPA hasdetermined that, based upon the exposure assumptions, human
exposure to sitecontaminants through the current and future
pathways... would not result insignificant increases in
carcinogenic risk". The Proposed Plan goes on to say,"EPA has
further determined that there are no significant risks to public
healthposed by exposure to noncarcinogenic contaminants in Middle
Marsh and theadjacent wetland area." EPA properly concluded that
Middle Marsh posesinsignificant excess risks to human health.
Indeed, application of reasonableand realistic risk assessment
assumptions (as discussed elsewhere in theseComments) re-affirms
that risks at the site to human health are negligible.
B. The risk assessment used In calculating human exposure to
PCBs isbased upon several unrealistic assumptions
A number of assumptions embodied in the health risk assessment
appear to beunrealistic given the nature of the Marsh. Despite all
these unrealisticassumptions, the estimated human health risks are
still within an acceptablerange. This underscores that the no
action alternative is fully protective ofhuman health at the Marsh.
Specific examples of the unrealistic assumptionsare discussed
below.
• Maximum exposure estimates are inappropriate and
areinconsistent with Superfund guidance
The risk assessment in the SRI calculates both average and
maximum exposurescenarios. These are distinguished by the manner in
which exposure pointconcentrations are estimated. Average exposures
utilize arithmetic averageconcentrations, an approach which
implicitly assumes that (1) eachmeasurement is weighted equally and
(2) people will randomly contact soils andsurface water throughout
the contaminated area (either Middle Marsh or the
golfcourse/wetland). EPA, in its use of the maximum exposure
approach, utilizesthe maximum detected concentration for each
pollutant. This assumptionignores the exposure averaging effect
which in fact results when people movearound through areas of
varying degrees of contamination.
-
Augusts, 1991Page 2
There are three problems with the maximum exposure approach.
Rrst, itincorrectly assumes that the individual is exposed to the
highest pollutantconcentrations for the duration of the exposure
period, a highly improbablescenario. Individuals do not, in fact,
remain stationary, and thus would beexposed to varying levels of
contamination, and to no contamination, during anexposure period.
More importantly, the maximum concentrations of allpollutants were
not detected at the same sample locations. Since the
maximumexposure scenario assumes constant contact with the highest
concentration ofeach pollutant, the individual must also be assumed
to be exposedsimultaneously at numerous locations; which is, of
course, physically impossible.
The second problem with using the highest detected
concentrations for theexposure point concentrations in the maximum
exposure case is that thisapproach is inconsistent with Superfund
guidance. Volume I of EPA's RiskAssessment Guidance for Superfund
(EPA, 1989a) specifies the use of the 95thpercentile upper
confidence limit of the arithmetic mean (UCL). Concerning thetwo
cases in the SRI data for which computation of the UCL is
performed, theUCL is lower than the maximum detected concentration.
If, as the 1989Guidance provides, UCLs had been used as the basis
for exposureconcentrations in estimating the reasonable maximum
exposure, the overallestimates of risk would have been
substantially lower.
Finally, it is unreasonable to assume that humans will be in
contaminated areasof Middle Marsh during the entire maximum or
average exposure period. Theseapproaches do not consider that a
person's actual movements would generallybe limited to
non-contaminated areas such as most of the golf tees and
fairways.
The health risk assessment inappropriately relies on PCB data
fromboth the 1989 and 1991 Remedial Investigations for Middle Marsh
andtherefore overestimates the maximum risks
Unlike the ecological risk assessment, which uses only data
collected in the1991 SRI, the health risk assessment also includes
the PCB data collected inMiddle Marsh during the 1989 Rl of the
Sullivan's Ledge site. Since PCBs weredetected at higher
concentrations in the 1989 study, the maximum exposureapproach,
using both sets of data, are seriously overstated.
Use of arithmetic averages of sampling data for several
substancesare unreliable; these averages likely do not reflect
actual conditions
The summary statistics of Tables 5-2 and 5-3 of the SRI indicate
that, incalculating the arithmetic averages of several substances,
EPA excluded asignificant proportion of the samples, because the
detection limits of nondetectswere too high relative to the maximum
detected concentration. This was donefor polycyclic aromatic
hydrocarbons such as acenaphthylene, for which 23 of24 potential
samples were excluded from the average (following Guidance in
USEPA, 1989a). (This problem apparently results from interferences
in the
-
Augusts, 1991Page3
laboratory analyses). There are at least two important
deficiencies in thisapproach.
Rrst, this approach demonstrates that there is considerable
uncertaintyregarding the actual concentrations of polycyclic
aromatic hydrocarbons andother compounds that are difficult to
distinguish in the laboratory. Page 5-13 ofthe SRI describes
difficulties in distinguishing benzo(b)fluoranthene
andbenzo(k)fluoranthene; it is likely that similar difficulties
were encountered inidentifying other compounds. If this is the
case, then the data uncertaintiesshould have been evaluated and
excluded from the risk assessment if theunderlying analytical data
were suspect. EPA, however, does not appear to haveundertaken such
a review.
Second, by reason of the systematic exclusion of low
concentration data, thereis an automatic bias toward indications of
higher than actual averageconcentrations. This adds still another
layer of bias to the analyses.
Exposure frequencies for contaminants in the Middle Marsh
areoverestimated, and future land-use assumptions are inconsistent
withthe ecological risk assessment
Current land-use scenarios 0"ables 5-9, 5-12, and 5-14 in the
SRI) assume thatexposures to contaminants in Middle Marsh occur 28
days per year. The futureland-use assumptions arbitrarily double
this estimate to 56 days per year. Thesefrequencies are highly
unrealistic.
Middle Marsh is a thickly vegetated area with limited access;
there is noevidence of pathways transecting the Marsh upon which
humans might walk.The muddy character of the Marsh is likely to
dissuade potential hikers, andnature enthusiasts are likely to find
nearby Apponagansett Swamp moreinteresting. The presence of golfers
will inhibit trespassers during the daytimehours, and the absence
of light will discourage the presence of people duringthe
nighttime. Realistically, the presence of people (particularly the
samepeople) likely would be limited to, at most, a very few
occasions of limitedduration per year.
The assumed frequency of exposure under the future land-use
scenario in theMiddle Marsh is double that of the current land-use
scenario, to account for thepossibility that Middle Marsh may be
dry. However, State and Federal lawsrelative to the protection of
wetlands and City zoning land use ordinances wouldgenerally
prohibit the erection of any structures in the Marsh. This
premise,however, is fundamentally inconsistent with the ecological
risk assessment,which assumes the continued presence of a
wetland.
-
August 5, 1991Page 4
• The assumed levels of exposure to contaminated soils and
surfacewaters overestimate realistic human exposures
SOIL: The risk assessment assumes a soil ingestion rate of 100
mg per day foradults. This rate is considerably greater than the
values typically found in theliterature. A review contained in
EPA's Exposure Factors Handbook (EPA,1989c) lists adult ingestion
estimates of 50 mg/day and 61 mg/day. In thathandbook, values are
derived for categories of adults who could be expected toingest
significant amounts of soil; consequently, even those ingestion
ratesreflect higher-than-average consumption rates for adults.
Further, any humanswho may be present in Middle Marsh would not
engage in activities expected toresult in significant soil
ingestion. Thus, the ingestion rate of 100 mg/dayassumed in the
risk assessment is much greater than would be reasonablyexpected
and adds yet another layer of excessive bias to the assessment.
SURFACE WATER: The risk assessment's surface water exposure
factorassumes that individuals would be exposed to the waters of
the Middle Marshcontinuously for 8 hours in each of the
hypothetical 28 exposure events per year.Further, it assumes direct
body contact (e.g., submersion of an arm) for theentire period. To
satisfy these assumptions, all golfers would have to beretrieving
golf balls all day, instead of playing golf. This is ridiculous
since nogolfer would put more than an occasional hand or arm into
the water (or mud)for only a moment to recover a golf ball. Given
the inaccessibility of the Marsh,its limited recreational value,
and the unlikelihood that golfers would be exposedover more than a
small part of their body for more than a moment, this
exposureassumption adds yet another layer of bias to an unrealistic
assessment.
C. The human health risk assessment for PAHs in the SRI is based
onunrealistic and inaccurate factors.
The SRI adopts a carcinogenic potency of 11.5 (mg/kg-d)"1 for
numerouspolycyclic aromatic hydrocarbons (PAHs). This value is
derived specifically forbenzo(a)pyrene, but is applied to other
PAHs for which the SRI apparentlyassumes potency estimates to be
unavailable. Cancer potency factors appliedin the SRI do not
reflect the current available data and yield a higher
excesscarcinogenic risk than is accurate.
ICF-Clement Associates (1987), under contract to EPA, has
estimated thecarcinogenic potencies of several pertinent PAHs
relative to the potency ofbenzo(a)pyrene. These relative potencies
are listed in Table 1. The relativepotencies are generally much
less than one for all chemicals exceptdibenz(a,h)anthracene.
If these relative potencies had been applied in the SRI, the
excess carcinogenicrisks estimated in Chapter 5 of the SRI would
decline. Table 2 compares theestimated risks in the SRI with
calculations using the cited relative potencies.Refining only this
one detail - i.e., assigning each PAH its specific potency -
-
August 5, 1991Page 5
reduces the estimated excess cancer risks by at least 20% for
the maximum riskestimate for the current Middle Marsh scenarios,
and by at least 60% for theaverage risk estimate for the future
Middle Marsh scenario.
These recomputed lower risks to human health once again
demonstrate thatunacceptable risks to human health simply do not
exist at the Middle Marsh andare a further indication that no
remedial action should be implemented in theMiddle Marsh.
SUMMARY EPA has concluded correctly that there are no
unacceptable risks to humanhealth due to exposures to contamination
in Middle Marsh. EPA has made this findingdespite the fact that the
human health risk assessments are based upon unrealisticexposure
scenarios and assumptions, and out-of-date cancer potencies. If the
health riskassessment had been done correctly with realistic
assumptions, it would havedemonstrated even more decisively that no
unacceptable risks to human health are posedby the Middle
Marsh.
-
August 5, 1991Page 6
II. THERE IS NO UNACCEPTABLE RISK TO THE ENVIRONMENT FROM PCB
OROTHER CONTAMINATION IN MIDDLE MARSH
A. The concentrations of PCBs measured in Middle Marsh are not
highand have significantly decreased since the measurements
reportedin the 1989 Remedial Investigation report for Sullivan's
Ledge site
In 1989, EPA initially decided that no remedial action was
appropriate for MiddleMarsh. The Proposed Plan for the Sullivan's
Ledge site released by EPA inJanuary 1989, specifically explains
why EPA selected the "no action" alternativefor Middle Marsh:
"After assessing the advantages and disadvantages of theremediation
options for Middle Marsh, EPA has determined that the benefits tobe
obtained by excavating sediments would be exceeded by the
adverseenvironmental impacts. Therefore, EPA is proposing no action
(Option A) forMiddle Marsh." (See Page 21). This determination was
based on the fact thatany remedial action would destroy habitat and
potentially resuspend andredistribute contaminants throughout the
marsh, resulting in adverseenvironmental impacts that would
outweigh any benefit of remedial action.
The decrease in the concentrations of PCBs in the sediments
(discussedhereafter) which has occurred since EPA's initial "no
action" decision, providesfurther justification for EPA's no action
decision. EPA, however, has unjustifiablychanged its mind and
decided that an intrusive remedial action should beimplemented,
even though PCB concentrations have decreased considerably inthe
Middle Marsh. EPA's reversal of its initial no-action decision is
inconsistentwith the observed decrease in PCB concentrations.
In 1989, the highest concentration of PCBs detected in the
sediments was 60mg/kg. In the SRI, the highest concentration was
reported to be 28 mg/kg. Atthe sampling station where PCBs were
originally detected at 60 mg/kg, theconcentrations were reported in
the SRI to have decreased to 20 mg/kg.Similarly, at stations where
the PCB concentrations were reported to be 28 and10 mg/kg in the
1989 study, the concentrations were reported in the SRI to
havedecreased to 12 and 2.2 mg/kg, respectively.
Table 2-9 of the SRI describes the locations of stations sampled
during the 1988Remedial Investigation for the Sullivan's Ledge site
as compared with samplingstations reported in the SRI. Stations
labeled with an ME were sampled byMetcalf and Eddy in 1990 as part
of the SRI and stations labeled with MM weresampled by Ebasco in
1988. According to Table 2-9 in the SRI, there are sevensampling
stations which are located near each other and can be
considered"paired": ME-1 and MM-5; ME-11 and MM-20; ME-12 and
MM-10; ME-15 andMM-12; ME-16 and MM-15; ME-17 and MM-12; and ME-27
and MM-25.
The PCB concentrations in samples taken from these pairs of
sampling stationsare presented in Table 3. As shown in Table 3, the
concentrations of PCBs insix of the seven sampling stations have
decreased an average of 54% during
-
August 5, 1991Page?
the last three years. Only one station pair showed a slight (4%)
increase inRGBs. This marginal difference between the measured
values at MM-12 andME-17 is within the acceptable criteria of
duplicate analysis and analyticallyrepresents essentially the same
PCB concentrations (USEPA Data ReviewWorkgroup, 1988). Thus, NO
statistically significant increases in PCBcontamination were
observed at the Middle Marsh site between 1988 and 1990.Instead, a
significant decrease was observed.
The reduction of observed PCB concentrations is especially
significant when oneconsiders that the sampling program in the SRI
was specifically designed todetermine the distribution of PCBs in
what EPA defined as the "hot spots." Page2-62 of the SRI
specifically identifies, as its objectives, "defining the
partitioningof contaminants among media and vertical distribution
in previously identified'hot spots'" and "determining the
persistence and extent of previously identified'hot spots." The
sampling plan, biased towards finding the highestconcentrations of
PCBs, instead showed a substantial decrease. The biasednature of
the sampling plan makes the observed PCB decrease even
moresignificant.
There are several physical, biological and chemical mechanisms
occurring onthe site that work to degrade the PCBs originally
deposited there. EPAacknowledges only transport as a possible
reason for lower PCB values. In fact,there are many reasons. One
mechanism of natural attenuation isbiodegradation, anaerobic and
aerobic. Dissolution and volatilization are alsoprocesses that may
have had, and may continue to have, a significant role indecreasing
PCB levels. Research in the New Bedford Harbor, for
instance,concluded that a substantial amount of PCB that entered
the water columnsubsequently volatilized to the atmosphere. Models
developed by Thibodeauxand ASA found that as much as 50 percent
would volatilize from the watercolumn. The New Bedford Harbor
Estuary and Lower Harbor/Bay FeasibilityStudy (Ebasco, 1990)
concluded that: "Volatilization appears to be the mostsignificant
process occurring .... Approximately 86 percent of the original 283
kgwhich migrates from the sediment into the overlying water column
is removedfrom the system in the 95-day simulation through
volatilization."
Volatilization would also be expected to occur more readily in
the shallow watersor periodically inundated areas within Middle
Marsh. The presence ofbioturbation in the aquatic and wetland
sediments would substantially enhancethe rate of volatilization
over what would be predicted by simple diffusiveprocesses
alone.
In addition to ignoring the reductions in overall PCB levels
seen at Middle Marsh,EPA failed to consider that bacterial
degradation would cause preferentialdechlorination of the
co-planar, and thus potentially more toxic, PCB
congeners(MacFarland and Clarke, 1989). Reductive dechlorination -
and resultingdetoxification -- is a concept well recognized by EPA,
yet Region I failed toexamine whether this natural process is
occurring in Middle Marsh and thereby
-
Augusts, 1991Pages
substantially reducing the potential for toxic effects to
endemic organics. Dr.Anna Yoakum (Yoakum, 1989 & 1990) has
documented that dechlorination isoccurring in both New Bedford
Harbor and Silver Lake samples. EPA (Lake etal., 1989) has
corroborated the occurrence of anaerobic biodegradation in
NewBedford Harbor samples. The failure of EPA to examine whether
this naturalprocess is occurring in Middle Marsh is a fundamental
flaw in EPA's ecologicalrisk assessment in the SRI.
B. Not all elevated concentrations of contaminants appear to
beoriginating from the Sullivan's Ledge site
The SRI acknowledges on Page 3-5 that "...at some locations, low
levels ofcontamination in the Sullivan's Ledge area may be due to
influences other thanflooding such as atmospheric deposition or
runoff from nearby highways".
Nevertheless, the SRI assumes that all contamination in Middle
Marsh originatesfrom the Sullivan's Ledge site through flooding of
the Unnamed Stream. Thispremise is undermined by the fact that the
locations of elevated concentrationsof contaminants are not always
within the flooded areas predicted by the model.As an example, the
PCB concentration in sample SL-51 was 10.0 mg/kg. Thatsample
location is a golf tee which is above the area predicted to be
inundatedby the 100-year flood. In addition, as indicated by
Figures 3 and 4 in the SRI,the highest concentrations of lead were
detected in the Middle Marsh atlocations approximately 250 to 300
feet to the east and west of the UnnamedStream. The topographic map
included in the SRI indicates that these locationsare within
drainage swales through which the Marsh receives runoff from
areas(roadways) other than the Sullivan's Ledge Site.
These observations suggest that there may be other sources of
contaminationof the Middle Marsh besides the Unnamed Stream, such
as airbornecontamination and road runoff. This assumption is
supported by a studyperformed by Environmental Science and
Engineering in 1978 entitled"Environmental Assessment of
Polychlorinated Biphenyls (PCBs) Near NewBedford, Ma. Municipal
Landfill" which describes measurements of PCBs in airto be in
excess of 1.0 microgram/cubic meter in ambient air over the
landfill.
C. EPA's withdrawal of the original "no action" remedy appears
to havebeen based on an erroneous comment, the contents of which
werenot even addressed In the SRI
The agency originally justified withdrawal of the no action
alternative in part uponconcerns raised by Kenneth Carr, U.S. Fish
and Wildlife Service (seeAdministrative Record, Memorandum from
Kenneth Carr to Jane Downing, datedDecember 8,1988). Mr. Carr's
memo raises questions about the protectivenessof a no action remedy
to benthic organisms, stating that the no action
-
August 5, 1991Page 9
determination "has a probability of 0.975 that it will result in
long-term impactsto benthic fauna." The Carr memorandum fails to
support the Agency's proposalto withdraw the no action remedy for
the following reasons:
The Carr memo incorrectly infers that statistical probabilities
relatingto the Sediment Quality Criterion (SQC) correspond to
probabilitiesof harm to benthic organisms
The documentation of the SQC (EPA 1989d) lists three values for
PCBs in freshwater: 19.5, 3.87, and 99.9 jig PCBs/g carbon. These
criteria are based upondifferent values of the partition
coefficient that relates pore water concentrationsto sediment
concentrations. The value of the 19.6 ng PCB/g carbon is basedon
the best estimate of the partition coefficient, while the values of
3.87 and 99.9jig PCB/g carbon are based on the lower and upper
limits, respectively, of the95th percentile confidence interval of
the partition coefficient. In actuality, the0.975, 0.5, and 0.025
probabilities cited by Carr reflect the likelihood that theSQC will
underpredict or overpredict pore water concentrations. For
example,the use of the mean value of the SQC (19.5 jig PCB/g
carbon) implies that theSQC will underpredict pore water
concentrations half of the time and overpredictpore water
concentrations the other half of the time. These probabilities thus
donot express a likelihood of harm to benthic organisms or, for
that matter, to anyorganisms. Estimation of probabilities of harm
would require a significantly moredetailed analysis that uses
additional data on the variability of organismresponse to pore
water concentrations. Thus, Carr's memo misinterprets themeaning of
the probabilities that pertain to the SQC.
The SQC for PCBs is designed to protect mink, not
benthicorganisms
The Carr memorandum discusses what Carr believes is the
probability of harmto benthic organisms. The memo incorrectly
assumes that the SQC for PCBswas designed to protect benthic
organisms. The SQC for PCBs, however, isintended to protect mink,
not benthic organisms. If protection of benthicorganisms were the
goal, an entirely different criterion would have to bedeveloped
that is based on toxicity data specific to benthos. The toxicity
valueslisted in the Ambient Water Quality Criteria Document for
PCBs (EPA, 1980)demonstrate that benthic organisms can tolerate PCB
concentrationsconsiderably higher than the 0.014 jig/l value on
which the mink-based SQC isbased. Consequently, an SQC designed to
protect benthos would be lessstringent than the 19.5 jig PCB/g
carbon criterion that is used in the SRI. As aresult, the Carr
memorandum provides absolutely no justification for EPA'sreversal
of the "no action" decision.
In summary, EPA's withdrawal of the original "no action remedy
appears to havebeen based on a total misinterpretation of the SQC
for PCBs and the underlyinginformation on which it is based. EPA's
reliance on the Carr memorandum asa basis to abandon the no action
alternative in Middle Marsh cannot be justified.
-
August 5, 1991Page 10
D. The SRI and FS studies tor Middle Marsh assume that the
PCBconcentrations pose an unacceptable risk to female mink In
theMarsh. However, there Is no evidence that any mink, female or
male,inhabit the Marsh
The SRI, page 2-53, acknowledges that the Middle Marsh is only
"...moderatelysuited to semi-aquatic carnivores as represented by
the mink (Mustela vison),"and that the developed area (that is, the
golf course) would "...somewhat limitoptimal habitat (for the
mink)..." During the SRI's field studies, no mink wereever observed
in the Middle Marsh, and the SRI did not provide any evidenceof
mink habitation. Nevertheless, EPA has designed a cleanup criterion
basedentirely on protecting mink. EPA has failed to explain why it
placed so muchemphasis on mink, when the presence of mink has not
been established.
E. Ranges and territorial habits of the mink would indicate
that, at most,only one mink would feed In the Marsh
Even if suitable habitat for mink were present at Middle Marsh,
a mink is notlikely to inhabit the Marsh. Nowak and Paradiso (1983)
demonstrate thatpopulation densities for mink range from one to
eight individuals per square kmand that females generally have
smaller home ranges (approximately 20 to 50acres for females versus
up to 2000 acres for males). These ranges are also inagreement with
data derived by the USDA (1987). Mink, by nature, areextremely
territorial; this means that their territories seldom overlap. The
extremeterritorial nature of mink suggests that, in such a small
area as Middle Marsh, itis highly unlikely that more than one mink
(if any) would be present.
The SRI uses 20 acres as the typical residence range of a mink;
yet, accordingto Nowak and Paradiso (1983), this is the lowest
range and applies only tofemale mink; the ranges of male mink are
much larger. Female mink require acontiguous habitat for nesting
purposes, and the 13 acres of the Middle Marshare not sufficient.
The 65% residence time assumed in the SRI applies only ifthe female
mink is willing to cross the golf course to forage in
theApponagansett Swamp. This is a very unlikely scenario, given the
reportedshyness of the mink (Grzimek, 1972).
In addition, according to Grzimek, mink have highly developed
day vision. Thissuggests that mink are probably more active by day
than is usually thought andnot entirely nocturnal. Thus, in light
of their shyness, female mink are even lesslikely to inhabit the
Marsh given the activity on the golf course during the
daylighthours.
-
August 5, 1991Page 11
In sum, Middle Marsh is not a likely habitat for mink because of
its small size;its proximity to the active golf course; and, the
availability of more suitablehabitat elsewhere (e.g.. Apponagansett
Swamp which is much larger and lessaffected by development). More
realistic, verified ecological assumptions woulddemonstrate that
unacceptable risks to the environment do not exist in
MiddleMarsh.
F. The habitat ranges for mink and other animals which EPA
considersto be potentially present are applied inappropriately
andinconsistently across the Marsh in the computation of the
ecologicalrisk. Therefore, the exposure predictions are excessive
and biased
Even if the Middle Marsh were a suitable habitat for mink, there
is a serious errorin the method by which the cleanup criterion is
applied in the SRI. The cleanupcriterion is based on an estimate of
the PCB concentration in sediment that mayinduce toxicity in mink.
However, the SRI errs in applying this criterion on apoint-by-point
basis to determine potential exposure.
Mink and other animals are likely to obtain their food supply
from a variety ofpoints in a habitat. Consequently, a contaminant
concentration averaged overthe entire feeding range (including
uncontaminated areas) should be used toestimate exposure, rather
than merely assuming a maximum concentration atall locations. The
SRI erroneously assumes that the mink will only feed at the"hot
spots" within their residence ranges. Mink, however, are
opportunists,feeding wherever there is suitable food. Hence, mink
food chain exposuresmust be averaged over the entire residence
range, and not just related to hotspot areas of the greatest
contamination.
Another example of a failure adequately to consider habitat
range is illustratedin the selection of the bioaccumulation factor
(BAF) for earthworms, which serveas a food source for carnivorous
birds. Pages 4-19 of the SRI list BAF valuesof 0.16 and 0.29 for
the east and west banks of the Unnamed Stream. The SRIproceeds to
select the more conservative value of 0.29 for use in the
exposureassessment. However, because birds can be expected to
obtain food from bothsides of the stream, an average value should
have been used.
This area-averaged approach should have been used by EPA in
applying theterrestrial food chain criteria to the species listed
in Table 4. Because the rangeof each of the predatory species were
not considered in constructing realisticexposure scenarios within
the Middle Marsh, the exposure predictions are notreliable.
-
August 5, 1991Page 12
G. The mink's average dietary concentration of PCBs. If obtained
solelyfrom food sources In Middle Marsh, will be lower than the
dietaryconcentration used In the derivation of the sediment quality
criterion
An average dietary concentration of PCBs can be estimated for
animals inMiddle Marsh using information on the animal's diet,
bioaccumulation factorsmeasured in Middle Marsh, and the
area-averaged concentrations as estimatedin Appendix A. The
following illustrates how the dietary concentration of PCBscan be
calculated for mink assuming it were present in the Marsh.
Thecalculation assumes a realistic varied diet for the mink, taken
from the entireMarsh area (i.e., averaged).
The composition of the mink's diet, taken from Grzimek, is based
upon thestomach content of North American mink. Bioaccumulation
factors (BAFs)measured in Middle Marsh are taken from Table 4-4 of
the SRI for the speciesfor which they are available. BAFs for
animal species for which measured valuesare available are
conservatively extrapolated from the BAFs for other species.For
example, BAFs for rabbits, moles and the "other" category are
assumed tobe 0.09, the highest value measured for a terrestrial
animal (Deer mouse) inMiddle Marsh. The BAF for fish and
crustaceans (0.038) is assumed to be theratio of the factors used
to construct the sediment quality criteria, 0.64 ppmdivided by 19.5
jig PCB/g carbon. The plant BAF of 0.017 is obtained fromTravis and
Arms (1988).
Terrestrial and aquatic PCB concentrations are assumed to be 5.3
ppm and 31.9jig PCB/g carbon, respectively, which are the
area-averaged PCB concentrationsestimated in Appendix A. The
dietary contributions in the right-most column ofTable 5 are the
product of the dietary fraction (normalized by 100%), the BAF,and
the soil/sediment concentration. Summed together over all the
foodsources, the total of 0.50 ppm represents an estimate of the
dietary PCBconcentration that the mink would experience if it
obtained aH of its food fromthe Marsh, an unlikely scenario. This
value is smaller than the dietaryconcentration of 0.64 ppm that is
used in the derivation of the SQC for PCBs.If one assumes, as the
SRI suggests, that mink obtains only 65% of its foodfrom the Marsh,
and that the remaining 35% is from non-contaminated sourcesof the
habitat, then the average dietary concentration of the mink would
be 0.50ppm times 0.65, or 0.33 ppm, which is about one-half of the
SQC basis of 0.64ppm. In either case, a Middle Marsh diet would
provide a mink with a muchlower "body burden" of PCBs than assumed
by the EPA.
-
August 5, 1991Page 13
H. The food chain presented In the SRI risk assessment Is based
uponthe erroneous assumption that mink eat contaminated trout in
theMarsh. The SRI does not provide any evidence of trout (and
anyother fish) being present In Middle Marsh.
The food chain route driving the exposure assessment assumes
direct ingestionof contaminated fish by mink. The aquatic food
criterion is applied for a floodedarea, designated as Area 1 in the
SRI (which is predicted to be sporadic inMiddle Marsh) when mink
can obtain fish from the aquatic habitat. Theterrestrial food
criterion presumably applies year-round and is based oningestion of
small mammals. Each of these criteria assume that, in a
givenseason, the mink's diet consists only of a single type of food
(all terrestrialmammals or all trout).
In actuality, the mink has a varied diet, as documented by
Grzimek, who reportsthe mink's diet as rodents (voles, muskrats)
(47%), lagomorphs (rabbits andhares) (4%), moles (2.5%), frogs
(2.5%), fish (19%), crustaceans (16.5%), insects(7%), and plants
(1%). It is, therefore, highly unlikely that the mink will
derive100% of its diet in any one season from a single food source
such ascontaminated fish.
In addition, EPA's consultants have observed that large areas of
water which arecapable of supporting a fish population do not exist
at the site. The onlypermanent water bodies on-site are the golf
course water hazards and theUnnamed Stream. Observations of
conditions in the stream and water hazardsillustrate that trout are
not likely to live in those conditions because of 1) lowstream
velocity, 2) mucky bottom sediments and, 3) low dissolved
oxygen(Raleigh, 1982).
Volume II of EPA's 1989 Risk Assessment Guidance indicates that
it isappropriate to eliminate exposure pathways which have not been
found to exist.Section 4.2 of this Guidance provides:
"Following collection of existing data, the technicalassistance
group should be in a position todetermine the nature and extent of
ecologicalassessment that will be necessary of the site. If
noecological exposure pathways have been revealed inthis initial
review, little or no additional work may beneeded. Alternatively,
certain exposure pathwaysmight be eliminated from further study
while othersmight require more data. For instance, if there is
nosurface water on the site and no opportunity forcontaminants to
reach surface waters off the site,further data on aquatic effects
would very likely bepointless, even though concern about exposure
toterrestrial organisms might warrant extensivesampling and
testing."
-
August 5, 1991Page 14
The site-specific reconnaissance of the Middle Marsh area and
observations ofthe Unnamed Stream as it traverses the subject
property found that in-streamconditions are not expected to be
suitable for the maintenance of a sustainablefish, and in
particular trout, population. Most notably, stream temperatures
areexpected to be too high and dissolved oxygen levels too low to
maintain a viabletrout population. Moreover, the bottom conditions
are the antithesis of preferredtrout habitat. Rather than sand and
gravel, the Unnamed Stream exhibitsextensive silt deposits
unsuitable for feeding and/or spawning. In addition,areas of deep
water are lacking, as are requisite pool and riffle areas.
Stableaverage annual water flows in the Unnamed Stream are also
insufficient. Withoutany actual demonstration that trout do, in
fact, exist on-site, the observed in-stream conditions in
conjunction with the scientific literature regarding fishhabitat
(Raleigh, 1982; Raleigh et al, 1984 and Raleigh et al, 1986)
stronglysuggest that the Unnamed Stream does not contain conditions
suitable for thegrowth and survival of trout. In addition, the
water hazards are also unlikelyhabitat for trout because they are
too warm, have muddy bottoms and no swiftcurrents to sustain a
trout population. EPA's consultants did not recordobservations of
trout in the golf course water hazards in the SRI.
The emphasis in the Guidance is on construction of plausible
exposurepathways through reasonable justification. That process has
simply not beenfollowed by EPA in the evaluation of the aquatic
food chain pathway for the minkin the Middle Marsh.
Hence, even if female mink were present at the site, and even if
they wanted toeat trout 19% of the time, they could not do so
because trout do not likely existin Middle Marsh due to lack of
adequate habitat. Table 2-2 of the SRI indicatesthat no fish
inhabit the Middle Marsh, and thus the appropriateness of the
foodchain assumptions that are used to derive the sediment quality
criterion arerefuted by EPA's own data. More realistic, verified
food chain assumptionswould further demonstrate that unacceptable
risks to the environment do notexist in Middle Marsh.
I. The SQC methodology Is applicable only if contaminated
sedimentsare submerged for sufficient periods of time to establish
anequilibrium between the sediments and the overlying water
column.This condition Is not true for most of Middle Marsh where
there Is NOtruly aquatic environment.
The Sediment Quality Criteria (SQC) approach is used by EPA for
assessing thedangers of hydrophobic chemicals that can
bioaccumulate in the food chain.The SQC methodology, which assumes
a pollutant equilibrium between waterand solids, allows the water
concentration to be implied from a measurement ofthe pollutant in
the sediment solids. Pg. 4-21 of the SRI points out that"sediment
quality criteria apply to permanently flooded or truly
aquaticenvironments with obligate aquatic species." As discussed
below, no part ofMiddle Marsh meets this criterion, and thus
application of the SQC in MiddleMarsh is inappropriate.
-
August 5, 1991Page 15
Cowardin et al. (1979) found that Area 1 in Middle Marsh is not
a "truly aquatichabitat," as classified by EPA, because the surface
waters are not usually greaterthan 2-m (about 6.6 feet). Nor is
Area 1 a riverine system, which is defined as:
"all wetlands and deepwater habitats contained withina channel,
with two exceptions: (1) wetlandsdominated by trees, shrubs,
persistent emergents,emergent mosses, or lichens, and (2) habitats
withwater containing ocean-derived salts in excess of0.5°/00. A
channel is 'an open conduit eithernaturally or artificially created
which periodically orcontinuously contains moving water, or which
formsconnecting link between two bodies of standingwater'."
The aquatic areas in the Middle Marsh appear to be the Unnamed
Stream andassociated water hazards, not any of the areas targeted
by EPA for remediation,including Area 1.
Wetlands are defined for regulatory purposes by the U.S. Army
Corps ofEngineers and U.S. Environmental Protection Agency as:
"Those areas that are inundated or saturated bysurface or ground
water at a frequency and durationsufficient to support, and that
under normalcircumstances do support, a prevalence of
vegetationtypically adapted for life in saturated soil
conditions.Wetlands generally include swamps, marshes, bogs,and
similarly areas." (EPA, 40 CFR §230.3 and ACOE33 CFR §328.3)
Based on this definition, EPA's classification of Area 1 as a
wetland appears tobe appropriate, while the classification of this
area as "truly aquatic," at least onthe basis of recognized and
accepted definitions, is unwarranted and unjustifiedbecause of the
lack of flooding frequency and sufficient depths of water.
Numerous site visits by personnel from NUS Corporation, E.G.
Jordan, andEbasco Services resulted in no observations of trout (or
any other fish) in theUnnamed Stream, Middle Marsh or the
Apponagansett Swamp. Site visits byconsultants for EPA and
participating party representatives did not identify thepresence of
any open water or flowing streams that could support fish
presentthrough most of the wetland areas. It is highly unlikely
that the areas designatedfor remediation are inundated
sufficiently, or for long enough duration, to allowfor the
establishment of a balanced population of aquatic organisms
includingfish (particularly trout), or of equilibrium conditions
between contaminatedsediment and the water.
-
August 5, 1991Page 16
Aquatic conditions also would not be generated by floods. The
duration of thefloods described in the SRI would inundate the Marsh
for, at most, several hoursor a few days. Even a 100-year flood
would not last for more than a few days,which is much less than the
several months' duration needed to establishchronic toxicity
conditions for fish. Therefore, the SQC methodology is
notappropriate to apply to Middle Marsh.
J. The sediment quality criteria methodology, a highly
controversial andcomplex method for evaluating risk, produces
extremelyunpredictable and Inaccurate results
Sediment quality criteria ("SQC") methodology is unproven and
unreliable. Themethodology must account for many complex mechanisms
yet it contains manysimplifying assumptions in order to be applied.
There are great uncertaintiesassociated with the assumptions,
particularly those involving application of theequilibrium
partitioning method ("EqP"), discussed further below. The
EPA'sScience Advisory Board (EPA, 1990) has expressed significant
reservationsabout the application of SQCs, observing that there are
many uncertaintiesabout the assumptions of the approach and the
water quality criteria upon whichmany of the SQCs are based,
particularly as related to PCB uptake and releasefrom sediments.
The Science Advisory Board also cautioned against theinappropriate
use of water quality criteria as the basis for SQCs. The Boardwarns
that water quality criteria based on residue concentrations, as is
the PCBcriterion, have limited applicability.
In the Middle Marsh SRI, EPA has used the SQC methodology in
exactly theway that the EPA Science Advisory Board rejects. The
Middle Marsh SRIconsiders the PCB water quality criterion which is
based upon residue levels andnot effects levels. The Science
Advisory Board's concern that the "leaps of faith"in the
assumptions needed to link sediment quality to effects levels by
usingresidue values are unsupportable makes the SRI methodology
untenable.
There is a limited amount of data available in the Middle Marsh
SRI that can beused to test the partitioning assumptions of the
SQC, but even this limited dataindicates that the SQC methodology
cannot be relied upon. The SQC used inthe SRI predicts that a
sediment concentration of 19.5 ng per g organic carbonwill
correspond to a pore water concentration of 0.014 jig/« (the value
of theAWQC). Larger or smaller values should scale linearly. A test
of thisassumption requires simultaneous measurements of (1) the
concentrations inpore water and sediments, and (2) the organic
carbon content of the sediment.Such measurements are available for
11 samples from Middle Marsh.
In Figure 1, PCB measurements in pore water samples from Middle
Marsh areplotted against the expected dissolved concentrations, as
predicted by theequilibrium partitioning (EqP) method for a given
sediment concentration andorganic carbon measurement. The solid
line indicates equal measured andexpected concentrations.
-
August 5, 1991Page 17
Clearly, the measured concentrations are very different from
those predicted. Sixof the 11 samples have dissolved concentrations
below the detection limit of0.05 ng/f, and they are plotted using
square symbols on Figure 1. Three ofthese nondetects (ME24, ME29,
and ME02) are not inconsistent with the EqPmethod; the expected
concentration (from EqP) is below the analytical detectionlimit —
Le^, these samples are and should be nondetects. Laboratory results
forME11, ME01, and ME17, however, are not consistent with the EqP
method; themethod predicts dissolved concentrations in excess of
the detection limit, yet thelaboratory reported nondetect. Note
that for ME17, the expected concentrationis >0.3 |ig/( — six
times the detection limit.
For samples with measured concentrations in excess of the
detection limit(ME03, ME04, ME14, ME15, and ME36-indicated by
crosses), the EqP methodunderestimates some sample concentrations
(ME03, ME04, and ME15) andoverestimates others (ME14 and ME36).
Thus, the EqP method for the SQC produces results that differ
significantly frommeasured data from actual Middle Marsh samples.
Expected and measuredconcentrations differ by more than a factor of
ten in several samples. Thisvariance indicates that the method does
not accurately predict dissolution ofPCBs into the water column at
Middle Marsh and thus is extremely unreliable.
K. The SQC methodology Improperly assumes that aquatic
organismssuch as trout will be exposed to pore water concentrations
ofcontaminants
As noted above, the application of the equilibrium partitioning
method ("EqP")to food chain exposures requires that a series of
simplifying assumptions bemade which may not be applicable to
conditions in Middle Marsh. Essentially,EqP assumes a partitioning
of the contaminant between solid sediment particlesand the pore
water that fills the interstitial spaces. These pore
waterconcentrations are then assumed to be available to aquatic
organisms, a falseassumption with the possible exception of its
application to some of the mostminute benthic organisms.
The Sediment Quality Criterion for PCBs is based on the Ambient
Water QualityCriterion of 0.014 ^g/l, which, again, is based on the
assumption that a minkingests contaminated trout. The PCB
concentrations in water accessed by trout,however, are likely to be
much lower than pore water concentrations. PCBs inthe pore water of
a marsh or stream must diffuse from and mix with the watercolumn.
Once in the water column, PCBs volatilize rapidly, with a half-life
ofabout 10 hours, and are also subject to photolytic degradation
(EPA, 1979).Since these rates of removal are typically faster than
the rate at which sedimentscan supply PCBs, concentrations in the
water column are usually substantiallylower than those in pore
water. The data collected in Middle Marsh is consistentwith this
phenomenon where the highest pore water concentration observed
is4.4 fjg/\, while the only detected surface water concentration in
filtered samplesis reported at 0.01 /^g/l (Appendix E.4, SRI). This
limited data set suggests that
-
August 5, 1991Page 18
the PCS sediment quality criterion, which is based upon pore
waterconcentrations, is overstated with respect to estimating
exposures to mink.
The SRI correctly concludes that volatile and semi-volatile
organlcs,heavy metals, and pesticides In the Middle Marsh
sediments, porewaters, and surface waters pose no threat to the
environment
According to the SRI, 13 soil and sediment samples were analyzed
for volatileorganic compounds and 48 soil and sediment samples were
analyzed for semi-volatile compounds. In addition, 10 core samples,
8 pore water samples and 6surface water samples were analyzed for
both volatile and semi-volatilecompounds. The data from this
sampling indicate that the concentrations ofvolatile and
semi-volatile organic compounds in soils and sediments aregenerally
within the range of background concentrations. In the pore water
andsurface water samples, volatiles and semi-volatiles were found
at levels belowor near detection limits.
To determine the presence of heavy metals, 34 soil samples, 10
core samples,8 pore water samples and 6 surface water samples were
collected and analyzed.While the data from the soil sampling
indicated some elevated concentrationsof lead and zinc, the data
from pore and surface water sampling shows thatmany of the heavy
metals were near detection limits and were below ambientwater
quality criteria.
Based on this data, the SRI correctly concludes that no excess
ecologicalhazards are present in the Marsh as a result of measured
non-PCBcontaminants. As Pg 4-1 of the SRI points out, "volatile
organic compoundswere detected infrequently and at concentrations
below detection limits in allmedia.." and therefore "...volatile
organics are not considered a threat to wildlifein the study area".
Additionally, the SRI on page 4-2 goes on to state
(regardingsemi-volatiles); "...semi-volatiles in both wetland areas
appear to be within therange of background concentrations from the
literature that are typically foundin soils near highways". For
these reasons and the fact that the semi-volatilesare at or near
the detection limit in pore water and surface water, the SRI onpage
4-2, concludes that "...semi-volatiles are not considered a hazard
to wildlifein the study area".
Several heavy metals were detected somewhat above background
levels insediment/soil samples in the Marsh, but surface water and
pore water metalsdata, when compared with the Ambient Water Quality
Criteria (AWQC), revealthat the dissolved metals concentrations
were near or below the AWQC for lead,zinc and other metals. The
SRI, pg 4-3, concludes that "due to the low waterconcentrations,
heavy metals have not been evaluated as a hazard to site
biota."
For the reasons discussed in the SRI, EPA properly concluded
that volatile andsemi-volatile organics, heavy metals and
pesticides pose threat to theenvironment in Middle Marsh.
-
Augusts, 1991Page 19
SUMMARY The ecological risk in Middle Marsh is based upon highly
inappropriate food chainexposures to mink, a species not even found
in the Marsh. The ecological risk assessment isdriven by the
Sediment Quality Criteria, a methodology inappropriately applied to
Middle Marshconditions. A realistic ecological risk assessment,
completed for species known to inhabit theMarsh, would show that no
unacceptable risks to wildlife are present due to PCB
contaminationand, therefore, that the "no action" alternative is
the appropriate remedy. The EPA correctlyconcludes that there are
no unacceptable risks to the environment from other
contaminants.
-
August 5, 1991Page 20
III. THE FEASIBILITY STUDY CONTAINS UNREALISTIC ASSUMPTIONS
ABOUTTHE LIKELIHOOD OF SUCCESS OF THE PREFERRED REMEDY,
INCLUDESUNNECESSARY STEPS IN THE PREFERRED REMEDY, AND
DEMONSTRATESNO REASONABLE JUSTIFICATION FOR THE CONTINGENCY
REMEDIALALTERNATIVE.
A. The remedial schedule proposed in the FS is unrealistic.
Remediationof the Marsh will be seriously Jeopardized by the
remediation of theUnnamed Stream (during the remedial action for
the first operableunit) if the actions are not performed
simultaneously
Even assuming that all pertinent permits can be obtained, the
three-month timeframe estimated for completion of remediation
efforts under the proposed Planis unrealistic. As indicated in
Figure 9-2 of the FS, wetland restoration activitiesare scheduled
to commence within one to two weeks of the initiation
ofsoil/sediment excavation and to be completed within approximately
two weeksfollowing excavation. Given the extent of site
preparation, excavation, watertreatment, and wetland and site
restoration, this is totally unrealistic due to theinherent
constraints associated with excavation in a confirmed
wetlandenvironment. The assumed excavation rate of 64 cubic yards
per hour appearsunattainable due to the small working areas, and
the presence of large rootsystems from the existing trees.
In addition, the timing of excavation of the Unnamed Stream
(from Operable UnitOne) will greatly affect the schedule for Middle
Marsh. Problems with equipmentaccess, resuspension of sediments,
and contractor interactions (Sullivan's Ledgesite vs. Middle Marsh
contractors) will adversely affect the already overlyoptimistic
schedule in the FS for Middle Marsh. For example, work in
theUnnamed Stream for Operable Unit One, optimally, should be
performedsimultaneously with excavation in Middle Marsh so that
disturbance of thewetland, and road building, only occur once.
Restoration efforts for eitherOperable Unit will be hampered or
nearly impossible if not performedsimultaneously.
B. There is no reasonable justification for EPA's proposed
contingencyalternative.
EPA's Proposed Plan states that implementation of the Preferred
Alternative isdependent on the Middle Marsh excavation being
performed prior to capping ofthe Disposal Area (Operable Unit 1 -
Sullivan's Ledge Site). However, thisstatement has no engineering
or risk assessment basis. Cap design andconstruction for Operable
Unit 1 can readily be coordinated with Operable Unit2 activities,
so that a portion of the cap (about 1 of 12 acres) is not
completeduntil the excavated sediments can be placed in the
disposal area.
Delay in completing a small portion of the cap will result in no
significant impactto human health or the environment because
appropriate storm water controlsand diversion ditches will be in
place and functional, and overburden ground
-
Augusts, 1991Page 21
water will be intercepted by the perimeter trench. Soil under
the portion of theSullivans' Ledge disposal area which may not be
capped, awaiting the Marshsediments, will already have been
excavated, solidified and stabilized, thusposing no unacceptable
exposure risks.
Any increase in cap construction costs attributable to delay is
insignificant whencompared to the $5M cost differential between the
preferred and contingencyalternatives.
In the unlikely event that the disposal area is not ready to
receive the excavatedMarsh sediments due to a delay in the remedy
at the Sullivan's Ledge disposalarea, there is no significant risk
to human health or the environment fromtemporary storage of the
excavated Marsh sediments, until the disposal area canaccommodate
the same. Sediments can be placed on a lined and bermedreceiving
pad, and covered, to prevent drying and blowing of
particulates.Rainwater could be diverted and the storage site
protected from animal orhuman damage with construction of a proper
fence.
C. Backfilling the disturbed wetlands, as required in the
preferredremedy, is not absolutely necessary and requires
destruction ofwetlands elsewhere in order to obtain the most
suitable materials.
The FS would require replacement of the entire volume of
sediments removedfrom the wetlands area. In question is the source
of the replacement fill materialfor each of the wetland areas which
would be excavated. Although the FSalludes to a variety of
potential sources, the actual source(s) of suitable materialshould
be identified well in advance. Ideally, soils obtained from other
wetlandswould increase the likelihood for restoration success. If
not available locally, thefeasibility of finding a source of
suitable replacement fill and obtaining permitsfor its excavation,
is very questionable.
Placement of fill in the remediation areas will also result in
problems withturbidity and will cause an increase in sedimentation
in downstream areas. Itmay well be environmentally preferable to
revegetate the excavated wetlandswithout replacing the excavated
sediments. This would result in less disturbanceof the wetland and
portions of the remediated area may develop standing water,due to
the lowering of the ground surface by 18 inches, allowing for
greaterdiversification of the Middle Marsh environment.
Wetland restoration will also cause harm to other wetlands. As
described in theFS, the soils excavated from the wetlands may be
replaced with wetland soilmade from soil mixed with organic
material and sediments dredged from a lakeor pond. Ideally, the
replacement soils should be obtained from a similarwetland that is
to be excavated for by development. It, thus, appears that
onewetland has to be destroyed in an attempt to restore another.
But this remedynot only destroys the Middle Marsh wetland, it will
cause far reaching damageto another.
-
August 5, 1991Page 22
D. There Is no Justifiable reason to manage the cleared
vegetation as ahazardous waste
The feasibility study (FS) suggests that "Contaminated stumps
and loosevegetation should be contained with other contaminated
materials for treatmentand/or disposal" (FS, p. 7-20). If handled
properly (e.g., not dropped intocontaminated mud), these materials
should not contain significant concentrationsof PCBs and could be
managed as solid waste under 310 CMR 19.100-143 ofthe Massachusetts
Landfill Design and Operational Standards.
The translocation of PCBs from soils/sediments to plants is
small. An uptakefactor can be used to relate the concentration of
PCBs in soil/sediment to theconcentration in plant tissue.
[^vegetation = UP13*6 factor * [^soil/sediments
Travis and Arms (1988) report an empirical value of 0.017
(non-dimensional) forthe PCB uptake factor. This relationship can
be applied to data from the MiddleMarsh. For the current highest
point-value of PCBs measured in sediment/soilsamples (28 mg/kg,
sample ME22), a concentration of 0.5 mg/kg would becalculated in
plants using the above relationship. This value is lower than anyof
the terrestrial cleanup criteria derived in the SRI. This
worst-case value is alsoconsistent with non-detected results for
measurements of PCBs made in rosehips and grass seed heads (at a
detection limit of 0.1 mg/kg) (p 2-24 of theSRI).
Therefore, the theoretical predicted maximum concentrations of
PCBs invegetation are consistent with the observed levels for
vegetation, demonstratingthat the cleared vegetation should not be
managed as a hazardous waste.
E. Production rates and treatment technologies stated in the FS
areunrealistic and likely unattainable
The FS assumes the use of a 100 ton per day treatment unit for
treating PCBcontaminated excavated sediment. A more realistic
treatment rate for a site withthis volume of sediments is 25 to 40
tons/day due to the size of equipmentspecified by EPA to be
mobilized. This would increase the estimated timenecessary to treat
the estimated 5,200 yd3 of contaminated sediments fromthree to six
months, and possibly longer. This increased treatment time
greatlyimpacts the overall remedial schedule and increases the cost
of the remedysubstantially.
Since the PCB treatment could take longer than the 3 months
estimated in theFS, the estimate of 6 months (Section 9.3.6) for
total remediation could be closerto 9 months, possibly interfering
with the mating season of the Spotted Turtle,a species of special
concern.
-
Augusts, 1991Page 23
In addition, the treatment technology to be implemented under
the contingencyremedy may not be capable of achieving the required
reduction in PCB levels.The contingency remedy calls for sediments
to be treated on-site with solventextraction and then be replaced
back into the excavated areas.
Although the FS evaluates (Section 9.3.2) the overall protection
of theenvironment using solvent extraction in regard to reduction
of PCBs, it does notappear to have addressed the issue of residual
toxicity associated withbyproducts from the BEST preferred solvent
extraction treatment method, thedesignated EPA-selected
treatment.
It should be noted that the BEST process removes oils, not
specifically PCBs.The BEST process may not yield significant PCB
reduction in a soil sample withPCBs but low oil and grease.
Therefore, if the contingency alternative isretained, the KPEG
process should have been included by EPA as an alternativetreatment
technology for treatability testing given the difficulties noted
above withrespect to the BEST process.
It is entirely possible that no known treatment technology may
be successful atthe site. Remediation of Area 1 (based upon the
normalized PCB exceedancesat ME5 and ME19) would require that
actual PCB concentrations not exceed 0.4ppm and 1.6 ppm. These are
levels which may not be attainable with ANYknown treatment
technology.
SUMMARY The timetable for implementation of the preferred remedy
is likely to be at leastsix months, not the three months
erroneously estimated in the FS. The FS does notdemonstrate an
unreasonable exposure, or engineering requirement, which justifies
theimplementation of the proposed contingency alternative. Instead,
the need for the contingencyalternative is unsubstantiated, and
that alternative will fail because there are currently no.
knowntreatment technologies which can attain the PCB levels
required by the cleanup criteria in MiddleMarsh.
-
August 5, 1991Page 24
IV. THE PREFERRED ALTERNATIVE WILL SIGNIFICANTLY HARM
THEWETLANDS, WITHOUT ANY ASSURANCE THAT THE WETLANDS CAN
BEADEQUATELY RESTORED.
A. The proposed plan would require excavation of more
wetlandacreage than necessary, even accepting the SQC In the
SRI.
Figure 6-1 of the FS identifies four areas to be remediated. The
extent of theareas to be remediated is considerably larger than
necessary dictated by theecological remediation criteria proposed
in the SRI. However, the criteria, whenappropriately applied to the
entire Marsh on an average basis, define smallerareas to be
remediated. Additionally, following the same area averaging
logic,there is no need for remediation at all in Areas 2, 3 and 4
to meet the mink-based criteria.
Table 6-2 of the FS summarizes the various ecological cleanup
criteria that arederived in the SRI. As described in the SRI, the
proposed target concentrationof 15 mg/kg total PCBs is used to
establish the extent of Areas 2, 3, and 4,while a target
concentration of 19.5 jig PCB/g carbon is used to determine
theextent of the remediation required in Area 1. These cases are
discussed, in turn,below.
• Areas 2, 3, and 4
In these areas, the proposed target concentration of 15 mg/kg
appears to bebased on the terrestrial feeding habits of mink, for
which a contamination levelof 14 mg/kg may be harmful according to
the SRI. Assuming female mink maylive in the Middle Marsh, it would
not be exposed, on average, to contaminationin excess of the 15
mg/kg criterion. On average is emphasized because, as theSRI points
out, the female mink requires a range of at least 20 acres for
feeding.Since the mink will obtain food from all points of the
site, the average PCBconcentration over the entire feeding area
should be used to compare to thecriterion. This site-average PCB
concentration is calculated to be 5.3 mg/kg, thederivation of which
is presented in Appendix 1.
Since the site average concentration of 5.3 mg/kg is smaller
than theremediation criterion of 15 mg/kg, remediation is not
required to meet thiscriterion. As the average PCB concentrations
across the site area in Areas 2, 3and 4 are below this cleanup
criterion, no remediation should be required inthese areas.
• Area 1
The cleanup criterion used in the SRI to estimate the extent of
remediationrequired in Area 1 is the interim sediment quality
criterion of 19.5 ng PCB/gcarbon. As mentioned above, this
criterion should be applied on an areaaverage basis, since a female
mink, if present, could obtain food throughout thearea and fish, if
present, would be expected to swim throughout the aquatic
-
August 5, 1991Page 25
environment, if present. An area-average calculation is
presented in AppendixA. The area to be remediated in Area 1, would
be 300 m2, or about 0.06 acres,which is considerably smaller than
the area of 0.36 acres suggested in the FS.Cleanup of this smaller
area (which is heavily weighted by one sample ME5)would result in
an average concentration in Area 1 of 16.5 jig PCB/g carbon,which
is below the criterion.
However, the overriding point with respect to Area 1 is that EPA
has determinedthe need for remediation based upon only one sample.
Even if one could relyon the accuracy of the analysis for PCBs or
TOG, which is doubtful, it isunreasonable, and scientifically
unacceptable, to base a remedy decision on asingle extreme sample.
In most instances, compliance with a certain criterionis achieved
where the mean or the median is lower than the criterion. The
above-described area-averaged analysis described above demonstrates
that anextreme value (in this case a low value for TOC at ME5) is
driving the remedyof Area 1. This indicates that the EPA's decision
to remediate Area 1 is basedupon an unreasonable reliance on an
outlying data point, which is not goodscience.
B. The proposed plan will disturb and/or destroy the wildlife It
Issupposed to protect
The proposed remedy is an extremely destructive approach for
remediating thelow levels of contamination in Middle Marsh. The
remedy requires excavationof approximately 18 inches of sediment in
the designated areas. The upperlayers of this sediment are teeming
with small organisms that will be destroyedby the excavation. All
vegetation, nesting areas, food sources - in short, theentire
habitat - will be completely destroyed in the remediation areas.
The shymink, even if present in the Marsh, would be driven out of
the Marsh during themonths of road building, excavation and
restoration activities, and would remainout of the Marsh afterward,
because the post-remediation habitat would remainunsuitable for
mink to survive. Stream dredging, stream diversion anddewatering
would dry up and destroy all aquatic life. The restoration
measureswould need considerable time - many years at least - to
become reestablished(trees to grow, grasses and bushes to become
reestablished), during which timethe Marsh will take on a vastly
different character.
Middle Marsh likely will never become reestablished to the same
extent and typeof biological diversity as existed prior to
remediation. Because forested wetlandswill, in part, be altered by
the proposed plan, and will not be replaced in kind fora period
many of years following initial restoration activities, the
functions andvalues of Middle Marsh relative to wildlife may well
experience a permanent netdecline. This is particularly true with
respect to the Spotted Turtle, a State-listedspecies of special
concern, known to inhabit Middle Marsh, which could beextremely
difficult to reestablish.
The proposed plan describes impacts to approximately 1.9 acres
of wetlands.In fact, this acreage only relates to the areas to be
excavated and not other
-
August 5, 1991Page 26
areas that will be disturbed due to associated activities
required by the proposedremedy. For example, additional acreage
would be adversely impacted byconstruction of roadways and sediment
basins. These activities, no matter howcarefully performed, will
damage a considerable area of the Marsh andpotentially mobilize
contaminated sediments into non-contaminated areas.
C. The proposed plan poses a substantial and unreasonable danger
ofdestroying the habitat of the spotted turtle, a species of
specialconcern in Massachusetts
Unlike mink, spotted turtles, a species of special concern in
Massachusetts, areknown to inhabit at least a portion of Middle
Marsh. The FS (pages 9-35) statesthat, in the event spotted turtles
are discovered in the proposed remediationareas (it is doubtful
that heavy equipment operators will look for them), they areto be
caught and translocated to uncontaminated areas of the wetland.
Thisproposed mitigation measure raises serious concerns.
The availability and extent of suitable habitat for spotted
turtles elsewhere on-sitehas not been documented. Whether or not
other suitable habitats exist toaccommodate the spotted turtles, in
addition to other spotted turtles that mayalready use the receiving
area, is also unknown. The effects of translocation onindividual
turtles and the effects of proposed activities on turtle movement
bothwithin and outside Middle Marsh, are also unknown. If suitable
sites areunavailable, the breeding success of translocated spotted
turtles will beadversely affected. In the long term, the spotted
turtle habitat conditions wouldsuffer prolonged and likely
permanent change, particularly given the extendedlength of time
required for disturbed forested wetlands to become
reestablished.
D. The proposed wetlands restoration plan is inadequate and
notconsistent with existing wetland species
The proposed remedy would ostensibly alter 1.9 acres of
wetlands, all of whichare associated with areas of soil/sediment
excavation. All of this impacted areawould have to be restored. The
actual wetland impact acreage would be muchhigher. Other areas
which would be disturbed under the proposed remedy, e.g.,the
proposed staging area and access roadways, were not included in
therestoration acreage estimate but should be included for purpose
of the remedialalternatives evaluation. While all such areas may
not be wetlands, they areadjacent to wetlands. Measures to minimize
impacts to adjacent wetlands wouldneed to be implemented, such as
in the forested wetland adjacent to theproposed staging/treatment
area.
The proposed planting plan in the restoration plan lacks
biological diversity. Theexisting diversity of plant species in
each of the proposed excavation/restorationareas are considerably
higher than proposed in the restoration plan. Increasingthe
diversity of proposed plantings with species characteristic of
theexcavation/restoration areas would, therefore, be more
consistent with existingconditions.
-
August 5, 1991Page 27
Furthermore, the planting plan proposes to introduce species
which may beunsuitable for the location. For example, the overall
revegetation plan fordisturbed wetland areas calls for the
introduction of red oak (Ouercus rubra) andarrowhead (Sagittaria
spp.). According to Reed (1988), red oak is an uplandspecies. This
species exists in the vicinity of Middle Marsh, but generally not
inthe low-lying areas like those to be remediated. Its potential
for survival in wetenvironments with high ground water tables is
far reduced relative to the nativered maple (Acer rubrum). In
contrast, arrowhead is characteristic of pond edgesand shallow
waters. As such, it may not be suitable for revegetative
purposesunless suitable water conditions, including ponded waters,
are established.Arrowhead is not listed in the inventory of
existing plant species on sitecontained in Table 2-2 of the SRI.
Thus, its use in Areas 2 and 4 appearscontrary to the stated
objective of providing similar vegetation to that which isremoved
during remediation activities.
SUMMARY The preferred alternative will cause more harm than good
to the environment inthe Marsh because of long-term wetlands
damage, potential destruction of the spotted turtlehabitat, and
potential mobilization of contamination as a result of site
activities.
-
August 5, 1991Page 28
V. THE FEASIBILITY STUDY, BASED UPON QUESTIONABLE DATA
ANDMODELLING STUDIES, FAILED TO ADEQUATELY INVESTIGATE
ALLPOTENTIALLY APPROPRIATE REMEDIAL ALTERNATIVES, AND FAILED
TOINCLUDE SIGNIFICANT FACTORS IN EVALUATING THE PREFERRED
ANDCONTINGENCY ALTERNATIVES.
A. The data which Is relied upon In the FS regarding PCB
concentrationsin surface water and pore water samples contain
severaldiscrepancies.
The purported need for remedial action is based on reported
exceedances ofAmbient Water Quality Criteria (AWQC) for PCBs in the
surface water and porewater samples. The SRI reports that, in pore
water samples, the dissolved PCBconcentrations ranged from
undetected to 4.4 \ig/t, and the total PCBconcentrations ranged
from undetected to 29 \ig/t. In the surface watersamples, the
dissolved PCB concentrations ranged from undetected to 0.19
t, and the total PCB concentrations ranged from 0.98 to 4.9
\ig/t.
The highest concentration of dissolved PCBs in the pore water
samples wasreported to be 4.4 \ig/l. However, PCBs were not
detected in the unfilteredsample from the same location. It is
impossible for PCBs to be present in adissolved or filtered sample
when they are not present in a total or unfilteredsample from the
same location. In addition, the highest concentration of totalPCBs
in the pore water was reported to be 29 \ig/t, but this result was
notfound in the analytical tables included in Appendix E. For
surface watersamples, pg 3-31 states that PCBs were detected in
filtered surface samples -values of 0.19 /^g/« and 0.077 fjg/t are
listed. Appendix E-4, however, showsnondetects for all samples
except SL15, for which the PCB concentration isestimated to be 0.01
vg/l. It should be noted that the detection limit forsamples SL01,
SL04, SL15 and SL17 appears to be 0.01 ttg/t, a value that isbelow
the AWQC of 0.014
In another example, for the unfiltered sample ME04, Aroclor 1254
was notdetected (detection limit: 0.05 ng/J) and Aroclor 1260 was
found at aconcentration of 1 .7 jig/t. However in the filtered
sample, Aroclor 1260 was notdetected (detection limit: 0.05 jig/«)
while Aroclor 1254 was found at 1.1 \ig/t.Since Aroclor 1254 was
not detected in the unfiltered sample, one would notexpect to find
it in the filtered sample. Similarly, for sample ME29, Aroclor
1254is detected in the unfiltered sample, and Aroclor 1260 is found
(at a lowerconcentration) in the filtered sample.
The discrepancies in the data call into question the reliability
of the data and, inturn, the need for remedial action based on
exceedances of the ambient waterquality criteria (AWQC).
In addition, review of the analytical data for water samples in
the appendices ofthe SRI indicates that ail of the PCB
concentrations reported by the analyticallaboratory are labeled
with the suffix "J": this indicates that the reported
-
August 5, 1991Page 29
concentration is an estimated value that the laboratory has
tentatively identified.Page 5-15, of the 1989 Volume I Interim
Final Manual for Risk AssessmentGuidance, states that if the "J"
qualified data contribute significantly todetermination of the
estimation of potential risk, then the report should notepotential
uncertainties associated with the data and proper caveats should
beattached. This was not done for the critical PCB data in Middle
Marsh.
B. The TOG Analytical Method produces inconsistent, unreliable
resultswhich are not reproducible to even an order of magnitude In
thelaboratory
EPA's reliance on the total organic carbon (TOC) measurements
(orassumptions of average values for the Marsh) is suspect. The
areas to beremediated are defined by a "normalized" PCB value,
which is derived bydividing the PCB concentration by the TOC value.
If the TOC value is unreliable,then the normalized PCB value is
unreliable and should not be used.
EPA analytical test methods have not been properly developed for
TOC, andaccording to laboratories performing TOC analyses for EPA,
the analysesproduce extremely variable results with "duplicate"
samples varying an order ofmagnitude and up to three standard
deviations (personal communication, R.Foster, RAI, July, 1991; M.
Perkins, RAI, July 1991).
Table 2-12 in the SRI indicates TOC analyses were performed by
CLP-SAS testprocedures. This test method uses very small sample
sizes (10-100 mg) andtends to yield results of low precision
(Analytical Resources, Inc., 1991).Analytical variability of
duplicates of samples with relatively high TOC (greaterthan a few
percent) is generally high, possibly up to a full order of
magnitude.This likely accounts for some of the variability in TOC
values observed insamples from Middle Marsh. TOC data from the
Middle Marsh range from 7,700PPM (SRI Table 3-1) to 480,000 PPM.
Specifically, remediation in Area 1 isdriven by the abnormally low
TOC value for ME5 (22,000 PPM).
PCB concentrations in Area 1 range from 0.325 to 11 mg/kg. When
normalizedby organic carbon, however, values range from 0.68 to 218
VQ PCB/g carbon.This increase in range could well be the result of
uncertainties in themeasurement of organic carbon.
C. The basic assumptions of the hydroloaic computer models are
notIncluded In the SRI or available for review, either by EPA or
thepublic. The limited Information on the hydroloqlc computer
modelingwhich is available Indicates that the models may not have
beencalibrated correctly
The computer models used to predict the flood flows and the
spatial extent offlooding are critical components of the SRI. EPA
assumes, based on thismodeling, that the source of contamination to
Middle Marsh is from theSullivan's Ledge disposal area. In
addition, the models are used to predict the
-
August 5, 1991Page 30
extent of flooding so that the sampling program can be designed
to investigatethe most likely areas of high contamination. The
analytical results of thesampling plan then determine the areas to
be remediated. Hence, the modelsplay an integral part in the Middle
Marsh remedy.
The referenced models are very sensitive to assumptions
regarding inputs suchas basin characteristics, stream routing
coefficients, and stream roughnessfactors (US Army Corps of
Engineers Users Manuals for HEC-1 and HEC-2,January 1985 and
September 1982a). Yet, information concerning data inputsto the
models is not included in the SRI. This information is critical for
anindependent reviewer to determine if the Unnamed Stream was
modeledproperly in the SRI. EPA should have required this
information in order to verifythe EPA consultant's work i