Methods and Tools for the
                    Evaluation of Monitored
                    Natural Recovery of
                    Contaminated Sediments:
                    Lake Hartwell Case Study
Office of Research and Development
National Risk Management Research Laboratory

                                          EPA/600/S-10/006/September 2010/
                                                              for the
Office of Research and Development
National Risk Management Research Laboratory


Purpose	1
Introduction	1
Site Background: Lake Hartwell/Twelvernile Creek	2
Research Study Phases	2
Methods to Assess MNR Processes	3
  Chemical Lines of Evidence	4
  Biological Lines of Evidence	6
  Physical Lines of Evidence	12
Conclusions	17
References	19

                                              Purpose  and  Introduction

The National Risk Management Research Laboratory
(NRMRL) of the U.S. Environmental Protection
Agency's (U.S. EPA's) Office of Research and
Development (ORD) has been conducting research
to develop methods and tools for the evaluation of
monitored natural recovery (MNR) of sediments
contaminated with polychlorinated biphenyls
(PCBs), polycyclic aromatic hydrocarbons (PAHs),
mercury, and other legacy pollutants. This research is
supporting a broad, national research program focused
on contaminated sediments in U.S. waterways. This
Research Summary provides a synopsis of a multi-
year, interdisciplinary research  project conducted by
ORD. specifically NRMRL and the National Exposure
Research Laboratory (NERL), at the Sangamo-
Wcslon, Inc./Twelvc-Milc Creek/Lake Hartwcll PCB
Contamination Superfund Site in Pickens County,
South Carolina.  The methods and tools described in
this Research Summary are comprised of quantitative
approaches for characterizing naturally-occurring
mechanistic processes that are necessary to manage risk
using MNR.

The information developed in this project is expected
to be used as a reference for site managers and Federal,
Stale, and local regulators who  may be considering
MNR as a site remedy or monitoring the progress of
MNR at a contaminated sediment site(s).  The methods
and techniques evaluated and optimized in this study-
also provide the broader sediment  community with an
approach for characterizing environmental processes
controlling the risk associated with contaminated

Sediments are often the ultimate receptors of
contaminants in aquatic systems. According to an
estimate by U.S. EPA, approximately 10% (-1.2 billion
cubic yards) of the sediment underlying the country's
surface water is sufficiently contaminated to pose
potential risks to humans and wildlife (USEPA, 1998).
To manage the risk imposed by these sediments, it is
critical to understand the fundamental mechanisms
responsible for contaminant transport and fate. These
mechanisms include chemical, biological, and physical
processes responsible for the risk associated with the
sediment.  Metrics or approaches to characterize the
effects of sediment remediation are needed to provide
quantitative measures of success.
MNR is one approach for managing the risk associated
with contaminated sediments. With MNR, the
contaminants are left in place and existing physical.
chemical, and biological processes that are expected to
contain, destroy, and/or reduce their bioavailability or
toxicity are monitored to assure effective management
is progressing as predicted. These processes must
be demonstrated to be viable, and occurring at an
acceptable rate, to  attenuate the risk of the contaminant
remaining in place. Similarly, an effective and
quantitative monitoring plan must be able to demonstrate
definitively that these processes are progressing as
predicted and  that risk is being reduced over time at the
anticipated rate.
MNR relies on a few primary mechanisms for mitigating
contaminant risk to human and ecological receptors.
First and foremost is a reduction in the contaminant's
availability to receptors. This reduction can be
accomplished  either  through a physical process of
isolating the contaminated sediment itself or through
a chemical process of contaminant sequestration or
sorption. Physical isolation typically results from
deposition of sediments not contaminated with  the
compound-of-concern (COC).  Generally, exposure to
contaminated  sediment is expected to occur either in the
water column  during transport or, more commonly, at the
water/sediment interface once deposited.  Depending on
the water body, contaminant exposure in the near surface
sediment would be expected to occur in the biologically
active zone of the sediment.  This biologically active
layer can vary from as little as a 1-3 cm to as much as
1 3 m in thickness depending on the habitat (Clarke,
2001.). The deposition of sediment, defined as
accretion, is a natural on-going process in most water
bodies. If the  source of the COC has been effectively
managed (a term often referred to as source control), the
contaminated  sediment will become physically isolated
from the primary receptors in the aquatic system.
However, sediment accretion rales vary spatially and
temporally, are dependent  on the  hydrologic and physical
conditions at the site, and are subject to change with
alterations within the watershed or in the site conditions.
MNR requires confirmation that a site continues to be a
deposition-dominated site  over time without significant
mixing of the  near-surface sediment.

Another important mechanism in MNR is often the
chemical sequestration, or sorption, of the contaminant to
the sediment matrix. With organic COCs, this sorption
process is generally most favorable with hydrophobic,
non-volatile contaminants binding to organic matter
within the sediment. With inorganic COCs and less,
commonly,  organic COCs, the binding of the contaminant
may occur with the inorganic clay and mineral structures
of the sediment. In either instance, this sequestration
or sorption  of the contaminant has been shown, in some
cases, to be effective in reducing the ability of a receptor
(human or ecological) to take up the contaminant if
exposed.  This reduced exposure may reduce the risk of
the COC. However, studies have shown the sequestration
of contaminants, whether organic  or inorganic, and any
reduced availability are related to  site-specific conditions
(sediment type, organic matter content and structure,
oxidation/reduction state, presence of sulfide, etc.).
These conditions, therefore, should be evaluated on a
site-specific basis during initial design for MNR and
should be incorporated into the long-term monitoring
plan Lo assure site conditions remain viable for continued
reductions in contaminant availability  (Magar, 2009).
The biological degradation, or biodegradation, of the
COC should also be evaluated where considering
MNR.  If these processes are determined to be important
mechanisms for either changing the availability or
toxicity of the  COC or reducing the inventory of the
COC, biodegradation evaluations  should be included as
part of the long-term monitoring plan.  Biodegradation is
dependent on the physical/chemical characteristics of the
contaminant itself and the viability of microfauna capable
of biodegrading the COC. Biodegradation can reduce or
increase the toxicity of COCs depending on the biological
process and/or the specific COC.  Finally, site conditions
can inhibit or control the environment  needed for optimal
biodegradation rates.
In 1994, U.S. EPA Region 4 issued a Record of Decision
(ROD) (USEPA, 1994) establishing MNR as the remedy-
of choice to remediate Operable Unit #2 (OU2) of the
Sangamo-Weston, Inc./Twelve-Mile Creek/Lake Hartwell
PCB Contamination Superfund Site (Figure 1). OU2
consisted of the contaminated sediments in Lake Hartwell
and Twelve-Mile Creek (TMC). OU2 was used as a
field location by U.S. EPA-ORD to develop, test, and
validate methods and tools to support the evaluation of
MNR of PCB-contaminated sediments. A comprehensive
set of measurements and analyses were conducted over
a number of years to develop chemical, biological, and
physical lines of evidence to document  the progress of the
MNR remedy in reducing PCB concentrations in surface
sediments and the associated ecosystem.

Sangamo, Inc. owned and operated a capacitor
manufacturing plant in Pickens, South Carolina, from
1955 to 1987. During operations, the facility used
several varieties of fluids that contained PCBs.  Waste
was disposed of on the plant site and at six satellite
disposal areas within a 3-mile radius of the plant.  PCBs
were also discharged with industrial wastewater effluent
into Town Creek, a tributary of TMC. TMC is a major
tributary of Lake Hartwell. Between 1955 and 1977,
the average quantity of PCBs used by Sangamo-Weston
ranged from 700,000 to 2,000,000 Ib/yr. An estimated
3% of the quantities received and used by the plant were
discharged into Town Creek, resulting in an estimated
cumulative discharge of 400,000 Ib of PCBs. PCB use
was terminated in 1977, prior to a U.S.  EPA ban on its
use in January 1978  (USEPA 1994).
Lake Hartwell is a U.S. Army Corps of Engineers
(USAGE) reservoir located in the northwest corner of
South Carolina along the Georgia state line (Figure 1).  It
is bordered by Anderson, Pickens, and Oconcc Counties
in South Carolina and Stephens, Franklin, and Hart
counties in Georgia. It was created between 1955 and
1963 when the USAGE constructed Hartwell Dam on the
upper Savannah River, 7 miles from the confluence of
the Seneca and Tugaloo Rivers. Lake Hartwell extends
49 and 45 miles up the Tugaloo and Seneca Rivers,
respectively. At full pool elevation (660 ft MSL), it
covers nearly 56,000 acres of water with a shoreline of
962 miles (USEPA,  1994).
U.S. EPA-ORD conducted research at the Lake Hartwell/
TMC site between 2000 and 2008. This research evolved
in phases as the study progressed and more information
was required.
Phases I and II of the project were completed in 2000-
2002 and consisted of chemical and age dating analyses
of sediment core samples collected longitudinally and
laterally on 10 transects previously established by U.S.
EPA Region 4 for ongoing monitoring at Lake Hartwell
(USEPA, 1994; USEPA, 2004). These studies focused
on the use of chemical analyses of surface sediments
and sediments at depth to characterize the decline of
surface scdimcnl concentrations over lime (Brenner,
Magar et al., 2004).  Additionally, evidence of anaerobic
dechlorination of higher-chlorinated PCBs was shown to
be occurring at depth in the sediment, albeit, at very slow
rates (Magar, Johnson et al., 2005).

Results from Phases I and IT along with the required ROD
annual monitoring data (USEPA, 2004)  were evaluated

in developing the subsequent studies at Lake Hartwell
and TMC. Surface sediment PCB concentrations were
declining at predicted rates;  however, reductions in
fish tissue did not appear to be declining similarly as
projected. The result was an expansion of the project
within ORD that encompassed Phases III-V over the
next 6 years.
Phase III (2002-2003) served as a proof-of-concept effort
in the development of innovative tools for monitoring
the mechanisms responsible for the transport and  fate
of PCBs in this system. Measures during Phase III
focused on the elements that comprise the Lake Hartwell
and TMC ecosystem including environmental matrices
(sediment, benthos, and water column) and food web
components (fish, macroinvertebrates, plankton, and
organic matter). Additional  studies concentrated on the
diffusion and advection of PCBs from contaminated
sediment/pore water into the water column,  as well as
surrogate biological measures using passive samplers.
Phase IV (2004-2005) and V (2005-2008) of these
studies focused on extending selected data sets  for
characterizing long-term trends and further validating
selected methods and tools from prior studies at Lake
Hartwell/TMC and other sites.  Phase IV focused on
characterizing seasonal variability of the mechanisms
monitored. Phase V consisted of further food web and
biotic studies along with studies of surface sediment
stability and mixing.
Methods to Assess MNR Processes

MNR relies on several primary mechanisms to exist
and progress at a predictable rate. These mechanisms
generally involve naturally-occurring chemical,
biological, and physical processes. Often, these
processes are unrelated mechanistically with one another
and may occur at different temporal and spatial scales.
As such, multiple lines of evidence may be required
to demonstrate success or progress of MNR toward
the remedial objectives or goals. These site-specific
lines of evidence are broadly categorized as chemical,
biological, and physical lines of evidence. Through a
                        Figure 1.  Location of Lake Hartwell and Twelvemile Creek Site.

weight-of-evidence approach, the progress of MNR can
be evaluated.
Conceptual Site Model

Prior to evaluating a remedy for a contaminated
sediment site, a conceptual site model (CSM) should
be developed to identify the key processes affecting the
transport and fate of the contaminant and receptors at
risk. Once the designed remedy has been completed,
verification of remediation effectiveness is required.
MNR may require long-term monitoring to verify that
the naturally-occurring processes necessary to manage
the contaminant risk are proceeding as predicted. The
CSM again should be used to assist in the choice of
critical sample matrices and the periodicity required
to demonstrate progress.  Figure 2 shows the CSM
generated for Lake Hartwell to aid in the development of
monitoring tools for documenting the lines of evidence
needed to characterize the progress of the remedy.  This
CSM was utilized by U.S. EPA researchers to guide
the development of sampling tools and methods briefly
described in this Research Summary document.
Chemical Lines of Evidence

In most cases, direct measurement of the COC in surface
sediments is necessary to document its reduction over
time.  At Lake Hartwell, surface sediment was defined
as the top  10 cm of the sediment. This depth was
established in the Remedial Investigation /Feasibility
(RI/FS) study and was followed throughout this  research.
Surface sediment samples were collected for all  phases
of the research in Lake Hartwell and, where available,
in TMC. Direct measures of the COC are a primary
line of evidence in MNR, and several approaches were
evaluated  at Lake Hartwell.
  Unconlaminated Sediment load from Lake Harwell
  tributaries (e.g., Keowee and Seneca Rivers) and
          forest and undeveloped land
                      PCB Volatilization and
                     Atmospheric Deposition
                                       Large Ftsti (*£., Hybrid   Small Fish IS.Q..
                                       and Largemoirlh Bass|  sh iners and shad)
                                                                                               Flow Inward
                                                                                               Lake Harnwefl
4                                                                                                 Suspended
                                                                                              sedmenl transport
                    Figure 2. Conceptual Site Model for Monitored Natural Recovery as applied
                    to Lake Hartwell (adapted from(Magar 2009).

When conducting a site investigation, it is important to
ask: "What defines surface sediment?" The definition
of surface sediment can be based on the depth of
expected mixing due to hydrological events (high flow
condition, propeller wash, etc.) and the biological
active zone (BAZ) (MacDonald, Ingersoll et al., 2000;
Eggleton and Thomas, 2004).  The BAZ is inhabited
by infaunal organisms including microbes, meiofauna,
macroinvertebrates, and other organisms  (i.e., early life
stages offish or amphibians) that spend all or part of
their lives associated with sediments.  The community
of organisms present generally depends on the physical
and chemical characteristics of the water  body as
determined by the watershed.  The depth  of the BAZ
varies depending on sediment substrate characteristics
(including particle size fractions, organic matter
content, consolidation, and pore water geochemistry).
These characteristics control the organisms present.  In
freshwater systems, the BAZ typically spans the top 20
to 40 cm of the surface sediment (Clarke, 2001; Nogaro,
Mermillod-Blondin et al., 2009). The  majority of benthic
organisms will usually be associated with the upper 15
cm. However, certain invertebrate and/or amphibian
species can extend the BAZ during a portion of their life
history (e.g., up to 100 cm below the sediment surface)
(Fleeger, Tita et al., 2006). Understanding the depth of
the BAZ is critical in designing sediment and biological
sampling plans for MNR sites.
Sediment Concentrations and Age Dating.
Surface sediment concentrations were measured over
the duration of the studies and combined with the data
on sediment concentrations for samples collected under
the ROD.  In addition, core sampling was completed at
selected transects to characterize the deeper sediments
that reflect changes in historic surface sediment
concentrations over time (Brenner, Magar et al., 2004).
Results indicated that age dating sediments can provide
a historic record of the surface sediment concentrations
and, in this case, demonstrated a trend toward reduced
concentrations of total PCBs (t-PCBs) over time (Figure
3). Researchers provided predictions of recovery time
to reach human and ecologically significant surface
sediment concentrations based on historic sediment
deposition rates (Table 1).  In addition, researchers also
0-5 cm
5-1 Ocm
15-20 cm
20-25 cm .
25-30 cm
E 30-35 cm
0 35-40 cm
•%_ 40-45 cm
o 45-50 cm
50-55 cm
55-60 cm .
60-65 cm
65-70 cm
70-75 cm
75-80 cm
80-85 cm
i 	 t





i i


i 1


; = 1


                                            20            30            40
                                                Concentration (mg/kg dry weight)
            Figure 3.  Total PCB (t-PCB) Vertical Concentration Profile in Core L with Age Dated Sediment.

Table 1. Required Sediment Deposition (in cm) to Achieve Selected t-PCB Surface Sediment Concentration Goals.
Sedimentation to
Achieve 1 mg/kgw
t-PCBs (cm)
2.7 ±3.0
Sedimentation to
Achieve 0.4 mg/kgw
t-PCBs (cm)
10 + 4.7
Sedimentation to
Achieve 0.05 mg/kgw
t-PCBs (cm)
32 ±13
       W ROD Surface sediment cleanup goal (U.S. EPA, 1994).
demonstrate exposure or effects. These metrics can use
organisms that range from fish that have both ecological
and human health risks to benthic infauna that can
demonstrate both a direct impact of the COC or a link to
higher trophic levels that are impacted.
In reviewing ROD-mandated monitoring data for Lake
Hartwell, two issues became evident: 1) sampling longer-
lived fish resulted in a bias based on gender, and 2) data
appeared to indicate that persistent organic pollutants,
like PCBs, may require as long as two generations of the
targeted fish species (10-16 years) to show statistically
significant changes in tissue concentrations. As a result
of these observations, biological indicators of exposure
were evaluated at Lake Hartwell that have shorter life
spans and, therefore, were hypothesized to reflect shorter
temporal responses to PCB exposure.
Macrobenthos Body Burden. One indicator
that was adapted to evaluate the long-term recovery
of contaminated sediments and the associated biota
Table 2.  Mean Time-Weighted Average Concentrations of Total Dissolved PCBs as Measured Using Different SPMD
Samplers (water column: WCSPMD; sediment surface rack: RSPMD; and submerged benthic dome: DSPMD) at
Three Lake Hartwell Sites (Schubauer-Berigan 2010 - submitted).

t-PCBs (ng/g SPMD/d)

= sooo
          WCSPMD: r= 0,9768, p = 0,000006; y= 962.050135 + 138.391967*x
          RSPMD: r= 0.9114, p = 0.0006; y= 2542.5249 + 54,6599965*x
 ~ 3000
3 2000

                                          Time (days)
  A  bkG
O: r= 0.9114, p = 0.0006; y= 2542.5249000 + 54.6599965004.x
M/N: r- 0.9244, p = 0.0004; y= 1924.16116 + 45.194861 B'x
                                          Time (days)
        Figure 4. Regression Analyses (using an exponential fit) of the Time Course
        Experiments Comparing WCSPMDs and RSPMDs at Site T-O (top panel) and a
        Comparison of the RSPMDs Deployed at T-O, T-M/N, and BKG (bottom panel)
        (Schubauer-Berigan, 2010 - submitted).

following remediation was the measurement of COC
body burden in aquatic macrobenthos. Researchers
utilized artificial substrates (Hester-Dendy substrates) to
collect macrobenthos for harvesting and quantifying the
COC body burden.

Aquatic macrobenthic organisms were examined as an
indicator of recent exposure to PCBs. A majority of the
benthic invertebrates, such as midge larvae, annelids
(aquatic worms), and other larvae, have life cycles
that last 30-90 days (Pennak,  1978; Merritt, 2008).
Contaminant tissue levels, referred to as body burden
concentrations, in macroinvertebrates represent very
recent contaminant exposure levels. At Lake Hartwell,
the approach was first to determine if macroinvertebrates
collected on artificial substrates, such as Hester Dendy's
(HDs)  (Klemm 1990) could discriminate between
                                                   various contaminated sites and compare the body burden
                                                   concentrations to sediment concentrations. In 2002,
                                                   20 nine-plate HDs were attached in four tiers of five
                                                   to inverted wire cages.  After 4 weeks of deployment
                                                   in the water column, HDs were retrieved, dissembled,
                                                   sorted, preserved (4°C), and analyzed for total PCBs by
                                                   summing PCB congeners as total PCBs (t-PCBs).  Figure
                                                   5 compares t-PCB concentration in water, sediment, and
                                                   macroinvertebrate tissue. Similar trends were noted
                                                   in sediment t-PCB concentrations for macrobenthos
                                                   body burden between sites, but some differences in the
                                                   magnitude of PCBs were evident at site T-M/N.

                                                   These results indicate that macroinvertebrates can be used
                                                   to assess sediment contamination among sites that have
                                                   different contamination levels of PCBs, but additional
                                                   validation over time is needed to demonstrate a more
                • T-M/N
            Figure 5. t-PCB Concentrations in Water (ug/L or ppb) and Sediment and Macroinvertebrate
            Tissue (ng/g wet weight or ppb).

            Figure 6. t-PCB Wet Weight Tissue (ng/g or ppb) for Two Indicators Used at Lake Hartwell:
            Collections of Indigenous SSLAF, Whitefin Shiners (Cyprinella nived) and 14-day Deployed
            Adult (8-12 months old) Fathead Minnows (Pimephales promelas).
rapid response to changes in sediment concentrations than
measured in higher trophic level fish.
Small Short-Lived Adult Fish (SSLAF). SSLAF
were evaluated as an indicator of shorter-term responses
within the indigenous fish populations as a short- and
a long-term monitoring tool. As opposed to long-lived
larger fish, SSLAF are expected to exhibit more rapid
response to changes in the COC exposure in the system.
The Cyprinidae (minnow family) is the largest of all
fish families.  Minnows are very important components
within the food web because they are food to larger sport
fish (e.g., largemouth bass, striped bass, etc.). There are
several species of minnows and shiners across North
America that are omnivores preying on insects, algae,
detritus, and microcrustaceans.  These fish reach maturity
in 4-6 months and live 1-3 years.  Many have good
site fidelity (do not forage or migrate great distances)
and may be excellent indicators of recent (1-3 years)
contaminant exposures in sediment and water. Therefore,
SSLAF should be a good biological indicator to track
exposure changes in shorter periods of time and space
than do longer-lived fish.
Figure 6 illustrates the wet weight tissue concentrations
measured in two SSLAF indicators at Lake Hartwell:
indigenous whitefin shiners, Cyprinella nivea, and 14-
day deployed adult (8-12 months old) fathead minnows,
Pimephales promelas. Whitefin shiners and 14-day
deployed adult fathead minnows showed similar trends
in tissue concentrations to the macroinvertebrates and
related well to sediment t-PCB concentrations (i.e.,
higher tissue concentrations were measured at T-0 than at
T-M/N as reflected in the sediment concentrations).
Additional studies are being conducted to compare results
of SSLAF to traditionally top carnivore, longer-lived
adult fish, as are typically used for long-term monitoring.
This shorter-term indicator may provide more rapid
indications of improvements in sediment contaminant
concentrations and ecological recovery and with less

Food Web Studies to Identify Biological
Indicators.  In an effort to begin to integrate a number
of biological indicators and measures and explore
alternative indicators of ecological recovery, food web
studies were also conducted at TMC and Lake Hartwell.
Several studies were conducted to characterize the
aquatic and riparian food webs associated with the
TMC-impacted area.  These studies combined a number
of novel techniques used in contaminated small stream
systems, including: stable  isotope analyses to predict
trophic position and contaminant levels (Walters, Fritz et
al, 2008), congener and chiral PCB chemistry to indicate
uptake and biological processing (Dang, Walters et al.,
2010), and riparian export of PCBs to terrestrial predators
(Walters, Fritz et al., 2008; Walters, Mills et al., 2009).

Food web studies in TMC and Lake Hartwell revealed
that t-PCB body burden was largely explained by an
organism's trophic  position (i.e., the relative position
in the food web as determined by stable isotopes of
nitrogen, 515N) and lipid content (Figure 7A, (Walters,
Fritz et al., 2008; Walters-in review)). Ongoing high
levels of PCBs were found in food web organisms
throughout TMC suggesting that TMC could be
an ongoing source of PCBs to the Lake Hartwell
food web. This hypothesis was supported through
ecosystem modeling for Lake Hartwell demonstrating
that contaminated detritus loaded to the system was
incorporated in the food web, thereby maintaining high
levels of PCBs in fish even though concentrations in
lake sediments were declining (Rashleigh, Barber et al.,

Congener and chiral analyses were used in concert to
identify patterns in transformation within TMC and Lake
Hartwell.  Biological and physical factors both strongly
influenced congener composition. Higher-chlorinated
congeners (HCCs) tend to be more hydrophobic and
recalcitrant than lower-chlorinated congeners, and their
proportional contribution to t-PCBs increased with
trophic position in the food web (Walters, Fritz et al.,
2008; Walters - in review).  Likewise, the proportion
of HCCs increased with downstream distance from the
Sangamo - Weston site, indicating that lighter, less-
chlorinated compounds were lost to the system via
volatilization (Walters, Fritz et al., 2008). In contrast,
patterns in chirality in TMC were unrelated to distance
from the Sangamo-Weston plant. Rather, biological
factors were important regulators in these types of
transformations.  Chirality  of PCB signatures varied
among organic matter types such as fine benthic organic
matter (FBOM), coarse particulate organic matter
(CPOM), and benthic algae, suggesting that distinctive
microbial communities among these organic matter types
have unique biotransformation processes (Dang, Walters

I   1000

03    100
          I* =0,56
                                  O organic mailer
                                  Q macrolnvertebrates
                                  A fish
                      trophic position
^- 1000 •
^ 100 '
« n «
^^^*«^ *
f*~9-^ * * » •
m ~^^.
^ ^*^
• • «^^^^^
• ""~^
V spiders
• herptiles
• .
o r-0,65
        -27      -26      -25      -24
                                        -23      -22
                   aquatic insect comumption
g   1200
                 300       600       900       1200
               Sediment ZPCBs ng/g wet
  Figure 7.  A) 515N Predicts EPCBs in the Stream Food
  Web; B) 513C Predicts EPCBs in the Riparian Food
  Web (open symbols are outliers excluded from model);
  C) Sediment PCB Concentrations Strongly Predict
  Spider PCB Concentrations Along a PCB Gradient in
  Lake Hartwell.

Riparian studies of aquatic-to-terrestrial flux of PCBs
focused on three key questions: 1) Which species in
diverse riparian predator communities are at highest risk
of exposure?, 2) Are concentrations in predators linked
to concentrations in sediments?, and 3) How far inland
do these fluxes extend? PCB concentrations in riparian
predators along TMC were largely driven by their
dependence on adult aquatic insects (Fig. 7B) (Walters,
Fritz et al, 2008). That is, predators that consume a
high proportion of aquatic insects had much higher PCB
concentrations than predators relying on uncontaminated,
terrestrial prey.  Ambient PCB sediment concentrations
were highly correlated with riparian spider concentrations
(Fig. 7C), suggesting that riparian spiders are effective
indicators of ecosystem recovery  as well as indicators
of potential risks to  other terrestrial species such as
arachnivorous birds (Walters, Blocksom et al., 2010).
Concentrations in spiders along the shoreline were high,
in some cases > 10-fold higher than calculated wildlife
risk values for arachnivorous birds. However, the spatial
extent of this contamination and risk was small and
limited to less than 5 m of the lake edge. Aquatic insect
prey were limited to this narrow band of terrestrial habitat
close to the lake margin, and spiders beyond this zone
demonstrated consistent and low levels of PCB exposure
(Walters, Mills etal, 2009).
Physical Lines of Evidence
The accumulation of progressively cleaner sediment
was documented early in the research conducted at Lake
Hartwell (Brenner, Magar et al., 2004).  Using age dating
coring techniques in conjunction with PCB chemistry,
researchers demonstrated that accumulation of cleaner
sediment over time reduces surface sediment contaminant
concentrations.   The age dating approach was used to
predict the time necessary for surface sediment
concentrations to recover to a remediation target
concentration. This approach and how this was
further developed at Lake Hartwell (as well as other
contaminated sediment sites) has been previously
described (USEPA, 2008).

Sediment Stability. The stability of the deposited
sediment is a key component for successful
MNR remedies.  It is critical that clean sediment
accumulation over the contaminated sediment is
stable and not subject to scour, thereby preventing
exposure of the contaminated sediment in the future.
One technique for measuring sediment stability
consists of simulating erosion events using any
number of commercially available sediment flumes
devices.  A diagram of a SedFlume used to evaluate
Lake Hartwell sediment stability is shown in Figure
8. SedFlume is a straight flume with an open bottom test
 section through which intact sediment cores are inserted.
 The core has a rectangular cross-section (10 cm x 15 cm)
 and is less than 1 m in length.
 The measures of erosion rates for sediments as a function
 of shear stress and depth below the sediment surface are
 briefly described here.  The undisturbed sediment core
 was inserted into the flume until the sediment surface
 was even with the bottom of the SedFlume channel. The
 flume was run at a specific flow rate corresponding to a
 particular shear stress (McNeil, 1996). Erosion rates are
 estimated by measuring the core length  at different time
 intervals as shown in Table 3.

 The critical shear stress of a sediment bed, icr, is defined
 as the shear stress at which a very small, but accurately
 measurable, rate of erosion occurs.  For SedFlume
 studies, this rate of erosion has been practically defined
 as 10-4 cm/s.  This rate represents 1 mm of erosion in
 approximately 15 minutes.  Since it is difficult to measure
 icr exactly at 10-4 cm/s, erosion rates were determined
 above and below 10-4 cm/s. The icr was then determined
 by linear interpolation. The technique produces a icr
 measurement with at least 20% accuracy (McNeil, 1996;
 Roberts, Jepsen et al., 1998).
 In addition to erosion rate measurements, samples were
 collected to determine the water content, bulk density,
 and particle size distribution of the  sediments.   Bulk
 density was determined by water content analysis using
 methods outlined in  (Hakanson, 2002).  The water
 content, W, was  determined (Table  3)  and then used to
 calculate bulk density, pb, (Table 3). Lastly, particle
 size distributions were determined using laser diffraction
 analysis.  The method was valid for particle sizes between
 0.04 and 2000 jam.  Fractions over  2000 jam were
 weighed to determine the weight percentage greater than
 2000 (am.

TOP VIEW      ,_       .„„__      _,_  .,	
  Figure 8. Schematic of the SedFlume Devise Used to
  Measure Sediment Stability.

Table 3.  Sediment Stability Parameters Measured and Computed for Lake Hartwell.
Bulk Density, pt,
(wef dry weight)
Water Content
Panicle Size Distribution
Erosion Rale
Critical Shear Stress, v
MH - Mu
Distribution ofparticle sizes by volume
percentage using laser diffraction
E = Az/T
Shear stress when erosion rate equals 1 0"
unit less
Detection Limit
Same as water content
0, 1 g in sample weight ranging
from 1 0 to 50 g
0.04 jim - 2000 jim
Az > 0,5 mm
T > 1 5 s
(Mo lO.ON/nr
This value is interpolated as
described in the text.
           W= water eo mem
           M,r dry ueighi ul"sample
           r = lime
           ps - density uf sediment (2.65 g/tm.l)
.•I/,,. = wet ueighl of sample
A/  iimuunl ofsedinienl eroded
pw = density of water (I g/cm3l


% —


\ '•

~ \


I ^
- . )
• . !

— m— •.turn'
i. antm'
* •IM«'
t- ••••M|
— f— k turn'

                                           25 -
                                                       NT       10        la         tO
                                                           Erosion Rate (em's!
Figure 9. Picture of Core TN-SSE-01 Aligned with SedFlume Erosion Rate Data.

                                            Erosion Rate (cm/sj
    Figure 10.  Erosion Rates (cm/s) as a Function of Depth (cm) for Three Cores on the 0 Transect at a Shear Stress
As an example, Figure 9 depicts the results for the
SedFlume erosion analysis of a core, TN-SSE-01. The
sediment found in this core consisted of fine silt over
a sand/silt mixture over a stiff, silty sand layer. Small
worms were visible near the surface of the core, and no
other organisms were observed in the sediments.  Core
TN-SSE-01 contained an easily eroded sandy layer
from 6 to 18 cm depth.  The bulk density measurements
increased markedly here with a corresponding increase in
particle size. Below this sand layer was a silty sand layer
that became progressively more resistant to erosion (stiff)
with depth. This trend was evident by the decrease in
bulk density, particle size, and erosion rates at the deepest
depths of the core.
To show the variable nature of sediments with respect to
depositional characteristics and related stability within a
small area (< 3-m radius), the average erosion rates (at
the 1.6 N/m2 shear stress condition) for three replicate
cores collected at Transect 0 are discussed below (Figure
10). The erosion rates demonstrate the same trend for
all three replicate cores  with a pronounced stiff (more
erosion resistant) layer in the 10 to 15  cm interval. The
magnitude of shear stress in the shallower and deeper
                                       regions of the cores show some variation, but this is
                                       likely due to heterogeneity in the sediments supported by
                                       corresponding particle size and bulk density differences
                                       and rapid sediment losses ("blowouts") in the core caused
                                       by gas pockets.  Measurement  of sediment stability is
                                       important to understand the permanence of the natural
                                       isolation that occurs during the accumulation of cleaner
                                       sediment over historically contaminated sediment

                                       Porewater Exchange through Groundwater/
                                       Surface Water Interactions.  The movement of water
                                       (sediment porewater)  into and  out of sediment can be
                                       controlled by the flux of water  from the groundwater
                                       to the surface water.  This groundwater/surface water
                                       interaction and various means of measuring porewater
                                       PCS concentrations were also  investigated at Lake
                                       Hartwell. This displacement of the sediment porewater
                                       to the overlying  water column  and the associated
                                       receptors was considered a possible pathway to
                                       provide continued exposure to  the aquatic system. To
                                       evaluate this mechanism, methods to characterize the
                                       groundwater/surface water interaction and porewater
                                       measurement methods were tested and analyzed. The

     Upland Well
      Segmented Core
      (20 Segments)
      Porawttw Centnlugo
      (3 Segment*)
      Poiewaler fork
      (6 Points}
      (6 Spikes)
                                                                                               GC sc PWE
  Figure 11: Piezometer and In Situ Sediment Deployments for Phase 4 and Phase 5
approach presented here uses a weight-of-evidence
approach to characterize the sediment porewater to
quantify the degree of mixing between groundwater and
surface water, and with differential depth sampling, to
quantify the magnitude and direction of the porewater
movement. The approach relies on the collection of
porewater, groundwater, and surface water samples and
stable isotope data (oxygen and hydrogen) to define the
proportion of groundwater and surface water in sediment
porewater.  Where available, traditional data  (hydraulic
head differential, conductivity, and temperature) were
used as additional lines of evidence to corroborate the
Three transects were monitored using vertically-
stratified, clustered piezometer wells for quantifying
the groundwater-surface water interaction. Two
contaminated transects (T-0 and T-N) and a background
transect (BKG) were monitored continuously for 18
months. At each transect, a series of well clusters was
installed to evaluate the vertical hydraulic gradient.
Figure 11 shows the well clusters relative positions
compared to the shoreline and transects.  One well was
screened in the water column to measure surface water
elevations.  A second well was screened just below the
sediment/water interface  (shallow). A third well was
screened at mid-depth (mid-level), and the last well was
screened at the  deepest point (deep).  Each well was
instrumented with a pressure transducer/temperature/
conductivity datasonde (600LS, YSI, Yellowsprings,
Ohio). Additionally, the wells were sampled to
                                                   characterize the porewater for PCBs and stable isotopes.
                                                   Three additional porewater methods were co-located with
                                                   these well clusters to allow comparison between the four
                                                   porewater measurement approaches.

                                                   The wells were used to measure head potential (the
                                                   potential for water to move from one point to another
                                                   based on pressure or head differentials) at different
                                                   sediment depths.  The water level in the wells at each
                                                   depth within the cluster was monitored to calculate the
                                                   head  difference between the three depth increments.
                                                   The surface water level was an important parameter in
                                                   determining the direction (gaining from the lake or losing
                                                   from  the aquifer) and magnitude of head potential.  This
                                                   reservoir along with the entire Savannah River basin
                                                   suffered drought conditions during these studies, with
                                                   surface water elevations varying as much as 25 ft. This
                                                   monitoring method resulted in observed head potential
                                                   trends that varied as much as +2.0 ft. Typically, the
                                                   head  potential between the  deepest well and the surface
                                                   water was less than 12 in., indicating the local aquifer
                                                   and surface water were connected and did not take
                                                   long to  equilibrate.  Figure  12 illustrates an example of
                                                   differential heads for Lake Hartwell sediment over time.
                                                   The differential head measurements indicated that in the
                                                   shallow sediment depths (10-25 cm) the sediment was
                                                   hydraulically connected directly to the surface water.
                                                   This observation indicates the sediment is unconsolidated
                                                   and free to exchange with the water column. In the
                                                   mid-depth wells (80-130 cm), the degree of connectivity
                                                   depended on the sediment characteristics at the transect.

                                                                        Shallow well
                                                                        Mid-level well
                                                                        Deep well
10          15           20
        Time (days)
            Figure 12. Head Potentials Shown for Vertically Stratified Wells in Lake Hartwell Sediment.
At T-0, the mid-depth was still connected with a slight
dampening effect.  At T-N, less connectivity was noted
due to a more consolidated sediment. The deepest
screened wells (180- 250 cm) indicated some connection
to the water column, but the deepest screened wells did
not exhibit day-to-day variations measured in the water
column, but did exhibit the general declining trend of the
surface water.

Sediment Porewater Exchange Using Stable
Isotopes.  Another approach to support the exchange
of sediment porewater with the overlying surface waters
was stable isotope  measurements of porewater for
oxygen and hydrogen. Variations in the stable isotope
ratios in natural waters are widely studied to interpret
hydrological processes. Isotopes of hydrogen and
oxygen, being inherent in the water molecule, can be
effective conservative tracers in understanding movement
and mixing within the hydrologic compartments. An
abundance of literature exists on the application of stable
isotopes in hydrologic investigations (Clark I, 1997;
Kendall, 1998; Machavaram, Whittemore et al, 2006)
and is only briefly  described here.
The ratios of most  abundant stable isotope species
of water molecules (HDO/HHO, HH180/HH160)
                             in the water cycle are governed primarily by global
                             precipitation/evaporation processes and are secondarily
                             affected by local thermodynamic processes such as
                             evaporation and mixing. Due to the differences in the
                             vapor pressure and kinetic diffusivity of lighter and
                             heavier isotopes, a preferential transfer of one isotope
                             over the other occurs during phase changes. During
                             evaporation, the lighter isotopes of H and 160 are
                             preferentially transferred, thus causing the  vapor to be
                             isotopically 'lighter' (depleted)  and the residual liquid
                             to be ' heavier' (enriched). The reverse is true during
                             condensation where D and 180 are  preferred. This
                             'isotope fractionation', which causes the ratios to change
                             between the liquid and vapor phases, is governed by
                             temperature, humidity, and the extent of transfer between
                             the phases. Fractionation, in turn,  results in distinct
                             isotopic ratios for various hydrologic compartments that
                             can be used as conservative tracers of water to investigate
                             the interactions between compartments.  Figure 13
                             illustrates the application of this approach to show the
                             different isotope end-points for surface waters compared
                             to aquifer-influenced porewater samples. Mixing models
                             are applied to  estimate the contribution of groundwater
                             and surface waters to the sediment porewater (Mills 2010
                             - in preparation).

16 -
14 -
? 12-
10 -
A _^


               -6.5      -6.0      -5.5      -5.0       -4.5      -4.0      -3.5       -3.0
                                                 8180 %o
Figure 13.  Mean d-excess values for Each Zone Measured and Plotted Against the Oxygen Isotope Data (BG =
Background and O = Trasect O).
Porewater PCB Concentrations. Porewater in
contaminated sediments is a concern from the standpoints
of the direct exposure of sediment dwelling organisms
and the release of contaminated porewater to the
overlying water, resulting in subsequent exposure to
both humans and wildlife.  Research has shown that
measuring porewater in contaminated sediments can
be difficult and often is biased by the measurement
technique (Ehlers, 2006; Tomaszewski and Luthy, 2008).
One aspect of the research conducted at Lake Hartwell
was to compare four methods of measuring porewater
PCB concentrations. The first method was a traditional
technique of elutriating the existing porewater from core
samples collected at depth. The second, third, and fourth
methods, respectively, consisted of direct sampling of
the porewater using lysimeter-type samplers inserted
at depth, lysimeter points with SPMDs deployed for
28 days, and SPMDs deployed in the subsurface wells
used for head measurements. Further description and
comparison of these methods are being developed for

Advective Porewater Transport. Advective
transport of the COC with  porewater through
contaminated sediments has the potential to be a
significant long-term contaminant source to surface
sediments and the overlying water column and,
ultimately, the resulting human or ecological receptors.
Once in the water column or surface sediments, the
contaminants may enter the food chain or a direct
pathway to human or ecosystem exposure.  Continued
research is needed to define the magnitude of advective
transport mechanisms and determine under what
conditions this mechanism may pose a threat to managing
the risk of contaminated sediments.
Research conducted at Lake Hartwell and TMC was
initiated to provide technical support to U.S. EPA Region
4 and to develop mechanistic-based tools and approaches
to evaluate MNR and other contaminated sediment
remediation strategies. The direct technical support
provided to the Region 4 Project Manager resulted in
modifications to the Lake Hartwell and TMC long-term
monitoring plan to demonstrate MNR progress. Those
changes included modifications to established fish tissue

monitoring protocols (a balance of male and female fish),
addition of congener specific analyses on a subset of
samples, and incorporation of bivalve monitoring in Lake
Hartwell (USEPA, 2004).

In addition, research initiated at Lake Hartwell has been
further expanded to support additional contaminated
sediment sites undergoing remediation by other
management strategies such as capping and dredging.
By basing the tools and approaches on sound science,
conceptual site models, and mechanistic principles,
combined  with technical support as needed, these
approaches and measures can generate reliable and useful
data to support a wide variety of site conditions, COCs,
and remediation strategies.

Point-of-Contact. Marc A. Mills, Ph.D.,

U.S. Environmental Protection Agency, Office of
Research and Development, National Risk Management
Research Laboratory, 26 W. Martin Luther King Drive,
Cincinnati, Ohio 45268

Tel. (513)  569-7322,

A project of this scope and scale requires a large team
of researchers and support staff. The authors, Marc
Mills, James Lazorchak, Joseph Schubauer-Berigan,
and David Walters, thank the following people for their
significant contributions to this research.  From EPA:
Richard Brenner, Paul dePercin, Michael Griffith, Ken
Fritz, Brent Johnson, Eric Kleiner, Terry Lyons, Frank
McCormick, David Raikow, Brenda Rashleigh, Paul
Wernsing, and Craig Zeller. From Battelle Memorial
Institute: James Abbot, Greg Durell, Jon Eastep, Eric
Footc, Greg Headington, Jennifer Ickcs, Carol Pcvin, and
Shane Walton. From ENVIRON:  Victor Magar. From
Clemson University:  Ryan Otter and Anthony Sowers.

Booij, K., F. Sniedes, et al. (2(306). "Environiiiental
Monitoring of Hydrophobia Organic Contaminants:
The Case of Mussels versus Semipermeable Membrane
Devices." Environmental Science & Technology 40(12):
3893 390(3.
Brenner, R. C., V. S. Magar, et al. (2004). "Long-Term
Recovery of PCB-Contaminated Surface Sediments at
the Sangamo Weston/Twelvemile Creek/Lake Hartwell
Supcrfund Site." Environmental Science & Technology
38(8): 2328-2337.

Clark I, F. P. (1997). Environmental Isotopes in
Hydrogeology. Boca Raton, Florida, Lewis.

Clarke, D. C., M. R. Palermo, andT.C. Sturgis. (2001.).
Subaqueous Cap Design: Selection of Bioturbalion
Profiles, Depths, and Rates. DOER Technical Notes
Collection. Vicksburg, Mississippi, U.S. Army Engineers
Research and Development Center.
Dang, V. D., D. M. Walters, et al. (2010).
"Transformation of Chiral Polychlorinated Biphenyls
(PCBs) in a Stream Food Webf." Environmental Science
& Technology 44(8): 2836-2841.
Rggleton, J. and K. V. Thomas.  (2004). "A Review of
Factors Affecting the Release and Bioavailability of
Contaminants During Sediment Disturbance Events."
Environment International 30(7): 973-980.
Ehlers, G. A. and A. P. Loibner. (2006). "Linking Organic
Pollutant (Bio)availability with Geosorbent Properties
and Biomimetic Methodology:  A Review of Geosorbent
Characterisation and (Bio)availability Prediction."
Environmental Pollution 141(3): 494-512.

Fleeger, J. W., G. Tit a, et al. (2006). "Does Bioturbation
by a Benthic Fish Modify the Effects of Sediment
Contamination on Saltmarsh Benthic Microalgae and
Meiofauna?" Journal of Experimental Marine Biology
and Ecology 330(1): 180-194.
Gale, R. W. (1998). "Three-Compartment Model
for Contaminant Accumulation by Semipermeable
Membrane Devices." Environmental Science &
Technology 32(15): 2292-2300.
Hakanson, L. and M. Jansson. (20(32). Principles of Lake
Sedimentology. Blackburn Press.

lluckins, J. N., J. D. Petty, and K. Booij. (2(3(36).
Monitors of Organic Chemicals in the Environment -
Semipermeable Membrane Devices. New York, Springer.
Kendall, C. and J. J. McDonnell. (1998). Isotope Tracers
in Catchment Hydrology. Amsterdam, Elsevier.

Klemm, D., P. Lewis, et al. (1990). Macroinvertebrate
Field and Laboratory Methods for Evaluating the
Biological Integrity of Surface Waters. U. 0. E. S. M. S.
Laboratory. Cincinnati, OH.
MacDonald, D. D., C. C. Ingcrsoll, et al. (2000).
"Development and Evaluation of Consensus-
Based Sediment Quality Guidelines for Freshwater
Ecosystems." Archives of Environmental Contamination
and Toxicology 39(1): 20-31.
Machavaram, M. V., D. 0. Whittemore, et al. (20(36).
"Precipitation Induced Stream Flow: An Event Based
Chemical and Isotopic Study of a Small Stream in the
Great Plains Region of the USA." journal of Hydrology
330(3 4): 470 480.

Magar, V. S., D. B Chadwick, et al. (2009). Technical
Guide: Monitored Natural Recovery at Contaminated
Sediment Sites, ESTCP.
Magar, V. S., G. W. Johnson, et al. (2005). "Long-Term
Recovery of PCB-Contaminated Sediments at the Lake
Hartwell Supcrfund Site: PCB Dechlorination. 1. End-
Member Characterization." Environmental Science &
Technology 39(1(3): 3538-3547.

McNeil, J., C. Taylor, and W. Lick. (1996).
"Measurements of Erosion of Undisturbed Bttom
Sediments with Depth:." Journal of Hydraulic
Engineering 122: 316-324.

Merrill, R. W., K. W. Cummins, and M. B. Burg. (2008).
Aquatic Insects of North America. Dubuque, Iowa,
Kendall/Hunt Publishing Company.
Mills, M. A., M. Machavaram,  and E. J. Kleiner. (2010
  in preparation). "Characterizing the Groundwater
Surface Water Interactions in Sediment Porewater:  A
Stable Isotope Approach." journal of Environmental

Nogaro, G., F. Mermillod-Blondin, et al. (2009).
"Ecosystem Engineering at the  Sediment-Water Iterface:
Bioturbation and Consumer-Substrate Interaction."
Oecologia 161(1): 125-138.

Pennak, R. W. (1978). Fresh Water Invertebrates of the
United States. New York Chichester Brisbane Toronto,
John Wiley.

Rashleigh, B., M. C. Barber, et al. (2009). "Foodweb
Modeling for Polychlorinated  Biphenyls (PCBs) in the
Twelvemile Creek Arm of Lake Hartwell, South Carolina,
USA." Ecological Modelling 220(2): 254-264.

Roberts, J., R. Jepsen, et al. (1998).  "Effects of Particle
Size and Bulk Density on Erosion of Quartz Particles."
journal of Hydraulic Engineering 124(12): 1261-1267.

Schubauer Berigan, J. P., E. A. Foote and V. S. Magar.
(2010 - submitted).  "Using SPMDs  to Assess Natural
Recovery of PCB Contaminated Sediments in Lake
Hartwell, SC: I. A Field Test of New In Situ Deployment
Methods." Soil and Sediment  Contamination.

Tomaszewski, J. E.  and R. G.  Luthy. (2008). "Field
Deployment of Polyethylene Devices to Measure
PCB Concentrations in Pore Water of Contaminated
Sediment." Environmental Science  & Technology 42(16):
6086 6091.

USEPA. (1994). Superfund  Record  of Decision:Sangamo-
Weston/Twelvemile Creek/Lake Hartwell Site, Pickens,
Georgia: Operable Unit 2.

USEPA. (1998). EPA's Contaminated Sediment
Management Strategy. Washington D.C., Office of Water:
USEPA. (2004). Five-Year Review Report for the
Sangamo Weston/Twelve Mile Creek/Lake Hartwell
PCB Contamination Superfund Site - Operable Unit Two
Pickens, Pickens County, South Carolina. R. 4. Altanta,
USEPA. (2008). Use of Sediment Core Profiling in
Assessing Effectiveness of Monitored Natural Recovery.
Cincinnati, Ohio, USEPA ORD: 8.
Walters, D. M., K. A. Blocksom, et al. (2010). "Mercury
Contamination in Fish in Midcontinent Great Rivers
of the United States: Importance of Species Traits and
Environmental Factors." Environmental Science &
Technology 44(8): 2947-2953.

Walters, D. M., K. M. Fritz, et al. (2008). "Influence of
Trophic Position and Spatial Location on Polychlorinated
Biphenyl (PCB) Bioaccumulation in a Stream Food
Web." Environmental Science & Technology 42(7):
Walters, D. M., K. M. Fritz, et al. (2008). "The Dark
Side Of Subsidies: Adult Stream Insects Export Organic
Contaminants To Riparian Predators." Ecological
Applications 18 (8): 1835  1841.
Walters, D. M., M. A. Mills, and B. S. Cade. (2010
- in review). " Bioaccumulation Patterns of PCBs
in a Temperate, Freshwater Food Web and  Their
Relationship to the Octanol-Water Partition Coefficient."
Environmental Science & Technology.

Wallers, D. M., M. A. Mills, el al. (2009). "Spider-
Mediated Flux of PCBs from Contaminated Sediments
to Terrestrial Ecosystems and Potential Risks to
Arachnivorous Birdsf." Environmental Science &
Technology 44(8): 2849-2856.

United States
Environmental Protection
   PERMIT NO. G-35
Office of Research and Development (8101R)
Washington, DC 20460

Official Business
Penalty for Private Use