Ecological Risk Assessment for Contaminated Sites - Chapter 8 doc

8

Remedial Goals

Diseases desperate grown by desperate appliance are relieved, or not at all

.
—William Shakespeare,

Hamlet

The first step of a feasibility study in the RI/FS process is to identify remedial goals
(also called remedial action objectives) for protecting human health and the envi-
ronment. The remedial goals specify contaminants and media of concern, potential
exposure pathways, and cleanup criteria (EPA, 1990a). The criteria are typically
concentrations of chemicals in the specified media that are expected to protect human
health and the environment adequately, based on risk assessments of the specified
routes of exposure. Chemical concentrations are the usual criteria because they are
the single line of evidence used in a human health risk assessment. However,
ecological risk assessment offers the possibility that remedial goals can be defined
more broadly than chemical criteria. These remedial goals are developed iteratively,
beginning with the DQO process in the problem formulation and ending with site-
specific goals that are set by regulatory agencies, and agreed to by site managers if
the site is a U.S. federal facility. The remedial goals are the basis for the selection
of candidate remedial alternatives by engineers and site managers. Remedial goals
must specify a receptor and exposure route, because the EPA acknowledges that
protection can be attained by actions that decrease exposure as well as by decreasing
concentrations of chemicals in environmental media (EPA, 1990a).
The risk assessor’s primary input to risk management is proposed cleanup
criteria, alternatively termed

preliminary remediation goals

(EPA, 1991d),

treatment
endpoints

(Alexander, 1995),

corrective action goals

(ASTM, 1999), or

remediation
objectives

(CCME, 1996a). The term

remedial goal options

(RGOs), used by some
EPA Regions, is preferable to the other four, because it emphasizes (1) that reducing
risks from contamination to minimal levels is only one of the risk manager’s options
when making a remedial decision and (2) that risk assessors may present multiple
options for remedial goals, based on different levels of risk, different endpoints, or
different definitions of the remedial action objective. Thus, we use the term RGO
throughout this chapter. In this chapter, the term

preliminary remedial goals

(PRGs)
is restricted to toxic concentrations of individual chemicals that are generically
derived and serve as default RGOs (Section 8.1). Thus, the following definitions
apply: preliminary remedial goals are concentrations in media that are starting points
for developing cleanup targets; remedial goal options are the assessors’ recommen-
dations concerning ways that remediation might achieve protection of the assessment
endpoints; and remedial goals are the ultimate cleanup targets set by risk managers
that engineers attempt to achieve. The final rule for the

National Oil and Hazardous
Substances Pollution Contingency Plan

states that the EPA sets remedial goals (EPA,
1990a). The EPA and states set these goals at many sites; however, for U.S. federal
© 2000 by CRC Press LLC

facilities, these goals are negotiated between site managers and the EPA, with a high
degree of involvement by risk assessors.

8.1 PRELIMINARY REMEDIAL GOALS

PRGs are upper concentration limits for specific chemicals in generic soils, waters,
or sediments that are anticipated to protect human health or the environment. The
following discussion is focused on the development and use of PRGs for ecological
endpoints. PRGs are more generic than RGOs and can be used as a starting point
for the development of RGOs for remedial investigations at multiple facilities, sites,
or units. There are two important questions to the risk assessor: (1) Which chemical
concentrations in environmental media should be used as PRGs? (2) How may PRGs
be modified to generate site-specific RGOs?

The EPA has published guidance entitled “Risk Assessment Guidance for Super-
fund: Volume I—Human Health Evaluation Manual, Part B” (RAGS), which is a
useful aid in developing PRGs intended to protect human health (EPA, 1991d).
However, no guidance is available in the United States on how to develop PRGs
based on ecological risk or even what level of protection is analogous to the 10

-6

risk for human carcinogens. (See Section 9.2.3 for a discussion of balancing health
and ecological risks.) PRGs should not be higher than any numerical applicable or
relevant and appropriate requirements (ARARs) for the chemical of concern. For
ecological endpoints, the only ARARs are National Ambient Water Quality Criteria
(NAWQC) that are available for many contaminants in surface waters. Other gov-
ernment entities have published guidelines that are recommended for use as PRGs.
The Canadian Council of Ministers of the Environment (CCME), for example, has
published a protocol for the derivation of soil quality guidelines (CCME, 1996b)
and guidelines for 20 chemicals derived using the protocol (CCME, 1997). These
guidelines are intended to be used or modified by site managers as remediation
criteria (CCME, 1996a, b, 1997); thus, they are PRGs.
Risk assessors may use existing PRGs (often national or state guidelines) or
derive PRGs that are less generic. In the latter case a risk assessor may, for example,
derive soil PRGs using generic toxic doses and site-specific wildlife food uptake
factors (LMES, 1997). Many ecological risk assessors do not distinguish between
screening benchmarks (Section 4.1.8) and PRGs; however, PRGs are useful if
(1) more than one assessment endpoint that is exposed to a single medium (e.g.,
piscivorous birds and fish in water) requires protection; (2) screening benchmarks
are no observed adverse effects levels (NOAELs) and therefore cannot serve as PRGs
(PRGs should be minimal effects levels unless the endpoint is a threatened or
endangered species); or (3) multiple screening benchmarks for an endpoint exist. In
the latter case, the selection of general PRGs serves to assure that particular bench-

marks (e.g., among aquatic or sediment toxicity benchmarks) are consistently the
starting points for developing final remedial goals. It is acknowledged that screening
ecotoxicity benchmarks are biased to avoid eliminating contaminants that are pos-
sible contributors to risk; thus, excessive cleanups could result if they were used as
final remedial goals (TNRCC, 1996). Sheppard et al. (1992) provide a short history
of the development of “generic guidelines” that are applicable to remediation of a
© 2000 by CRC Press LLC

broad range of sites. Most of these focus on health effects, but environmental effects
are incorporated in others.
When considering the use of existing PRGs, the risk assessor must be aware of
(1) the intended use and (2) how they were derived. For example, ecotoxicity
benchmarks are often not appropriate PRGs, particularly if they represent no-effects
levels. Also, if PRGs for an arid state in the United States are derived using arid
soil data, they are not appropriate PRGs for humid locations. Similarly, PRGs derived
from toxicity data for organisms that are not related to endpoint species should not
be used. For example, PRGs based on toxicity to osprey should not be used for
remediation of a small stream. Toxicity tests on which the PRGs are based vary,
depending on needs of the institution or facility that developed them. For Canadian
soils, test endpoints include mortality, reproduction, growth, development, behavior,
activity, lesions, physiological changes, respiration, nutrient cycling, decomposition,
genetic adaptation, and physiological acclimatization (CCME, 1997). Since the
PRGs may be used as default remedial goals, the assessor should be aware of the
level of conservatism associated with the goals.
In general, existing PRGs correspond to small effects on individual organisms,
and these chemical concentrations would be expected to cause minimal effects on
populations and communities. Far more studies that concern effects on individual
organisms are available than those that demonstrate effects at higher levels of
organization. PRGs developed in this manner may not be sufficiently protective of
individual organisms among threatened and endangered populations if they are also

sensitive species. Because these species are protected at the individual level, remedial
goals for such species should be developed ad hoc and should be based on NOAELs.
PRGs may apply to one of three environmental media: surface water, sediment
(including pore water), and soil. At hazardous waste sites where cleanup is contem-
plated, it is unlikely that an air source is the major contributor to the ecological risk
and even less likely that air will be remediated. Similarly, ecological PRGs for
groundwater exposure have not been developed. Groundwater contamination has
greater consequences for human health than for nonhuman organisms, data on
microscopic and other small biota of groundwater are scarce, and regulatory agencies
do not typically advocate their protection. Although contaminants of potential con-
cern at a site can be identified based on concentrations in wildlife food or in the
assessment endpoint organism’s tissues, ultimately one of the three abiotic media is
remediated. Therefore, the media for which PRGs have been developed do not
include “foods” and are limited to surface water, sediments, and soil. In addition,
indirect effects of contamination (such as avoidance of contaminated food) are not
typically considered in the derivation of PRGs (CCME, 1996b).

8.1.1 PRG

S



FOR

S

URFACE

W

ATER

In the United States, PRGs for surface waters should be at least as conservative as
NAWQC, unless the particular NAWQC are based on effects on humans. Numerous
other benchmarks for aquatic toxicity may also serve as PRGs. At Oak Ridge
National Laboratory, PRGs for surface waters are chosen by comparing the ORNL
benchmarks for screening toxicity of contaminants to aquatic life (chronic NAWQC
© 2000 by CRC Press LLC

or secondary chronic values; Suter and Tsao, 1996) with those for toxicity to
piscivorous wildlife (LOAEL; Sample et al., 1996b). The lower of the two bench-
marks is the PRG (LMES, 1997). It should be noted that this PRG may be too
conservative in the case of small streams, where avian or mammalian piscivores
may not forage. The PRG for a particular chemical cannot be assumed to protect
piscivorous wildlife if information on the toxicity is not available. As in the risk
assessment for aquatic organisms, the filtered concentration of a chemical in water
should be assumed to be more representative of exposure than the total concentration.
The only exception would be in the drinking water concentrations used in the
calculation of PRGs for piscivorous wildlife.

8.1.2 S

EDIMENT

Both the concentrations of chemicals in the solid phase of sediments and concen-
trations in the pore water are relevant to the exposure of benthic (sediment-inhab-
iting) organisms, and PRGs may be developed for both media (LMES, 1997). If
PRGs are available for both sediment and pore water, the PRG that is determined
by the remedial investigation to be the best estimate of risk to sediment biota should

take precedence. At ORNL, PRGs for sediments have been defined as the lowest
concentration of six types of sediment toxicity benchmarks (LMES, 1997). Most
PRGs for nonionic organic chemicals are based on equilibrium partitioning. In the
United States, PRGs for chemicals for which sediment quality criteria have been
proposed should be at least as low as those values.
Few peer-reviewed publications exist in which PRGs are derived. However, one
is worth mentioning. Comber et al. (1996) provide sediment guideline values (which
may be used as PRGs) for dioxins and dibenzofurans. They derived values using
two independent sets of parameters: (1) an aquatic guideline value corresponding
to a toxic level for rainbow trout fry and the K

oc

and (2) a toxic residue level for
benthic organisms and the sediment–organism uptake factor. The values for 2,3,7,8-
TCDD were within a factor of 6 of each other.

8.1.3 S

OIL

Standard benchmarks for soil do not exist in the United States, although they are
currently being developed by the EPA. At ORNL, PRGs for soil are protective of
avian and mammalian wildlife, plant communities, and soil invertebrate communities
(LMES, 1997). Microbial processes are not included. The EPA and state regulatory
agencies in the United States rarely make remedial decisions based on the protection
of soil invertebrates, and to the authors’ knowledge have never based a decision on
protecting microbial processes. Thus, regulatory priorities and precedents should be
considered when the assessor is selecting or developing PRGs for use at a particular
site. The CCME adds nutrient cycling processes to the ecological receptors of

concern for the derivation of its effects-based soil quality guidelines or PRGs
(CCME, 1996b). Soil PRGs should be derived under the assumption that plants and
soil invertebrates are exposed through direct contact with the chemical in soil and
that wildlife are exposed primarily through ingestion. Interestingly, the CCME
(1996b) assumes that PRGs developed for soil-dependent organisms should be
© 2000 by CRC Press LLC

protective of wildlife exposed via ingestion and dermally, except in the case of a
few specific chemicals (e.g., molybdenum and selenium).
The CCME (1996b) provides a methodology for estimating the threshold effects
concentration (TEC) using single-chemical toxicity tests, where the first method is
preferred to the second, and the second is preferred to the third: (1) weight-of-
evidence analysis (percentile of the combined effects and no-effects distributions of
concentrations), (2) extrapolation from the LOEC, or (3) extrapolation from the
EC50 or LC50. A microbial effects concentration (nitrogen fixation, nitrification,
nitrogen mineralization, respiration, and decomposition) is compared with the TEC.
If the microbial concentration is lower than the TEC, the geometric mean between
the two is used as the guideline for soil contact. As stated above, guidelines for soil
contact are assumed to protect wildlife exposed through ingestion (CCME, 1996b).
An exception is ingestion by herbivores in an agricultural scenario (CCME, 1997).
A methodology for calculating soil PRGs for terrestrial wildlife has been devel-
oped for the Oak Ridge Reservation (LMES, 1997). PRGs were calculated as a
concentration in soil that would result in an estimated dose by all oral routes equal
to the contaminant-specific and species-specific LOAEL (Figure 8.1). Exposure
estimates were iteratively calculated using varying soil concentrations and soil-to-
biota uptake models. The soil concentrations were manipulated to produce an expo-
sure estimate equivalent to the wildlife endpoint-specific and contaminant-specific
LOAEL. Because different diets may dramatically influence exposures, and sensi-
tivity to contaminants varies among species, PRGs were developed for six species
present on the Oak Ridge Reservation: short-tailed shrew, white-footed mouse, red

fox, white-tailed deer, American woodcock, and red-tailed hawk. For each chemical,
the PRG for each of the wildlife species was compared, and the lowest soil concen-
tration was selected as the final wildlife PRG. Estimates of oral exposure to con-
taminants were generated using a generalized exposure model (Section 3.10). Among
the 18 chemicals and six wildlife species for which PRGs were derived, the final
PRG for protection of wildlife was always based on either the short-tailed shrew or
the American woodcock (LMES, 1997).
Reports in which PRGs are compiled or derived do not typically recommend a
soil depth to which the quantities should apply (e.g., CCME, 1996b). The relevant
depth of exposure is left to the assessor to determine. The considerations related to
the depth of sampling for ecological risk assessments (e.g., Section 3.4.2) apply
here.

8.1.4 M

ODIFICATION



OF

PRG

S

The specific, numerical PRGs that an assessor uses are less important than how they
are modified (below) or how ultimate RGOs are chosen (Section 8.2). PRGs that
are not ARARs or based on ARARs may be modified during the remedial investi-
gation and feasibility study using site-specific data (EPA, 1991d). These modified
PRGs may be recommended by assessors as RGOs for the site. The use of the same

remedial goal at sites with varying soils or exposure pathways would result in
variable risks (see Labieniec et al., 1996, for an analysis of this risk variability with
respect to human health). The Canadian Council of Ministers of the Environment
© 2000 by CRC Press LLC

(CCME, 1996a) suggests that generic guidelines (PRGs) should be modified if (1) a
particular site has high background concentrations of a chemical; (2) contaminants
may move from soil to groundwater or air; (3) toxicological data used to derive the
guidelines are not relevant to the site (different receptors, complex mixtures of
chemicals); or (4) land uses necessitate modification of PRGs. The EPA emphasizes
that multiple chemicals and multiple exposure pathways may justify the modification
of PRGs (EPA, 1990a). California adds that if multiple media contribute to exposure,
the media should be assumed to be in equilibrium for the development of remedial
goals (Cal EPA, 1996).
In summary, modifications of PRGs may be based on:
• Land-use assumptions;
• Exposure assumptions and habitat considerations (e.g., fraction of land
that is suitable habitat);

FIGURE 8.1

Procedure for calculation of PRGs for soil based on toxicity to wildlife (LMES,
1997)
© 2000 by CRC Press LLC

• Environmental assumptions used to derive PRGs (e.g., water hardness,
soil pH, and organic content);
• Absence of type of organism (e.g., wide-ranging predator) which the PRG
is intended to protect (or exclusion from list of assessment endpoints);
• Synergistic, antagonistic, or additive effects of multiple pollutants;

• Form of chemical different from assumption in derivation of PRG (e.g.,
nonaqueous-phase liquids);
• Exposure from multiple media;
• Impacts of contamination of one medium on another (EPA, 1991d);
• Impacts of remediation of one medium (such as sediments) on contami-
nation of another medium (such as surface water);
• Effects of remediation on organisms and their habitat (Chapter 9);
• Desirable level of protection;
• Background concentration of element higher than PRG; or
• Indirect effects of contamination.

8.1.5 L

AND

U

SE

Land-use scenarios play a different role in ecological risk assessments than they do
in human health risk assessments. For human health risk assessments, remediation
depends on the land-use scenario because land use determines human exposure.
Exposure pathways for humans can change, for example, depending on whether the
land is industrial, agricultural, or residential. Soil ingestion by children may occur
in residential areas and not in industrial areas; plants for human consumption are
not grown on industrial sites; and inhalation of soil particles would be more signif-
icant in agricultural than in residential areas. Therefore, because humans engage in
different activities in different locations, exposure depends on land use, and
risk-based remediation should depend on land use. In ecological risk assessments,
three issues must be considered with respect to post-remedial land use.

1. What habitat will occur on the site given the proposed land uses?

Plants and most animals are more likely than humans to engage in all
activities on a particular site. Therefore, land-use scenarios are important
primarily because they determine which receptors will find habitat on the
site. Land use determines habitat, which determines the presence of end-
point populations and communities, which determines exposure and risk.
Even for migratory species, the question arises: Will the site provide
habitat for the species during the time it might spend on the site? There-
fore, for an industrial site, one might develop a soil PRG for a continued
industrial use scenario based on leaching and runoff to an off-site stream,
because there would be no on-site ecological endpoint receptors. However,
if the site is to be converted to a park with a plant community, birds, and
small mammals, the soil PRG should be protective of on-site as well as
off-site endpoint receptors. In some cases, as in human health risk assess-
ment, land use may eliminate habitat for some activities but not all. For
example, waterfowl may rest on and drink from an industrial pond during
© 2000 by CRC Press LLC

migration but would not breed there. Therefore, a PRG might be developed
for that limited exposure.

2. Will land use affect the value of ecological endpoint receptors?

In
some land uses, endpoint populations or communities are not as highly
valued as they would be with other uses. For example, earthworms in soil
adjacent to a factory are less valued than in agricultural or natural land
uses where maintenance of soil texture and fertility is important. These

differences in value could result in different degrees of protection. An
example of this concept is the identification of “designated uses” for
receiving waters under the Clean Water Act. The various designated uses,
which vary among states, require different water qualities, which in turn
require different levels of pollutant regulation.

3. Will land use affect the sensitivity of endpoint receptors?

Land use
may modify the species composition of an endpoint community or the
life stages of an endpoint population on the site. For example, streams in
industrial, agricultural, or residential areas are physically disturbed and
tend to have little riparian vegetation and highly variable flows, resulting
in low species richness. Similarly, stream reaches with no spawning hab-
itat (e.g., suburban channelized reaches) may contain adult fish but very
few or no fish eggs or larvae. Such stream reaches may require less
protection, because (a) sensitive species or life stages are absent, (b) no
amount of waste remediation will result in recovery, and (c) the land use
effectively precludes habitat restoration.
For any of those reasons, risk managers may decide that different levels of
protection should be associated with different land uses, thus modifying the PRG
for a medium. However, it should be noted that no particular land-use scenario is
assumed for the application of ARARs (e.g., NAWQC) to a site.
Protection levels for different land uses may be determined generically for a
nation or other political entity. Soil quality guidelines developed by the CCME
(1997) and used as PRGs are specific to different land uses: agricultural, residen-
tial/parkland, commercial, and industrial. Agricultural lands include agricultural
edge habitats. The residential/parkland use assumes that the land serves as a buffer
zone between residences but is not broad wilderness. Commercial land includes
managed systems, such as cultivated lawns and flowerbeds (CCME, 1996b). How-

ever, it is assumed that the normal range of activities on commercial lands does not
rely as much on ecological services as agricultural or residential/parkland. Although
the same receptors are generally examined in the derivation of the PRGs, the level
of protection for commercial and industrial land uses is lower than for the other two
(CCME, 1997).

8.2 REMEDIAL GOAL OPTIONS

Differing quantitative or qualitative definitions of RGOs are possible because of the
multiple lines of evidence that are available in ecological risk assessment. Conven-
tionally, an RGO is defined as a concentration of a particular chemical that constitutes
© 2000 by CRC Press LLC

a threshold for unacceptable risk. The risk assessor modifies the PRG to derive the
RGO. Media with chemical concentrations below the RGO are assumed to be
acceptable, but those with concentrations above the RGO may be remediated. Indeed,
the EPA definition of a remedial goal is a concentration of a chemical in an envi-
ronmental medium (EPA, 1990a).
Alternatively, RGOs may be defined in terms of media toxicity test results (Office
of Emergency and Remedial Response, 1994a). That is, one may specify that areas
where a particular test endpoint (e.g., >20% mortality of earthworms) or any one of
a set of test endpoints is exceeded are candidates for remediation. For example, an
RGO may be defined in terms of toxicity if the risk assessment identifies a medium
as toxic without clearly isolating the cause. Potential reasons for the use of a toxicity
RGO would be inadequate data on concentrations of all chemicals, poor relationship
between bioavailability at the site and in laboratory media, failure of any of the
chemicals measured in the medium to be toxic by themselves, or high variance in
the relative contributions of individual chemicals to toxicity at different sample
locations. In such cases an appropriate RGO could be a direction to remediate all
toxic areas. This was the case for a sediment depositional area in Bear Creek in Oak

Ridge, where the sediments were unambiguously toxic (i.e.,

Hyalella azteca

survival
was reduced by 37% relative to reference and control), but a causative agent could
not be determined from the available chemical data. Alternatively, an RGO may be
to perform a toxicity identification and evaluation (TIE) procedure and remediate
the chemicals that are identified by the TIE to be causing the toxicity (Section 4.2).
Finally, one may specify that areas where biological surveys indicate levels of effects
in exceedence of some measure of effect (e.g., dead plants or fewer than

x

earth-
worms per square meter) are candidates for remediation.
Derivation of chemical-concentration-based RGOs should incorporate site-spe-
cific exposure and effects data, to the extent practical. One way to do this is to derive
a site-specific no apparent effects level (SSNAEL) for an environmental medium
(Jones et al., 1999). The SSNAEL is the highest measured concentration of each
chemical at which toxicity was never observed in standard tests of the ambient
medium from the contaminated site. If the tests are sufficiently sensitive, the RGO
for a particular chemical should not be lower than the concentrations which were
shown to be nontoxic in the site medium. This approach can also be used for survey
data (e.g., species richness or abundance of benthic invertebrates), provided that the
surveys are sufficiently sensitive and the measured chemical concentrations are
representative of the exposures associated with the observed effects. An example of
the application of this approach to the assessment of risks to benthic invertebrates
in Poplar Creek, TN is presented by Jones et al. (1999). The recommended RGO
was the higher of the generic probable effects level from the literature (e.g., the

effects range–median; see Section 4.1.8) and the SSNAEL.
As the word

options

in the phrase suggests, multiple RGOs may be provided to
the risk manager as alternative levels of protection. If the best basis for remediation
is unclear, the assessors may provide a set of concentrations or other criteria from
which the risk managers could select or derive the final remedial goals. For example,
RGOs for water might include (1) the chronic NAWQC for the chemical that is
believed to cause significant toxic effects on an endpoint community, (2) a threshold
© 2000 by CRC Press LLC

for toxicity of that chemical in a toxicity test such as the EPA subchronic

Ceri-
odaphnia

test performed with site water as the diluent, (3) chemical concentrations
derived by performing a TIE on the contaminated water (Section 4.2), (4) a concen-
tration of a chemical bioaccumulated to toxic levels in fish tissue but not detected
in water, and (5) a requirement that toxicity be eliminated, as determined by one or
more specified test endpoints. Combinations of these types of RGOs may be used.
For example, to confirm that apparent effects are due to contamination, risk managers
may require that areas to be remediated show some level of toxicity and have some
minimum level of a chemical that is the primary contaminant of concern.
In the Netherlands, site-specific risk assessment is not used to determine whether
or not remediation is required at a particular site. Under the Dutch Soil Protection
Act, remediation is required if “serious soil contamination” is present, i.e., if risk-
based intervention values are exceeded (Swartjes, 1997). The intervention value

incorporates both human health and ecotoxicological screening criteria; the ecotox-
icological criterion is the hazardous concentration 50 (HC50), the concentration at
which 50% of species is assumed to be protected.
Risk assessment is used to determine the priority for remediation of sites where
concentrations exceed the intervention values. The prioritization of sites may be
based on the factors above, such as the diversity of ecological receptors, soil char-
acteristics, and results of ambient media toxicity tests.
The CCME (1996a) provides five alternative procedures for developing RGOs
(termed

site-specific objectives

). They are:

• Adopt a PRG (termed generic guideline) directly as site-specific objec-
tive

. This alternative provides a conservative level of protection for eco-
logical and human receptors under known land uses.

• Modify a PRG.

PRGs may be modified if receptors or other site conditions
are somewhat different from the assumptions used in deriving the PRG
(see Section 8.1.4). For example, if specific toxicity data used to derive
the generic guidelines are not relevant to the site, PRGs may be recalcu-
lated without them. Data used in the calculation of PRGs may be elimi-
nated for ecological receptors not present at the site, but the adjusted data
set must retain values for plants, vertebrates, and invertebrates from fam-
ilies that are or could be represented at the site. Also, properties of the

medium such as organic carbon content may be used to modify the PRGs.

• Develop RGO using risk assessment.

If the site is considerably different
from the assumptions used in the derivation of PRGs, risk assessment is
recommended. More specifically, risk assessment is recommended: (a) if
critical habitats are on or near the site; (b) if a large degree of uncertainty
is associated with the fate and transport of contaminants (e.g., periodic
flooding); (c) if sensitive populations or endangered species are present;
(d) if a large degree of uncertainty is associated with the fate or toxicity
of contaminant mixtures or metabolites; or (e) if multiple sources of
contamination or exposure pathways exist and were not considered in the
derivation of PRGs.
© 2000 by CRC Press LLC

• Derive site-specific objective with the CCME (1996b) protocol.

If
minimal acceptable data requirements are met, RGOs may be developed
for contaminants for which no effects-based guidelines exist.

• Consult other jurisdictional options.

Government entities may have
particular requirements or may recommend the use of background levels.
Final remedial goals for different environments may vary over orders of mag-
nitude. Although some variance is due to political or policy differences, others are
based on real differences in bioavailability and toxicity among sites. Zhang (1992)
shows that the final remedial goals for arsenic and pentachlorophenol in soils in

Records of Decisions for U.S. Superfund sites have ranged from 1.1 to 300 mg/kg
for arsenic and 0.0012 to 83,000 mg/kg for pentachlorophenol. These goals have
typically been based on human health exposure scenarios. The wide ranges in values
reflect the large number of approaches to the derivation of remedial goals (Sheppard
et al., 1992).

8.3 SPATIAL CONSIDERATIONS

Soil concentrations that constitute remedial goals are usually applied on a point-by-
point basis, rather than averaged over the area of exposure (Bowers et al., 1996).
Although this criticism of risk assessment by Bowers and colleagues is based on
human health risk, the same statistical argument applies to wildlife assessment
endpoints. Remedial action objectives for wide-ranging wildlife can be met through
the iterative use of geographic information systems (GIS, Clifford et al., 1995).
Wildlife are exposed to multiple contaminant concentrations at multiple locations,
and some contamination may be irrelevant if it occurs in a nonhabitat area. Optimal
remedial alternatives for different areas may be different. Clifford et al. (1995)
suggest that future contaminant concentrations at different locations in the particular
medium be estimated, based on what each remedial alternative can reasonably
achieve. Then the associated residual risk can be determined through the use of GIS.
Thus, if only chemical concentrations in soil were available as the line of evidence,
the RGO could be defined in terms of risk reduction.
An approach for determining remedial goals in a spatial context is presented by
Sample (1996). At large contaminated sites such as the Oak Ridge Reservation and
component watersheds, population-level risks to wildlife are determined by perform-
ing Monte Carlo simulations of the average contaminant exposure over the entire
contaminated site. It may be assumed that wildlife are equally likely to forage at
any location at the site and that each site contributes equally to risk. This approach
is biased (wildlife are not likely to use all areas equally), but in the absence of
comparative habitat-use data, the procedure is adequate. To determine which sites

should be remediated to reduce the estimated population-level risk to an acceptable
level (e.g., proportion of population experiencing exposures greater than the LOAEL
is <20%), exposures at the most contaminated sites (potential remedial units) are
set to the estimated exposure at background, to represent the results of remediation.
Sites are added to the remediated set until an acceptable level of risk has been
achieved. Exposures at sites recommended for remediation are set to background
© 2000 by CRC Press LLC

because inorganic contaminants that present risks are frequently present at low
concentrations in uncontaminated background soils which are likely to be used to
remediate the contaminated sites. (For organic contaminants, such as PCBs, exposure
goals should be set to zero. Clean soils should have unmeasurable concentrations
of most organic toxicants.) This method assumes that the primary means of reme-
diation is the removal or capping of soil, rather than the incomplete reduction of
contamination by biological or chemical means.
The approach described above focuses on the most contaminated sites, with the
goal of reducing population-level risk to an acceptable level. As a consequence,
some sites, where point estimates of exposure indicate risks may be present, will
not be recommended for remediation. While these sites may present a hazard to
individuals that use these sites extensively, their remediation is not required to
prevent risks at the population level. However, if threatened and endangered (T&E)
species are present in the area for which these RGOs are being developed, remedi-
ation should be considered for these areas. An example of the application of this
method to the Bear Creek watershed in Oak Ridge may be found in DOE (1996a).

8.4 HUMAN HEALTH

As stated above, the EPA provides guidance for developing PRGs related to human
health risks associated with different chemicals (EPA, 1991d). The text above has
addressed the development and use of ecological PRGs, specifically. It is advisable

to develop remedial goals for ecological endpoints separately from those for human
health, and to compare the two when all site-specific considerations (such as those
discussed in Section 8.1.3) have been factored into the final ecological remedial
goals (e.g., CCME, 1996b). The ecological risk assessor’s role is to provide the risk
manager with remedial goal options based on ecological risks. However, the human
health and ecological risk assessors should have a common understanding of the
differences in exposure and sensitivity of the endpoints that are responsible for
discrepancies in the two sets of PRGs.
© 2000 by CRC Press LLC