Ecological risk assessment of contaminated soil
J. M. WEEKS1,* AND S. D. W. COMBER
2
1The Centre for Environment, Fisheries and Aquaculture Sciences, Lowestoft Laboratory, Pakefield Road,
Lowestoft, Suffolk NR33 0HT, UK2WRc-NSF Ltd., Henley Road, Medmenham, Marlow, Buckinghamshire SL7 2HD, UK
ABSTRACT
The basis for an ecological risk assessment based on meeting the needs of recent UK and EU
legislation is described. The background to the framework and the legislative driver and relevant
definitions of harm are provided, prior to an overview of the proposed ecological risk assessment
process, which has been broken down into a Tiered approach. Tier 0 requires the establishment of a
conceptual site model, where potential contaminant-pathway-receptor linkages are sought and,
assuming they are identified, lead on to higher Tier assessments. Tier 1 relies largely on chemical
analysis of soil contaminant levels and comparison with soil quality guideline values to assess the
likelihood of harm. In some cases biological screening assays may also be undertaken within this Tier.
Based on a weight of evidence approach, should data Tier 1 indicate harm or leave uncertainty, then
Tier 2 biological testing is undertaken using assays relevant to the site of interest. In situations where
harm is identified under Tier 2 then Tier 3 is reserved for establishing the extent of harm within the
ecosystem. Finally the use of the `weight-of-evidence' approach to generate scientifically robust
conclusions regarding the harm (or potential for harm) within the ecosystem is briefly outlined. The
framework discussed is currently being adopted by the UK Environment Agency, with implementation
expected in 2005. The UK scheme compares favourably with comparative schemes operating in other
countries possessing the merits of being iterative, tiered, flexible with agreed exit points subject to
satisfying defined criteria and so speeding the decision-making process.
KEYWORDS: ERA, soil, risk assessment, toxicity testing.
Introduction
ECOLOGICAL risk assessment is an increasingly
important part of the decision-making process for
managing contaminated land; historically such
processes have been commonly associated with
uses in aquatic environments. In the UK, changes
to statutory regulations have been the driving
force.
Risk assessments may be used to evaluate
environmental problems arising from historical
and ongoing activities (retrospective risk assess-
ment) and, in some cases, those associated with
future activities (prospective risk assessment). In
the case of the contaminated land regime we are
largely concerned with retrospective risk
assessment.
Principles of environmental risk assessment
The principles of risk assessment are simple. An
ecological risk assessment (ERA) for toxicants in
soil, for example, relies on the comparison of
some exposure estimate for each chemical of
interest with a corresponding toxicity threshold
for representative organisms or other relevant
endpoints such as survival, growth or reproduc-
tive potential. This comparison is typically
accomplished by the derivation of a hazard
quotient, which is simply the exposure estimate
divided by the toxicity threshold. In the derivation
of these thresholds, uncertainty factors or other
toxicity extrapolation methods are often applied
* E-mail: [email protected]
DOI: 10.1180/0026461056950274
Mineralogical Magazine, October 2005, Vol. 69(5), pp. 601±613
# 2005 The Mineralogical Society
to translate an endpoint of interest into a toxicity
threshold (Byrns and Crane, 2002).
However, there are limitations to the use of
screening approaches, especially where exposure
and toxicity remain unquanti®ed (i.e. exposure
and effects data are lacking) or where biota are
exposed to complex mixtures of chemicals (which
is often the case). As well as making assessments
on the basis of exposure and toxicity of individual
substances, useful bene®t can be gained through
undertaking biological testing of the whole soil, or
by incorporating evidence gained from ecological
surveys (sometimes referred to as the `triad
approach') (Chapman, 1986; Rutgers et al,. 2000).
Biological tests may include the use of
ecotoxicity tests or biosensors in the laboratory
or in situ. Biological assessment can be made at
varying levels of biological organization
including ecological function or structure, where
effects may be assessed at the sub-cellular level,
through effects on individual species, up to effects
at the population, community or ecosystem levels.
The use of biological tests in ERA is attracting
research interest and is seen as a useful
supplement to chemical analysis in decisions
related to contaminated land (Ferguson et al.,
1998; CIRIA, 2002). To date in the UK, such
biological toxicity tests have only been used in a
regulatory role in the aquatic environment, but
increasing their potential for use within risk
assessments of contaminated land is being
realized. This application in the terrestrial
environment represents a signi®cant development
of the use of biological testing for regulatory
purposes (Weeks et al., 2004).
Legislative drivers
Soil is now protected in its own right in the UK
and many other countries within Europe. In the
UK, the primary legislation for contaminated land
in England and Wales is set out in Part IIA
(S78A-S78YC) of the Environmental Protection
Act 1990, as inserted by Section 57 of the
Environment Act 1995 (DETR, 1995); the
Contaminated Land (England) Regulations 2000
(DETR, 2000a); and Statutory Guidance
contained in DETR Circular 02/2000 (DETR,
2000b). This new regime came into force in
England on 1 April 2000 and is the ®rst UK
legislation to deal exclusively with land contam-
ination. Part IIA is based on the polluter-pays and
riskmanagement principles and it only applies to
existing contaminated land not being dealt with
under different legislative regimes such as
Integrated Pollution Prevention and Control
(IPPC).
The legal de®nition of contaminated land in the
UK is provided in statutory guidance contained in
Section 78A(2) (DETR, 2000b) as `̀ any land
which appears to the local authority in whose area
it is situated to be in such a condition, by reason
of substances in, on or under the land, that (a)
signi®cant harm is being caused or there is a
signi®cant possibility of such harm being caused;
or (b) pollution of controlled waters is being, or is
likely to be caused''. The new regime is
profoundly restricted by the de®nition and is
subject to interpretation. The regulatory regime
set out in Part IIA involves a tiered approach that
involves (1) problem identi®cation, (2) risk
assessment, (3) determination of appropriate
remediation requirements, (4) consideration of
costs, (5) establishment of who should pay, and
(6 ) imp l emen t a t i on and r emed i a t i on
(Environment Agency, 2000). The Contaminated
Land (England) Regulations 2000 (DETR, 2000a)
set out further requirements, particularly in
respect of the content of remediation notices.
Designated areas
The design of the risk-assessment framework is
primarily aimed at protecting designated areas.
Sites of national and sometimes international
importance for nature conservation are noti®ed as
Site(s) of Special Scienti®c Importance (SSSI) by
the Government's statutory conservation agencies
(English Nature, Scottish Natural Heritage and the
Countryside Council for Wales). SSSIs presently
cover more than 7% of land in England and Wales
and the protection of such sites for their wildlife
value is a priority. Many SSSIs are showing
damage or neglect; a review in 2003 of SSSIs in
England, Scotland and Wales indicated that in
England, 58% of SSSI land by area was in
favourable or recovering condition, leaving 42%
in an unfavourable condition (English Nature,
2003).
A review of international approaches to
ecological risk assessment demonstrates that
similar approaches are taken, usually on a tiered
basis, and that different levels of protection or
tolerable effects may be derived. The basis,
limitations and bene®ts of selected approaches
adopted by different countries are presented in
Table 1. Many countries, including the UK, are
only at the stage of developing or agreeing
602
J. M. WEEKS AND S. D. W. COMBER
TABLE 1. International approaches to ecological risk assessment for land contamination (after Smith et al., 2005).
Country/basis Limitations or benefits
The Netherlands
Soil Protection Act 1994, as amended. The Soil Protection Act aims to prevent, restrict or remedy
changes of soil properties, which entail a reduction of or a
threat to the functional properties the soil has for man, flora
and fauna.
Suitable for use approach, HC50 applied to land use types.
Maximum permissible soil concentrations defined. A log
distribution of NOEC data provides species sensitivity dis-
tributions (SSD). The serious risk concentration is derived by
the 50thpercentile of the SSD.
Assumes it is acceptable for 50% of the organisms involved to
be exposed to contamination levels greater than their NOEC*.
Toxicity of pollutants for land use specific ecological
parameters are mostly unavailable and are highly arbitrary.
Circular on target values and intervention values for soil
remediation 2000.
Remediation is geared to the desired end use.
Intervention values, indicative levels for serious contamination
and target values apply equally to aquatic sediment.
Germany
Federal Soil Protection Act 1999 and accompanying Soil
Protection Ordinance designed to protect and restore soil
functions.
Soil protection has to be considered in relation to the
anthropogenic use of the soil. Establishes what constitutes soil
contamination.
Soil organisms have not yet been considered.
Italy
Maximum admissible concentrations derived. If MAC values
are exceeded, then the land is classed as contaminated and
clean-up liability is initiated.
Ecological criteria not yet used in the development of MAC
values.
Sweden
Contaminated land assessment is related to hazard and levels of
pollutants, taking into account their migration potential, site
sensitivity and protective value. General framework and
guideline values which do not pose unacceptable risks to the
environment, to indicate the degree of contamination on a site,
to develop clean-up goals and to evaluate clean-up results.
Denmark
Use of NOEC, LOEC and EC, extrapolation of laboratory and
field data with use of application factors as a safety margin.
Aim to protect function and structure of soil.
Assumes log-normal distribution. Only a fraction of species
present in the ecosystem are usually covered by toxicity data.
This therefore assumes that protection of those species will be
sufficient to protect ecosystem function and structure.
UK
Tiered approach for assessing soil contaminants only on the
basis of the legal definition of harm as contained in Part IIA.
Uses a weight-of-evidence approach. Soil screening values used
for effects assessment.
Consultation framework. Existing soil screening values are
limited though they are being developed.
USA
Tiered approach to ecological risks associated with contami-
nated soil. Developing a programme to develop ecological (risk)
soil screening levels as toxic reference values (Oak Ridge
Toxicological Benchmarks).
Standardization is minimized. Based mainly on characteristics
and need of decision makers. Comparable to human health
risk-based screening concentration (RBC) values.
Canada
Recommended Canadian soil quality guidelines; interim envir-
onmental quality criteria for contaminated sites;
Guidelines for ecological risk assessment; Protocol for the
derivation of soil quality criteria.
Includes site specific data and predictive modelling to derive
quantitative information on complex ecosystem responses
including chronic effects and interactions between chemicals
and ecosystem level studies.
Australia
Derivation of ecological risk-based cleanup goals to protect
terrestrial receptors.
Background concentrations used as default for cleanup criteria
if site specific data is unavailable. Ecological receptors
selected to include different trophic levels. Lack of national
policy to require criteria to be applied.
* NOEC ÿ No Observable Effects Concentration of the substance being tested.
suitable frameworks and the need for a common
European framework in the form of a conceptual
model has been identi®ed (CLARINET, 2001).
ERA approach
The approach adopted by the Environment
Agency's ecological risk assessment framework
consultation (Environment Agency, 2003) follows
the conventional approach to environmental risk
assessment and management in the UK, as
contained in the revised guidelines. This provides
a tiered framework for characterizing the source,
pathway, receptor relationship (DETR, 2000a).
All three components of a linkage must be present
for a risk to exist.
The USEPA (United States Environmental
Protection Agency, 1998) de®ned ecological risk
assessment as `̀ a process that evaluates the
likelihood that adverse ecological effects may
occur or are occurring as a result of exposure to
one or more stressors''. An alternative but similar
UK de®nition of ecological risk assessment is `̀ an
evaluation of the likelihood of adverse effects on
organisms, populations, and communities from
chemica ls present in the envi ronment ''
(Environment Agency, 2003). This latter de®ni-
tion recognizes that ecological effects can occur at
different levels but the de®nition lacks speci®c
reference to ecological function.
ERA requirements
An ecological risk assessment has two funda-
mental requirements: information on exposure
and information on an effect(s) on a site-speci®c
basis. Exposure data typically consist of pore
water, soil quality and tissue concentrations in the
receptor and must include (ambient) reference or
control areas. A fundamental concept is that
increased exposure time increases toxicity (or
decreases the LC50; the concentration of a
chemical that kills 50% of the test population).
TRIAD approach
Ecological risk assessment requires an integrated
approach based on a number of measures and
criteria (Rutgers et al., 2000). The triad approach
has often been adopted as a basis for interfacing
chemical, toxicological and ecological informa-
tion (Chapman 1986, Chapman 1992, Breure and
Peijnenburg, 2003). In the triad approach,
toxicology is assessed by the use of bioassays as
experiments in which organisms are exposed to
site-speci®c ®eld samples and ecological
responses are observed under standard conditions
(van Straalen, 2002).
Ecological surveys
Ecological assessment can also play an important
role in helping judge the biological status of a site
at higher levels of biological organization.
Typically such surveys address the biological
diversity at a site and the abundance or rareness of
species. Such surveys may entail assessments of a
wide range of taxa, or be constrained to measures
of abundance of certain taxa that are known to be
sensitive to change, or have a key ecological or
conservation role.
Ecotoxicity testing
Standardized ecotoxicity tests (e.g. described as
formal OECD or ISO test guidelines) were
originally designed to assess the toxic effects to
a range of test species of new and existing
chemicals, such as plant protection products (i.e.
pesticides) added to soils, to enable a regulatory
risk assessment. These are usually used as part of
a prospective assessment of risk. Such tests are
also used in the derivation of toxicity data which
forms the basis for the deviation of chemical
thresholds, e.g. Environmental Quality Standards
(EQSs).
Biological tests are used within the risk
assessment process to assess soil and water
samples directly when taken from potentially
contaminated sites to generate sitespeci®c risk
assessments. This `Direct Toxicity Assessment'
(DTA) approach has been the subject of
considerable research in the UK and elsewhere
in the context of controlling emissions to
controlled waters and in some countries has a
clear regulatory role. Direct toxicity assessment
offers a number of advantages over chemical-
speci®c approaches but until the paper by Weeks
et al. (2004)m had not been applied in a
regulatory context for the testing of whole soils.
The approach is the same as adopted for use in
regulatory testing of water or ef¯uent samples.
Numerous single-species toxicity test (bioassays)
are available to measure contaminant exposure
and its effects on different biological components
of the soil ecosystem. Tests can be conducted
either ex situ in the laboratory or in situ in the
®eld, and the results of either may be used to
604
J. M. WEEKS AND S. D. W. COMBER
assess the risk of effects of contaminated soil on
ecological receptors.
Biological testing is not without its drawbacks.
Biological tests are prone to problems of
heterogeneity of contaminant levels in the actual
test soils taken from a contaminated land site. A
more fundamental dif®culty is the identi®cation
of the chemical(s) actually responsible for any
measured toxicity. Whilst this may not matter in
identifying whether adverse impacts are likely (as
the measure of toxicity is cumulative to all
toxicants present in a given sample), it becomes
important when assessing remediation options,
when the identity of toxic contaminants can be
extremely useful in guiding remediation and
redevelopment. An additional criticism
commonly levelled at biological testing schemes
is the possibility for variability within and
between testing laboratories that can give rise to
differences in interpretation about the risks posed
at a particular site. However, this is unlikely to be
any greater than the current variation demon-
strated in chemical analysis of samples by
commercial laboratories. In the UK the proposed
use of DTA for the control of wastewater quality
(Environment Agency, 2000) assumes that any
variability in the measurement is understood and
accounted for in the decision-making process, i.e.
that the level of any measures to constrain
variability are suf®cient to ensure ®tness-for-
purpose without being over-restrictive.
A framework for carrying out an ERA has been
presented by Byrns and Crane (2002) with
modi®cations, notably the addition of a new tier,
Tier 0, and the use of toxicity screening at Tier 1.
This framework now lends itself to adoption for
the assessment of contaminated land (Weeks et
al., 2004). Below, we outline the key principles in
the framework and describe the biological testing
activities involved at tiers 1 and 2 and the way in
which decisions are made at each level.
The ecological risk assessment framework
The UK ERA framework is based on a review of
schemes used in other countries, notably the US,
Australia, Canada and the Netherlands. A key
element of these schemes is the reliance placed on
a tiered framework, i.e. one in which progression
through a series of tiers re¯ects greater re®nement
in the quality and quantity of information
gathered and progressive reduction in uncertainty.
Initial decision-making is based on conservative
assumptions so that truly benign sites may be
eliminated from further investigation with a high
degree of con®dence. Subsequent data gathering
is intended to make more realistic assessments of
the risk of harm to ecoreceptors, ensuring that
resources are devoted to issues where they are
most likely to be needed.
T|er 0
The ®rst step of the ERA process is Tier 0 (see
Fig. 1). Activities at this tier aim to determine
whether or not a site falls under the UK's
legislative considerations. It involves the devel-
opment of a Conceptual Site Model (CSM)
describing what is currently known about the
site, its geographical limits, and identi®es
potential contaminants, pathways and receptors.
The level of detail that is required in the risk
assessment will be in¯uenced by many factors but
particularly:
(1) the sensitivity of the site (the receptors
present);
(2) the inherent hazardous properties of the
contaminants (toxicity);
(3) risks of exposure of receptors to contami-
nants (the existence of pathways between
contaminants and receptors, and chemical proper-
ties such as persistence); and
(4) the potential for contaminants to bioaccu-
mulate through food chains.
T|er 1
Having identi®ed at Tier 0 that a site might fall
under the Part IIA conditions and that there are
indeed at least potential contaminant-pathway-
receptor linkages, the next step (Tier 1) is to
evaluate whether these are realised and to
determine whether further evaluation is
warranted. Thus Tier 1 is essentially a step to
screen out those sites where an unacceptable risk
is unlikely to be realised. This decision is based
on conservative assumptions, effectively
according bene®t of doubt to the environment.
Two consequences follow from this. The ®rst is to
minimize the risk of false negatives (i.e. failing to
detect sites that truly pose a risk to ecological
receptors). However, in doing so, it is important
not to increase the incidence of false positives (i.e.
progressing a site that does not actually pose an
unacceptable risk to ecoreceptors) to a level that
the process becomes inef®cient because it fails to
exclude genuinely benign sites. Second, the
assessment is generic in nature, since work of
RISK ASSESSMENT OF CONTAMINATED SOIL
605
direct relevance to assessment endpoints at a
particular site is addressed in higher tiers.
At Tier 1, the emphasis is on a chemical-
speci®c risk assessment. Essentially, concentra-
tions of known or potential contaminants present
in soil are compared against thresholds for
individual chemicals, referred to generically as
soil quality guideline values (SQGVs). These are
threshold concentrations of chemicals below
which no adverse effects on the speci®ed receptor
are expected. They are sometimes referred to as
soil-screening guidelines or soil-screening values.
In addition, some toxicity screening may usefully
be incorporated as a means of reducing the
chances of missing contaminants that are not
covered by SQGVs.
Chemical contaminant data (i.e. contaminants
of potential concern emerging from the Tier 0
assessment) are obtained by chemical analysis
and a reasonable worst-case estimate of the
predicted environmental concentration (PEC) for
each contaminant made. In some cases, existing
chemical data will be available and should be
used unless there is reason to believe they are no
longer relevant (e.g. residues data for volatile
compounds). The PEC is compared with the
predicted no effect concentration (PNEC). Where
SQGVs are available, these are used as the PNEC
FIG. 1. Schematic of tiered ERA approach.
606
J. M. WEEKS AND S. D. W. COMBER
but where they are not, PNECs can be derived ab
initio from ecotoxicity data. The UK Environment
Agency is proposing to use the EU Technical
Guidance Document (TGD) on Risk Assessment
(EC, 2003) methodologies for deriving PNECs as
the basis of soil screening values (SSVs)
(Environment Agency, 2004).
This is therefore a simple quotient approach to
risk assessment, similar to that adopted in many
other regulatory regimes. If this step can be
performed and no adverse risk is indicated for all
the chemicals identi®ed within potential pollutant
linkages at Tier 0 (i.e. the PEC/SQGV ratio is <1)
then no further work is necessary unless toxicity
screening (see below) reveals evidence of
toxicity. Although SQGVs have only currently
been derived for a very limited set of substances,
it is clear that by using the recommended
approach set out in the EU TGD is likely to
generate very conservative values, thus increasing
the likelihood of false positives. In addition, the
SQGVs are based on total concentrations rather
than the bioavailable or toxic fraction of the
substance, which is also likely to lead to
overestimates of the potential impact of the
contaminants on the receiving ecosystem. For
some metals (Cu, Zn, Cd Pb), an approach
accounting for soil chemistry has been proposed
and the EU Risk Assessment Guidance is
currently being adapted speci®cally for metals.
Consequently, there is likely to be a strong driver
requiring the need for Tier 2 biological testing.
It has to be accepted that the toxicity of a
chemical is strongly governed by the physical and
chemical properties of a soil, as well as the
intrinsic properties of the chemical itself.
Consequently, soil contaminant concentrations
might exceed SQGVs, resulting in prediction of
negative effects on the soil ecosystem. A more
realistic evaluation of toxic effects may be
achieved by carrying out speci®c toxicity assess-
ments of the soils at a site by the use of biological
testing. The biological signi®cance (as determined
by toxicological testing) of chemical residues is
considered further at Tier 2 (see later).
There is a judgement to be made about the
signi®cance of any exceedances of SQGVs. An
isolated or very small exceedance would carry
less weight than large exceedances of several
SQGVs in a large number of samples. However,
uncertainty at this level of assessment is high and
it would not be prudent to automatically exclude
such sites. It is inappropriate to provide restrictive
rules about the magnitude and number of
exceedances required before progressing to
Tier 2. It is better to use the information to
prioritize resources. So, where several sites
compete for resources, attention would be
focussed on those where exceedances were
large, where residues in a large number of
samples exceeded SQGVs, and/or where there
are exceedances for persistent or bioaccumulative
or toxic (PBT) compounds. An exception to this
would be when only a very small number of
samples are available (i.e. spatial coverage of the
site is poor) or only few determinands are
represented. Under these circumstances, there
may be a need for further sampling and analysis.
If this is not possible for practical reasons (e.g. the
site is a nature reserve where disturbance caused
by sampling might itself cause harm), any
exceedance would be suf®cient to trigger progres-
sing to the next Tier.
Toxicity screening at T|er 1
The framework proposed by Byrns and Crane
(2002) suggested that sites thought to be
contaminated with complex mixtures of
substances would `side step' Tier 1 and advance
immediately into Tier 2. The logic behind such an
approach is that relying on SQGVs may be
inadequate and would fail to deal with interac-
tions between substances that might give rise to
more-than-additive effects (i.e. there is a risk of
false negatives). On the other hand, most sites
would fall into this category and it is dif®cult to
develop criteria for what constitutes a `complex
mixture'. Under this rule, Tier 1 could become
redundant for most sites.
Although the emphasis at Tier 1 is on chemical
data, some rapid biological screening (e.g.
bacterial bioluminescence testing such as
Microtox2: see Kwan and Dutka, 1992), could
usefully be introduced at this stage. By incorpor-
ating Microtox? testing into Tier 1, the risk of
false negatives should decline because toxicolo-
gically signi®cant mixtures that are not detected
using conventional chemical-speci®c SQGVs may
well be detected using this bioassay. Toxicity
screening would therefore serve as a backstop to
reduce the incidence of false negatives rather than
as a means of understanding the biological
signi®cance of any chemical contamination.
Where the outcome of any Microtox2
testing is
negative (i.e. the results indicate no toxicity) but
SQGVs indicate exceedance then progression to
Tier 2 is warranted.
RISK ASSESSMENT OF CONTAMINATED SOIL
607
However, while bacterial biosensor testing such
as Microtox2appears a useful addition to SQGVs
in Tier 1, concerns over its robustness and
sensitivity mean that, at present, regulatory
authorities will not accept decisions on negative
effects based solely on Microtox2
information.
Thus, where a SQGV is not available for Tier 1,
even though Microtox2
data may have been
obtained and indicate no effects, the risk
assessment should proceed to Tier 2.
Decision making
The main decision to be made at the end of Tier 1
is whether or not to advance a site to Tier 2.
Progression and consequently more detailed
evaluation would be required if either chemical
or toxicity screening tests indicated a possible risk
to biota. Any of the following circumstances
would trigger progression to Tier 2:
(1) absence of a soil screening or appropriate
guideline value (e.g. No SQGV) for any chemical
present
(2) when a soil screening value is available but
soil-screening testing (e.g. using MicrotoxTM
) is
negative despite chemical concentrations
exceeding the guideline
(3) the PEC/SQGV ratio is >1 for one or more
contaminants; regardless of whether or not soil-
screening test indicate that toxic contaminants are
present
(4) there are insuf®cient data to assess the risk;
and
(5) screening ecotoxicity tests (MicrotoxTM
)
suggest that toxic contaminants are present but the
chemical analyses have, as yet, failed to detect
them.
Where chemical data indicate a potential risk,
then MicrotoxTM
testing is not essential because a
decision to progress to Tier 2 will already have
been made on the basis of the chemical data alone.
However, in these circumstances, some useful
guidance about spatial variability in toxicity may
still be gained from using MicrotoxTM
testing as a
screen to aid in the location of contaminant
hotspots for subsequent soil remediation. It
follows that a site would exit the ERA framework
if all these criteria are not met.
At this point, the risk assessor can re®ne their
understanding of the CSM in the light of the
information gained at Tier 1. For example, it may
be possible to con®rm some contaminant-
pathway-receptor linkages whilst at the same
time, excluding others. The outcome of the
chemical assessments and any biological
screening should be reported, along with any
assumptions that have been made in reaching a
decision about whether or not a site should be
progressed to Tier 2. However, the risk assessor
should also document areas of outstanding
uncertainty and gaps in the available data. For
example, contamination at a single location might
exceed a SQGV whilst others all fall below. If a
very large number of samples have been taken the
chances of identifying an extensive `hotspot' (or
several `hotspots') are higher than if few locations
have been sampled. Subsequent work may then
focus on more intense sampling but concentrate
on chemical characterization for only a small
number of contaminants.
T|er 2
The aim of Tier 2 is to enable a decision to be
made about whether or not receptors of concern
are actually at risk of harm, now or under the
proposed use of the site. It is therefore primarily
concerned with the biological signi®cance of
contaminants that are present at the site. This is
expected to be the most detailed level of
assessment applied at the majority of contami-
nated sites.
At Tier 1, decision-making was predominantly
based on chemical-speci®c data. At Tier 2, the
emphasis shifts toward biologically-based deci-
sion-making. This is because we are now more
explicitly concerned with the assessment
endpoints de®ned at Tier 0 and also with the
biological signi®cance of chemical residues
present at the site. This biological information
may be obtained through toxicity testing,
ecological assessments, or a combination of both.
Chemical analysis
Further chemical analysis may or may not be
merited at Tier 2. As explained above, there are
circumstances in which a better understanding of
the bioavailability of contaminants identi®ed at
Tier 1 would improve our understanding of the
level of risk faced by ecoreceptors (Janssen et al.,
2003).
For example, particularly high levels of certain
substances may have been measured at Tier 1 but
these are not re¯ected in biological impacts from
ecological surveys or toxicity testing. This could
be explained by sequestration or speciation of the
chemical(s) at the site of concern to an extent that
608
J. M. WEEKS AND S. D. W. COMBER
they are not bioavailable or do not occur in a toxic
form at suf®cient levels to cause adverse effects.
For metals, in most cases it is the toxicity of the
free ion that is of greatest concern, however, the
proportion present in this form is generally very
small compared with the total concentration.
Factors such as pH, organic carbon concentration,
mineralogy, presence of inorganic ligands and the
redox potential of the soil all serve to amend a
metal's speciation and hence toxicity. Partitioning
and associated kinetics also control bioavail-
ability, with a general rule of thumb being the
greater the af®nity of a metal for the particulate
phase, and the longer the contact time, the less the
bioavailability and toxicity. Such cases could be
addressed by site-speci®c chemical analysis,
probably on soil extracts that vary in extraction
ef®ciency; ranging from extraction techniques to
remove all residues to aqueous extracts, intended
to assess levels of readily bioavailable residues ÿ
to weight-of-evidence approaches. Several
different extraction techniques have been
published (Janssen et al., 2003). These include
relatively simple single reagent extraction (KoÈrdel
et al., 2003), through to the use of a series of
chemicals to attempt to isolate discrete fractions
(e.g. exchangeable, carbonate, redox, organically
bound and residual; Tessier et al., 1979) through
to speci®c mixtures simulating gut and gastric
juice for estimating uptake by humans (Ruby et
al., 1996).
Conversely, Tier 1 assessment might have
revealed evidence of modest exceedances of
SQGVs at only a small number of sampling
locations (perhaps only one). In this case, it is
reasonable to concentrate efforts on better under-
standing the extent of any `hotspots' through
more intensive sampling but against a limited
range of determinands (determinands dependent
on the former use of the site and the type of
contamination likely to occur on such sites). It
might be argued that this is best regarded as an
intensi®cation of Tier 1 activity.
Examples of further chemical issues that might
be addressed at Tier 2 could include:
(1) Are the chemical residues bioavailable?
(2) Is the site subject to elevated natural
background levels of contaminants? (3) Are the
receptors being exposed at levels that cause harm?
(4) If a complex mixture of chemicals is
present, what is the biological signi®cance of
this combined toxicity?
(5) What is the spatial extent of contamination
(relative to distributions or home ranges of key
receptors) and is there evidence of any particular
`hot spots'?
(6) Can the contaminant(s) that are responsible
for adverse effects be identi®ed, thereby focussing
any remediation toward the critical sources of
contaminants or pathways.
Where toxicity may clearly be evident, but
cannot be linked directly to available chemical
evidence, more detailed studies at Tier 3 (ecolo-
gical monitoring and longer term assessment of
harm) could be warranted (Chapman et al., 2003).
Biological testing
Because the main emphasis is to understand the
biological signi®cance of the contaminants that
were present at Tier 1, emphasis is placed on the
use of biological tests and, where they are
available, on the results of ecological surveys.
The ®rst step is to review existing ecological
data to determine whether, on the basis of these
alone, there is suf®cient evidence of adverse
impacts to conclude that there is a risk to
ecoreceptors (i.e. obvious signs of damage,
decrease in species diversity and abundance).
These data could be collected as part of the site
walk-over. If so, further testing may be circum-
vented and risk management options considered.
In many cases, such data will not be available, or
they do not yield compelling evidence of impact.
It is at this point that soil ecotoxicity tests are
employed.
The suite of biological tests highlighted by
Crane and Byrns (2002), and for which current
data and practical experience exists, is modest,
and can therefore only be regarded as a subset of
the potential measures that could be used at a
contaminated site. However, these tests include
measurement endpoints that could act as surro-
gates for assessment endpoints of direct relevance
to soil ecosystems.
Some further steps can be taken to improve the
relevance of decisions made on the basis of
biological tests and the assessment endpoints
identi®ed at Tier 0. For example, if ecoreceptors
are de®ned in terms of particular species of
concern, such as a protected plant species, then it
would be prudent to ensure that plant tests are
incorporated into the suite of toxicity tests.
Indeed, further re®nement may be possible, e.g.
if the species of concern are families such as
Orchidae or Liliaceae, the range of monocotyle-
donous species might be extended at the expense
of dicotyledonous species. Similarly, if a key
RISK ASSESSMENT OF CONTAMINATED SOIL
609
assessment endpoint arises from the fact that the
site supports a breeding area for a butter¯y
species, emphasis would be placed on tests with
insects, or on tests using potential prey/food
species. If a contaminantpathway-receptor linkage
can be made to agricultural land, tests with
earthworms, plants and indicators of soil func-
tioning may be more useful. As a default, the
battery of biological tests should include at least
tests of soil function, plant growth and earthworm
mortality.
Decision making
Weight of evidence
Although the concept of `harm' is understood
intuitively by most people, it presents dif®culties
in a regulatory ERA because it is laden with value
judgements and perceptions that make it open to
interpretation. The problem is further exacerbated
by the fact that we must combine information
from a variety of sources (chemical, ecotoxicolo-
gical and ecological) in a way that enables the risk
assessor to make an overall judgement about the
risk of signi®cant harm at a particular site. If
acceptability was determined simply by compli-
ance with a single threshold this would be a
relatively trivial matter, perhaps as simple as a
pass/fail decision. However, where a triad
approach is adopted ÿ as advocated here ÿ the
determination of harm will not usually be
straightforward. Instead, it requires a judgement
that integrates all the available information
leading to a decision based on the weight of all
the available evidence.
Under the proposed ERA framework, a
decision about the acceptability (or otherwise)
of a site is based on an assessment of the risks
identi®ed at Tier 0. In this risk assessment,
predicted or measured levels of contaminant
exposure are compared with those deemed to be
acceptable (i.e. SQGV) or direct effects being
measured through testing and comparison with
controls. When biological testing is used, the
biological tests effectively integrate both expo-
sure and effects in a single assessment. This
contrasts with the chemical-speci®c approach,
where they are derived independently. Therefore,
decision-making based on biological tests is
simply a question of whether or not an adverse
effect has been reliably detected and whether this
is of suf®cient magnitude to warrant further
investigation. When both sorts of data are
available, possibly with ecological data as well,
there is a real challenge in coming to a decision
that is transparent and auditable. Below, we
outline an approach based on that advanced by
Suter (1993) that sets out to meet this objective.
In reality, a site is assumed not to pose a risk of
signi®cant harm unless evidence is available to
suggest otherwise. However, there is no guarantee
that effects would be seen from conducting
standard tests at a site that was truly contaminated
because the tests may not be sensitive to the
toxicants present or test durations may be too
short to elicit an effect. The relatively late
discovery of the adverse effects of tributyl tin
(TBT) to marine molluscs and endocrine
disrupting effects (EEA, 2001) of a range of
synthetic substances are examples where true
risks remained undetected because of reliance on
a small suite of standard tests. It follows that risk
of such a false negative is reduced by deploying a
large range of species representing as wide a
range of potential receptors as possible.
In the tiered approach outlined above, signi®-
cant linkages (or the possibility of signi®cant
harm occurring) will have been identi®ed in the
Conceptual Site Model and the problem formula-
tion stage. The subsequent risk characterization
determines whether these risks are signi®cant for
each identi®ed receptor, and then attempts to
determine the magnitude of the risk (i.e. the
extent of any effects) and the associated
uncertainties. In a weight-of-evidence approach,
all available data (e.g. from chemical analyses,
toxicity testing and other available data) are used
to estimate the likelihood that signi®cant effects
are occurring or are likely to occur and to describe
the extent of these effects (Suter, 1993).
The process of weighting the evidence effec-
tively estimates the level of risk that is most
likely, given all the available data. If the
assessment endpoint is de®ned in terms of a
threshold, such as a difference between control
and treatment group responses of >20%, then the
process can be performed in two steps:
(1) Examine the outcome of each individual
test result independently and draw a preliminary
conclusion ÿ has the measured response exceeded
the minimum level of response in treatment
groups above which a signi®cant effect may be
concluded?
(2) Determine whether the results taken
together indicate that it is likely or unlikely that
a risk of signi®cant harm will arise. If there is no
bias in the assessment that affects all lines of
evidence, and all tests yield consistent outcomes,
610
J. M. WEEKS AND S. D. W. COMBER
then it is reasonable to draw a clear conclusion (of
an adverse effect or no adverse effect). However,
if there are inconsistencies between tests, then a
process of weighting must take place.
Suter et al. (2000) describe this process in
detail and consider a number of factors that will
affect this assessment, such as:
(1) Relevance of the test ÿ more weight is
given to measures of effect that are more directly
related to the assessment endpoint. In other
words, where there is concern about a particular
plant species, then a test with a closely related
species would be more relevant than, say, testing
using invertebrates.
(2) Exposure-response ÿ more weight is given
to data that demonstrate a clear relationship
between magnitude of exposure (concentration)
and effects, e.g. from dose-response tests.
(3) Temporal scope ÿ the test should consider a
range of temporal variances relevant to the site
and its future intended use.
(4) Spatial scope ÿ testing is performed on
samples that are representative of the area of
concern.
(5) Quality ÿ more weight is given to data
generated using standardized protocols and to
studies that are properly replicated, executed and
interpreted.
TABLE 2. Application of weight-of-evidence approach to interpreting data generated at a ®ctitious
contaminated site (Suter et al., 2000).
Test option (evidence) Result
(test outcome)*
Explanation
Soil analyses/single chemical
tests (Tier 1)
+ High concentrations of Total Petroleum Hydrocarbon
in soils were reported, literature data suggest that
such levels are likely to be toxic to soil organisms.
Significant adverse effects on earthworms would
therefore be expected. Relevant toxicity data for
other detected soil compounds were unavailable
Earthworm acute toxicity test
(Tier 2)
ÿ Upon testing, the contaminated soil did not reduce
survivorship of the earthworm Eisenia fetida; other
sublethal effects were not determined
Body residue data (Tier 3) Ô Concentrations of PAHs in earthworms were seen to
be elevated relative to concentrations measured in
worms from control sites; but this elevation did not
impact on the survivorship of the worms.
Biological survey data (Tier 2) ÿ Soil microarthropod taxonomic richness was within
the range of reference soils of the same type, and did
not correlate with the range of soil petroleum
components present
Final Decision ÿ Although the earthworms were shown not to be
sensitive to the soils, such tests, when coupled to the
data from the biological surveys, presented sufficient
weight of evidence compelling the argument that
toxicity is not present for these soils. These results
are considered to be more reliable than using single
chemical toxicity data estimated from the measured
soil analytical data.
Risks to higher trophic levels as a result of chemical
uptake via the food chain were not measured, as Tier
3 testing was not indicated.
* Results of the risk characterization for each line of evidence and for the weight of evidence approach
+ indicates that the evidence is compelling and consistent with a signi®cant biological effect (according to de®ned
test criteria)
ÿ indicates that the evidence is inconsistent with the occurrence of a signi®cant biological effect
Ô indicates that the evidence is too ambiguous to interpret
RISK ASSESSMENT OF CONTAMINATED SOIL
611
(6) Quantity ÿ more weight is given to a large
quantity of data than to a small body of data.
(7) Uncertainty ÿ a line of evidence that
estimates the assessment endpoint with low
uncertainty should be given more weight, e.g.
test species and routes of exposure are relevant to
assessment endpoints of concern (this also relates
to relevance and quantity of data).
Suter et al. (2000) recommend a simple scoring
system of + or ÿ to summarize test results. A+ is
assigned if test data are consistent with signi®cant
adverse effects, and aÿ if test data do not reveal
signi®cant effects. If the data are ambiguous (e.g.
a positive response was seen but test validity
criteria were not met, or no response was seen in
the positive control), aÔnotation is assigned. The
®nal conclusion is not based simply on the
relative number of + or ÿ signs but on the
reliability of the conclusions drawn from various
lines of evidence. This still leaves the ®nal
decision to a process of expert judgement,
although it attempts to make the reasoning
transparent.
Table 2 illustrates this type of weight-of-
evidence approach for a ®ctitious site; essentially,
each type of data is scored and the reasoning for
the score explained. The outcome in this case is
that the soil is not suf®ciently contaminated to
warrant further action.
Summary and conclusions
Ecotoxicity tests have been shown to be capable
of demonstrating biological effects at sites that are
subject to chemical contamination. They therefore
have a useful role to play in demonstrating
whether or not chemical residues found at Tier 1
are of biological signi®cance. This is important
because this relates directly to the concept of
`signi®cant harm' required under Part IIA
Regulations in the UK. It may not be possible to
draw such conclusions on the basis of chemical
data alone because the bioavailability of chemical
residues is not usually known.
We have already highlighted the merits of the
MicrotoxTM
test as a screening test at Tier 1. In
addition, a range of biological tests has been
demonstrated as worthy of inclusion at Tier 2
(Weeks et al., 2004). Higher plant tests (OECD,
1984) may usefully be adopted with modi®cation.
In addition, a sublethal variant of the earthworm
test is likely to yield useful information although
the acute test should not be adopted. Whilst a
nutrient cycling test is considered useful, the
nitrogen mineralization test would ®rst need
signi®cant development to make it a practical
proposition.
The weight-of-evidence approach illustrated
above provides a useful way of integrating
biological data generated at Tier 2 with other
types of data and for making transparent the
decisions that are drawn as a result.
Acknowledgements
The authors would like to thank the UK
Environment Agency for funding this work.
References
Breure, T. and Peijnenburg, W. (2003) Developments in
soil protection in The Netherlands. Journal of Soils
and Sediments, 3, 248ÿ249.
Byrns, T. and Crane, M. (2002) Assessing risks to
ecosystems from land contamination. R&D technical
Report P299. Environment Agency and Sniffer,
Bristol, UK.
Chapman, P.M. (1986) Sediment quality criteria from
the sediment quality TRIAD: an example.
Environmental Toxicology and Chemistry, 5,
957ÿ964.
Chapman, P.M. (1992) Sediment quality triad approach.
In: Sediment classi®cation methods compendium.
EPA 823-R-92-006, September 1992.
Chapman, P.M., Feiyue, W., Janssen, C.R., Goulet, R.R.
and Kamunde, C.N. (2003) Conducting ecological
risk assessments of inorganic metals and metalloids:
current status. Human and Ecological Risk
Assessment, 9, 641ÿ697.
CIRIA (2002) Biological methods for the assessment of
contaminated land ÿ case studies. CIRIA research
project 625. CIRIA, London.
CLARINET (2001) Sustainable Management of
Contaminated Land. Proceedings of the
CLARINET Final Conference, June 2001, Vienna,
Austria, Federal Environment Agency, Austria.
ISBN 3-85457-594-7.
Crane, M. and Byrns, T. (2002) Ecotoxicological and
Biological test methods for the Assessment of
Contaminated Land. Environment Agency R&D
Technical Report P5-066/TR.
DETR (1995) Department of Environment Transport
and the Regions. Contaminated land research report
2, Guidance on preliminary site inspection of
contaminated land (two volumes).
DETR (2000a) Department of Environment Transport
and the Regions. Contaminated land environmental
protection act 1990: Part IIA DETR Circular 02/
2000.
612
J. M. WEEKS AND S. D. W. COMBER
DETR (2000b) Department of Environment Transport
and the Regions. Guidelines for environmental risk
assessment and management. The Stationery Of®ce,
London.
EC (2003) European Commission Technical Guidance
Document (TGD) on risk assessment of chemical
substances following European Regulations ad
Directives, 2nd edition. European Chemicals Bureau.
EEA (2001) European Environment Agency report No
22 on Late lessons from early warnings: the
precautionary principle 1896ÿ2000. EEA,
Copenhagen, 2001.
English Nature (2003) England's best wildlife and
geological sites ÿ the condition of Sites of Special
Scienti®c Interest in England in 2003. English
Nature.
Environment Agency (2000) Secondary model proce-
dure for the development of appropriate soil
sampling strategies for land contamination.
Environment Agency, Bristol, UK.
Environment Agency (2003) Ecological risk assess-
ment: a public consultation on a framework and
methods for assessing harm to ecosystems from
contaminants in soil. Environment Agency, Bristol,
UK.
Environment Agency (2004) Derivation of soil screen-
ing values. Environment Agency, Bristol, UK.
Ferguson, C., Darmendrail, D., Freier, K., Jensen, B.K.,
Jensen, J., Kasamas, H., Urzelai, A. and Vegter, J.
(editors) (1998) Risk assessment for contaminated
sites in Europe. Volume 1. Scienti®c basis. LQM
Press, Nottingham, UK, 165 pp.
Janssen, C.R., Heijerick, D.G., De Schamphelaere,
K.A.C. and Allen, H.E. (2003) Environmental risk
assessment of metals: tools for incorporating
bioavailability. Environment International, 28,
793ÿ800.
Kwan, K.K. and Dutka, B.J. (1992) Evaluation of
toxichromotest direct sedimentary toxicity testing
procedure and Microtox solid-phase testing proce-
dure. Bulletin of Environmental Contamination and
Toxicology, 49, 656ÿ662.
KoÈrdel, W., Konstantin, T. and Hund-Rinke, K. (2003)
Mobility and bioavailability of pollutants: interna-
tional methods standardization. Journal of Soils and
Sediments, 3, 235ÿ236.
OECD (1984) Guideline for the testing of chemicals:
208 Terrestrial plants, Growth Test. Organisation for
Economic Co-operation and Development, Paris,
France.
Ruby, M.V., Davis, A., Schoof, R., Eberle, S. and
Sellstone, C.M. (1996) Estimation of lead and
arsenic bioavailability using a physiologically based
extraction test. Environmental Science and
Technology, 30, 422ÿ430.
Rutgers, M., Faber, J., Postma, J. and Eijsackers, H.
(2000) Site-speci®c ecological risks: a basic
approach to the function-speci®c assessment of soil
pollution. Reports of the Programme on Integrated
Soil Research, 28, Programme Bureau Soil
Research, Wageningen, Germany.
Smith, R., Pollard, S.J.T., Weeks, J.M. and Nathanail,
C.P. (2005) Assessing signi®cant harm to terrestrial
ecosystems from contaminated land. Soil Use and
Management Journal, special issue on contaminated
land (in press).
Suter, G.W. (1993) Ecological Risk Assessment. Lewis
Publishers, Boca Raton, Florida, USA.
Suter, G.W., Efroymson, R.A., Sample, B.E. and Jones,
D.S. (2000) Ecological Risk Assessment for
Contaminated Sites. Lewis Publishers, Boca Raton,
Florida, USA.
Tessier, A., Campbell, P.G.C. and Bisson, M. (1979)
Sequential extraction procedure for the speciation of
particulate trace metals. Analytical Chemistry, 51,
844ÿ851.
United States Environmental Protection Agency (1998)
Guidelines for ecological risk assessment. EPA/630/
R-95/002F, April 1998.
Van Straalen, N. (2002) Assessment of soil contamina-
tion ÿ a functional perspective. Biodegradation, 13,
41ÿ52
Weeks, J.M., Sorokin, N., Johnson, I.J., Whitehouse, P.,
Ashton, D., Spurgeon, D.J., Hankard, P., Svendsen,
C. and Hart, A. (2004) Biological test methods for
contaminated land. Stage 2 ÿ A demonstration of the
use of a framework for the ecological risk assessment
of land contamination. Environment Agency,
Bristol, UK.
[Manuscript received 30 November 2004:
revised 26 June 2005]
RISK ASSESSMENT OF CONTAMINATED SOIL
613
