February 1992
                        ON A
                   Risk Assessment Forum
             U.S. Environmental Protection Agency
                   Washington, DC 20460
                                             Printed on Recycled Paper

       This document has been reviewed in accordance with U.S. Environmental Protection Agency
policy and approved for publication.  Mention of trade names or commercial products does not
constitute endorsement or recommendation for use.


Acknowledgements  .:....	vi

Foreword	 yii

Preface	 viii


       1.1.    Scope and Role of the Peer Review Panel	1
       1.2.    Peer Review Process	1


       2.1.    Overall Recommendations  	4
       2.2.    Specific Recommendations to Improve the Ecological
             Risk Assessment Paradigm	4
       2.3.    Importance of Verification, Monitoring and Research  	7
       2.4.    Issues Related to Overall Framework	8


       3.1.    Issues Recommended for Discussion in the Framework
             Document	11
       3.2.    Other Recommendations of the Panel 	14


       4.1.    Recommendations of the Panel	15
       4.2.    Other Recommendations  	18



       6.1.    Adequacy of the Framework Document under Stated Definition	22
       6.2.    Need for Expansion of the Definition of Risk Characterization	24
       6.3.    Treatment of Expanded Definition in the Framework
             Document	24


       List of Participants
       List of Observers .

      Draft Framework for Ecological Risk Assessment
      Issues Papers	
.  B-l

                                    LIST OF FIGURES
Figure 1.      EPA's Draft Ecological Risk Assessment Paradigm	3

Figure 2.      Ecological Risk Assessment Framework  	6

Figure 3.      Relationship Among Research, Monitoring and Verification,
              and Improved Ecological Risk Assessment Tools	.  • • • 8

Figure 4.      Problem Definition Component	•	12

Figure 5.      Characterization of Stress Component	16

Figure 6.      Characterization of Ecological Effects Component	 20

Figure 7.      Risk Characterization Component	26


List of Participants	•	2


       Many individuals contributed to this report. James Fava, Lawrence Barnthouse, Mark Harwell,
Kenneth Reckhow, and James Falco prepared the chairperson's summary and, prior to the workshop,
worked with the U.S. Environmental Protection Agency (EPA) staff to plan the meeting. William van
der Schalie, the Risk Assessment Forum coordinator for ecological effects, coordinated all workshop
and document development activities with the assistance of William Wood, associate staff director for
the Forum. Susan Brager of Eastern Research Group, Inc., an EPA contractor, worked with Dr. van
der Schalie, Dr. Wood, and Dr. Fava in all phases of this project, providing administrative and
logistical support.


       In 1986, EPA issued five guidelines for health risk assessment (51 Federal Register 33992-
34054, September 24, 1986).  Based on a 10-year effort, these guidelines set forth risk assessment
principles, concepts, and methods for cancer, developmental effects, mutagenic effects, exposure, and
chemical mixtures. Since then,-EPA has developed guidance in other health risk areas, for a total of
nine guidelines for health risk assessment in various phases of development. During this period,
individual EPA programs have generated program-specific guidance for ecological effects, but none
has been developed for EPA as a whole.

       At the behest of the EPA Risk Assessment Council, EPA's Risk Assessment Forum  has
sponsored three activities related to the development of Agency-wide ecological risk assessment
guidelines: (1) compilation of case studies to illustrate the state of the practice in ecological
assessments; (2) preparation of a long-term plan for developing ecological risk assessment guidelines;
and (3) development of a framework for ecological risk assessments that will offer a simple  and
flexible structure for conducting and evaluating ecological risk assessments at EPA.  The proposed
framework also is intended to contribute organizing principles for future ecological risk assessment
guidelines and is expected to evolve with experience.

       A seven-person EPA workgroup, chaired by Susan Norton, Donald Rodier, and Suzanne
Marcy, prepared a draft EPA report entitled "Framework for Ecological Risk Assessment" (framework
draft), which incorporates comments from EPA reviewers on earlier drafts.  The framework  draft also
benefitted from insights gained at a February 1991 National Academy of Sciences (NAS) Conference
held at the Airlie House in Warrenton, Virginia, and from recommendations of EPA's Science
Advisory Board.

       Ecologists and  ecotoxicologists from academia, consulting firms, and Government (State and
Federal) brought expertise in a wide range of relevant disciplines to the Risk Assessment Forum's May
1991 peer review workshop for the framework draft.  EPA did not expect to cover all of the many
principles, concepts, and methods that  are important for ecological risk assessment in this one
workshop.  Rather, EPA asked for expert opinion on the logic, scientific validity, and utility of the
principles proposed in the framework draft as general guidance for EPA risk assessors. EPA expected
the workshop participants to develop consensus on some parts of the document and useful
recommendations for change on others. Members of the public and EPA scientific staff attended the
workshop as observers.

       The workshop was highly productive.  Most significantly, the 20 peer reviewers,  some of
whom had participated in the February 1991 NAS workshop and EPA's April 1991 ecological risk
strategic planning workshop, agreed during the discussion that the basic elements of the ecological risk
assessment process, rather than substantive guidance, should be the focus of the final framework
report.  Accordingly, the peer reviewers developed information and identified issues  to assist EPA in
this task.

                                                                Dorothy E. Patton, Ph.D.
                                                                Risk Assessment Forum


       On May 14,1991, EPA convened a 3-day workshop in Rockville, Maryland, for discussion
and peer review of the draft report "Framework for Ecological Risk Assessment" (56 Federal Register
20223; May 2,  1991). The framework draft and the workshop were part of a new EPA program for
developing risk assessment guidelines for ecological effects.

       This workshop report highlights issues and recommendations developed at the Rockville
Workshop, which was chaired by Dr. James Fava of Roy F. Weston, Inc.  The report features the
chairperson's summary of the workshop findings and includes a copy of the framework draft that was
the subject of the Rockville peer review.  Based on the highly constructive and useful suggestions
presented in the chairperson's summary, EPA is simplifying and streamlining the framework draft for
publication in early 1992 as a step toward future development of EPA guidelines for ecological risk


       The U.S. Environmental Protection Agency (EPA) has established a program through its Risk
Assessment Forum to develop ecological risk assessment guidelines. As part of this effort, JEPA
developed a draft technical framework document for ecological risk assessment The technical
framework will provide general guidance and promote consistency within EPA on the basic principles
for conducting ecological risk assessments.  The proposed framework also is intended to provide
organizing principles for future ecological risk assessment guidelines in specific subject areas.


       In order to improve the technical basis for ecological risk assessment guidelines, EPA
requested an independent peer review of the draft "Framework for Ecological Risk Assessment," which
was prepared by the Risk Assessment Forum (EPA, 1991). A panel of 20 experts (see table  1)
participated in the peer review. These individuals represent expertise in a wide range of disciplines
and experience in ecological risk assessment.


       The peer review of the draft framework involved three steps. First, the draft framework
document was mailed to each reviewer; each reviewer then prepared comments that were distributed to
all reviewers.  Second, a peer review workshop was held to obtain an independent review of the logic,
scientific validity, and utility of the principles that were proposed in the framework document as
general guidance on ecological risk assessment Consensus was reached on some parts of the
document, and recommendations for change were made for other parts. Third, a written report was
prepared, summarizing the results of the workshop and presenting the panel's recommendations to
EPA.  This report represents that third step.

        EPA's draft "Framework for Ecological Risk Assessment" presents an ecological risk
 assessment paradigm based on the National Academy of Sciences (NAS) human health risk assessment
paradigm, which was published in 1983 (NRG, 1983). The ecological risk assessment paradigm
presented in the draft framework consists of four components (figure 1):  conceptual framework,
hazard assessment, exposure assessment and risk characterization. One of the panel's
 recommendations is to use more ecologically relevant terms for three of these components:  "problem
 definition/scoping" replaces  "conceptual framework," "characterization of ecological effects" replaces
 "hazard assessment," and "characterization of stress" replaces "exposure assessment" These suggested
 terms are used throughout this summary report.

         Because the peer reviewers' charge was to conduct an independent review of the draft
 framework and make suggestions to improve the framework, the panel's comments and
 recommendations follow the draft framework's organization. The panel's overall recommendations are
 presented hi section 2; comments on the conceptual framework (problem definition/scoping)
 component are presented in section 3; suggestions on  exposure assessment (characterization of stress)
 and hazard assessment (characterization of ecological  effects) are presented in sections 4 and 5; and,
 finally,  comments on the risk characterization component are discussed in section 6.

                                Table 1. List of Participants
Workshop Chair

       •      James Fava, Roy F. Weston, Inc.

Workshop Topic Area Leaders

       •      Lawrence Bamthouse, Oak Ridge National Laboratory
       •      James Falco, Battelle Pacific Northwest Laboratory
       •      Mark Harwell, University of Miami
       "      Kenneth Reckhow, Duke University

Other Participants

              William Adams, ABC Laboratories
              John Bascietto, U.S. Department of Energy
              Raymond Beaumier, Ohio Environmental Protection Agency
              Harold Bergman, University of Wyoming
              Nigel Blakeley, Washington Department of Ecology
              Alyce Fritz, National Oceanic and Atmospheric Administration
              James Gillett, Cornell University
              Michael Harrass, U.S. Food and Drug Administration
              Ronald Kendall, Clemson University
              Wayne Landis, Western Washington University
              Ralph Portier, Louisiana State University
              John Rodgers, University of Mississippi
              Peter Van Voris, Battelle Pacific Northwest Laboratory
              James Weinberg, Woods Hole Oceanographic Institution
              Randall Wentsel, U.S. Army Chemical Research, Development and Engineering Center

                   Conceptual Framework

          Stressor and Environmental Characterization
              Endpoint Identification and Selection
                 Conceptual Model Formation
Hazard Assessment

Hazard Identification
 Stressor Response
Exposure Assessment
                   Risk Characterization
 Figure 1. EPA's Draft Ecological Risk Assessment Paradigm


        The peer review panel would like to express appreciation for the opportunity to participate in
 the peer review of the draft "Framework for Ecological Risk Assessment."  The panel is pleased to
 provide assistance in preparing a framework that EPA and the Risk Assessment Forum have indicated
 will be the underlying foundation from which ecological risk assessment guidelines will be developed.
 The individuals within EPA who prepared the draft framework should be commended for an excellent
 job, and the panel would like to recognize the performance of those individuals at EPA who prepared
 the draft document for review.

        While the panel was mostly positive about the draft framework document, members were in
 consensus that there are issues associated with ecological risk assessment that can only be addressed
 more broadly throughout EPA. These issues and the panel's recommendations are presented below.


        EPA's management structure and approach to ecological risk assessment should be
 examined.  During the workshop, there was considerable discussion about the importance of
 conducting ecological risk assessments in an integrated fashion.  Because of the importance of
 effectively using interdisciplinary teams to perform ecological risk assessments, the panel believes that
 the current EPA organization should be critically examined to identify ways to foster and facilitate the
 effective implementation of consistent and comprehensive ecological risk assessments.  For example,
 the current separation of the exposure and hazard assessment technical staffs in different branches,
 which might inhibit such integration, should be carefully considered.

        Ecological risk assessment terminology should be standardized throughout EPA.  A
 terminology workgroup should be formed to provide Agency-wide consistency.  Once EPA produces
 the framework, strategic plans, and guidelines documents, the terminology for ecological risk
 assessment used by other agencies will probably follow EPA's lead.


        One of the first and major areas of discussion by the panel was Ihe  draft ecological risk
 assessment paradigm  developed by EPA.  The panel agreed that, while using the National Academy of
 Sciences (NAS) human health risk assessment paradigm as a basis was a good idea, that model should
 not be adopted directly.  There are many  important differences between ecological and human health
 risk assessment.  For  example, ecological risk assessment must include consideration of stressors other
 than chemicals for evaluation. Also, ecological risk assessment concerns multiple species, populations,
 communities, and ecosystem levels, while human health risk assessments focus on one species.

       EPA should develop a paradigm for ecological risk assessment and then, if desired,
 reference how the NAS human health risk assessment is similar to or is related to the ecological
 risk assessment. Consistent with this  recommendation, the panel strongly believes that the definitions
 for the components of the ecological risk  assessment paradigm should be ecologically based, not
human health based.  Also, the panel recommends that EPA refer to the ecological risk assessment
paradigm as a framework.  Throughout the remainder of this report, the panel will refer to the overall
ecological risk assessment paradigm as the ecological risk assessment framework.

       After much discussion, the panel developed a modified ecological risk assessment framework.
The modified framework is based on the one developed by EPA (see figure 1).  Because the panel
recognizes EPA's desire to maintain similarity with the NAS human health risk assessment framework,
the modified framework presented here (figured) should be viewed as an evolution of the
NAS framework, recognizing distinct differences between human health and ecological risk
assessment  The panel would like to emphasize that the modified framework is presented to EPA for
review and consideration.  As EPA evaluates and develops the Technical Framework for Ecological
Risk Assessment, the panel's modified framework should be used as the basis for that development.

       Several major distinctions were incorporated into the modified framework.  For ecological risk
assessment, Hie framework must be able to incorporate both chemical and nonchemical stressors.
Thus, the panel suggests the term "characterization of stress" as a replacement for the term "exposure
assessment," which was used in the draft framework.  (Exposure assessment is generally perceived as
referring to chemical exposure.)  Second, ecological risk assessments must be able to incorporate
potential effects at various levels of biological organization (e.g., population, community, and
ecosystem).  The panel recommends that EPA use  the term "characterization of ecological effects
rather than "hazard assessment," as used in the draft framework.  Using the suggested term will help
eliminate the inconsistent use of the term "hazard assessment," which has come to have several
different meanings.

        Third, while ecological risk assessment must be considered a scientific and technical process,
the panel recognizes that policy influences ecological risk assessment.  The reviewers believe that
policy input would be appropriate for the components of problem definition/scoping and risk
 characterization.  However,  me analysis sections of Hie characterization of stress and ecological effects
 must be allowed  to proceed without policy influence. The panel agrees that risk assessors have  the
 responsibility to present scientific and technical  recommendations as a key  output of ecological risk
 assessment.  The panel also agrees that the final decision-making and risk management must take other
 factors into account that are not addressed during the ecological risk assessment. Therefore, the panel
 supports the need for a risk management component outside the ecological risk assessment framework.

        Fourth, the draft framework describes two distinct components for hazard assessment and
 exposure assessment.  The panel believes strongly that these two components  are not separate activities
 that can proceed side by side, but are, in fact, interrelated.  The modified framework indicates this
 interrelationship  as such, by including the two within one box with a dotted line between them.  The
 acceptance and recognition of this important element of ecological risk assessment have implications
 for how the assessments are performed. Along with an understanding of this  interrelationship, the
 panel recommends that for  ecological risk assessment to proceed, a team approach with qualified
 individuals is required.

    Ecological Risk Assessment
        Problem Definition/Scoping
                           Risk Management
Figure 2. Ecological Risk Assessment Framework


       EPA should adopt a long-term program to verify, monitor, and conduct research to improve
ecological risk assessment.  The reviewers spent considerable time discussing the importance of using
verification, monitoring, and research to improve the framework and tools of ecological risk
assessment.  The interrelationship among these components is shown hi figure 3.

       The panel debated whether verification and monitoring should be included within the
ecological risk assessment framework.  Reviewers strongly agreed that verification and monitoring are
essential to (1) determine the overall effectiveness of the ecological risk assessment framework; (2)
provide feedback to adjust and improve the framework in future years; (3) provide feedback to
evaluate the effectiveness and practicality of current policy and help adjust policy, as necessary; and
(4) provide feedback on scientific analysis of the framework to help identify requirements for new or
improved scientific tools. The foregoing interrelationships underscore the adaptive nature of ecological
risk assessment, which cannot be cast in concrete but, rather, must involve learning about and living
with natural systems.

        Because overall verification is not included within the ecological risk assessment framework,
EPA must develop and perform the necessary verification within its other programs. Whether
current efforts for  monitoring (e.g., the Environmental Monitoring and Assessment Program [EMAP])
are adequate is unknown. However, the panel believes  that other forms of verification (e.g., quality
assurance/quality control (QA/QQ) are inadequate to meet the specified needs.  Focused efforts are
needed to evaluate and verify the effectiveness of the ecological risk assessment framework as it is
being applied.  For example, in the Superfund program or the Toxic Substances Program, sites and
decisions must be  revisited and their effectiveness must be determined.

         The framework document should discuss the dichotomy  between the ideal ecological risk
 assessment, as outlined in  the developed framework, and the practicalities of present assessments by
 EPA. Clearly the need to balance the framework between ideal and practical considerations is critical
 to the successful use of the framework.  Presenting ecological risk assessment too idealistically,
 without clarification of what is currently practicable, will establish unrealistic expectations. On the
 other hand, presenting ecological risk assessment only as it is practiced today will establish
 expectations mat are too low, will reduce the likelihood of scientific advances, and will fail to meet
 the anticipatory needs of EPA in addressing the priority ecological risks identified by the Science
 Advisory, Board (SAB, 1990).

         To fill the gap between the ideal and current risk assessments, EPA should establish a
 research program that is focused to meet the ideal ecological risk assessment framework.  The
 relationship  among research, monitoring and verification, and improved ecological risk assessment
 tools is presented in figure 3.  The panel strongly recommends that EPA use the developed framework
 as the foundation not only for ecological risk assessment guidelines, but  also  for focused long-term
 research, monitoring, and verification programs. Much of ecological risk assessment also requires a
 more substantial commitment to basic research in both the biological and physical sciences. While
 there are many documents that discuss the research needs of ecological risk assessment, one document
 that might be useful to review is "Research Priorities in Environmental Risk Assessments" (SETAC,

        V Verification J'
                     I     Risk    I
                     V Assessment/
  Figure 3. Relationship Among Research, Monitoring and
Verification, and Improved Ecological Risk Assessment Tools


       The framework document should be limited to the discussion of the first and second levels
of flowcharts (see figures included in this report).  Because of Hie evolving science and application of
ecological risk assessment, more detailed presentations of methods should be included in the guidelines
to be developed.  A more detailed discussion of the state of the practice also should be part of the
various guidelines and technical support documents proposed as part of the ecological risk assessment
strategic planning process.

       The framework document should provide guidance in addressing ecological risk assessment
as a tiered approach. The panel also discussed the use of tiered approaches during the analysis phase
of ecological risk assessment.  Ecological risk assessments differ in the breadth and depth of the
required data collection and analysis. A tiered approach would offer the opportunity to approach
ecological risk assessments from basic data sets (e.g., LD50 and LC50) or from requirements  for more
complex data sets (e.g., field studies).  Of critical concern are the triggers (or decision points) that
would move an ecological risk assessment from a lower case data requirement,  to a more sophisticated
study,  to full-scale field studies providing complex interactive data.  Movement through the tiers
according to perceived data needs for evaluating certain ecological risks must be carefully considered.
Also, field validation using at least one appropriate model system (e.g., stream, lake, cropland,
rangeland) should be designed and implemented to test the ecological risk assessment process. In
other words, how confident (or how uncertain) are the predictions that a certain chemical or
nonchemical stressor might cause an aberration in the environment?  Examples  of some triggers to
more extensive data collection and analysis are the following:

        •      Specific regulatory requirements.

        •      Serious consequences associated with potential impacts.

        •      Lack of adequate data for extrapolations needed to proceed from the measurement
               endpoints to the assessment endpoints.

        •      Potential for irreversible consequences.

        The state of the practice of ecological risk assessment constitutes an expert judgment process
with a range of methodologies, including qualitative and quantitative tools. Another topic of
discussion was the use of the quotient method in ecological risk assessment.  The panel was concerned
about  the limitations of the quotient method and strongly recommended that EPA develop or expand
its research program to investigate innovative ways to integrate data for making risk characterizations.
The quotient method should be used in ecological risk assessment only in the context of wise use of
professional judgment  and should not exceed professional judgment during decision-making.  Another
suggestion was the addition of a range of values in the numerator and denominator used hi the
quotient method.

        Given the importance of professional judgment in the final decision-making process, the panel
recommended that EPA incorporate language into the final framework document to express the need
for qualified environmental scientists and ecologists to participate in ecological risk assessments.

        The issues associated with uncertainty should be given much greater attention throughout
 the framework document. There are many different components of uncertainty, ranging from
 fundamental lack of understanding about ecological systems and anthropogenic stresses, to
 measurement error and natural stochasticity.  Some aspects are reducible with further research; others
 are not. Consequently, environmental decision-making must proceed in the presence of uncertainties.
 The presence of uncertainties must be recognized throughout the ecological risk assessment process,
 and the types and magnitudes of the  various components of uncertainty must be identified at each step
 in the process.

        These discussions should be  carefully worded to convey the sense of uncertainties without
 implying that little  or nothing is known about ecological systems and their responses to anthropogenic
 activities.  Discussion of uncertainties should include such issues as the use of sensitivity analyses,
 availability of qualitative and quantitative methods, precision versus accuracy, and QA/QC

         The framework document should be written to ensure that it will not be outdated quickly,
 unless research illustrates that the basic framework needs revision. The panel emphasizes that the
 ecological  risk  assessment framework document should be conceptually sound in order to be the basis
 for future ecological risk assessment  efforts, including guidelines. The panel strongly agrees that
 ecological  risk  assessment performance will improve greatly over the next 5 to 10 years as research
 enhances understanding of basic ecological processes and improves the tools.  The scientific tools will
 be improved and our basic understanding will be enhanced.  Thus, the guidelines, not the framework,
 will need to be revised.

        The framework document should clearly state its objectives, the intended audience, and
 where the framework fits into the overall ecological risk assessment strategy.  The document should
 be written so that it can be understood by decision-makers and risk assessors.  It is recommended that
 professional editing be used to highlight the important points and terms in the document. One
 approach that should be considered is having sidebar information that could help communicate critical
 definitions  and  concepts without disrupting the flow of the text.  Good examples of this approach are
 found in the policy-makers' summary document produced for the EPA Reports to Congress on global
climate change  and  in the summary document of the  Science Advisory Board's "Reducing Risks"
report (SAB, 1990).


       The peer review panel decided that the term "problem definition/scoping" better represents the
initial component of the framework than does the draft framework document's term "conceptual
framework."  There is a clear need to examine the nature of the specific environmental problem at
hand initially and then to develop a conceptual model of the stress and its ecological response/recovery
relationships for affected ecological systems.  This is the objective of the problem definition        •
component as recommended by the panel.

       The panel recommends adoption of the problem definition component illustrated in figure 4
and recommends that the framework report elaborate on this component in the context of the overall
revised ecological risk assessment framework discussed above. The following section suggests issues
and concepts  for elaboration in the framework document.


       Ecological risk assessment, unlike human health risk assessment, must address a diverse set
of ecological systems, from tropical to arctic environments, deserts to lakes, and estuaries to alpine
systems.  For each type of ecological system, there is a wide array of specific properties that may be
of concern with respect to human activities. These properties may cut across biological organizational
levels, from the reproductive viability of a particular population with special ecological or economic
importance, to community-level concerns for biodiversity, to ecosystem-level issues such as primary
productivity or nutrient processing, and landscape-level issues of maintaining the spatial heterogeneity
of a mosaic of ecosystems.

       Human activities result in a plethora of environmental stresses, some of which are strictly
xenobiotic (e.g., releases of certain pesticides), while others consist of changes in the frequency or
intensity of natural physical conditions (e.g., global climate change or increased ultraviolet radiation
from stratospheric ozone depletion). Typically, multiple anthropogenic stresses occur simultaneously,
with potential interaction among stresses or in the response processes of biological systems.
Ecological risk assessment may occur over much wider temporal and spatial scales than those for
human health risk assessment.

       Because of the great diversity of ecosystems, ecological components, and anthropogenic
stresses,  ecological risk assessment must be flexible and adaptive, and the specific data needs,
analytical methodologies used, and interpretations made will vary considerably for different
environmental problems.  The initial step in ecological risk assessment, therefore, must be detailed
problem  definition, in which the characteristics of the ecological systems and anthropogenic stresses
are sufficiently specified to guide the particular ecological  risk assessment.

Problem Definition/Scoping Process for Ecological Risk Assessment
                              Policy Goals
                                                  Endpoint Selection
                                                    1 - Assessment
                                                    2 - Measurement
                Figure 4. Problem Definition Component

       The problem definition/scoping phase of ecological risk assessment is closely linked to the
policy context of the environmental problem.  The problem may be either stress-specific (such as the
risks of a new chemical or of global climate change) or effects-driven (such as when forest damage is
observed, and a risk assessment is conducted to explore possible causes).  The policy connection
comes through identifying the environmental problem of concern and, in some cases, through
specification of the regulatory endpoints at issue. Here the term "regulatory endpoint" is defined as
the legislative, regulatory, or judicial norm for decision-making (e.g., the Clean Water Act regulatory
endpoints of "maintenance of a balanced indigenous population" and "unreasonable degradation of the
marine environment").     .                          .               ,          .   . . .,

        The problem definition component should identify the particular ecological systems that are
at risk from a particular stress.  The stress must be examined sufficiently to define its direct and
indirect target ecosystems, including consideration of such issues as the spatial extent of the stress, the
potential for transport across ecosystem boundaries, the transformations of the stress, and other factors
that relate to where in the environment the stress may occur.  Using expert judgment with this and
similar stresses will identify those ecological systems and ecological components that might be
adversely affected by the stress.

        Once the ecosystems of concern are specified, the problem definition component should
include selection of the appropriate endpoints for conducting the ecological risk assessment.  Two
types of endpoints are required:  assessment or ecological endpoints (i.e., the specific properties of
each ecosystem at risk that are used to evaluate the state or change in state of the ecological system);
and measurement endpoints or indicators (i.e., those aspects of the ecological system that are measured
to characterize the assessment endpoints).  First the assessment endpoints must be identified, with
explicit attention to organizational hierarchy of the ecological system.  A suite of assessment endpoints
is usually necessary, covering the species-, community-, ecosystem-, and landscape-level concerns for
the health of the at-risk ecosystem.  The selection of assessment endpoints relates in part to the policy
interests (e.g., to specified regulatory endpoints or fo public concerns); thus, changes in assessment
endpoints must be related ultimately to changes in things about the ecosystem that humans care about
(anticipating the so what? question). But  assessment endpoints are actually characteristics of the
ecological  systems and, thus, must reflect  ecological importance. Changes in the selected endpoints
would constitute changes in the health of the  ecosystem.  Moreover, the assessment endpoints selected
are in part a function of the specific stress of concern; for example, a chemical stress suspected of
causing avian eggshell thinning would logically have raptor population viability as one ecological
assessment endpoint.

        Similarly, measurement endpoints or indicators should be specified in the problem definition
component.  Each selected assessment endpoint should have one or more indicators that can be used to
characterize the state or change of state of the endpoint. These are the items that are actually
measured in a monitoring scheme or that represent data sought from a historical data base. The
indicators may be selected also for relevance to the  policy concerns, although that is not necessary.
Indicators  will be somewhat stress-specific, as is the case for endpoints. The ecological risk
assessment guidelines should specify criteria for selecting both assessment endpoints and associated

       The problem definition component should include a conceptual model, in which initial
consideration of the stress-specific and ecosystem-specific situation is used to establish a working
concept of how the stress might be imposed on the environment (i.e,, a conceptual model of the
stress regime) and how the ecological systems might respond and recover when exposed to the stress
(Le., a conceptual model of the ecological response/recovery regime). The conceptual model, perhaps
comprising a set of testable hypotheses and assumptions, becomes the basis for entering the analysis
phase of the ecological risk assessment (see figure 2).


       In addition to the above-described framework for the problem definition component, the panel
has several other recommendations for the framework development team.  Many of the
recommendations apply to the overall framework paradigm, not just the problem definition component.

       The problem definition component should include a discussion of uncertainties.  As an
example, uncertainties about the ecosystems that might be  at risk from a particular stress may, through
the expert judgment process, result in inclusion of ecosystems with less obvious risks.  As experience
is gained for a particular stress type, the selection of ecosystems for examination in an ecological risk
assessment may eliminate some ecosystems from consideration. Similarly, uncertainties may suggest
that additional endpoints or indicators are required for evaluation of ecological risks. The initial
conceptual model itself is limited by the uncertainties, and as the ecological risk assessment proceeds,
or as experience is gained from similar ecological risk assessments, the conceptual model may be
refined and improved. The adaptive aspect of ecological risk assessment must be explicitly recognized
in the framework.

       The framework document should specify an initial set of criteria for selecting ecological
assessment endpoints and measurement indicators for ecological risk assessment. Examples of the
criteria that should be specified for endpoints are ecological importance, relevance to regulatory
endpoints and/or public concerns, stress-specificity or susceptibility, predictability, and the purpose or
needs of the particular risk assessment. Examples of criteria that should be specified for indicators are
relevance to the assessment endpoint, signal-to-noise ratio, early-warning ability, stress-specificity, ease
or economy of measurement, availability of historical data, and predictability.

       The framework document should include  crisp examples  of a few types of stresses, including
at least one xenobiotic chemical example and one nonchemical stress example, such as climate
cJiange or habitat alteration.  The selected examples should be used throughout the framework
document to illustrate what is meant at each stage, from problem definition, endpoint and indicator
selection, through stress and recovery regime analyses, to risk characterization.


       The peer review panel concluded that the term "characterization of stress" was preferred over
the term "exposure assessment," because it better incorporates both chemical and nonchemical
stressors.   The panel developed a flowchart for the characterization of stress component (see figure 5).

       Characterization of ecological stress requires consideration of a number of aspects.  For
chemical stressors, source characterization, which usually results in defined distribution and rates of
release of chemical contaminants into the environment, must be considered. For other stressors,
biological, chemical, or physical changes that are the causes of ecosystem, stress are defined.
Modification of the stressor by ecosystems also should be evaluated. For chemical stressors, such
phenomena include transformation reactions that these chemicals may undergo and ecosystem
conditions such as pH, which may alter reaction rates.

       For all stress, ecosystems  stressors must be characterized. This must be done in conjunction
with ecological effects assessment because of the close relationship between these aspects of ecological,
risk assessment. Both abiotic and biotic features of ecosystems must be characterized.

       Source and ecosystem characteristics are employed in model development, selection,  and
verification.  These models are used to characterize the behavior of chemical stressors in the
environment  as well  as to define the opportunity for exposures to ecosystems. For other stressors,
these models are used to predict the extent and severity of the stress and the ecosystem  components
exposed to the stressors.

       Finally, the results of modeling studies are evaluated and  organized to establish a stressor
.profile that can be input to the risk characterization component (see figure 2). It should be
emphasized that stressor analysis requires the skills of a multidisciplinary team.  Such assessments
should be carried out in conjunction with technical staff characterizing ecological effects to ensure that
all overlapping aspects are addressed and to minimize duplication of effort.


       Stress characterization should include description of scaling phenomena and the definition
of heterogeneity of the ecosystem and ecosystem bounds.  Both temporal and spatial aspects of
scaling should be included.  Under scaling phenomena, it should be emphasized that spatial  extent as
well  as a temporal variation in stressor should be characterized.  Time frames of interest include the
time  required to complete a nutrient cycle, the  life span of individual species, and the life span of the
ecosystem itself.

        Characterizing the heterogeneity of the ecosystem should be emphasized.  Heterogeneity in
ecosystems is a major confounding factor in characterizing the behavior of contaminants and other
stressors, and, in many cases, heterogeneity is a major factor in the maintenance or diminution of a
species in a particular ecosystem. When migratory species are of interest, characterization of all
occupied ecosystems and the role each ecosystem plays in the development of the species should be

                                     Problem Definition
              Characterization of Stress
Source or
Physical Change
/ *
/ Modification \
( of Stress by )
\ Ecosystem /
Abiotic i Biotic
                                        Stress Model
                                                                             (See Figure 6)
                                   Risk Characterization
                         Figure 5. Characterization of Stress Component

       A synopsis from EPA's Exposure Guidelines should be included.  Summaries of guidance
should be included in conjunction with describing the sources of uncertainty and methods for
describing these uncertainties, with descriptions of point-of-contact measurements, scenario evaluation,
and reconstructive assessment approaches. Given that the role of measurement and modeling of
ecosystem stressors is different from measurement and modeling of human exposures, these different
roles must be identified. •

       The adaptation of these approaches to nonchemical stressors should be specified.  The
meaning  of a physical change that is analogous to a source should be described.  The role of
approaches such as measurement, scenario evaluation, and reconstructive assessment should be
presented. The role of measurement techniques, as well as that of scenario evaluation, seems to be
straightforward. The role of reconstructive assessment techniques will require the discussion of field
studies and biomonitoring techniques.
                            • -
       The framework should state that expert judgment is required to assess  most stressors.
Expert judgment should be used to formulate assumptions in developing the scenario evaluation
approach and analogies from previous assessments when interpreting limited data using measurement
and retrospective assessment techniques.  The use of simplified techniques, such as estimation of a
single exposure value in the quotient method, requires the maximum amount of expert judgment. The
framework should clearly state that when such methods are employed, the assessment should include
detailed descriptions of the logic and the assumptions made in describing an appropriate exposure

       The framework should state that gaps in stressor assessment technology exist and should
provide a brief description of a commitment to further guidelines development, implementation of
better methods, and research in the areas of source characterization, ecosystem characterization,
and stress model development and verification.  It is clear that EPA has a variety of plans  and
programs to address these issues. To  provide a perspective for the framework, a summary statement
of these  efforts and the direction that EPA intends to pursue would show how major gaps are expected
to be filled.

       A stressor uncertainty characterization section should be added. This section should describe
the sources of uncertainty in source or physical change characteristics, ecosystem characteristic model
assumptions, and resultant stressor profile.  The framework should specify that both qualitative and
quantitative methods can be  used. It would be appropriate to note that when professional judgment is
used, qualitative descriptions of uncertainty may be crucially important. In most cases, including those
in which professional judgment is used, sensitivity analysis  and/or simulation techniques should be
included as part of the assessment.  For quantitative assessments, precision and accuracy of estimates
should be characterized.

        The framework should indicate  that every assessment should include QAJQC requirements.
This can be accomplished through the statement of data quality objectives required of studies
conducted within EPA.  The framework  should incorporate standard language about the scope and
intent of those objectives.


       Because chemical stressors are not all equally available, bioavattabittty should be considered
as an aspect of chemical stressor assessment. For such stressors, this will require consideration of
transformations that the chemical undergoes and, for inorganic elements, the species of chemical
present.  Consideration of bioavailability may require estimation of dosage in some assessments.

       Chemical stressor exposure pathways from the source to the organism!system should be
defined.  For nonchemical stressors, the relationship of a physical or other change should be related to
the occurrence of a stressor that affects an organism/system.


       After much discussion, the panel agreed that the term "hazard assessment" in the draft
framework should be changed to the term "characterization of ecological effects." This would      ;,
eliminate concerns about the inconsistent use of the term "hazard assessment," and could more
accurately reflect the broader levels of biological organization required in ecological risk assessment
The reviewers developed a proposed flowchart for the characterization of ecological effects component
(see figure 6).

       Characterization of ecological effects refers  to the determination of the relationship between.
the stressor and the endpoints identified during problem definition.  Characterization will involve both
observation in the field (biotic/abiotic ecosystem characterization) and experimentation in controlled
settings.  Both observation and experimentation contribute data and scientific understanding to the
development, selection, and verification of simulation models.  The relationship between
observation/experimentation  and model development/testing may be iterative; unsatisfactory model
verification may result in additional field or laboratory studies.  Once the model is verified, it may be
used in stressor-response characterization, which concerns the relationship between the stressor and
assessment endpoints. The process of characterizing ecological effects probably also will involve
activities concerned with characterization of stress.

       The workgroup on characterization of ecological effects developed a series of
recommendations that would improve the framework document.  These recommendations are
summarized in the following paragraphs.

       The risk characterization section of the framework document should present a more detailed
discussion of uncertainty.   The importance of uncertainty analysis in ecological risk assessment was
discussed.  Even though error calculations may occur in only one step, uncertainty is important hi all
components of risk assessment. As such, it should  be discussed in all sections.

       Statements should be added that identify (and reference) statistical issues and methods so
that specific guidance is provided in subsequent technical manuals. The discussion of statistical
methods, with frequent reference to regression analysis, ignores many difficult questions with regard to
selection and application of statistical methods when data are error-contaminated and models are non-
normal and/or nonlinear.

       Technical guidance and detail should not appear in the framework document unless they
are necessary for identification of otherwise unfamiliar methods.  Even in those cases, the discussion
should be brief.  The dilemma concerning avoidance of technical details while maintaining clarity of
explanation could be solved by using "boxed-in" examples of statistical methods, applications, etc.

       Field confirmation of an effect caused by a stressor should be assessed whenever possible,
since it is the field response of the assessment endpoint that is ultimately of consequence.  In many
instances, experimentation provides strong support for causality, whereas field observation provides
evidence of correlation.  Both are important, as noted in Hill's criteria. Still, it is important to
recognize the limited support for causality that correlation alone provides.  The framework document

                                     Problem Definition
                                           V *?;
  (See Rgure 5)
                        Characterization of Ecological Effects
Abiotic  i  Biotic
                                      Effects Model
                                  Risk Characterization
           Figure 6. Characterization of Ecological Effects Component

should include a brief discussion of the importance of field verification.  A brief discussion also
should be added to address laboratory-to-field extrapolation and its role in ecological risk assessment.
Additional guidance and discussion on nonchemical stressors are needed.

       It is important to build flexibility into the framework to allow for future methods
development.  Flexibility in the framework would permit consideration of key unresolved issues, such
as how multiple stressors with interactions an4 synergisms should be addressed. The panel believes
that the framework should be flexible enough to allow new methods and approaches to be incorporated
in the various guidelines and technical documents without requiring alteration of the basic framework


       The 1983 NAS report, "Risk Assessment in the Federal Government," defines risk
 characterization as

       ...the process of estimating the incidence of a [health] effect under the various conditions of
       human exposure described in exposure assessment.  It is performed by combining the exposure
       and dose-response assessments. The summary effects of the uncertainties in the preceding
       steps are described in this step.

 The draft EPA framework document accepts the NAS definition but expands it to include ecological
 effects and consequences. This definition implies the risk characterization section in the framework
 document should discuss:

       "     Integration of exposure and stressor-response information.

       »     Summarization of uncertainties.

       •     The relationship between "consequences" and "effects."

       In their premeeting comments,  a significant number of reviewers suggested that risk
 communication is Inadequately addressed in the draft framework document, and that communication
 between scientists and managers is an important aspect of risk characterization. In light of these
 comments, the discussion groups were  asked to answer three questions:

       •     How adequately does the framework document address risk characterization under the
              stated definition?

       •     Should the definition be expanded to explicitly include communication to decision-
              makers or to the public at large?

       •     If an expanded definition is needed, how should the expansion be addressed in Hie
              framework document?


Effects and Consequences

       There was substantial confusion concerning the meanings of the terms "effect" and
 "consequence" and the relationships of these terms to "assessment endpoints."  The terms "effect" and
 "consequence" are used nowhere else in the document, and their use in the risk characterization section
creates apparent conflicts between different sections.

       The conceptual framework development section defines assessment endpoints in terms of
immediate relevance to decision-making (e.g., acres of wetland lost, decline in biodiversity).  Given
this definition, the consequences discussed in the risk characterization section should be synonymous
with assessment endpoints. However, the discussion of stressor-response assessment in the hazard  '
assessment section appears to include extrapolation to assessment endpoints.  This would imply that
effects, as defined in the risk characterization section, are synonymous with assessment endpoints and
that consequences are some new and previously undefined type of endpoint.

       The discussion groups were unable to resolve this problem.  One group suggested substituting
the term  "occurrence" for "consequence," but later agreed that this change does not solve the problem.
The final consensus was that, if "consequence" is retained as the form in which risks are
communicated to managers, then consequences should be expressed in terms of assessment endpoints.
The hazard assessment and risk characterization sections, which both appear to discuss extrapolation to
assessment endpoints, must be reconciled.  As stated above, this area was not resolved by the peer
review panel.  However, we  believe it is critical for EPA to reach a consensus on how best to proceed
in this area. The comments  above are meant to be helpful as EPA proceeds with its considerations.


       The consensus among both risk characterization workgroups was that summarization of
uncertainties is a critical component of risk characterization. The value of scientific information in the
risk'assessment is conveyed in the uncertainty analysis.  Scientific uncertainty is present in all risk
assessments. It does not prevent management and decision-making; rather, it provides a basis for
selecting among alternative actions and for deciding if (and what) additional information
(experimentation and/or observation) is needed.

       The reviewers recommended recasting the uncertainty  discussion in the risk characterization
section to include discussion of sources of uncertainty, methods of characterizing uncertainty, methods
of propagating uncertainty, and presentation of uncertainty.

       Sources of uncertainty.  The sources of scientific uncertainty in ecological risk assessment
include inadequate scientific knowledge, natural variability, measurement error, and sampling error
(e.g.,  standard error of an estimator). In actual practice, uncertainties that should be addressed
specifically include mis-specification of models used (e.g., excessive aggregation of variables and
inappropriate assumptions), error in parameter estimates, errors in the specification of initial
conditions, and errors in expert judgment

       Methods of characterizing uncertainty.  In some situations, uncertainty in an unknown quantity
(e.g.,  a model parameter or a measurement endpoint) may be estimated by means of standard measures
of statistical variability. Model errors can sometimes be estimated from a measure of goodness-of-fit
(predictions versus observations).  In many situations, however, judgmental estimation of uncertainty is
the only  option. This alternative is acceptable since methods of eliciting uncertainty estimates are
available from experts.

       Methods of propagating uncertainty.  For model-based assessments, e.g., Monte Carlo or Latin
Hypercube simulation, first-order error analysis and response-surface analysis often are used to
estimate the influence of parameter errors on uncertainty concerning model predictions.  Software for
performing these analyses is now widely available.

       Presentation of uncertainty. The group noted the importance of properly communicating
uncertainty to decision-makers but did not offer specific comments on the adequacy of the framework
document's treatment of this topic.

Exposure-Response Integration

       The consensus of the group was that the risk characterization section places far too much
emphasis on the "quotient method." The discussion of integration should be more general, and it
should be revised to eliminate the confusing distinctions among responses, effects, and consequences.
The material on the quotient method, if retained,  should be condensed and put in a box as an example.


       Workshop participants all recognized the  need for communicating ecological risk information
to decision-makers and to the public at large.  There was not, however,  uniform agreement about the
role of risk analysts in this process. During the discussion, the group reached a consensus that
communication should be included in the definition of risk characterization, but in a carefully
circumscribed way. Risk characterization should include expression of risks in terms of assessment
endpoints of direct management relevance and communication to risk managers of the ecological
implications of alternative management actions.  Communication to the  public at large (magnitude and
significance of the risks and rationale for actions  taken) is the responsibility of the risk  manager.


       Figure 7 is a flowchart for the risk characterization component of the framework.  As
envisioned by the group, risk characterization includes several intervening steps between formal
exposure/effects integration (termed "Consequence Analysis" in the figure and taken to  include
quantitative  and qualitative uncertainty analysis) and communication to  risk managers.  These steps are
as follows:
Expression of the quantitative results as a "consequence distribution," in which the
range of possible ecological responses is presented as a function of probability of
occurrence (quantitative or qualitative);

Interpretation of the ecological significance of the consequences (in narrative form);

Description of the ecological consequences (either quantitative or qualitative) of the
action alternatives available to the risk manager.

        As appropriate, all three steps would include discussion of uncertainty and weight-of-evidence
determinations.  After the range of possible consequences and their relationship to both society's risk
goals and action alternatives are conveyed to tfie risk manager, the risk assessor's responsibility ends.
The risk manager must integrate relevant non-ecological considerations, make a decision, and then
communicate the decision to the public.


                             Consequence* Analysis
j, f
Risk Management
                                                       •Other Inputs
                                  V Monitoring /

'Consequence is used in this figure, even though the peer review panel did not reach
a consensus on its use.

                     Figure 7. Risk Characterization Component


National Research Council (NRG), 1983. Risk Assessment in the Federal Government: Managing the
       Process.  National Research Council, National Academy Press, Washington, DC.

Science Advisory Board (SAB), 1990.  Reducing Risks:  Setting Priorities and Strategies for
       Environmental Protection. U.S. Environmental Protection Agency, Washington, DC.

Society of Environmental Toxicology and Chemistry (SETAC), 1987. Research Priorities in
       Environmental Risk Assessment. Published by SETAC.
U.S. Environmental Protection Agency (EPA), 1991.  Draft Framework for Ecological Risk
       Assessment. Prepared by the Risk Assessment Forum.






                          U.S. Environmental Protection Agency

TUESDAY. MAY 14.1991

8:30 a.m. -11:30 a.m.       OPENING PLENARY SESSION

8:30 a.m.     Welcome
             Dorothy Patton, Chair, USEPA Risk Assessment Forum

8:45 a.m.     Workshop Purpose and Objectives
             James Fava, Roy F. Weston, Inc.

9:00 a.m.     Ecorisk Activities of the NAS Committee on Risk Assessment Methodology
             Lawrence Barnthouse, Oak Ridge National Laboratory

9:15 a.m.     EPA's Strategic Planning Workshop for Ecorisk Assessment
             Mark Harwell, University of Miami, Rosentiel School of Marine and Atmospheric

9:30 a.m.     Ecorisk Paradigm: Highlights
             Susan Norton, USEPA

9:45 am.     Ecorisk Paradigm: Issues Presentation
             James Fava

10:00 a.m.    BREAK

10:15 a.m.    Ecorisk Paradigm: Discussion
             James Fava
11:30 a.m.    LUNCH

12:30 p.m. -1:30 pan.
12:30 p.m.    Conceptual Framework Development: Highlights
             David Mauriello, USEPA

12:45 p.m.    Conceptual Framework Development: Issues Presentation
             Mark Harwell

TUESDAY. MAY 14.1991 (cont)
1:30 p.m. - 4:30 p.m.
1:30 p.m.     Conceptual Framework Development: Discussion (Two Workgroups)
             Mark Harwell and Lawrence Barnthouse

3:00 p.m.     BREAK

3:15 p.m.     Conceptual Framework Development: Discussion (cont.)

4:30 p.m. - 5:30 p.m.        PLENARY SESSION

4:30 p.m.     Conceptual Framework Development: Summary
             Mark Harwell and Lawrence Barnthouse

5:30 p.m.     ADJOURN

5:30 p.m.     Reception - Pool Terrace


8:30 a.m. - 9:30  a.m.        PLENARY SESSION

8:30 a,m.     Hazard Assessment:  Highlights
             Donald Rodier, USEPA

8:45 a.m.     Hazard Assessment  Issues Presentation
             Kenneth Reckhow, Duke University, School of Forestry and Environmental Studies

9:15 a.m.     Exposure Assessment: Highlights
             Anne Sergeant, USEPA

9:30 a.m.     Exposure Assessment: Issues Presentation
             James Falco, Battelle Pacific Northwest Laboratory
10:00 a.m.    BREAK

10:15 a.m. - 12:15 p.m.
10:15 a.m.    Hazard Assessment: Discussion
             Kenneth Reckhow

             Exposure Assessment: Discussion
             James Falco

WEDNESDAY. MAY 15,1991 (cont.)
12:15 p.m.

1:15 p.m. •

1:15 p.m.

1:45 p.m.

2:15 p.m.

2:30 p.m.

3:15 pjn.

3:30 p.m.

3:30 p.m.

3:15 p.m.
   Hazard Assessment: Summary
   Kenneth Reckhow

   Exposure Assessment:  Summary
   James Falco

   Risk Characterization:  Highlights
   Michael Brody, USEPA

   Risk Characterization:  Issues Presentation
   Lawrence Barnthouse

5:00 p.m.
   Risk Characterization:  Discussion (Two Workgroups)
   Lawrence Barnthouse and Mark Harwell
5:00 p.m. - 5:30 p.m.

5:00 p.m.
                PLENARY SESSION
5:30 p.m.
   Risk Characterization: Summary
   Lawrence Barnthouse and Mark Harwell


8:30 a.m. -12:00 p.m.       CLOSING PLENARY SESSION

             James Fava

9:30 a.m.
   James Fava

THURSDAY. MAY 16,1991 (cont)
9:35 a.m.     Resolved Issues
             Mark Harwell

10:15 a.m.     BREAK

10:30 a.m.     Unresolved Issues
             Lawrence Barnthouse

11:15 a.m.     Recommendations to EPA
             James Fava

12:00 p.m.     ADJOURN

                               LIST OF PARTICIPANTS

                          U.S. Environmental Protection Agency

William Adams
ABC Laboratories
Columbia, MO

Lawrence Barnthouse
Oak Ridge National Laboratory
Oak Ridge, TN

John Bascietto
U.S. Department of Energy
Washington, DC

Raymond Beaumier
Ohio Environmental Protection Agency
Columbus, OH

Harold Bergman
University of Wyoming
Laramie, WY

Nigel Blakely
Washington Department of Ecology
Olympia, WA

James Falco
Battelle Pacific Northwest Laboratory
Richland, WA

James Fava
Roy F. Weston, Inc.
West Chester, PA

Alyce Fritz
National Oceanic and
Atmospheric Administration
Seattle, WA

James Gillett
Cornell University
Ithaca, NY
Michael Harrass    ,  ;    .
U.S. Food and Drug Administration
Washington, DC

Mark Harwell
University of Miami
Miami, FL

Ronald Kendall
Clemson University
Pendleton, SC

Wayne Landis
Western Washington University
Bellingham, WA

Ralph Poitier
Louisiana State University
Baton Rouge, LA

Kenneth Reckhow
Duke University
Durham, NC

John Rodgers
University of Mississippi
University, MS

Peter Van Voris
Battelle Pacific Northwest Laboratory
Richland, WA

James Weinberg
Woods Hole Oceanographic Institution
Woods Hole, MA

Randall Wentsel
U.S. Army Chemical Research,
Development and Engineering Center
Aberdeen Proving Grounds, MD

                                  LIST OF OBSERVERS

                           U.S. Environmental Protection Agency

Sidney Abel
Office of Pesticides and Toxic Substances
U.S. Environmental Protection Agency
Washington, DC

Larry Bowers
Law Environmental, Inc.
Kennesaw, GA

Michael Brody
Office of Policy, Planning and Evaluation
U.S. Environmental Protection Agency
Washington, DC

Janet Burns
Office of Emergency, Remedial Response
U.S. Environmental Protection Agency
Washington, DC

Jeff Butuinik
Cleary, Gottlieb, Steen & Hamilton
Washington, DC

Chao Chen
Office of Health & Environmental Assessment
U.S. Environmental Protection Agency
Washington, DC

Damon Choppie
Bureau of National Affairs
Washington, DC

Charlotte Cogswell
Goldberg-Zoino &
Associates Geoenvironmental
Newton, MA
Thomas Dillon
U.S. Army Corps of Engineers
Vicksburg, MS

Donald Enye
FMC Corporation
Princeton, NJ

John Festa
American Paper Institute
Washington, DC

Jerry Frumkin
Government Institute
Rockville, MD

John Gentile
Environmental Research Laboratory
Office of Environmental Processes & Effects
U.S. Environmental Protection Agency
Narragansett, RI

S. Ian Hartwell
Maryland Department of the Environment
Annapolis, MD

John Helm, HI
Office of Toxic Substances
U.S. Environmental Protection Agency
Washington, DC

Eliz Hirsch
ICI  Americas, Inc.
Wilmington, DE

D. Eric Hyatt
EMAP Integration and Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC

Paul Klauman
Lockheed Engineering and Science Company
Washington, DC

Norman Kowal
Environmental Criteria & Assessment Office
U.S. Environmental Protection Agency
Cincinnati, OH

Steven Kroner
Office of Solid Waste
U.S. Environmental Protection Agency
Washington, DC

Ronald Landy
Office of Technology Transfer and
Regulatory Support
U.S. Environmental Protection Agency
Washington, DC

Rick Linthurst
Office of Research & Development
U.S. Environmental Protection Agency
Washington, DC

David Mauriello
Office of Toxic Substances
U.S. Environmental Protection Agency
Washington, DC

Monte Mayes
The Dow Chemical Company
Midland, ML

Margaret McVey
ICF, Inc.
Fairfax, VA

John Meier
Environmental Monitoring Systems Laboratory
 U.S. Environmental Protection Agency
 Cincinnati, OH
Charles Menzie
Menzie-Cura and Associates, Inc.
Chehnsfbrd, MA

Eugene Mones
Unilever Research U.S., Inc.
Edgewater, NJ

Ralph Northrop
Office of Pesticides and
Toxic Substances
U.S. Environmental Protection Agency
Washington, DC

Susan Norton
Office of Health and Environmental
U.S. Environmental Protection Agency
Washington, DC

Edward Odenkirchen
Environmental Impact Section
Food  and Drug Administration
Washington, DC

Dorothy Patton
Risk Assessment Forum
U.S. Environmental Protection Agency
Washington, DC

Kevin Reinert
Science Management Corporation
Valley Forge, PA

Donald Rodier
Office of Toxic Substances
U.S. Environmental Protection Agency
Washington, DC

Philip Ross
Office of Federal Activities
U.S. Environmental Protection Agency
Washington, DC

Louis Scarano
Environ Corporation
Arlington, VA

 Anne Sergeant
 Office of Health & Environmental Assessment
 U.S. Environmental Protection Agency
 Washington, DC

 Bill Shade
 Rohm & Haas Company
 Spring House, PA

 Michael Slimak
 Office of Environmental Processes and
 U.S. Environmental Protection Agency
 Washington, DC

 Rick Stevens
 NOR-AM Chemical Company
 Wilmington, DE

 Jerry Stober
 Office of Personnel Management
 U.S. Environmental Protection Agency
 Atlanta, GA

 Greg Susanke
 Office of Pesticides and Toxic Substances
 U.S. Environmental Protection Agency
 Washington, DC

 Gregory Toth
 Environmental Monitoring Systems Laboratory
 U.S. Environmental Protection Agency
 Cincinnati, OH

 William van der Schalie
 Risk Assessment Forum
 U.S. Environmental Protection Agency
 Washington, DC

 Stephanie Weinstein
 Jellinek, Schwartz, Connolly and Freshman
 Washington, DC

Molly Whitworth
Office of Policy, Planning and Evaluation
U.S. Environmental Protection Agency
Washington, DC
Chris Wilkinson
Technical Services Group
Washington, DC

Keith Williams
U.S. Army
Environmental Hygiene Agency
Aberdeen Proving Ground, MD

William Wood
Risk Assessment Forum
U.S. Environmental Protection Agency
Washington, DC

Dick Worden
U.S. Environmental Protection Agency
Washington, DC

I.E. Young
Union Carbide Chemicals &
Plastics Company, Inc.
Bound Brook, NJ

Maurice Zeeman
Office of Toxic Substances
U.S. Environmental Protection Agency
Washington, DC



                                       PEER REVIEW WORKSHOP DRAFT
                         Prepared for the
                      Risk Assessment Forum
               U.S.  Environmental Protection Agency
                          Washington,  DC
                            March 1991

Susan Braen Norton
Donald J. Rodier
Suzanne Macy Marcy
                         Technical Panel
          Michael Brody
          Anne Sergeant
David Mauriello
Molly Whitworth
William van der Schalie
William P. Wood
                  DRAFT -  DO NOT CITE OR QUOTE


                    DRAFT: 00 NOT CITE, QUOTE, OR DISTRIBUTE
                        TABLE OF CONTENTS
1.   INTRODUCTION  .	    1

     1.1. Intended Audience  	    1
     1.2. Definition of Ecological Risk Assessment   	    1
     1.3. Applications of Ecological Risk Assessment.  .  .  . '.    2
     1.4. Document Background and Ancillary Activities   ...    2
     1.5. Purpose and Scope  of the Framework Document  ....    3
     1.6. The Ecological Risk Paradigm	    4
     1.7. Ecological  Risk   Assessment  Issues  for   Future
          Consideration	    8
     1.8. Organization	11


     2.1. Stressor and Environmental Characterization    ...   12
     2.2. Endpoint Identification and Selection  	   13
          2.2.1.    Purpose  and Needs, of the Assessment  ...   15
          2.2.2.    Ecological Relevance  	   17
          2.2.3.    Susceptibility	   18
          2.2.4.    Practical Constraints    	   19
     2.3. Presentation of the Conceptual Model  and Evaluation
          Approach	20


     3.1. Hazard Identification 	   23
     3.2'. Stressor-Response  Assessment	   25
          3.2.1.    Types  of  Data  and  Analyses  Used  in
                    Stressor-Response Assessment 	   25
          3.2.2.    Extrapolation Methods for
                    Stressor-Response Assessments 	   26


     4.1. Estimating Exposure 	   30
          4.1.1.    Exposure 'Scenario Evaluation 	   30
          4.1.2.    Reconstructive Exposure Assessment  ...   31

     5.1. Assess the Likelihood of Adverse  Effects
Basic Concepts	33
Quotient Method of Ecological Risk
Characterization  	  34
Additional Approaches  	   36
     5.2. Describing the Consequences  of  Identified Risks



7 .

5.2.2. Methods for Describing the Consequences
5.3. Uncertainty in Ecological Risk Characterization . .
5.3.2. Reducing Uncertainty in Ecological Risk

5.4. Communicating the Results of the Ecological Risk


                     DRAFT:  00 NOT CITE, QUOTE, OR DISTRIBUTE

     In  1984,  the United  States Environmental Protection Agency
 (EPA)  organized the Risk Assessment Guidelines program to ensure
 scientific quality  and  technical consistency in the Agency's  risk
 assessments.  The first group  of five  guidelines,  issued  in 1986,
 focused  on evaluating  risks  to human  health.    In addition  to
 concerns for human health,  there has been an increased awareness  in
 the public, private, and governmental sectors of society regarding
 ecological  issues.    These  issues  include  global warming,,  acid
 deposition, a decrease  in biological diversity, and the ecological
 impacts  of xenobiotic   compounds such  as  pesticides  and toxic
 chemicals.  This Framework for Ecological Risk Assessment is the
 first  agency-wide   statement   of general  principles  to  guide
 ecological risk assessment.  It is intended to foster a consistent
 Agency approach for conducting  ecological risk assessments, help  to
 identify key  issues and research needs,  and provide  operational
 definitions for terminology.  . In addition,  it will serve as the
 foundation for  future subject-specific guidelines.

 1.1. Intended Audience

     This guidance  is  intended for risk assessors  in the Agency,
 and other persons who either perform work under Agency contract  or
 sponsorship  or  who  are  subject to  Agency  regulations1.    Risk
 managers in the Agency,  other Federal agencies, and state and local
 agencies may also benefit from  this guidance since it clarifies the
 terminology and methods used by assessors.

 1.2. Definition of Ecological  Risk Assessment

     Ecological  risk assessment evaluates  the  likelihood  that
 undesirable ecological  effects  may  occur or  are  occurring as  a
 result of exposure-to one or more stressors.   The  term stressor  is
defined here as any physical,  chemical or biological  entity  that
can induce an adverse effect.  Adverse ecological effects encompass
a wide range of disturbances,  ranging from an  increase in the
normal mortality rate  in  individual  organisms to  reductions  or
deviations from normal  ecosystem structure and function.
     1  It is preferable  that scientists with ecological  training
perform and  interpret  ecological risk assessments.  Those who  do
not have this background may find the standard texts listed  in the
references helpful (e.g., Odum, 1983; Krebs,  1985;  Ricklefs,  1990;
and Pianka,  1988).



                    DRAFT: 00 MOT CITE. OJOTE, OR DISTRIBUTE
     Risk is a function of two major elements,  hazard and exposure.
Hazard  refers to the type  and magnitude  of  effect  caused  by a
stressor.   It  is  usually evaluated  by  identifying  levels  of a
stressor associated with  effects  observed in  laboratory or field
studies.  Exposure refers to the co-occurrence of a stressor with
an ecological component (e.g., individual, population, community,
or ecosystem).  It  is usually determined by measuring or estimating
the amount  of the stressor  in  environmental  compartments (e.g.,
air  soil,  water).   An  adverse effect is  likely to occur in  the
field only  if exposure  approaches or exceeds  a  levels associated
with the adverse effects  identified in the hazard assessment.  A
probabilistic statement about the  likelihood of adverse effects  can
be made when stochastic estimates  of the two elements are provided.

     The current state  of the  art in ecological risk assessment
permits only  limited potential for developing  stochastic estimates
of both  the hazard and  exposure elements.  Thus, ecological risk
assessments  often  are deterministic  in  nature and likelihood is
expressed as a semi-quantitative comparison of exposure and hazard.
In  some instances,  such as evaluating  current or  past risks,
quantifying hazard and exposure may be difficult, and qualitative
risk estimates or  opinions  are often employed.  Even though  such
estimates may be qualitative, they are still considered  to be  risk
estimates in this  document.

1.3. Applications  of  Ecological Risk Assessment.

     Ecological risk assessments  play  a  fundamental  and  often
pivotal  role for addressing ecological effects.  Ecological  risk
assessments can be used  to  define problems,  set priorities,  and
serve  as a basis for  regulatory actions.   The ecological  risk
assessment  process  is  flexible  enough that  it can be  used to
predict future risks or assess adverse effects that are occurring
or  have  already  occurred.     An example  of  the  former is  an
evaluation   of  a  new   chemical   not  yet manufactured.    Such
assessments are often referred to as predictive risk assessments.
Examples of the latter include evaluation of hazardous waste sites,
eu?r?phication of aquatic systems, and oil spills.  These types of
assessments  are  commonly  referred  to  as  retrospective   risk
assessments.  Although the types of data and analyses may differ,
the elements of the risk assessment paradigm described in Section
1.6 are used in both types of assessments.

 1.4.  Document Background and Ancillary Activities

      As part of the present effort  in ecological risk assessment,
meeting were  held in the  spring  and  summer  of  1990  tc. «»view
 important scientific issues  (Gentile et al., in press).  Experts i,.

                     DRAFT:  00 NOT CITE, QUOTE. OR DISTRIBUTE
 ecology  and  ecological  risk  assessment  met  to  discuss  the
 ecological risk assessment paradigm,  uncertainty issues in hazard
 and exposure assessment, and population modeling.  Representatives
 from state  and federal  agencies described  how ecological  risk
 assessments  are  conducted in  their organizations,  and the  EPA
 Science Advisory Board  provided an informal consultation  on  the
 development of ecological  risk assessment guidelines.

      Based  in  part  upon  these meetings  as  well  as  extensive
 discussions with EPA managers  and scientists and outside experts,
 EPA has initiated a three-part program to develop ecological risk
 assessment guidelines.  Two efforts are underway in addition to the
 framework guidance document:

      Compilation of Ecological Risk Assessment Case Studies.  Peer-
      reviewed case studies  illustrating the "state-of-the-practice"
      in ecological risk assessment are being compiled  by  six  EPA
      work  groups   chaired   by  personnel  from  the   Regions,
      Environmental   Research   Laboratories,    and   Headquarters.
      Selected case  studies represent  a wide range of programmatic
      tasks and ecosystem types.  Individual case studies will  be
      compiled  into  an  overall  report   that will   include   a
      description of each study; a "tools" section that will contain
      a cross-referenced listing of ecological risk methods,  models,
      and assessment  schemes  used  in  the  case studies;  and  a
      discussion of issues related to ecological  risk assessment and
      research needs.  The report will provide interim assistance in
      performing ecological  risk  assessments   until   additional
      specific guidelines can be developed.

      Plans for  Future Guidelines.  A work group has been formed to
      create a work plan for long-term  (1991-1998) development  of
      ecological risk guidelines. This  group will  coordinate with
      other Agency ecological risk assessment activities, including
      the core research program and the Ecological Monitoring  and
      Assessment Program (EMAP).  Based on  scientific  feasibility
      and EPA's  program priorities,  the work group will recommend
      specific subject  areas for future  ecological  risk assessment

1.5.  Purpose  and Scope of  the  Framework Document

      This framework is intended to convey the general principles of
ecological  risk assessment and provide a  foundation  for  future
subject-specific  guidelines.    It will also  foster a  consistent
Agency approach for conducting  ecological risk assessments, help to
identify key issues and research needs,  and provide  operational
definitions  for terminology.    It is  not intended  to serve as  a

                    DRAFT: 00 MOT CITE, QUOTE, OR DISTRIBUTE
detailed  instructional  guide  or set of  rules.   The principles
discussed  here  apply  to  ecological  risk  assessments  at  the
individual,  population,  community,  and ecosystem organizational
levels.   The need  for  assessing risks at higher, organizational
levels  (i.e.,  communities  and ecosystems) has been highlighted
recently  (U.S. EPA,  1990a,b).   However, most operational methods
assess effects at  lower levels of ecological organization (i.e.,
individuals and populations) and these methods provide most of the
examples  discussed in this  guidance.   As  methods for assessing
risks  at  higher  organizational  levels  are developed, the Agency
will prepare more detailed guidelines.

     Risks posed  by introduced  exotic  species are not addressed
here because EPA does  not  have the  authority to regulate these
organisms.   EPA does have the authority to regulate  genetically-
engineered  organisms;  although the   risk  assessment  paradigm
described in Section 1.6 would conceptually apply to  genetically-
engineered organisms,  methods  for  evaluating  the  hazard  of and
exposure to such organisms are still being  investigated.  As more
experience is gained,  guidelines for  evaluating the ecological
risks  of  genetically-engineered  organisms  will  eventually  be

1.6. The geological  Risfe Paradigm

     This  guidance  represents the  first  Agency-wide  effort  to
identify and discuss the elements of ecological risk  assessment.
Figure 1 illustrates the elements of the ecological risk assessment
paradigm  described  in  this  framework.   Figure  2  presents the
paradigm in the context of  risk  management and policy concerns.

     The risk assessment paradigm published by the National Academy
of Sciences  (NRC "red book," 1983) is used as a foundation  for the
ecological risk assessment paradigm shown  in Figure 1. The Academy
identified "four  basic  elements of  risk  assessment:  1)  Hazard
Identification,   2)   Dose-Response   Assessment,   3)   Exposure
Assessment,,  and  4)  Risk  Characterization.   For the purposes of
ecological   risk,  assessment,  an   additional   step,  Conceptual
Framework  Development,  is  also  shown.   This  is analogous  to  a
preliminary  hazard identification that  identifies adverse  effects
associated with the stressor.  It is proposed here since ecological
risk  assessments,  unlike human health  assessments,  must  often
address the  risks of stressors to many species as well as risks to
communities  and ecosystems.  In  addition, there may be many ways  a
stressor  can  elicit adverse effects  (e.g.,  direct effects on
mortality  and growth,  or indirect effects  such as decreased food
supply).   A  systematic  planning  element helps identify  major
factors to be considered  in a particular  assessment in  order to


       Stressor and Environmental Characterization
          Endpoint identification and Selection
             Conceptual Model Formulation
Hazard Assessment
                                Exposure Assessment
Hazard Identification
Stressor Response
     Figure 1: Ecological Risk Assessment Paradigm


                  CONCEPTUAL FRAMEWORK

               Stressor and Environmental Characterization
                   Endpoint identification and Selection
                     Conceptual Model Formulation
         Hazard Assessment
          Hazard Identification
           Stressor Response
Exposure Assessment
                    RISK CHARACTERIZATION
                       RISK MANAGEMENT
Figure 2: Activities Associated with Ecological Risk Assessment


                     DRAFT:  00 NOT CITE, QUOTE, OR DISTRIBUTE
produce  a  scientifically-acceptable ecological  risk  assessment
relevant to risk management  decisions.

     There  are three elements of Conceptual Framework Development.
Stressor  and   Environmental  Characterization   describes   the
stressor's   potential   spatial  and • temporal  distribution   and
identifies  the ecological  components that  might be  exposed  to  it.
Endpoint Identification describes the types of effects that  may be
elicited by  a  certain stressor.    Conceptual  Model  Formulation
summarizes  plausible ways  a  stressor could cause  adverse effects.
Section  2 describes these  elements  in more detail.

     The four elements of the NAS paradigm complete  the ecological
risk assessment  paradigm.  The  term Hazard Identification is used
here  in  a  manner similar to NAS  (NRC,  1983) :   "the process of
determining whether exposure to an  agent can  cause  an increase in
the incidence of a health  condition (cancer,  birth  defect,  etc.)-
This element  characterizes the nature and strength of the evidence
of  causation."    Dose-Response Assessment  is  "the  process  of
characterizing  the relationship  between  the dose  of  an  agent
administered  or  received  and the  incidence of an  adverse  health
effect in exposed populations and estimating  the  incidence  of  the
effect as a function  of human  exposure to the agent."  The term
stressor-response.  rather than dose-response,  is  used  in this
guidance to include the great number of non-toxicological stressors
that impair ecological systems 1  Like the dose-response assessment
described by the  NAS  (NRC,   1983),  stressor-response  assessment
considers  the   intensity  and  pattern  of  exposure,  and  other
variables   (e.g.,  gender,  life-history  stage)  to   evaluate   the
responses elicited by  a particular  agent.

     In many ecological risk  assessments, the hazard  identification
and stressor-response assessments may be conceptually close  to  one
another.    In assessments  that  rely  on  laboratory  data,   the
information  used for hazard  identification (which  effects  are of
concern?) and stressor-response  (what is the magnitude  of   the
effect?)  may be  obtained  simultaneously.   In other  assessments,
information  linking the stressor  with the effect may be obtained
separately  from  the stressor-response assessment.   To accommodate
both approaches,  hazard  identification and stressor-response  are
retained  but are treated  as one element  of  the risk  assessment


                    DRAFT:' 00 MOT CITE, QUOTE, OR DISTRIBUTE
paradigm,  hazard  assessment2.    This  information  on  hazard  is
integrated with the exposure assessment to estimate risk.

     Exposure  assessment is  defined -by NAS  as  "the  process  of
measuring or estimating the intensity, frequency, and duration of
human exposures to an  agent currently present in the environment or
of  estimating  hypothetical exposures  that might arise from the
release  of  new  chemicals  into  the  environment."     Ecological
exposure assessment considers many of the same  concerns; important
issues  include  exposure  of multiple organisms,  exposure to non-
chemical  stresses,  and  the  timing  of  the- exposure  relative  to
important life cycle  attributes.

     Risk characterization  is defined by NAS  as "the process  of
estimating the  incidence of  a health  effect under  the various
conditions of  human exposure described  in exposure assessment."
This definition is applied  here  to ecological  effects,  but  is
expanded  to  include  a  discussion,  when  applicable,  of  the
ecological consequences of observed  or estimated adverse effects.
For example, a  risk assessment demonstrating  adverse effects on
aquatic invertebrates may also describe the ramifications of these
effects on other organisms such  as  fish.   Risk  characterization
includes a summary of  the strengths,  limitations  and uncertainties
of the data and models used to form conclusions.

1.7. Ecological Risfc Assessment Issues for Future consideration

     Incomplete  information  resulting from  gaps in  scientific
theory  is  common  in both  ecological  and   human  health  risk
assessments.     The  NAS  emphasized  the  need  to make  necessary
judgements and proceed with risk assessments under these  conditions
(NRC,  1983).  At the  same time,  it  is important to identify and
address critical  scientific issues  to fill  information gaps and
advance the risk assessment process.

     Table 1 presents a  number of issues of special significance
for future ecological  risk assessment guidelines. The orientation
of this document and  the  incomplete  development  of some subjects
     2 The combination of hazard identification and stress-response
assessment is called "hazard assessment" here.  Another definition
of hazard assessment is a quotient or margin of safety calculated
by comparing the toxicological endpoint of interest to an estimate
of  exposure  concentration.   The  latter definition  refers  to a
combination of stress-response assessment and exposure assessment.
This document uses only the  former definition.

                    DRAFT: 00 NOT CITE, QUOTE, OR DISTRIBUTE
Table 1:  Issues in Ecological Risk Assessment
     Conceptual Framework Development

     How do endpoints identified for a  risk assessment depend
     on spatial and temporal scale and the type of stressor?

     Are  there  endpoints  that  are  most  appropriate  for
     different types  of  assessment  (e.g., priority-setting,
     initial  evaluations of  risk,  evaluation of  remedial

     Hazard Assessment

     How should different types of evidence be weighed?

     What consideration should be given to data obtained from
     tests conducted with nonstandard procedures or conducted
     with  surrogate  species  for  which  there  is  little
     information on relative sensitivity?

     How  should  the  distribution  of individual  organism
     responses to the stressor be taken into account?

     What should be the  basis  for extrapolating among taxa,
     organizational levels,  and functional groups?

     What should be the basis for extrapolating the effects of
     physical perturbations?

     What is the nature of the stressor-response function at higher
     organizational levels (e.g., communities and ecosystems)?

     Exposure Assessment

     How should temporal and spatial  variation  in exposure
     (e.g.,  transitory vs.   resident  populations,  episodic
     exposure) be considered?

     How should multiple - stressors  and  multiple  routes  of
     exposure be considered?

     How is  exposure  influenced by  physical,  chemical,  and
     biological attributes of the environment?

     What are the  best attributes  of  physical disturbance for
     assessing exposure  (e.g., fragmentation, edge)?



                     DRAFT: 00 NOT CITE, QUOTE, OR DISTRIBUTE
Table 1 (continued)
     Risk Characterization

     How  is  the potential  for recovery  factored into  risk

     How can critical effects levels be incorporated into risk

     What  role  should  assumptions  play  in  reducing the
     probability of a false negative  (e.g., Type  II  error)?

     How are the overall results of the risk  assessment best

     What additional approaches  are available  for characterizing

     How can chemical  and non-chemical  stressors (e.g., habitat
     alteration) be combined in the risk characterization process?

     What alternatives  to  the quotient method are available  for
     risk characterization?

     General Issues

     What is the role  of risk management and policy in the risk
     assessment process?

     What information,  in addition to the points  raised above, is
     needed for conducting ecological risk assessments:

          1.    At regional and global scales?

          2.    At community and ecosystem levels?

          3.    For non-chemical stressors?


                     DRAFT:  00 MOT CITE, QUOTE, OR DISTRIBUTE
reflect  the lack  of  scientific knowledge  in some  areas.    For
example,  more  methods   are   currently  available  for  chemical
stressors  and individual-  or population-level  effects than  for
certain  non-chemical stressors and community-  or ecosystem-level
effects.   However,  risk assessment  guideline development  is  an
evolutionary process, so new approaches or methods for dealing with.'
these  issues  may be  incorporated  into future guidelines  as they
become available.

1.8. Organization

     The  remainder  of this  document  is arranged  sequentially.
Chapter 2 discusses conceptual framework development; this chapter
is particularly important for assessors to consider when endpoints
are not determined a priori  by statute or  other authority.  Chapter
3 discusses hazard  assessment, Chapter 4, exposure assessment and
Chapter 5, risk characterization.  Chapter 6 contains the glossary.


                     DRAFT:  00 NOT CITE, QUOTE, OR DISTRIBUTE
      Conceptual  framework  development  establishes  the  goals,
 breadth,  and  focus of the ecological risk assessment.  Its-product
 is  a conceptual' model that describes how a  stressor might  affect
 organisms,   populations,   communities  and  ecosystems   (i.e.,
 ecological components) in the natural environment.  This conceptual
 model is  evaluated further in the hazard identification,  stressor-
 response, and exposure  assessments.

      Conceptual framework development  begins  with the  review  of
 available information on the characteristics of the stressor and
 the  receiving environment, including the organisms,  populations,
 communities,   and  ecosystems  likely  to be exposed.    It also
 describes the  characteristics of the biological systems that might
 be  affected by exposure to the stressor'(i.e., endpoints).  Some
 endpoints are selected for further evaluation based on the purposes
 of  the  assessment,  ecological  relevance,   susceptibility,  and
 practical constraints.     Selected  endpoints  and  preliminary
 information on the stressor are  then integrated into the conceptual

      The  extent and detail of the development process depend on the
 purpose of the assessment  and the amount of  information  available
 on the situation under evaluation.  One conceptual  model may serve
 a  suite  of similar risk  assessments   (e.g.,  for  new  individual
 chemicals released to Water).

 2.1.  Stressor  and Environmental Characterization

      Conceptual    framework    development    begins   with   the
 identification of a stressor or  group of stressors.  In some  cases,
 an  effect observed  in the  field or  laboratory  can  be used  to
 identify  stressors that can be evaluated further.   In other  cases,
 the  description  of  a  source helps   identify   stressors.    A
 preliminary evaluation of  the characteristics  of the.'stressor and
 the receiving  environment  helps evaluate the spatial and temporal
 distribution of the stressor and identify the ecological components
 that may  be exposed to it.

     The  preliminary  evaluation  of  the  spatial  and   temporal
 distribution  of the  stressor uses available  information on its
 source and the factors  that influence its distribution  and fate.
This   includes  information  on  release   rates  and   patterns,
 physicochemical properties such  as water solubility and volatility,
 and environmental characteristics such as soil type  and climate.
The  physical  and chemical properties of the chemical  stressors
provide important insight into fate and distribution, which in turn


                     DRAFT:  DO MOT CITE, QUOTE, OR DISTRIBUTE
determines  which  ecological  components  might be  exposed.   The
components  evaluated in ecological risk  assessment are  discussed
below  and  can  include organisms,  populations,  communities, and

     The  amount  of  information  available  to  characterize the
temporal  and spatial  distribution of the  stressor and  identify
potentially exposed ecological components  varies greatly among risk
assessments.  In  site-specific assessments  (e.g.,  hazardous  waste
sites), the physical characteristics are best described using data
obtained from site investigations and sampling.  A general sense of
the expected physical  environment and biological community can be
provided  by using topographic  maps,  soil  maps,  remote sensing
techniques,  and vegetative-cover  or  ecoregional  maps.   Expected
populations and organisms, can be identified  by characterizing the
habitat  at the site.    Other  stressors  that may influence the
community should  also  be  identified at this  point.

     When  evaluating  stressors released to particular  habitats
(e.g., pesticides applied to  agricultural lands),  it  is  important
to consider- both  exposed and adjacent areas.   In addition, the
organisms using exposed and adjacent areas may vary  in  different
regions, even though the  habitat may  be similar.

     In  other   assessments,  very little  is  known  about  specific
potential exposure points, but fate-and-transpprt data and general
release locations can be used  to describe generic or representative
exposure  settings (e.g., aquatic habitats).  In  these cases,  a
generic  or  surrogate  community  can  be  defined,  and  surrogate
organisms can be  used  to  represent populations.

2.2. Endpoint Identification  and Selection

     The second major step in  the planning process is the selection
of  the  characteristics  of  ecological   components  that can be
adversely affected by exposure to a stressor.  For  the purposes of
this document,  these  characteristics are called  endpoints.  An
endpoint  describes  the  change  in   the characteristic  (e.g.,
increased mortality),  the  ecological  component  that  is  affected
(e.g., trout)  (Suter,  1990a), and often  the spatial  scale  (e.g.,
long-terra population viability of.' a  species  within   its current

     Assessment and measurement endpoints are  often distinguished
in ecological  risk  assessments.   Measurement  endpoints are the
effects  that can be measured.   As  used in  this guidance, the
definition  encompasses both  the  characteristic  that  is  measured


 (mortality)  and the  quantitative summary  of  those measurements
 (e.g. , an LC50)3.

     Effects that are readily measured may  not be directly useful
 in risk management,  because the significance of the response is not
 always  evident.   Assessment endpoints  are useful  intermediaries
 that describe the environmental value to be protected and thus link
measurement endpoints to the risk  management process.  They are the
ultimate focus  of risk characterization.   In  the best case, the
assessment endpoint  can  be measured and then the measurement and
the assessment  endpoint  are the  same.   If  an assessment  endpoint
cannot be  directly measured, measurement  endpoints . are  selected
that can be related, either qualitatively or quantitatively, to an
assessment endpoint.

     Measurement and assessment endpoints are often categorized by
organizational  levels.   Organizational  levels include  individual
organisms,  populations that include many individuals of  the same
species, communities comprised of interacting multiple populations,
and   ecosystems  comprised  of   organisms  and   their  abiotic
environment.  Multiple units at one organizational  level  form the
next higher  level,  and  changes  at  one  level  may, influence what
occurs at adjacent levels.

     Each organizational level has both  structural  and  functional
attributes that may serve as endpoints.  Structural  attributes of
ecological systems include,  for example, the mass of individuals,
the age-class structure of populations, the number and distribution
of populations within a community, and the  biomass  of ecosystems.
Functional attributes  involve the flow  of mass and  energy  (e.g.,
respiration  rate of  individuals,  intrinsic rate  of increase of
populations, primary productivity  of communities,  and decomposition
and nutrient cycling rates of ecosystems) .  The  interaction between
structure and function is an area  of active research. For example,
it is  still -difficult to map functional attributes onto species
assemblages because organisms may perform more than one function,
and may perform different functions at different  life stages.
     3    Alternative  terminology   distinguishes  between   the
characteristic that is measured  (called a response indicator)  and
the  quantitative summary  of the  results  (called a  measurement
endpoint) (U.S. EPA,  1990b).



                     DRAFT: 00 NOT CITE, QUOTE, OR DISTRIBUTE
     Table 2 presents example measurement and assessment endpoints;
in  practice,  an  ecological component  would also be  specified4.
Endpoints   at   each  organizational  level  have  strengths  and
limitations.     Risk   assessments  are  not   confined   to  one
organizational level, and may use a suite of endpoints  at multiple
levels  for  different aspects  of the assessment*  Measurement and
assessment endpoints are selected by considering the purpose of the
assessment, ecological  relevance,  susceptibility  to the stressor,
and practical constraints.  These criteria are  discussed generally
below;  more detailed discussions  can be found in Suter  (1990a),
Kelly and Harwell (1990),  and U.S.  EPA (1990b).   Endpoint selection
relies  on professional  judgement; for this  reason, the rationale
for selection should be clearly  documented.
Purpose and Needs of the Assessment
     It is important to consider the purpose of the assessment when
selecting endpoints.   Assessment endpoints. vary with program and
need within the program,  for example, assessment endpoints selected
to  support a  decision  under  a  specific regulation  may differ
substantially  from those used  to support a  request for  further
testing.  The assessor may wish  to consult with the risk manager to
identify  assessment  endpoints  for  specific  regulatory  needs.
Measurement endpoints also vary  with the purpose of the assessment.
For  example^   a measurement  endpoint diagnostic of  a specific
stressor may  be preferred when the  assessment is based on field
observations and this causal evidence is  particularly important to
the risk management decision.

     Assessment endpoints may be  selected because they  are valued
by society  (Clements,  1983).  Examples include the maintenance of
commercially-   or  recreationally-important   populations  or  the
viability of an endangered or threatened species.  Other examples
are attributes of ecosystems that are valued for functional (e.g.,
.flood  water retention  by wetlands)   or  aesthetic reasons  (e.g.,
visibility in the Grand Canyon).  In some cases the adverse effect
is an increase in undesirable species; the risk assessment may then
focus  on factors that  favor  these organisms.
     4  Several  areas  of  active  research  may  provide  useful
measurement  endpoints in the  future.   These  include  bioraarkers,
which measure physiological and biochemical changes, and landscape-
level  studies and  models  (e.g.,  Costanza et  al.,  1990),  which
describe distribution patterns of communities and ecosystems.  For
these  measurements  to be  useful  as endpoints,  an  established
relationship  with  endpoints  like  those  shown  in  Table  2  is



                     DRAFT: 00 HOT CITE, QUOTE, OR DISTRIBUTE
Table 2:  Recommended  Endpoints   at  Each  Level  of  Ecological
Organism health
Growth* •
Birth rate
Death rate
Age-Size-Class Structure*
Deviation in
structure and
function from
Species shifts
Numbers of species*
Species dominance*
Trophic shifts
Deviation in
structure and
function from
unimpaired system
Productivity  (P/R ratio)*
Nutrient dynamics
Materials and energy flow*
#    Generic  examples are  shown  in this  table;  in practice,  an
ecological component  would also be specified  (e.g., mortality in
trout;  flood  retention by wetlands).    In addition, the  spatial
scale of the endpoint is often  specified.

*    Depending  on the goal  of  the assessment, these  measurement
endpoints may also- serve as assessment endpoints.


                    DRAFT: 00 MOT CITE, QUOTE, OR DISTRIBUTE
     When evaluating  populations  that are valued,  it is critical
that  measurement  endpoints  include  both  direct  and  indirect
effects.  Direct effects include changes in characteristics of the
valued population, such as increased mortality,  reduced growth and
development, or  impaired  reproduction.   Indirect effects include
similar effects on species upon which the valued population depends
for food or  habitat  (see  the discussion on ecological relevance,

     When  evaluating valued  characteristics of  communities and
ecosystems, it is important to recognize that measurement endpoints
at lower organizational levels may  not  adequately reflect adverse
effects on community  structure and  function.  Endpoints at higher
organizational levels are  difficult  to quantitatively predict using
measurement  endpoints at  lower organizational  levels because of
characteristics  of ecological  components that  confer resiliency.
For example, individuals may be able to tolerate or  compensate for
a stressor through some physiological mechanism.  Communities and
ecosystems can continue to function despite changes in components
when  many  components  provide similar  functions5.   For  these
reasons, measurement endpoints at  the same organizational  level may
be the most useful, or,  alternatively, a conservative  approach may
select endpoints on the basis of susceptibility  (see the discussion
on susceptibility  below).
Ecological Relevance
     Ideal  assessment  and measurement endpoints are  ecologically
relevant.     Ecologically-relevant   endpoints  influence  other
endpoints, both at the same organizational level and also at other
levels.  Changes in endpoints  at higher organizational levels or at
large spatial scales are  often ecologically  relevant  because they
can involve large numbers of  organisms,  populations,  communities,
and ecosystems.

     Changes in endpoints at  lower  organizational  levels can also
produce  wide-ranging  consequences.   Changes in  organisms   and
populations that  provide  important  functions in communities  may
result in  indirect effects on other community members.   Keystone
     5 While they enable a system to persist despite exposure to a
stressor,   resistance   mechanisms   are   not   without   costs.
Physiological tolerance mechanisms may require additional energy,
and may leave the organism less fit to compete or reproduce.  Long-
term  exposure  to  a  stressor  may  decrease  species  or  genetic
diversity;  genotypes  or species that compete well  in  the .face of
one stress may be more vulnerable to future, different stresses.



                    ORAfTt 00 MOT CITE, OJOTE, 08 DISTRIBUTE
species, for example, influence the abundance and distribution of
other  community  members;   effects  on  them  may  induce  changes
throughout the  community.   Community.interactions determine the
extent  to  which an  effect will be  manifested  in  more  than one
organizational  level.    Food-web and  trophic  relationships are
primary  characteristics  to  consider  when  evaluating  indirect
effects.  In addition, competition for key  resources  ('e.g.,  food,
nesting sites, mates) and important species-to-species  interactions
(e.g.,  predator-prey  relationships,  mutualism,   and commensalism)
should be considered.  A complete risk assessment will account for
important ecological relationships during ehdpoint  selection.
     During early or minimal stages of exposure to a stressor, the
most susceptible individuals and processes are  often the first to
be  affected6.     Because   ecosystems  consist  of  communities,
populations,  and  individual organisms,  if shifts are observed at
the ecosystem level,  it is  likely that significant changes have
already occurred  at lower  organizational levels  (Rapport et al.,
1985; Hermann, 1985; Kelly  and Harwell,  1990; ESA, 1991).  However,
because  of differences  in sensitivity  within  an  organizational
level,  adverse   effects  can   be  seen  simultaneously  at  the
individual,  population, community and  ecosystem  organizational

     Information on relative susceptibility  can be  used to choose
endpoints that are  among the first to be affected by exposure to a
stressor.   Alternatively,  this information can be  used to select
the  measurement  endpoints  that  best   represent  response  to  a
stressor.    Risk  assessments  based  on  endpoints   that  are not
susceptible  will   underestimate  the  risk  to  more susceptible
endpoints.  For example, if an organism is tolerant of a stressor,
risk  assessments   based  on  responses  of   that  organism  will
underestimate risk to  less-tolerant  organisms,  and  may  also
underestimate risk  to a community of more-sensitive or more highly-
exposed populations.

     Susceptibility to a stressor is a  function  of both  exposure
and sensitivity.  Relative sensitivity is often stressor-specific,
but  can  also vary with  classes  of  stressors  (e.g.,   narcotic
chemicals,  pesticides).   Data used  to evaluate sensitivity can
     6 Low  inputs  of some stresses  (e.g., water,  carbon dioxide)
can actually result in improvements in health or productivity under
some circumstances.  This "subsidy effect" is not usually seen with
inputs of toxic chemicals (Odum et al.,  1979).


                     DRAFT:  00 HOT CITE, QUOTE, OR DISTRIBUTE
 include chemical structure-activity relationships (SARs),  effects
 observed in  the  field,  and  laboratory stressor-response  data.
 Exposure to multiple stressors may make organisms more sensitive to
 the particular  stressor  being  assessed   (e.g.,  in  populations
 exposed to habitat loss and harvesting,  toxicological impacts  may
 have  greater  impact  on population  viability than  toxicological
 studies alone suggest). Although sensitivity directly influences
 the outcome  of  the  risk  assessment,  information  on relative
 sensitivity,  particularly at higher organizational levels, is often
 unavailable during the planning process.

     The second  aspect  of  susceptibility  is  exposure.    When
 evaluating endpoints at the  individual and population level,  it is
 particularly  important to  consider life  history  attributes,  since
 life stage may influence both the sensitivity of the organism  and
 the magnitude of exposure.   Organisms may not be present when  a
 stressor is introduced in the environment because they move  into an
 area for short periods  to feed, breed, or mate, or they migrate on
 a  diurnal  or  seasonal basis.   These organisms  tend to be less
 exposed than  those living continually in  the environment if  the
 rout©   of  exposure  is  similar.    However,  organism  use   of
 environmental resources further influences the extent of exposure:
 some species  use resources where  exposure  will be greatest  (e.g.,
 predators at the top of the food web will incur greater exposure to
 chemicals that bioaccumulate).

     When  little   is  known  about  susceptibility,   a  group   of
 endpoints  is  often evaluated  in the hazard identification  and
 stressor-response  assessment.   For example,  surrogate species  are
 often  chosen  to  represent different  trophic levels  or  taxonomic
 groups.  No single species is appropriate for every situation,  but
 surrogates  can  provide  useful  information  with  a  reasonable
 commitment of resources.   The  selection of surrogate species  is
 discussed in .U.S.  EPA  (1980,  1982,  1990c).
Practical Constraints
     A  number  of  practical  constraints  further  influence the
utility of measurement and assessment endpoints.   Ideal assessment
endpoints have a clear, unambiguous definition and can be predicted
or  measured.    Ideal  measurement  endpoints  have  low  natural
variability,  are easy and inexpensive  to measure, have standard
protocols, are  supported  by  an existing data series, and produce
scientifically defensible results  (U.S. DOI,  1987;  Suter, 1989).

     Endpoints  at lower organizational  levels are  often the most
readily  measured  in  controlled  laboratory settings.   However,
community and ecosystem endpoints may be the easiest way to detect



effects in field studies.  For example, microorganisms that break.
down organic matter  may  be  adversely affected by a stressor, but
the  effect 'may  be  difficult to  recognize  at  the  individual,
population, or community  level.  Inhibition of decomposition can be
evaluated at  the  ecosystem level by  predicting or observing the
accumulation of leaf litter on a  forest floor.  In addition,rthe
natural  variability  in  some endpoints at  lower organizational
levels may be larger than those at higher levels.  In these cases
it may  be easier  to detect  stressor-related  changes  at higher
organizational levels.

2.3.  Presentation of the Conceptual Model and Evaluation Approach

     The  information gathered  on  the  stressor, the  receiving
environment and endpoints  is  integrated into a conceptual model.
The conceptual model consists of a  series  of working hypotheses
regarding how the  stressor  might  affect ecological components of
the natural  environment7.   The  conceptual  model  summarizes the
hypotheses that will be evaluated  in the hazard  identification,
stressor-response,   and  exposure  assessments,  and  provides  a
foundation to  determine  whether the  assessment will  reflect all
logical ways a stressor could cause an adverse response and ensure
that important endpoints are considered.

     The  conceptual  model  can  be  presented  in narrative  or
schematic  form.   An  example  of  a schematic  diagram  is  shown in
Figure 3.  In this example,  decreased birth  rate in a hypothetical
aquatic population was  selected as  an assessment; endpoint.   The
diagram  illustrates  different measurement  endpoints  that can be
used to estimate decreased birth rate.  The  use of these diagrams
is also discussed  in Barnthouse et al.  (1982)  and  Rodier  (1990).

     Many hypotheses may be generated during conceptual framework
development; those that  are most reasonable  and quantifiable are
selected  for further evaluation.    Because  of data  gaps,  some
hypotheses will not be carried further in  the .assessment;  it is
important  that   these   hypotheses  are  noted  when  evaluating
uncertainty during Risk Characterization.   Professional judgement
is needed to  select the most  appropriate  risk hypotheses; the
selection rationale  should be documented.
     7 The term hypothesis, as used in this guidance, reflects this
preliminary   thought  process  for  relating   and   demonstrating
relationships  between  observed  or  predicted  effects  and  the
stressor  under  evaluation.    The NRC  "green  book" (1986)  also
describes  development  and use  of  working  hypotheses in  ecological



                       DRAFT: DO MOT CITE, QUOTE, OR DISTRIBUTE

death rate

birth rate




in mating

production of
eggs or sperm

of spawnii
                         growth of
                 in food
             Figure 3: Example flow diagram analysis of
                       decreased birth rate in a population


                     DRAFT: 00 NOT CITE, QUOTE, OR DISTRIBUTE
     Data are not always available to support the development  of  a
conceptual  model.    When  data  are  insufficient  for hypothesis
development,  it   cannot be  concluded that there  is no  risk;
iterative  evaluations  are conducted  to  identify  data  gaps  and
ensure that hypotheses are developed.


                     DRAFT; 00 NOT CITE, QUOTE, OR DISTRIBUTE

      Hazard  assessment  describes  the  relationship between  the
 stressor and the endpoints identified during conceptual framework
 development.     Hazard   identification  describes  the   causal
 relationship  between   the  stressor   and   the   assessment   and
 measurement endpoints.   Stressor-response assessment evaluates ±he
 relationships between  the  stressor and  measurement endpoints  and
 quantitatively extrapolates from these to assessment endpoints.

 3.1. Hazard Identification

      Hazard  identification  qualitatively   evaluates  the  causal
 relationship between  a  stressor  and  an  adverse effect3.    The
 information gathered during hazard  identification supports  and
 complements the stressor-response  assessment.   For example,  when
 the stressor-response relationship  is based  on laboratory studies,
 the hazard identification might gather data on effects that occur
 in the field.   When stressor-response  is  based  on observational
 field data  (e.g., biomonitoring), hazard identification might focus
 on causal evidence.

      Both controlled tests and observational studies may be used in
'hazard  identification.   Where test data are not  available  (e.g.,
 for chemicals yet to be  produced) , structure-activity relationships
 may  be  helpful  (Clements  et  al.,  1988;  Auer  et  al.,   1990).
 Evidence  provided by these  studies  is evaluated by considering the
 elements  of  statistical design  and  analysis, particularly  with
 respect to replication  and variability.   For example,  statistical
 methods  are  not very  powerful   (i.e.,  they cannot detect  small
 differences)  when replication is  low  and variability is high,  such
 as  in many  observational studies.    And statistical  significance
 does not  always   reflect   biological   significance;   important
 biological  changes may  not be  detected  by statistical  tests.
 Professional judgement  and statistical consultation are both used
 to  evaluate statistical and biological significance.

      Controlled  laboratory  and  field  tests  can provide  strong
 causal evidence  linking a stressor with a response,  and can also be
 used to  discriminate  between  multiple  stressors.    Data  from
 laboratory  studies tend to be less  variable than those from field
 studies because many environmental factors can be  controlled in the
     8 A  causal relationship  occurs  when an event,  condition  or
characteristic plays  an  essential role  in producing an  adverse
response  (Rothman,  1986).



                    DRAFT: 00 HOT CITE, QUOTE, OR DISTRIBUTE
laboratory.    However,   because  these   factors  are  controlled,
responses may differ from those in the natural environment.

     Observational field studies provide environmental realism that
laboratory' studies  lack.    However,  the  presence  of  multiple
stressors and other confounding factors  (e.g., habitat quality) in
the natural environment  can make it difficult to attribute observed
differences   to   specific  stressors.     Confidence  in  causal
relationships  can be improved  by carefully selecting comparable
reference sites,  or by  evaluating changes  along a gradient of the
stressor where minimal  differences in other  environmental  factors
are  apparent.    Potential  confounding factors must be addressed
during the analysis.

     Many   of   the   concepts   applied    to  evaluating  causal
relationships  in human  epidemiology can be  useful  for evaluating
observational  field studies.   Hill  (1965) suggested  that nine
aspects of an association be considered when evaluating causality.
Rothman  (1986) summarized these criteria as follows:

     1)  strength, a high magnitude  of  effect is associated with
     exposure  to the stressor;

     2)  consistency, the association is  repeatedly  observed under
     different circumstances;

     3)  specificity, the effect is diagnostic  of  a  stressor;

     4)  temporality, the stressor preceded the  effect in time;

     5)  presence of a biological gradient, a positive correlation
     between the stressor and response;

     6)  a  plausible mechanism of action;

     7)  coherence,  the  hypothesis does not conflict with knowledge
     of natural  history and biology;

     8)  experimental evidence; and

     9)  analogy, similar stressors cause  similar responses.

 Not all  of  these criteria  need   to   be satisfied,  but  each
 incrementally reinforces the argument for causality.  Addition
 negative evidence does not  rule  out a causal association but may
 indicate that knowledge of the association  is incomplete.


                     DRAFT: 00 NOT CITE, QUOTE, OR DISTRIBUTE
 3.2. Stressor-Response Assessment

      Stressor-response  assessment  quantifies  the  relationship
 between .-the  amount of  the stressor  and magnitude  of response.
 Ideally,  it quantifies the  relationship  between the stressor and
 the assessment  endpoint  identified during  conceptual framework
 development.   When the assessment and measurement endpoint are the
 same,  this analysis is straightforward.  When they are different,
 the relationship between  measurement  and assessment endpoints is
 quantified first,  and then extrapolations  are used  to  predict
 changes  in  the  assessment   endpoint.     In  some  cases,   the
 quantitative   relationship  between  measurement  and  assessment
 endpoints is not known, and  qualitative inferences are made during
 Risk Characterization (see Chapter 5).
Types  of Data and  Analyses Used  in Stressor-Response
      The specific experimental protocols and statistical analyses
used  to  assess  stressor-response  relationships  depend  on  the
assessment objectives and available methods. Since methods change,
this  discussion addresses the  strengths and limitations of general

      Like hazard  identification, stressor-response assessments can
be  based  gn  controlled   laboratory  and  field   studies   and
observational  field studies.   Stressor-response assessments  often
progress  from short-term, inexpensive tests that measure effects on
mortality to longer-term tests that evaluate sublethal effects such
as  reduced  growth, development, or  reproduction.   Similarly,  the
test  environment  may   progress  from  very  uniform  laboratory
conditions  to more realistic mesocosm and  field  trials.    The
decision  for proceeding to a more detailed  analysis can be based on
stressor-response information  alone (e.g., the  LC50 is at or  below
a threshold value), or can be based on preliminary risk estimates.

      Data from   these  studies  can  be  used  to  test  specific
hypotheses or conduct a  regression  analysis.  Hypothesis testing is
most often used to identify a NOEL or LOEL (no-observed-effect and
lowest-observed-effect  levels,  respectively).   Hypothesis  testing
is  a  commonly-used and  accepted approach, but  has  some  important
limitations:  1)   statistical  significance may  not correspond  to
biological significance; and 2) a poor design or testing procedure
can reduce  the apparent toxicity  of the  chemical  (Barnthouse  e.t
al.f 1986).  When using hypothesis testing,  the power of the  test
to  detect differences and  the level of statistical  significance
should both be reported.


                     DRAFT:  DO MOT CITE, QUOTE, OR DISTRIBUTE
      Regression analyses can generate stressor-response curves that
 can  evaluate  risk  at  different  exposure  levels.    Regression
 analyses have been applied to both chemical and physical stressors
 (for an  example of the latter,.see Turner's [1977] analysis of the
 relationship between wetland area and commercial shrimp harvest).
 For practical reasons, the results of stressor-response curves are
 often summarized  as  one reference point, for instance,  an  LC50 or
 ECSO (lethal  or effective concentration, respectively, in 50 percent
 of a test population).  Although useful, these  values  provide no
 information  about the slope  or shape  of  the  stressor-response
 curve.    When  the entire curve is used, or  when many  reference
 points  are  identified, the  difference in magnitude of  effect at
 different  .exposure   levels  can  be   reflected  in   the  risk
Extrapolation Methods for Stressor-Rasponse Assessments
      As  discussed above,  a stressor-response assessment  ideally
quantifies  the relationship between the amount of the stressor and
the/magnitude of change in the assessment endpoint.   This  section
describes   quantitative  methods  used   to   extrapolate   between
measurement and  assessment endpoints.  If quantitative methods are
not  available, measurement and assessment endpoints  may be linked
•qualitatively    (see   Chapter  5) .     The   rationale  for   any
extrapolations and- their associated uncertainty  should be  clearly

      Most  quantitative  extrapolation  methods assess  response  to
chemical  exposure at the  individual level.   The discussion below
addresses species-to-species, endpoint-to-endpoint, and laboratory-
to-field  extrapolations.  Much less is known about  extrapolating
among communities and ecosystems.  Models to  extrapolate  between
levels of  organization and evaluate indirect effects have rarely
been  applied.    Active  research in  these  areas  may   provide
quantitative extrapolation methods in the future.   Because these
models are  most often used  to  provide a  common   framework  to
describe and compare consequences of adverse  effects rather than as
predictive  extrapolation  methods,  they  are  discussed  under Risk
Characterization (Chapter 5).

      Species-to-Species Extrapolations

      The difference in response between species is often estimated
based on   relative  differences  in  other  attributes  such  as
physiology,  morphology,  or   life history.    The  factors  that
influence response vary  from stressor to stressor. If a stressor's
mechanism  of action  is known, it may be easier to  identify the
characteristics  that  influence response and perform more confident



                     DRAFT: 00 HOT CITE, QUOTE, OR DISTRIBUTE
extrapolations.  These general concepts can be applied to physical
as  well  as  chemical   stressors.     For   example,   interspecies
extrapolations for habitat alteration can be qualitatively based on
life history characteristics  such  as resource utilization.

     Statistical1 methods have great utility for species-to-species
extrapolations,  although  most  of   these   focus   on  evaluating
responses  of  aquatic  organisms  to  chemicals.  One approach to
species-to-species extrapolations simply calculates concentrations
corresponding to a  specific  endpoint (e.g., an LC50)  for  a number
of species (see Sloof and Canton, 1983; Chapman, 1983; and Mayer et
al., 1986 for aquatic examples).   Untested  species are  assumed to
fall within the same range  (i.e., it  is assumed that tested species
adequately  represent  the  response  of  untested   species) .    The
response range can be very  large, and increases as more species  are
included in the  study.   However, the confidence that the  response
of untested organisms  falls  within the range  also increases with
the number of species.

     Regression models can be used to reduce the confidence limits
and increase the utility of species-to-species extrapolations by
correlating taxonomic proximity with  variation in response (Kenaga,
1978; Suter et  al.,  1983, 1986, 1987;  Sloof, 1986;  von Straalen  and
Denneman,   1989) .      These   models  indicate  that   taxonomic
extrapolations have narrower prediction limits for  closely-related
species  than  for  distantly-related  species.   In addition,   the
prediction  limits  tend  to be narrower  for  structurally-similar
chemicals  (and  chemicals  with similar  mechanisms of  toxicity).
Exceptions may  be expected when  life-history  characteristics or
biochemical and physiological processes are very different between
closely-related  species.

     An area of active research is  toxicokinetic and toxicodynamic
modeling.  These models evaluate inter- and intraspecies variation
in  response to  chemicals and  may provide a  basis  for  more
mechanistic extrapolations in  the  future.

     Endpoint-to-Endpoint Extrapolations

     Endpoint-to-endpoint extrapolations are used  when  short-term
endpoints are used to predict  long-term  or  chronic effects  (e.g.,
an LCSO used to predict a  NOEL).  These extrapolations often include
temporal and lifestage components  and may combine  several chronic
endpoints.  All  of these components  are  integrated in an analysis
of  acute-to-chronic  ratios   or   a   regression analysis.     The
relationships derived are then applied to other species for which
only acute data are available.  The implicit assumption here is


that the  difference between  acute and chronic  toxicity remains
relatively constant between species.

     Because of  the many  sources of  uncertainty,  this approach
often yields very large ranges of acute-to-chronic ratios and wide
prediction  limits  in  regression  analysis   (see,   for  example,
Barnthouse  et  al.,  1990 and  Sloof et al.,  1986).   Endpoint-to-
endpoint extrapolations often vary because the'degree of response
differs between  tests.   For example,  results  from an acute test
will be expressed as  an LD50,  whereas results from a chronic.test
are often  expressed as  a  NOEL or  LOEL.   One way  to reduce the
confidence limits and increase the utility of these extrapolations
is to correlate endpoints separately ar standardize the degree of
response (Mayer, 1990; Mayer et  al., 1  6; Suter  et  al., 1985).

     Laboratory-to-Field Extrapolations

     The responses  of organisms  exposed  in  the  laboratory often
differ from those exposed under natural conditions;  laboratory-to-
field  extrapolations  evaluate  these differences.     Laboratory
predictions may overestimate field response if they do not account
for compensatory or regulatory mechanisms, adaptation to stress, or
reduced bioavailability under field conditions (van Straalen and
Denneman, 1989;  Suter et al.,  1985).   On  the other  hand, they may
underestimate  field  response  if  laboratory  conditions  do not
reflect actual field conditions,  account for the ecological cost of
adaptation, or identify other interacting stressors.

     When possible, factors that influence differences in response.
between   the  laboratory   and   field  should   be   incorporated
quantitatively into the stressor-response  assessment.  For example,
some   data   are   available   that   relate   specific  habitat
characteristics  to  changes in response to a  stressor (e.g.,  water
hardness and metal  toxicity).   Similarly,  some  data are available
to help predict responses to complex mixtures that are composed of
chemicals having the same mechanism of action  (see  Broderius and
Kahl,  1985; McKim et al.,  1987).

     Laboratory-to-field extrapolations  can greatly increase the
uncertainty of response estimates, but the direction of any bias  is
often unclear.  If the laboratory-to-field extrapolation appears  to
be  the major  component of  uncertainty  in  an assessment,  field
studies may be warranted.


                     DRAFT: 00 MOT CITE, QUOTE, OR DISTRIBUTE

     For  the  purposes of  this document,  exposure  assessment is
defined as the assessment of the spatial  and temporal distribution
of a stressor and its co-occurrence with components .of ecological
systems.  This definition is somewhat broader than that provided, in
the  Guidelines for  Exposure Assessment  (U.S.  EPA,  1991)  which
focuses on human  exposure  to chemicals.   The exposure guidelines
differentiate between exposure, the contact of a chemical with an
organism's outer boundary,  and dose,  the  amount of chemical within
the outer boundary  of  the  organism.   While these definitions are
useful for chemical exposure to organisms, the broader definition
better represents  exposure assessment for ecological components
(i.e., populations,  communities, and ecosystems)  where the boundary
of the system does not serve as a barrier.

     Many of  the  other concepts presented  in the Guidelines for
Exposure Assessment also apply to ecological  exposure assessment.
Important aspects  of ecological exposure  assessment include the

     Many  different ecological  components  within • a  particular
environment  may be exposed,  including   organisms,  populations,
communities and ecosystems.

     The timing  of  the  exposure  relative to the  life  stage and
seasonal  activity  patterns  of  exposed  organisms  can  greatly
influence the occurrence  of adverse  effects.    Even  short-term
events may  be significant if they  coincide with  critical   life

     The perception of, as well as direct contact with, a stressor
can cause adverse  effects.  For example, the perception of degraded
spawning habitat  may cause  animals  to avoid spawning  areas and
decrease reproductive success.

     Exposure assessments  are  most effective when the results of
the exposure and stressor-response assessments are comparable.  For
example, exposure estimates used to evaluate  acute effects should
be  averaged over  .short periods  of  time  to take  into  account
short-term,  pulsed  stressor  events.   Exposure  assessments for
chronic stressors  should  account  for both  long-term,  low-level
exposure and possible shorter-term, higher-level  exposure that may
elicit similar adverse chronic effects.  Other factors to consider
include  cumulative  effects  from  continuous  or  intermittent
exposures, the magnitude and frequency of exposure,  and the life-
history stage of exposed  organisms.  Particular attention should be
given to exposure during periods of reproductive activity, since



                    DRAFT: 00 NOT CITE, QUOTE, OR DISTRIBUTE
early life stages are often more sensitive to stressors, and adults
may also be more vulnerable during this time.

     The  description  and  analysis  of  uncertainty  in  exposure
assessments is  combined with other uncertainty  analyses in risk
characterization.    Sources   of  uncertainty  and  methods  .for
describing it are discussed in greater detail in Chapter 5 and in
the Guidelines for Exposure Assessment (U.S. EPA, 1991).

4.1. Estimating Exposure

     Guidance on specific methodologies for conducting an exposure
assessment is  beyond the scope  of this document.   However,  the
overview  of   the basic philosophy  and  concepts  of ecological
exposure assessment  presented below provides a  basis for method

     There are three approaches used to quantify  human exposure to
chemicals: point-of-contact measurements, scenario evaluation, and
reconstructive assessment (U.S. EPA, 1991).  The point-of-contact
approach uses  monitoring devices to measure the stressor at the
actual point  of contact while exposure  is occurring.  Point-of-
contact measurements are rarely used in ecological risk assessment
because  it is  difficult to  attach  monitoring  devices  to free-
ranging organisms.  The other two methods (scenario  evaluation and
reconstructive) are discussed below.
Exposure Scenario Evaluation
     The  scenario  evaluation  approach  to  exposure  assessment
consists of two  basic  elements.   First, the spatial and temporal
distribution of the stressor is measured or estimated.  Second, the
distribution of  the biological  component and its characteristics
that influence  exposure are evaluated.   The  two are combined to
evaluate  the  co-occurrence  of  the  stressor and  the ecological

     The first element of scenario evaluation measures or estimates
the stressor's spatial and temporal distribution.  The initial  fate
and  transport evaluation  conducted during conceptual framework
development  should  be used  to  focus measurement  and  modeling
activities. The measurement and modeling of chemical stressors are
discussed in detail in the Guidelines for Exposure Assessment (U.S.
EPA, 1991).  Non-chemical stressors such as increased flooding can
be evaluated with techniques from geology, hydrology, engineering,
and other relevant  fields.   Physical alterations can be evaluated
by ground reconnaissance, aerial photographs, or:satellite imagery,
depending on  the scale of  the disturbance.,-  Quantifying specific



                     DRAFT: 00 MOT CITE, QUOTE, 0« DISTRIBUTE
attributes  of  physical  alteration  (e.g.,   fragmentation,  edge
effects)  is an  area  of  active  research  that may  yield use.'ful
methods  in  the future.

     The presence of  one  stressor may  indicate  that others are
present.  For example, removal of riparian  (streamside) vegetation
alters habitat structure directly. However, removal can also cause
siltation and increase water temperature.  In this case the initial
stressor  (vegetation  removal)   has  additional   ramifications
(siltation  and temperature rise).  Similarly,  the discovery of one
chemical may provide good reason  to  test  for others in  the same

     The second  element  of scenario  evaluation  considers  the
spatial  and temporal  distribution of  the  ecological 'components
under evaluation.  It  should also consider the  characteristics of
these  components  that  influence their  co-occurrence  with  the
stressor,  such as  habitat,  food preferences,  and  reproductive
cycles.  Seasonal activities like migration and use  of  alternate
resources may substantially  influence exposure  and should also be

     Exposure  scenario  evaluations  use  .information   routinely
obtained by the  Agency  and are  therefore  cost-effective  for
ecological exposure' assessments.  The  assessor should be aware that
scenario evaluation  implicitly  assumes  that measured  or  estimated
stressor concentrations  accurately represent those at the actual
point  of contact.   In  addition, exposure  scenario evaluations
commonly assess  stressors  individually,   and  may  under-  or
overestimate exposure to multiple stressors and mixtures.   Scenario
evaluation  can be performed  with little or no data;  consequently,
the  underlying assumptions  and uncertainties  should be  clearly
Reconstructive Exposure Assessment
     Reconstructive  exposure  assessments  examine  organisms to
determine the  presence of, or  previous  exposure to, a  stressor.
Biochemical  or physiological evidence  (e.g.,  biomarkers)  may be
used to  evaluate exposure.  This  form of exposure  assessment is
most useful when a chemical or unique metabolite can be detected in
the exposed organisms.  Changes in certain enzyme systems (such as
mixed-function  oxidases)   must  be  interpreted carefully because
other unrelated  stressors  may induce similar changes.

     Retrospective exposure measurements  (including biomarkers) are
most useful for risk assessment when 1) they can be  quantitatively
linked to the.amount of stressor contacted by the organism;  and 2)



                      DRAfT: 00 MOT CITE, QUOTE, OR DISTRIBUTE-
the  relationship  between  the measurement and  an adverse  response
can  be  defined as  part of the  stressor-respbnse assessment.


                     DRAFT:  00 NOT CITE, QUOTE, OR DISTRIBUTE

      Risk Characterization evaluates the likelihood that an adverse
 ecological  effect  could  occur  as  a- result  of exposure  to  a
 stressor, and  may also address the significance or consequences of
 identified  risks.   It integrates hazard identification,  stressor-
 response assessment,  and  exposure assessment  to  evaluate  the
 endpoints selected  during  conceptual framework development.

      As   discussed   previously,   the  purpose  of   the   risk
 characterization determines  its  sophistication and depth.   Before
 proceeding, the assessor may wish to  review  the conceptual  model
 and  the  relationship between measurement and  assessment  endpoints
 to evaluate how adequately the data  meet the  assessment's  needs.

      The strength  of a risk assessment  depends  on its supporting
 data.  Because there are many interactions between organisms,  their
 environment, and introduced  stressors,  it may not be possible to
 answer  every  question  that  arises  in  a  particular assessment.
 Thus,  the  assessor  may  need  to  supplement the analysis  with
 assumptions or models to bridge interpretational or data  gaps that
 arise during  risk characterization.   Any  methods or assumptions
 used, and the rationale for their application, should be explained.
 Because  ecological risk assessment is an area of current  research,
 methodologies and assumptions evolve and change.  Future guidelines
 may   address  developing  technologies  that  might  be   used  for
 ecological risk assessments.  In the  interim, EPA encourages  the
 assessor to employ state-of-the-art  methods and assumptions.

      The four basic steps of ecological risk characterization are:

 1.    Evaluate  the likelihood of  adverse effects;

 2.    Describe  the consequences of  identified  adverse effects;

 3.    Assess the uncertainty associated  with the risk assessment and
      the evidence that supports  the  conclusions; and

 4.    Communicate the  results of  the  risk characterization.

These elements are discussed below.

5.1.  Assess the Likelihood of Adverse  Effects

5.1.1.    Basic Concepts

      Ecological  risk assessment compares  predicted or measured
environmental  concentrations or levels  of  the stressor with  the



stressor-response data.  Thus,  the hazard posed by the stressor is
compared with the exposure  to  the stressor in'order to determine
the likelihood of effects resulting from a  combination of the two.

     The degree of quantification in the comparison or integration
step  depends  on  the' available  data.   Most  ecological  risk
characterizations  .for  single-chemical   stressors   are  easily
quantified because the endpoints used to  determine hazard can be
measured in  a field or laboratory setting,  and exposure  can be
measured or  predicted.   Characterizations for multiple-chemical
stressors are harder to develop because hazard and exposure data
are often unavailable  and are difficult to determine empirically.

     The quality  and  quantity  of  stressor-response and exposure
assessment data  determine  how  the risk characterization  can be
presented.   When  variation  in  the stressor-response and exposure
assessments   is   quantified,   the  results,  can   be  presented
probabilistically (e.g., there is a 50 percent probability  of a 10
percent"mortality).  However, in most situations, data limitations
permit  only   qualitative risk  expressions  (e.g.,  the  LCSO will
probably be exceeded).

     Although desirable,  a  quantitative risk characterization is
not  mandatory  (nor  may  it  be  achievable)   for a  successful
ecological risk assessment.  Qualitative judgements based on the
best available data can be very useful. If qualitative categories
of  risk (e.g.,  high,  medium,  low) are used, it  is important to
define  the categories  clearly.

   .  state-of-the-practice  approaches, which  are  based  on  the
principle  of comparing hazard  with exposure,  are  presented in
subsequent sections..   Some methods use only a single measurement
endpoint such as  an LCgo.  If the  results of an assessment  are to
be  used  as   decision   criteria  (e.g.,  determine  the  need  _for
additional testing), a comparison of single  measurement  ertdpoints
may be  appropriate.  On the other hand, if the assessment compares
regulatory options  for mitigation, it  is desirable that -the risk
characterization compare several stressor levels to obtain-a range
of  values.   The latter approach provides better insight into the
magnitude or severity of hazard than a single measurement endpoint.
Quotient Method of Ecological Risfc Characterization
     A commonly-used method of ecological risk characterization is
called the Quotient Method (Barnthouse et al,  1986).   It compares
hazard with exposure and has  been used extensively for addressing
the  risks of pesticides  (U.S.  EPA,  1986)  and industrial chemicals
 (U.S. EPA, 1990d).  The  algorithm is given below:

                     DRAFT:  00 NOT CITE, QUOTE, OR DISTRIBUTE
            Exposure Value

        Stressor-Response Value
      Exposure estimates may be measured or estimated, and may need
 to  be  adjusted  to  account  for differences  in  bioavailability
 between laboratory  and  field  conditions.    The  frequency  and
 duration  of  the  field  exposure and the  exposure used  in  the
 stressor-response assessment  should  also be  comparable.    The
 Quotient Method implicitly assumes  that  the predicted  or measured
 exposure duration equals or exceeds  that of the toxicological tests
 used  to derive the stressor-response curves.

    .  Stressor-response  values  commonly  used  with  the  Quotient
 Met hod. include LCsos, ECsos,  LOELs, and maximum acceptable toxicant
 concentrations (MATCs).    When  needed,  stressor-response  values
 should  include  the  extrapolation  factors presented  in  Section
 3.2.2.  Often, the stressor-response values are adjusted to provide
 some  conservative measure of protection.  As an example,  one-tenth
 of an MATC might be used as the stressor-response value (U.S.  EPA,

      Interpretation   of   the   Quotient   Method    is    fairly
 straightforward.   The greater the  expected exposure  compared to
 stressor-response values,  the  larger the  quotient and  greater the
 risk  (i.e., greater  likelihood that the adverse  effects  described
 by the  stressor-response  value will occur).  The  Quotient  Method
 works best when the  ratio is either very low or very  high.   When
 the  ratio is near  1,  the  results  cannot be  interpreted  with
 certainty.  Professional  judgement  should be used in  such  cases,
 and additional hazard and exposure  data might be  sought.    The
 Quotient Method is most often applied by  comparing one value  from
 the stressor-response curve  to exposure levels.  An extension of
 this  method that provides greater  insight into the magnitude of
 expected effects  is  to compare many stressor-response  values.

      The Quotient Method  has several  advantages:   It is  simple,
 flexible,  and  amenable   to  the   data  obtained  in   standard
 ecotoxicological  tests  and  exposure  assessments.    Among  the
 disadvantages  are that it cannot easily  be  applied to multiple
 stressors  or  cumulative  effects,   and  it  cannot  predict  the
magnitude  of   any effect  except that  which  corresponds  to  the
 reference point used in the calculation.

      The Quotient Method addresses risks of direct effects that can
be quantified; these may or may  not be assessment endpoints.   For
 example,  a stressor may  not be directly  toxic to  a  fish of
 interest, but may be  toxic to the invertebrates it feeds upon.  The


                     DRAFT: 00 HOT CITE, QUOTE, OH DISTRIBUTE
critical  relationship  between mortality in aquatic  invertebrates
and  reductions  in a fish population may  be difficult for a risk
manager  to  recognize  unless  the  assessor  links  the  two  and
describes the consequences during risk characterization.
Additional Approaches
      The basic principle of integrating hazard and exposure can be
applied in many ways.  For example, exposure assessment models have
been used to determine how often a particular stressor-response or
other reference value will be exceeded in rivers and streams during
a one-year  season (U.S. EPA,  1988).   Suter  et al.  (1983) .treat
reference values,  such as  an  MATC,  as probability distributions
that  are  matched  against similar  exposure  distributions  to
determine the probability of exceeding the  reference.  Models can
combine many different  reference values with models  of exposure:
For example, Pearlstine et al. (1985)  combined hydrologic  models
with stressor-response data relating water level to tree growth to
estimate  the  response  of  a  bottomland  hardwood  community  to
different water flow regimes.

5.2. Describing the Consequences  of Identified  Risks
General Concepts
     Many ecological risk assessments evaluate endpoints that can
be directly measured or estimated using a closely-related surrogate
or quantitative  extrapolation  methods.   In these cases,  the need
for an evaluation of the consequences is reduced.   In other cases,
appropriate   extrapolation   methods  relating   measurement  and
assessment endpoints are  not available,  and an evaluation of the
consequences of a measurement endpoints's occurrence is an integral
part of the risk characterization process.

     Consequences  include indirect  effects,  effects at multiple
organizational levels,  and effects at greater spatial and temporal
scales.  Indirect consequences of an adverse  effect  are evaluated
using the logical structure established during conceptual framework
development  as  well  as  professional  judgement.    Interspecies
relationships  (e.g.,   predation)   and  resource   utilization  are
considered when  evaluating  indirect effects.   Effects on higher
organizational levels  depend on the severity of the effect, the
number of organisms affected,  the role of those organisms in the
community  or  ecosystem,   and  characteristics  that  influence
resiliency (see Section 2.3.2).

     The  implications  of adverse effects  to greater spatial and
temporal scales are usually  evaluated on a case-by-case basis and


 are influenced  by the spatial  and temporal distribution  of  the
 stressor.  The spatial extent of adverse effects can be compared to
 the overall  extent  of  the  ecological resource.    For  example,
 adverse effects  to a resource that  is small in scale  (e.g.,  acidic
 bogs)  may have a small spatial effect, but represent a significant
 degradation  of  the  resource  due  to  its  overall   scarcity.
 Immigration and  emigration  patterns  can be  used  to  evaluate  the
 implications of a local loss (e.g., destruction of a local heron
 rookery affects  heron abundance over a much larger area).   At  the
 ecosystem level, import and  export  functions  are considered (e.g.,
 destroying coastal wetlands can reduce nutrient  export to adjoining

     The effects of short-term exposure may have long-term impacts.
 The temporal extent of adverse affects depends  in part  on  the
 attributes of the  exposed  systems that influence resiliency  and
 recovery.   Ecosystem recovery  depends  on  physiological,  life-
 history,  and genetic-adaptation mechanisms9, and  is  difficult to
 predict.   However, some useful  generalizations  can be  drawn from
 recent reviews  (Cairns,  1990;  Poff and  Ward,  1990;  Kelly  and
 Harwell,   1990).    Recovery  depends  to a  large  extent  on  the
 existence of a  nearby  source   of  organisms to  immigrate  to  an
 affected system.  The source can be refugia within  the  affected
 system,  or a nearby  unaffected area.  If some individuals are in a
'latent or unsusceptible stage during exposure,  these individuals
 can provide-a source  of  immigrants.   Organisms  immigrating  from
 other  areas must be able  to reach the affected area," the  distance
 immigrating organisms can travel depends on their mode of transport
 (e.g.,  by wind,  water, self-propulsion)  and the characteristics of
 the habitat between the two areas.  Finally, the success  of  the
 immigrants depends on the chemical-physical environmental quality
 following exposure to the stressor (e.g.,  presence of  persistent

     In summary, the evaluation of consequences  may be critical for
 certain risk assessments  in which the relevance  of measurement
 endpoints  is  not  clear to risk  managers.    However,  not  all
 ecological risk  assessments  require an evaluation of consequences.
 In  cases  where  the stressor-response  data  indicate that  direct
 effects are the main concern and these endpoints can be measured or
 estimated  using  quantitative  extrapolation   methods,   further
 evaluation of consequences  may  be unnecessary.
      9  There  is  some  evidence  to  suggest  that these  adaptive
mechanisms  are influenced by historical  patterns of temporal and
spatial heterogeneity.



                    DRAFT: 00 NOT CITE, QUOTE, OR DISTRIBUTE
5.2.2.    Methods  for Describing the  Consequences of Identified

     There are twd basic methods for describing the consequences of
identified  risks:   1)  narrative statements  and 2) mathematical

     Narrative  statements  describe the  possible consequences if
adverse effects occur in the environment.  This  approach uses the
logical   structure   established   during   conceptual   framework
development   and   supporting  information   to  demonstrate  the
consequences of concern.

     When   sufficiently   supported,   mathematical  models  can
quantitatively describe the consequences  of identified risks. Most
often, they are used  to  provide a common framework for comparing
the consequences of adverse effects:  For example,  an assessor may
use laboratory data on the stressor's  effect  on  the mortality and
reproduction of individual organisms  to  evaluate its effect on a
population.     Mathematical   models  can  extrapolate  data  for
individual organisms to effects at  the population  level.

     The  applications of mathematical models to ecological risk
assessment are not discussed here, but this  topic may be addressed
in future guidelines.  In general,  models should be considered in
light of how well  they make use of  available  data  and support the
decision-making process.

     No single model  is suitable  for all risk assessment, and the
assessor  may  want to  apply  more than one  model  to a  system to
cross-check results.   The  most appropriate  model is the one best
able to address the assessment endpoints selected during conceptual
framework development.   Models used to  project risks to natural
populations  and   communities  can  be  divided   into  two  main

     a) Sinale-species population fdemographic) models  can be used
     to predict direct effects on a single population of concern.
     These are basically bookkeeping models  which balance natality
     and mortality factors for the  population in question.

     b)  Multispecies models  can  include  both  aquatic  food web
     models and terrestrial plant-succession or forest-gap models,
     and  may  be  used  for risk  assessments  at the population,
     community, or ecosystem level.  Such models  can assess both
     direct and indirect effects of the stressor on the population
     of  interest  and effects on  the  community or ecosystem as a
     whole.   Generally,  these are  large-scale mechanistic  models


                    DRAFT: 00 NOT CITE, QUOTE, OR DISTRIBUTE
     comprised of  a  coupled  system of difference or differential
     equations.  Both spatial and temporal dynamics can be modeled.

     In summary,  narrative explanations and mathematical models are
equally-valid  approaches  for  describing  and  evaluating  the
ecological  consequences  of an adverse effect.   Either  approach
should present the assumptions made and associated uncertainties.

5.3. Uncertainty in Ecological Risk Characterization
Characterizing Uncertainty
     In  this   step,   the  assessor  describes  the  sources  of
uncertainty in each step  of the risk assessment and the impact of
each on the risk assessment's conclusions.  Elements of uncertainty
include those  associated with particular  analyses,  methods, and
techniques (e.g.,  fate-and-transport modeling, extrapolations).  In
addition,  the  rationale  behind any assumptions should be clearly

     A useful approach to uncertainty characterization, suggested
by Holling (1978) , divides uncertainty  into three  major classes:

1) Events that can be predetermined, with known affects and  known
probabilities  of  occurrence.   This type of  uncertainty has also
been  described   as   quantitative  uncertainty  (Suter,  1990b).
Examples   include  natural   variability  (O'Neill,   1979)   and
experimental, measurement, and sampling errors.

2) Events  that  are partially describabla, but hav« unknown outcomes
or probabilities.  This  type of uncertainty refers to incomplete
characterization  of the   system in  question.   In  any assessment,
some variables  are  known but  excluded  by choice,  and others are
unknown and therefore inadvertently excluded.   This category  deals
primarily with the hypotheses  developed to explain the effects of
the  stressor  on the  system.    Insufficient  data  and. lack  of
fundamental, understanding about ecological processes are- examples
of this type of uncertainty.  Other sources are extrapolations from
reference  systems,  unpredictable  perturbations, and  indirect
effects (Harwell  and Harwell,  1989).

3) Events  for  which we have  no  experience or knowledge, or that
involva unknown processes of unknown form. This category includes
"true" uncertainties,  those for which  we lack and cannot acquire
information.  Such uncertainties may be manifested  as  inconsistent
results from alternative hypotheses or contradictory predictions
from alternative  models.


                     DRAFT: 00 HOT CITE. QUOTE, OR DISTRIBUTE
Reducing Uncertainty in Ecological Risk Assessments
     The  approach used to reduce  uncertainty in ecological risk
assessments varies  with  the  source of the uncertainty and how it
influences the assessment.

     Quantitative treatments of uncertainty deal primarily with the
first category of uncertainty (Finkel,  1990; Suter, 1990b).  The
effects  of this  type of error  on  risk assessments  have been
described by O'Neill  and Gardner (1979).   They can be quantified
using Monte-Carlo simulation or statistical uncertainty  analysis
(O'Neill et al.,  1982).   Probability  distributions  for parameters
and processes enable  the assessor to place statistical bounds on
the results of the assessment.  Measurement errors can be minimized
by  obtaining  data  in  accordance  with  accepted  methodologies
(published guidelines or  other validated  methods).  Computational
errors  should  be  minimized  by  adherence  to  good  laboratory
practices and quality assurance procedures.  Natural variability
can  be  acknowledged  and described,  but  it normally  cannot  be
controlled  or  minimized.    However,  it  may   be  described  by
appropriate statistical distributions and  the quantitative methods
described above.

     Uncertainty  due to  data  gaps   (Category  2 above)  is best
addressed  by  collecting additional  information.    Conceptual
uncertainty can be addressed during conceptual model  formulation by
development  of   alternative  risk   hypotheses  or  alternative
predicative process models.

     As noted by Suter (1990b) ,  one way to reduce uncertainty is to
use a combination of  modeling, observation, and  experimentation.
Where  a   strong  causal  relationship   has  been  established,
observational or  field studies  may  be the  most appropriate for
reducing  uncertainty,  particularly conceptual  uncertainty.   For
example,  effects  observed in  the laboratory can be verified by
outdoor  studies,  which  reduce  uncertainty about whether the
observed  effects  .will occur  in the  field.   Where  the  cause of
observed effects is less certain, controlled laboratory experiments
or modeling may help reduce conceptual uncertainty.  Once causality
is established,  appropriate  studies  can  be undertaken to reduce
quantitative uncertainty.
Presenting Uncertainty
     Quantitative  approaches  reduce  but  cannot  eliminate  the
effects of true uncertainty.  At best, such approaches place bounds
on  the  risk  assessment's  conclusions.    Differences  between
predictions  from  alternative models or  hypotheses may result  in

contradictory or ambiguous conclusions.  While such differences may
be resolved by further analysis, decisions are often made  in spite
of remaining uncertainties (Tversky and Kahneraan,  1974).   Peterman
(1990a,b) suggested  that  risk assessments include an estimate of
the  statistical  power  of  the analysis,  which  can be  used  to
determine  the  probability  of  rejecting  the   null  hypothesis
(concluding that  the stressor causes no effect when an effect is
indeed  present).    This  analysis  is  useful   for evaluating  the
results of several alternative models.

     There  is  not  always  enough  information  to address  all
questions that  arise during  a risk assessment,  so the  assessor
usually  makes  a  number  of  assumptions  based on  data  from  the
stressor-response  and  exposure   assessments.     Because  these
assumptions may introduce conceptual uncertainty, they should be
carefully explained  in the risk characterization.

     A  weight-of-evidence  approach  is  often  used  to  present
information supporting  the results  and  conclusions of  the risk
characterization.    Weight-of-evidence  analysis  evaluates  data
quantity and quality  as well as supplemental information (U.S. EPA,
19-90d) such as that  described  in Hazard Identification.   This can
include  data  from field  studies,  observations  of other adverse
effects  caused  by a particular stressor,  and data on  observed
effects for, similar  types of  stressors (e.g., those with similar
mode of action or belonging to the same chemical  class).

     The uncertainty analysis  should  present  the strengths  and
weaknesses  of  each  risk  assessment element  and  the  impact  of
associated uncertainties  on the risk assessment's conclusions.  A
clear presentation of how uncertainty could relate to final risk
management decisions can  provide  justification for obtaining or
requesting additional information to reduce uncertainty.

5.4. Communicating the Results of the Ecological  Risk Assessment

     The  presentation  of an  ecological  risk assessment  is  as
important as its scientific validity.  Ecological  risk  assessments
may  focus  on  endpoints  not  readily   understood by   managers
unfamiliar with  biological or ecological concepts.  Whether its
conclusion is a simple  quotient or  a sophisticated simulation, a
risk assessment may  be  misunderstood or  misinterpreted if poorly
presented.     Suggestions  for   communicating   ecological  risk
assessment results follow:

Clearly   establish-   the   relationship   between   endnoints	and


                     DRAFT: 00 NOT CITE, QUOTE, OR DISTRIBUTE
     In many cases the assessment endpoint is not directly affected
by  a  stressor but will be  severely impacted because of indirect
effects  due  to  loss  of  essential supporting  resources.   The
assessor should explain the ecological relevance of such risks even
if  only in a  qualitative discussion.

Provide a summary profile of the degrees of  risk.

     Risk assessments that  compare many stressor-response values to
exposure estimates  provide greater insight into the magnitude of
effects.   Summaries  may  be  presented  in tabular,  graphic,  or
narrative form.  When more  than one analytical method is employed,
all results should be discussed.

Restate the assumptions  and uncertainties  in the ecological risk

     Although assumptions and uncertainties are identified  in each
element of ecological  risk assessment,  those that most  influence
the conclusions should be repeated  when communicating the  results
of  the ecological  risk  assessment.    The  weight  of evidence
supporting the estimates should be  clearly summarized.

     Often, the data employed  to  conduct a  risk assessment are
flawed  or  incomplete.    Data deficiencies  should be  clearly
identified.   The assessor  should also  describe  the   potential
contribution  of more accurate  or  complete data  to reducing the
uncertainty in the risk estimates.

Place  risk  in  the  context of  severity,   including  temporal  and
spatial attributes.

     There is  no  universally-accepted scale  that  can be used to
compare ecological effects.  However, the considerations outlined
in  Section  2.2 can be used  to guide  discussion  of  direct and
indirect effects.  Discussions of the spatial  scale of  effects
relative to the extent of the resource  affected and the  time frame
of the estimated  effects add significantly to the value  of a risk


                     DRAFT: 00 NOT CITE, QUOTE, OR DISTRIBUTE

Commensalism—A  relationship  between two  species  in  which one
benefits from the association  and the other is  unaffected.

Community—An aggregate of multiple populations within a specified
location in space and time.

Direct  Effect—Any  adverse effect  induced  by a  stressor that
directly  affects  an  ecological  component of   concern   (e.g.,
individual, population).

Ecological Risk Assessment—The characterization of effects of one
or more stressors on  biotic  components  of ecosystems  including
communities, populations, or individuals.

Ecosystem—The biotic  community and abiotic environment within  a
specified location in space and time.

Endpoint—A characteristic of  an  ecological system that can be
affected by exposure to a stressor.

Endpoint. Assessment—The characteristic  of the ecological system
that is the focus of the risk  assessment.

Endpoint. Measurement—An effect on  an ecological  component that
can be measured and described  in some quantitative fashion.

Exposure—The  co-occurrence  of  a  stressor  (including  sensory
perception) with one or more ecological components.

Hazard—The intrinsic ability  of a substance or other stressor to
cause adverse effects under a  particular  set of circumstances.


                     DRAFT:  00 NOT CITE, QUOTE, OR DISTRIBUTE
 Hazard Assessment10—The overall process of evaluating the type and
 magnitude  of  adverse  effects  caused by   a  stressor.    Hazard
 Assessment consists of two steps -1)  Hazard Identification and 2)
 Stressor-Response   Assessment.      Hazard   Identification--the
 evaluation of the  causal  relationship between a  stressor  and an
 adverse effect.             .

 Indirect Effect—An adverse  effect elicited  by  a stressor  to  a
 ecological component via a reduction or change its food supply or
•other trophic level disturbances such as predator/prey imbalances.

 Keystone Species—A species whose function plays a  critical role in
 maintaining a particular community structure.

 Lowest  Observed   Effect  Level   f LOEL1.--The  lowest   amount  or
 concentration of  a stressor for which some  effect is observed.

 Maximum  Acceptable Toxicant   Concentration   (MATC)—The  maximum
 concentration at  which a stressor can be  present and not be toxic
 to  the  test organism.  The MATC  is  normally calculated as  the
 geometric mean  of  the  lowest: concentration for. which  an adverse
 effect was  observed and the highest  concentration that did not
 yield any adverse effects.

 Median  Effective  Concentration   (EC501_—The  concentration  of  a
 stressor in water that is estimated to be  effective in producing
 some  response,  other  than mortality, in 50 percent of  the  test
 organisms over  a  specific time  interval  (e.g., a. 48-hour daphnid
 ECSO) .                                 '•"."•;•

 Median Lethal Concentration (LC50)—The concentration of a stressor
 in water that is  estimated to be lethal to  50  percent of the test
 organisms over a specific time interval (e.g.,  a 96-hour fish LC50) '-.
      10  Hazard   assessment  currently   has  two   commonly-used
definitions: The first is the evaluation  of the intrinsic potential
of  a stressor to cause adverse effects  under  a  particular set of
circumstances.  This definition refers to the combination of hazard
identification  and  stressor-response  assessment.    The  second
definition is  a  quotient  or  margin  of   safety  calculated  by
comparing  the toxicological endpoint of  interest to an estimation
of  exposure concentration.  This definition refers to a combination
of  stress  response and exposure assessment.   This document refers
only to the  first  definition, which confines hazard assessment to
hazard identification  and stressor-response.


                     DRAFT: 00 NOT CITE, QUOTE, OR DISTRIBUTE
 Mutualism—A relationship between  two  species  where both benefit
 from the association and in fact cannot survive separately.

 No Observed Effect Level (NOEL)—The amount or concentration of a
 stressor that does not result in any adverse effect.

 Null Hypothesis—-A hypothesis  that states  there is no difference
 between parameters.  The null hypothesis is usually designated as

 Population—An aggregate of  interbreeding individuals  of a species
 within a specified location in space and time.

 Predictive Ecological Risk Assessment—The  characterization of the
 ecological  effects  of  a  stressor  prior  to  its  release  or

 Primary Productivity—The production  of energy  from  sunlight by
 green plants.

 Recovery—The ability of a population or community  to  partially or
 fully return to a  level  of  equilibrium that existed prior to the
 introduction of the stressor.

• Resiliency—The ability of a population or  community to persist or
 maintain itself in the presence of one or more stressors.

 Retrospective Ecological Risk Assessment—The characterization of
 the ecological effects of a stressor after or during its release or

 Risk  Characterization—The   evaluation  of  the  likelihood  that
 adverse ecological effects may occur as a result of exposure to a
 stressor,   including  an evaluation of  the  consequences  of  these

 Statistical Power—Defined  as  1-B where B  is  th&  .probability of
 failing to reject  the  null  hypothesis when  in  fact  the  null
 hypothesis is false.

 Stressor---Any physical,  chemical,  or  biological entity  that can
 induce an adverse response.

 Trophic Level—A functional  classification  of populations within a
 community that is based upon feeding relationships  (e.g., aquatic
 and terrestrial green plants comprise  the  first trophic level and
 herbivores comprise the second).


                     DRAFT: 00 NOT CITE, QUOTE, OR DISTRIBUTE
Stressor-Resnonse  Assessment—A quantifiable relationship between
the  amount  or concentration  of  stressor and  the  magnitude of
response  observed in a  test organism or  other  higher ecological
component  (e.g.,  population).

Xenobiotic—A  chemical  or  other stressor  that  does not  occur
naturally  in the environment.   Xenobiotics occur  as a result of
anthropogenic'activities such  as  the application of pesticides and
the discharge  of  industrial chemicals to air, land,  or water.


                     DRAFT:  00 NOT CITE, QUOTE, OR DISTRIBUTE

 Auer, C.M.; Nabholz, J.V.; and Baetcke,  K.P.  (1990). Mode of action
 and  the assessment  of chemical hazards in the presence of limiting
 data:  use of  structure-activity relationships  (SAR)  under TSCA,
 Section 5.  Environmental  Health Perspectives (87):  183-197.

 Barnthouse, L.W.;  DeAngelis, A.L.; Gardner,  R.H.; O'Neill,  R.v.;
 Powers, C.D.;  Suter, G.W.; Vaughan, D.S.  (1982).  Methodology for
 environmental  risk analysis.    Oak  Ridge  National  Laboratory,
 Publication No. ORNL/TM 8167,  Oak Ridge,  TN.

 Barnthouse, L.W.; Suter, G.W.  II; Bartell,  S.M.; Beauchamp,  J.J.;
 Gardner,  R.H.; Linder,  E. ;  O'Neill,  R.V.;  Rosen,  A.E.   (1986).
 User's  Manual  for  Ecological  Risk  Assessment.    Environmental
 Sciences  Division,  Oak  Ridge National Laboratory,  Publication No.
 2679, ORNL-6251.  Oak Ridge,  TN.

 Barnthouse, L.W.; Suter,  G.W. II; Rosen, A.E.   (1990).   Risks of
 toxic contaminants  to  exploited  fish populations:   influence of
 life  history,   data  uncertainty  and   exploitation  intensity.
 Environmental  Toxicology and Chemistry  9:297-311.

 Bormann,  F.H.   (1985).   Air pollution and  forests:   An ecosystem
 perspective.   BioScience 35(7):434-441

 Broderius,  S.; Kahl,   M.    (1985).    Acute  toxicity of  organic
 chemical mixtures to the fathead minnow.  Aquatic Toxicology 6:307-

 Cairns, J., Jr.  (1990).  Lack of Theoretical Basis for predicting
 rate and pathways of recovery.  In:  Yount, J.D.; Niemi, G, J.,  eds.
Recovery of Lotic Communities and Ecosystems Following Disturbance:
Theory and Application.   Environmental Management 14(5):517-526

Clements,  R.G., Ed.; Johnson,  D.W.; Lipnick, R.L.; Nabholz, J.V. ;
and  Newsome,   L.D.   (1988).  Estimating  toxicity  of  industrial
chemicals   to  aquatic  organisms   using   structure   activity
relationships.  EPA-560-6-88-001..  U.S.   Environmental  Protection
Agency, Washington, DC.  (available  from  NTIS, Springfield,  VA,

Clements,  R.G., ed.  (1988).   Estimating toxicity of industrial
chemicals   to  aquatic  organisms   using   structure   activity
relationships.  U.S. EPA.  Office of Toxic Substances.  Health and
Environmental  Review  Division.    Environmental  Effects  Branch.
Washington, DC.  EPA 560/6-88-001.

                    DRAFT: 00 NOT CITE, QUOTE, OR DISTRIBUTE
Clements, R.G.  (1983). Environmental effects of regulatory concern:
a  position  paper.  Unpublished  report  dated  Dec.   2,  1983.
Washington,  DC:   Environmental   Effects  'Branch,   Health  and
Environmental Review  Division,  Office of Toxic Substances, U.S.
Environmental Protection Agency  20460-0001.

Costanza,  R. ;  Sklar,  F.H. ;  and  White M.L.   '.(1990).   Modeling
coastal landscape dynamics.  BioScience 40:   91-107

Ecological Society  of America  (ESA).    (1991)  (In  Press).   The
Sustainable Biosphere  Initiative:   An Ecological Research Agenda
for the Nineties.  Ecology.

Finkel, A.M.  (1990).  Confronting Uncertainty in Risk Management:  A
Guide  for  Decision-Makers.  Center for Risk Management.  Resources
for the Future. Washington, D.C..

Harwell, M.A.;  and Harwell, C.C.  (1989). Environmental decision
making in the presence of uncertainty.  In: Ecotoxicology: Problems
and Approaches.  S.A. Levin, M.A.  Harwell, J.R.  Kelly, and  K.D.
Kimball, eds. Springer-Verlag. New York,  pp 517-540.

Hill,  A.B.   (1965).   The environment and disease:  Association  or
causation?  Proc. R. Soc. Med.   58:295-300.
Rolling, C.S. (1978). Adaptive Environmental Assessment and Manage-
ment.  John Wiley and Sons. New York,

Kelly,  J.R.; Harwell,  M.A.   (1990).    Indicators of  ecosystem
recovery.  In:  Yount,  J.D.;  Niemi, G.J.,  eds.   Recovery of Lotic
Communities  and Ecosystems  Following  Disturbance:   Theory  and
Application.  Environmental Management 14(5):527-546

Kenaga,  E.E.   (1978).   Test organisms and methods  useful  for the
early  assessment of acute toxicity of chemicals.   Environmental
Science  and Technology 12:1322.-1329

Krebs,   C.J.   (1985).    Ecology:  The  experimental  analysis  of
distribution and abundance.   3rd  ed.   Harper  and  Row,  New York.
800 pp.

Mayer, F.L.  (1990).  Predicting chronic  lethality of chemicals to
fishes from acute  toxicity test  data.    EPA/600/X-90/147.   July

Mayer, F.L.  Jr.; Mayer,  K.S.;  Ellersieck, M.R.  (1986).  Relation
of survival to other endpoints in chronic toxicity texts with fish.
Environmental  Toxicology and Chemistry 5:737-748.


Mckim,  J.M.;  Bradbury,  S.P.;  Niemi,  G.J.    (1987).   Fish  acute
toxicity  syndromes  and their use  in the QSAR approach to hazard
assessment.  Environmental Health  Perspectives  71:171-186.

National  Research Council  (NRG).   (1983).  Risk Assessment in the
federal  government:  managing the  process.    National  Research
Council.  National  Academy Press,  Washington, DC.

O'Neill,  R.V.  (1979).  Natural variability  as  a  source  of  error  in
model predictions.  In: Systems Analysis of Ecosystems.  G.S.  Innis
and  R.V.  O'Neill  (eds.).  International  Cooperative  Publishing
House. Burtonsville, Maryland, pp  23-32.

O'Neill,  R.V. and Gardner, R.H.  (1979).  Sources of uncertainty  in
ecological models.  In: Methodology in Systems Modelling and  Simu-
lation. B.P.  Zeigler, M.S. Elzas, G.J.  Klir, and T.I. Orens (eds.).
North Holland Publishing Company,  pp 447-463.

O'Neill,  R.V.;  Gardner,  R.H.; Barnthouse,  L.W.;  Suter II,  G.W.;
Hildebrand, S.G.; and Gehrs,  C.W. 1982. Ecosystem Risk Analysis:  A
new methodology. Environmental Toxicology and Chemistry 1:167-177.

Odura, E.P.  (1983).   Basic Ecology.   3rd ed.  Saunders College Pub.
Philadelphia. 574 pp.

Odura, E.P., Finn, J.T.; Franz, E.H.   (1979).  Perturbation theory
and the subsidy-stress gradient.   BioScience  29 (6):349-352

Pearlstine, L.; McKellar, H. ; Kitchens, W..  (1985).  Modelling the
impacts of a  River  Diversion on Bottomland Forest  Communities  in
the Santee River Floodplain,  South Carolina.   Ecological Modelling

Peterman, R.-M.   (1990a) . Statistical  power  analysis can  improve
fisheries research  and management. Canadian Journal of Fisheries
and Aguatic Sciences 47(1):2-15.

Peterman, R.M.  (1990b).  The importance of reporting  statistical
power: the forest decline  and acidic  deposition  example.  Ecology

Pianka, E.R.   (1988).  Evolutionary Ecology.  4th ed.   Harper and
Row,  New York.   468 pp.

Poff, N.L.;  Ward, J.V.   (1990).  Physical habitat template of  lotic
systems:    recovery  in  the  context  of  historical  pattern   of
spatiotemporal heterogeneity.  In:  Yount,  J.D.? Niemi,  G.J., eds.


                    DRAFT: 00 MOT CITE. QUOTE, OR DISTRIBUTE
Recovery of Lotic Communities and Ecosystems Following Disturbance:
Theory and Application.  Environmental Management 14(5):629-646
Rapport/. D.J.; Regier, H.A.; Hutchinson, T.C.   (1985).-  Ecosystem
behavior under stress.  The American Naturalist 125(5):617-640

Ricklefs, R.E.   (1990). Ecology.   3rd ed.  W.H. Freeman, New York.
'896 pp.

Rodier,  D.J.    (1990).   -Assessing  risks  to  populations.    In:
Chesapeake  Bay ambient toxicity workshop report.  Chesapeake  Bay
Program.  CBP/TRS 42/90.

Rothraan,. K.J.  (1986).  Modern epidemiology.  1st ed.  Little, Brown
and Company, Boston.   358 pp.

Schindler,   D.W.    (1987).    Detecting  ecosystem  responses  to
anthropogenic stress.  Can.  J. Fish. Aguat. Sci. 44(Suppl. l):6-25.

Sloof,  W.;   van  Oers,  J.A.; de Zwart,  D.    (1936).   Margins  of
uncertainty in  ecotoxicology  hazard  assessment.  Environmental
Toxicology  and Chemistry 5:841-852.

Slooff,  W.;  and  Canton,  J.H.    (1983).    Comparison  of  the
susceptibility of  11 freshwater species to  8 chemical compounds.
II.  (Semi)  chronic tests.   Aquatic Toxicology 4i271-282

Suter,   G.    (1989).    Ecological  endpoints.    In:  U.S.  EPA.
Ecological   Assessments  of Hazardous  Waste Sites:  A  field  and
laboratory  reference document.  Warren-Hicks, W.? Parkhurst, B.R.;
S.S.  Baker,  Jr.  (eds.).  EPA  600/3-89/013.  March 1989.

Suter II,  G.W.;  Rosen, A.E.;  and Linder, E.  (1986).  Analysis of
extrapolation  error.   In:   Barnthouse,  L.W.;  Suter,  G.W.  II;
Bartell, S.M.; Beauchamp, J.J.;  Gardner,  R.H.; Linder,  E.; O'Neill,
R.V.; Rosen, A.E.    (1986).   User's  Manual for Ecological Risk
Assessment.  Environmental Sciences Division,  Oak Ridge National
Laboratory, Publication No. 2679,  ORNL-6251.  Oak Ridge, TN.

Suter II,  G.W.; Rosen,  A.E. ;  Linder,  E.; and  Parkhurst, D. F.
 (1987).     Endpoints   for   responses   of  fish   to   chronic toxic
exposures.   Environmental  Toxicology and Chemistry  6:793-809

Suter, G.W. II.  (1990b). Uncertainty in  environmental risk  assess-
rneht. In:  von  Furstenberg, G.M., ed.  Acting  Under  Uncertainty:
Multidisciplinary Conceptions.  Kluwer  Academic Publishers.  Boston,
MA.  pp 203-230.


                     DRAFT:  00 NOT CITE, QUOTE, OR DISTRIBUTE
Suter,  G.W.  II;  Vaughan,   D.S. ;  Gardner  R.H.    (1983).    Risk
assessment by analysis of extrapolation error: A demonstration 'for
effects  of  pollutants  on  fish.   Environmental  Toxicology  and
Chemistry 2:369-378.

Suter, G.W.; Barnthouse, L.W.;  Breck, J.E.; Gardner, R.H.;  O'Neill,
R.V.  (1985) .  Extrapolating from the laboratory to the field:  How
uncertain are  you?  In:  Cardwell,  R.K.;  Purdy, R. ;  Bahner,  R.D.
eds.  Aquatic Toxicology and Hazard Assessment.  American Society
for Testing and  Materials,  Philadelphia,  PA.   pp.  400-413.

Suter,  G.W.  II.   (1990).    Uncertainty in  environmental  risk
assessment. •  In:  von  Furstenfaerg,   G.M.,   ed.     Acting  under
uncertainty:  Multidisciplinary  conceptions.     Kluwer.  Academic
Publishers, Boston,  pp  203-230.

Suter II, G.W.    (1990).   Endpoints for regional ecological  risk
assessments.  Environmental Management  14:9-23

Turner, R.E.  (1977).  Intertidal Vegetation and Commercial Yields
of  Penaeid  Shrimp.    Transactions  American  Fisheries   Society

Tversky, A., and D..Kahneman. (1974).  Judgement under uncertainty:
heuristics and biases.  Science 185:1124-1131.

U.S. EPA.   (1982).  Surrogate species workshop.  Office  of Toxic
Substances.  Health and Environmental Review Division.  Ecological
Effects Branch.   Washington, DC.

U.S. DOI.   (1987).  Injury to fish and wildlife  species.   Type  B
Technical information document.   CERCLA 301  Project.   Washington,
U.S.  Environmental   Protection  Agency   (U.S.  EPA).     (1980).
Guidelines for Deriving Water Quality Criteria for the Protection
of Aguatic Life and  its Uses.   Federal Register 45:   79341-79347

U.S.   Environmental   Protection   Agency   (U.S.   EPA).   (1988).
Probabilistic Dilution Model  3 (PDM3).  Draft Report dated May 31,
1988.   Washington,   DC:   Office   of  Toxic  Substances,   U.S.
Environmental Protection Agency.  Contract No. 68-02-4254.
(1991).   Guidelines  for exposure  assessment.   Draft


                    DRAFT: 00 MOT CITE, QUOTE, OR DISTRIBUTE
U.S.  Environmental  Protection  Agency   (U.S.   EPA).     (199fla)..
Reducing Risk:  Setting Priorities and Strategies for Environmental
Protection.  Science Advisory Board SAB-EC-90-021.

U.S. Environmental Protection Agency (U.S. EPA).  (1936).  Standard
Evaluation Procedure for Ecological  Risk Assessment,  by D.J.  Urban
and  N.J.  Cook, Hazard  Evaluation Division,  Office of  Pesticide

U.S.  Environmental  Protection  Agency   (U.S.   EPA).     (1990b).
Environmental  Monitoring  and  Assessment  Program.    Ecological
Indicators.  Office of Resesarch and Development.  EPA/600/3-90/060

U.S.  Environmental  Protection  Agency   (U.S.   EPA).     (1990c).
Technical Support  Document  for Water  Quality-based Toxics Control
Draft April 23, 1990.  Office of Water.   Washington, DC

U S  EPA.  (1990).  Draft.  Ecological Risk Assessment: A state [of-
the  Practice.  Office of Pesticides and  Toxic Substances.  Office
of  Toxic Substances.  Health and Environmental Review Division.
Ecological Effects Branch.   Washington,  DC.

van   Straalen,  M.M.;   Denneman,   C.A.      (1989)   (in  press).
Ecotoxicological   evaluation  of soil quality  criteria.   Ecoto.
Environ.  Safety.



    Framework Document Workshop

     Pre-Meetinq Issues Papers
  Ecorisk Paradigm
  Conceptual Framework Development
*  Hazard Identification and Stress-
  Response Assessment
  Exposure Assessment
  Risk Characterization

                      Pre-Meeting Issue Paper

      The proposed  paradigm  for  ecorisk  assessment  is modeled
      after  the National Research  Council paradigm  for human
      health risk  (NRC, 1983).   Is the modified paradigm presented
      in the framework document  appropriate  for ecorisk
      assessments/  or is another approach preferable?


1.    The proposed  ecorisk paradigm is quite similar to the NRC
health effects paradigm, but it also explicitly provides for an
initial planning process in ecorisk assessments.

la*   Comment on whether the elements of the paradigm proposed in
      the framework document are appropriate.  In what situations,
      if any/ would another paradigm be more appropriate?

Ib.   It has  been suggested that the conceptual framework
      development step be included in an expanded hazard
      identification section.  Please comment on this suggestion.

Ic.   Comment on the need to include a monitoring step (following
      risk characterization) in  the paradigm.

2.    The framework document uses  terms such as "risk assessment"
and "hazard"assessment" that either have no standard meaning or
are used inconsistently in the  ecotoxicology literature.

2a.   Comment on whether terminology related to the paradigm
      (including "hazard assessment1* and "risk assessment")  is
      clearly defined and used consistently  throughout the
      document.   Suggest alternate terms or definitions, if

2b.   Comment on whether the definitions used in the glossary are
      appropriate and recommend  any additional terms that should
      be added to the glossary.

3.    The conceptual framework development section _(and the
framework document as a whole)  emphasizes individual- or
population-level effects.   While this may reflect the present
orientation  of the Agency and the  state of  the science,  some
scientists feel that there should  be more emphasis on community-
and ecosystem-level measurements and effects.

3.   What should be added or changed in the framework document to
     place appropriate emphasis on community- and ecosystem-level

4.   The proposed ecorisk paradigm maintains a distinction
between the process of risk assessment (scientific analysis) and
risk management (which may include social, political, legal, and
economic/valuation factors) .   This distinction is also .made in
human health risk assessments.  However, there has been
considerable discussion as to the degree that policy or risk
management issues should influence the objectives that are
established at the outset of an ecorisk assessment.  Thus, the
relationship between ecorisk assessment and management may need
additional clarification in the framework document.

4a.  comment on the roles of risk management in the initial
     stages of an ecorisk assessment  (e.g. establishment of
     assessment goals and endpoint selection) .

4b.  What further explanations are needed, if any, of the role of
     policy considerations in the ecorisk process?

4c.  Comment on the role of valuation considerations, in any, in
     ecorisk assessments.  (Valuation concerns the assignment of
     monetary or societal values to ecological resources).

5.   The framework document distinguishes between those areas of
ecorisk assessment that are currently used and those areas that
are under development or show promise.

5.   Comment on whether the distinctions between the present
     s tat e-of- the- science and future research needs in ecorisk
     assessment are clearly and appropriately made.

6.   Some persons have commented that more examples should be
used throughout the document to illustrate points made.
6 .    Where are additional examples needed?
examples that could be used?
                                            Can, you suggest
7.   In the introduction, issues for consideration and future
development are listed.

7.   What issues, if any, should be added to the list?

                     Pfe-Meeting Issue Paper



     Conceptual framework development is proposed as the first
     step in an ecorisk assessment.  Does this section provide an
     appropriate description of the process/ and are relevant
     ecorisk issues adequately discussed?


8.   The proposed criteria for endpoint selection are (1) purpose
and needs of the assessment, (2) ecological relevance, (3)
susceptiblity, and (4) practical constraints.

8.   Which criteria, if any, should be added (such as type of
     stressor) or deleted?

9.   The framework document distinguishes between "assessment
endpoints" (the characteristics of an ecosystem that are the
focus of the risk assessment) and "measurement endpoints" (the
effects that are actually estimated or measured).

9a.  What changes/ if any/ do you recommend in the terms
     "measurement endpoint'* and "assessment endpoint"?

9b.  What changes/ if any/ do you recommend in the examples shown
     in Table 2?

10.  The conceptual framework development section concludes with
the presentation of a conceptual model that is further evaluated
in the exposure and hazard assessment sections.

10,  Comment on whether the development of a conceptual model is
     a useful approach for focusing an ecorisk assessment.   How
     should a conceptual model be used?

11.The importance of spatial and temporal scaling factors is
mentioned briefly in the conceptual framework development section
(2.1.1 - Exposure Settings and Pathways).

11.  What additional consideration, if any, should be given to
     spatial and temporal scales?

12.  Some decisions are based only upon preliminary information,
similar to that assembled during Conceptual Framework
Development.  Examples include the decision to consult with the
U.S. Department of the Interior on endangered species and the
decision to implement an emergency removal.

12.  Should the framework document acknowledge that Conceptual
     Framework Development can be used in this fashion?


                     Pr-e-Meeting Issue Paper



     This section includes discussions,of the establishment of
     the causal relationship between stressor and response
     (hazard identification) and the quantification of that
     relationship (stress-response).  Are the emphasis and level
     of detail provided for these topics appropriate?


13.  The framework document distinguishes between hazard
identification and stress-response  assessment, as is commonly
done in health risk assessments.  Hazard identification, which
establishes a cause-effect relationship between a stressor and a
response, references Hill's criteria.

%13a. Comment on the necessity of distinguishing between hazard
     identification and stress-response assessment.

13b. Discuss the usefulness of Hill's criteria for evaluating
     cause-effect relationships in  ecorisk assessments.

14.  The stress-response section describes both extrapolation
techniques as well as the uncertainty associated with the
extrapolation process.  Extrapolation may occur at several
levels:  across endpoints at a given level of biological
organization, across levels of biological organization, from the
results of laboratory studies to the prediction of effects in the
field, and across temporal and spatial scales.

14a. Are there other currently accepted extrapolation methods
     that should be added?

14b. Which portions of tha extrapolation or uncertainty
     discussions should be condensed or expanded?

3,4c. Where should uncertainty be discussed, in this section or
     under risk characterization?   why?

15.  While the framework document recognizes the significance of
non-chemical stressors such as habitat alteration, sedimentation,
and nutrient loading, there is only minimal guidance provided on
these topics as compared with threats associated with toxic
chemicals.  In addition, there is little information on how to
deal with multiple stressors (whether chemical or non-chemical)
or cumulative effects.

15a. How should the discussion of non-chemical stressors bo

ISb. What additional guidance should be presented eoneeraiag the
     risks of multiple stressors and cumulative effects?

                     Pre-Meeting Issue Paper

     Does ts» discussion in this section adequately eov@r th®
     major aspects of ecological exposure assessment?


16.  Exposure assessment is defined as the evaluation of the
temporal and spatial distribution of a stressor and its co-
•occurence with ecological components.  The framework document
highlights several aspects of exposure assessment, including the
evaluation of multiple kinds of organisms, the importance of the
timing of exposure, and the perception of and direct contact with
a stress.
16.  Are these considerations appropriate?
     information should be highlighted?
What additional
17.  Two methods of exposure assessment used in human health
assessment were applied to ecorisk exposure assessment.
17.  Comment oa ta® usefulness ©f ta®s® two mataods.  Waat other
     methods, if any, should be added?

18.  The framework document briefly discusses variables that may
modify or complicate exposure, assessments, such as the influence
of species-specific habitat and behavioral patterns,
bioavailability,  multiple stressors, time-varying exposures, and
multiple exposure routes.

18.  &r@ these topics adequately addressed ia view of the
     objective* and scope of th@ framework documeat?  If not,
     what would you expand and now?

19.  Methods for assessing uncertainty in exposure assessment are
deferred to chapter 5 of the framework document.

IS.  What additional discussions of uncertainty ia exposure
     assessment should be included ia this section or slsewhere,
     if any?

                      Pre-Meeting Issue Paper

      The risJc characterization process  is  defined to  include
      determination of the likelihood of adverse effects,
      evaluating the consequences  of the adverse effects,
      assessing uncertainties,  and communicating the results of
      the risk assessment.   Does this provide  a complete picture
      of  risk characterization  as  it is  applied to ecorisk


20.   The primary approach proposed in the  framework document for
evaluating the likelihood of adverse effects  is the quotient
method,  which provides a  point estimate or a  series of point
estimates of the ratio between exposure and effect levels.  Other
methods  may  be available  that  provide a more  complete picture of
risk  by  using a broader range, of  the stress-response curve or by
considering  a variety of  different exposure scenarios.

20&.*  Comment on the clarity and usefulness of the discussion of
      the quotient and other methods.

2Qb.  what are the alternatives to  the quotient method of risk
      characterization (particularly for higher levels of
      ecological organization) ,  and how  might  they b« incorporated
      into the framework document?

21.   Risk assessments generally provide an estimate of the
probability  of  an adverse effect.   The  concept  of  ecorisk
assessment in the framework document has been broadened to
include more  qualitative estimates  of risk, such as might be used
for comparative risk assessments or risk ranking exercises.

21.  Discuss whether ecorisk assessments should include
     qualitative as well as quantitative evaluations of risk.

22.  The  section on uncertainty (5.3) describes general sources
of uncertainty  and discusses reducing and presenting uncertainty.
22a. Is th» discussion of uncertainty clear?
     should be changed?
What, if anything,
22b. What, if anything, should be added to the discussion of
     weight of evidence?

23.  Ecosystem recovery  is only mentioned briefly in the
framework document  (section  5.4 -  Describing the Consequences of
Identified Risks).

23a. Should more information on ecosystem recovery be included in
     this document?  If  so,  specifically what should be added?

23b. should ecosystem recovery b*  discussed in any other

24.  The consequences of effects  (e.g.  indirect effects/ erfacts
at higher levels of organization,  and effects at other spatial
and temporal scales) are discussed in risk characterization.

24.  Is the discussion of ecological  consequences appropriately
     located?  Where else in the document should it be included?