United States
Environmental Protection
Agency
Office of Research and
Development
Washington DC 20460
EPA/630/R-96/002
September 1996
Peer Review Workshop
Report on Draft Proposed
Guidelines for Ecological
Risk Assessment
RISK ASSESSMENT FORUM
-------
EPA/630/R-96/002
September 1996
PEER REVIEW WORKSHOP REPORT ON
DRAFT PROPOSED GUIDELINES FOR ECOLOGICAL RISK ASSESSMENT
Prepared for:
U.S. Environmental Protection Agency
Risk Assessment Forum
401 M Street, SW.
Washington, DC 20460
Contract No. 68-D5-0028
Work Assignment No. 95-2
Prepared by:
Eastern Research Group, Inc.
110 Hartwell Avenue
Lexington, MA 02173-3198.
February 13,1996
Printed on Recycled Paper
-------
NOTICE
Mention of trade names or commercial products does not constitute endorsement or
recommendation for use. Statements are the individual views of each workshop participant; the
statements in this report do not represent analyses or positions of the Risk Assessment Forum or
the U.S. Environmental Protection Agency (EPA).
This report was prepared by Eastern Research Group, Inc. (ERG), an EPA contractor, as
a general record of discussions held during the Workshop on the Draft Ecological Risk ^
Assessment Guidelines. As requested by EPA, this report captures the main points and
highlights of the meeting. It is not a complete record of all details discussed, nor does it
embellish, interpret, or enlarge upon matters that were incomplete or unclear.
-------
CONTENTS
Foreword v
EXECUTIVE SUMMARY ..... ES-1
SECTION ONE—INTRODUCTION . 1-1
Welcome 1-1
Introductory Remarks by AA/ORD Robert Huggett 1-1
Workshop Objectives and Format .. 1-2
SECTION TWO—CHAIRPERSON'S SUMMARY OF THE WORKSHOP 2-1
SECTION THREE—WORK GROUP SESSIONS .3-1
Problem Formulation 3-1
Analysis 3-4
Risk Characterization 3-12
SECTION FOUR—OBSERVER COMMENTS 4-1
Day 1 Observer Comments 4-1
Day 2 Observer Comments 4-4
SECTION FIVE—NEXT STEPS 5-1
Where Does EPA Go From Here? 5-1
Who Is the Audience for the Guidelines? 5-2
What Are EPA's Goals in Terms of Guidance? 5-2
What Is the Relationship Between These Guidelines and
Existing Program Guidance or Program Guidance That
Might Be Developed in the Future? 5-4
Discussion 5-4
APPENDIX A—REVIEWER LIST A-l
APPENDIX B—CHARGE TO REVIEWERS B-l
iii
-------
APPENDIX C—WORKSHOP AGENDA C-l
APPENDIX D—PREMEEWNG COMMENTS D-l
APPENDIX E—REVIEWER WORK GROUP ASSIGNMENTS E-l
APPENDIX F—FINAL OBSERVER LIST , F-l
APPENDIX G—WORKSHOP CHAIRPERSON'S OPINION SURVEY G-l
IV
-------
Foreword
EPA's work to develop a set of Agencywide guidelines for ecological risk assessment
started in 1989, when the Agency's Risk Assessment Forum began holding a series of colloquia
to identify and discuss significant issues in ecological risk assessment. Based on these discussions
and on consultation with EPA's Science Advisory Board (SAB), the Risk Assessment Forum
decided to pursue a stepwise process for developing ecological risk assessment guidelines. This
process has involved development of:
• Framework for Ecological Risk Assessment, a widely used report that proposed
principles and terminology for the ecological risk assessment process.
• Case studies illustrating ecological risk assessment approaches.
• Issue papers highlighting important principles and approaches that EPA scientists
should consider in preparing ecological risk assessment guidelines.
• The Draft Proposed Guidelines for Ecological Risk Assessment currently being
reviewed.
The guidelines development process has emphasized peer review and consensus-building.
Indeed, many experts from academia, industry, consulting firms, all EPA program offices and
regions, and state and other federal agencies have participated in the development and review of
the Draft Guidelines and its predecessor documents.
To continue this emphasis on peer review and consensus-building, EPA convened a 2-day
workshop to discuss and peer review the Draft Proposed Guidelines for Ecological Risk
Assessment. The workshop, held December 6 to December 7,1995, in Washington, DC, brought
together 25 peer reviewers and more than 50 observers from academia, industry, state agencies,
EPA, and other U.S. and Canadian agencies. Workshop participants discussed the main topics in
the Draft Guidelines in three series of introductory plenary sessions, work group sessions, and
summary plenary sessions. Dr. William Smith of Yale University chaired the workshop, while
Dr. Dwayne Moore of Environment Canada and Dr. Richard Kimerle of the Monsanto Company
led the work group sessions.
This report highlights the main comments and recommendations arising from workshop
discussions. Based on the highly constructive and useful suggestions outlined in this report, EPA
will give serious consideration to workshop participants' opinions and revise the Draft Guidelines
accordingly. EPA will then seek both internal and interagency review of the Guidelines and
revise the document again. After submitting the revised document as proposed guidelines to the
Federal Register for public review and comment and to the SAB for review, EPA will publish final
guidelines. As the report that follows indicates, the workshop was a very productive effort that
will help EPA move forward in this ambitious development process.
William Wood, Ph.D.
Executive Director
Risk Assessment Forum
-------
EXECUTIVE SUMMARY
William Smith, Ph.D.
Workshop Chairperson
Professor of Forest Biology
Greeley Laboratory, Yale University
The EPA Workshop on the Draft Proposed Guidelines for Ecological Risk Assessment,
held December 6 to December 7,1995, in Washington, DC, brought together 25 experts charged
with generating a clear set of comments and recommendations for EPA to use in revising the
Draft Guidelines. The workshop also attracted more than 50 observers, who offered their
perspectives on the Draft Guidelines as well. Workshop participants generally agreed that the
Draft Guidelines represent an appropriate evolution toward a uniform strategy for conducting
ecological risk assessments—one that can be used not only by EPA personnel, but also by a
broader array of natural resource and environmental managers. Thus, the Draft Guidelines
represent an important step toward improving the scientific soundness, consistency,
comparability, and completeness of ecological risk assessments conducted in the United States
and elsewhere.
Workshop participants also agreed that the overall framework for and main elements of
ecological risk assessment proposed in the Draft Guidelines are sound. In their current form,
the Draft Guidelines are not (and should not be) a comprehensive technical manual. Rather, the
document promotes and guides selection of appropriate risk assessment techniques for specific
situations—including situations involving biological or physical stressors rather than chemical
stressors. Considering information on biological and physical stressors is appropriate given
increasing recognition of the relevance of these stressor types to ecological risk assessment. In
fact, reviewers recommend expanding the treatment of nonchemical stressors.
Opportunities to improve the Draft Guidelines include:
• Expanding the treatment of nonchemical stressors (as noted above).
• Emphasizing the importance of multiple stressors.
ES-1
-------
• Adding more material on the utility of a tiered approach to risk assessment.
» Presenting hypothetical or actual case studies that run through the entire
document (i.e., to illustrate important points in each section of the document).
The Draft Guidelines reviewers encourage EPA to move forward with the document as
quickly as is reasonable after considering the specific recommendations highlighted in sections 2
and 3 of this report.
ES-2
-------
SECTION ONE
INTRODUCTION
WELCOME
Dr. William Wood, Executive Director of EPA's Risk Assessment Forum, opened the
workshop by welcoming the guideline reviewers and workshop observers. He then introduced
Dr. Robert Huggett, Assistant Administrator (AA) of EPA's Office of Research and
Development (ORD), noting that Dr. Huggett is a longtime supporter of the Forum's efforts to
develop guidelines on ecological risk assessment.
INTRODUCTORY REMARKS BY AA/ORD ROBERT HUGGETT
Dr. Huggett began his remarks by noting the importance of the workshop to review
EPA's Draft Proposed Guidelines for Ecological Risk Assessment. In the past, he said, EPA has
spent most of its labor and financial resources on human health research and human health risk
assessment; the ecological risk assessment guidelines development effort is an important part of
EPA's move toward improving the balance between human health work and ecological work. In
addition, several pending regulatory reform bills would require the federal government to
conduct a risk assessment and a cost-benefit analysis for any mandate or action generating costs
of more than $100 million. Given that EPA will likely have to rely increasingly on risk
assessments to fulfill legislative requirements such as these, this review of the Draft Guidelines is
important to advancing EPA's ability to conduct sound ecological risk assessments. Dr. Huggett
emphasized that EPA will take reviewers' comments very seriously in revising the Draft
Guidelines.
Dr. Huggett went on to state that the ecological risk assessment guidelines development
effort also fits in with ORD's new strategic planning process, which is organized around the risk
assessment paradigm to ensure that EPA focuses its research on areas that will most reduce
uncertainties in risk assessments. Noting that ORD's strategic plan is available to the public, he
1-1
-------
invited workshop participants to comment on the document. A sound strategic planning process
is especially important now, he said, because EPA's fiscal year (FY)1997 budget will-.be
considerably less than the FY 1996 budget. ORD will use the new planning process to prioritize
EPA research efforts. Despite budget cuts, EPA hopes to increase resources devoted to the
extramural grants program—from $44 million ia FY 1995 to $85 million in FY 1996 to $100
million in FY 1997. So far, the response to EPA's advertisements for new fellowships has been
tremendous. . , •
In concluding his remarks, Dr. Huggett thanked the guideline reviewers and workshop
observers for their participation in an effort that is very important to EPA Reiterating that EPA
will consider comments on the Draft Guidelines very seriously, he urged the reviewers and
observers to be as candid as possible.
WORKSHOP OBJECTIVES AND FORMAT
Workshop chairperson William Smith thanked Dr. Huggett for his remarks and the
energy that Dr. Huggett has brought to the Agency. He the,n introduced the reviewers of the
Draft Guidelines (see appendix A), described the charge to reviewers (see appendix B), and
reviewed the workshop agenda (see appendix C). Dr. Smith explained that the primary objective
of the workshop was to generate a clear set of comments and recommendations that EPA can
use to revise the Draft Guidelines. The three main topics in the Draft Guidelines (problem
formulation, analysis, risk characterization) would be discussed separately, he said, in a series of
plenary sessions and work group sessions on each topic:
• An introductory plenary session with presentations summarizing key issues from
the perspective of the authors of the Draft Guidelines and common themes seen
in the reviewers' premeeting comments (reproduced hi appendix D).
» Concurrent work group sessions in which reviewers, divided into two groups (see
appendix E), would discuss their comments and recommendations.
• A followup plenary session in which the work group leaders, facilitated by the
workshop chairperson, would summarize the comments and recommendations of
their groups.
1-2
-------
Dr. Smith told observers (listed in appendix F) that they could offer comments during the
work group sessions at the discretion of the work group leaders and that they also would have an
opportunity to speak during plenary sessions at the end of each workshop day.
As a prelude to the plenary and work group sessions on individual Draft Guidelines
sections, Dr. Smith offered his own perspective on the document as a whole. He said that the
Draft Guidelines have the potential to advance the goal of a single strategy for ecological risk
assessment, and that such a strategy would in turn enhance the consistency, comparability, and
completeness of these assessments. To achieve these ends, the Draft Guidelines do not
necessarily need to address all types of stressors in detail; rather, he suggested, they must provide
a coherent framework and clear guidance for at least some major stressors. In addition, the
Draft Guidelines should clearly state the importance of being very explicit about the nature and
scope of individual ecological risk assessments (e.g., their boundedness over time and spacu, the
ecological level that they address, the societal values that they reflect). Moreover, because
ecological risk assessments and stressor/ecological resource monitoring are intimately linked, the
Draft Guidelines should emphasize the importance of effective monitoring.
Dr. Smith's remarks sparked a lively discussion of issues that were subsequently explored
in greater detail in the work group sessions. The next section of this report summarizes
overarching comments and recommendations that arose from these sessions, while section 3
summarizes comments and recommendations related to specific Draft Guidelines sections and
section 4 highlights the comments of workshop observers. Section 5 of this report discusses how
EPA plans to follow up on this workshop.
1-3
-------
-------
SECTION TWO
CHAIRPERSON'S SUMMARY OF THE WORKSHOP
William Smith, Ph.D.
Professor of Forest Biology
Greeley Laboratory, Yale University
Throughout the 2-day Workshop on the Draft Proposed Guidelines for Ecological Risk
Assessment, guideline reviewers and workshop observers representing many different disciplines
exhibited tremendous interest in the Draft Guidelines and commended EPA for its efforts in
developing the document. Indeed, although reviewers offered numerous suggestions for
improvement, workshop participants overall viewed the Draft Guidelines as a very useful
extension of the Framework Report and an appropriate next step in the evolution of specific
recommendations for conducting ecological risk assessments. The development of standard
guidelines represents an opportunity to achieve greater scientific soundness, consistency,
comparability, and completeness in ecological risk assessments. Directing resources to
"technology transfer" of the Guidelines would represent a wise investment by the Agency.
Overarching themes and recommendations that emerged from workshop discussions and
an informal opinion survey conducted at the end of the workshop (see appendix G) include the
following:
• The Draft Guidelines should not be a comprehensive technical manual. They
should, however, express preferences and/or provide criteria for choosing among
alternative techniques.
RECOMMENDATION 1. Place greater emphasis on selection criteria in the next
draft of the Guidelines by referencing seminal primary literature references, by
referencing conclusions of previously developed Issue Papers, and by using case
studies to highlight selection criteria for specific situations.
» Recognition of the importance of biological and physical stressors is increasing.
As a result, it is appropriate to use the Draft Guidelines as an opportunity to
present information about biological and physical as well as chemical stressors.
2-1
-------
• The present practice of risk assessment follows a phased approach that begins -
with broad initial screening and continues with more refined analyses. As a result,
it is appropriate to expand the discussion of the tiered approach to risk
assessment in the Draft Guidelines. Such a discussion could provide guidance on
deciding when data are sufficient to move to a next step, describe how the costs of
Type n errors are used, in tier decisions, and provide illustrative case studies.
RECOMMENDATION 2 Expand the discussion of the tiered strategy in the next
draft of the Guidelines by outlining the logic for progressing from one step to the next
and by illustrating this logic with case studies.
» Ecological resource exposures to stress typically involve concurrent or sequential
exposure to more than one stress factor. As a result, single-stress risk assessments
rarely reflect actual exposure conditions. This topic is sufficiently important to
justify a more detailed evaluation.
RECOMMENDATIONS. Develop an Issue Paper on multiple stressor interactions.
Rather than delaying publication of the Guidelines while the Issue Paper is being
developed, add information from the Issue Paper at a later time.
• Hypothetical case studies that are consistently continued through all sections of
the Draft Guidelines would represent a very effective training and technology
transfer tool.
RECOMMENDATION 4. Consider developing one or more case studies that are
carried through the Problem Formulation, Analysis, and Risk Characterization
sections of the document.
Reviewers found no major fault with the general organization of the Draft Guidelines,
nor with the main themes in the sections on the principal elements of the risk assessment
process. Specific recommendations for improving the Problem Formulation, Analysis, and Risk
Characterization sections of the Draft Guidelines are summarized in the next section of this
report.
2-2
-------
SECTION THREE
WORK GROUP SESSIONS
During the workshop, reviewers divided into two work groups, which separately reviewed
the three major sections of the Draft Guidelines. During plenary sessions, the work group
leaders presented the major comments and recommendations of their work groups. Following
the workshop, the workshop chair and work group leaders integrated the comments and
recommendations of the two work groups into a single summary on each major section reviewed.
These summaries are presented below.
PROBLEM FORMULATION
William Smith, Ph.D.
Professor of Forest Biology
Greeley Laboratory, Yale University
Problem formulation represents the first major element of the ecological risk assessment
process. More important, it is the foundation that supports and bounds the entire process.
Deficiencies at this stage will permeate the entire assessment and compromise the full process
and final product. The most critical steps of this phase include:
• Clear articulation of the question that initiates the risk assessment
• Identification and engagement of relevant parties to the assessment
• Identification of source/stressor characteristics
• Identification of ecological resource(s) at risk
• Identification of ecological effects
• Identification of assessment endpoints
» Development of a conceptual diagram
• Development of an analysis plan
3-1
-------
Clarity of Original Question
The clarity of the original question leading to the risk assessment process is critical.
Recognition of the importance of the initial question led the reviewers to make the following
recommendation:
RECOMMENDATION 5. For site-specific assessments, the initial question should provide
temporal, spatial, biological hierarchy, and human value boundaries for the assessment.
Identification and Engagement of Relevant Parties
Identifying parties relevant to the risk assessment, and involving these parties at the
beginning of and throughout the assessment, is extremely important. Social science techniques
for "extracting societal values" and identifying relevant stakeholders are well established. Legal
statutes, political considerations, historical knowledge, and self-identification (e.g., via the Federal
Register) are additional techniques for identifying relevant parties. Stakeholder "maps" should be
carefully developed. Including too many parties can slow and complicate the process, while
excluding relevant parties can undermine chances for acceptance of the assessment strategy.
RECOMMENDATION 6. Through examples, the Draft Guidelines should identify methods
for identifying stakeholders as well as the advantages and disadvantages of a smaller or larger
stakeholder map.
In most cases, it will be appropriate to think of the risk assessors and risk managers in
EPA's framework diagram as teams rather than individuals.
RECOMMENDATION 7. At the outset, the Draft Guidelines should very clearly articulate
the roles of assessors, managers, reviewers, and other relevant parties.
Identification of Assessment Type/Purpose
Labeling risk assessments as stressor-, ecological effects-, or societal value-initiated is of
limited usefulness. This classification contributes little of substance to the assessment and adds
3-2
-------
to the proliferation of noncritical technical distinctions. Organizing assessments by stressor type
(e.g., chemical, biological, physical) or as prospective or retrospective assessments appears to
have greater merit.
RECOMMENDATION 8. Eliminate or de-emphasize the origination designations in the next
draft of the Guidelines. Consider relying more heavily on case studies to illustrate the
differences between different types of assessments.
Risk assessments can be useful tools for hypothesis testing, but not all assessments involve
hypothesis testing.
RECOMMENDATION 9. Clarify the distinction between risk assessments for science
(hypothesis testing) and those for environmental management (decision-making). To avoid
confusion about what is meant by "hypothesis testing," consider adding a text box on the
difference between "risk hypotheses" in risk assessment conceptual models and statistical mitt
hypothesis testing.
Data Decisions
Development of a conceptual diagram should include a formalized procedure to guide
determination of how much data are required, elucidation and application of data "decision
rules," and data selection. This procedure should provide the opportunity for risk assessors, risk
managers, and other relevant parties to achieve consensus on data quality objectives (DQOs).
RECOMMENDATION 10. Through examples in the Problem Formulation section of the
Draft Guidelines, illustrate processes for defining DQOs.
More formal adoption of a tiered approach to risk assessment also will facilitate decision-
making on the quality and quantity of data needed in particular assessments.
3-3
-------
Introductory Material on Cross-Cutting Issues
Uncertainty is a cross-cutting issue that is fundamentally important to every step in the
risk assessment process. As a result, EPA should introduce this topic early in the Draft
Guidelines, covering the main types of uncertainty that are relevant throughout the risk
assessment process and providing an expanded treatment of uncertainty specifically associated
with laboratory and field studies.
RECOMMENDATION 11. Add an expanded introduction to uncertainty in the Problem
Formulation section of the Draft Guidelines.
Other cross-cutting topics that pervade the entire risk assessment process also should be
introduced early in the Draft Guidelines.
RECOMMENDATION 12 Consider adding specific introductory material on multiple
stressors, interactive and cumulative effects, and selection criteria for surrogates to the
Problem Formulation section of the Draft Guidelines.
Written material in the reviewers' premeeting comments might provide a useful starting
point for addressing some of these issues.
ANALYSIS
Dwayne Moore, Ph.D.
Senior Evaluator
Commercial Chemicals Evaluation Branch, Environment Canada
The Analysis section of the Draft Guidelines is impressive in covering a wide range of
stressor types and in raising key issues that affect the analysis of each. Covering such a broad
range of topics compromises the section's effectiveness, however, when it prevents in-depth
discussion of complex issues. This occurs in the subsections on physical, biological, and multiple
stressors. The subsection on chemical stressors is more complete, presumably because of EPA's
experience in this area.
3-4
-------
On the whole, the Analysis section is well organized, although some reviewers questioned
the need to discuss uncertainty throughout the document and several reviewers suggested adding
or deleting various text boxes and figures.
The most fundamental criticism of the Analysis section is that it provides "precious few
guidelines" (see, e.g., John Bascietto's written comments). Although the document does an
admirable job of raising complex issues (e.g., uncertainty, secondary effects, multiple stressors),
many of which have been ignored by other jurisdictions, it rarely provides guidelines or
recommendations on preferred approaches and methods—or even criteria for choosing among
available approaches and methods. Unless EPA takes this further (if difficult) step, program-
specific guidelines developed in the future will not be consistent with each other, nor will they
reflect the state of the art. Specific suggestions are provided below.
Organization
The Analysis section would be more useful if it were organized around the components of
ecological risk assessment rather than around the various types of stressors. With the new
organization, the section could explain how analysis fits into each component of ecological risk
assessment and provide examples, while still noting how analysis differs depending on the types
of stressors (biological, physical, chemical) involved.
RECOMMENDATION 13. Start the Analysis section by describing the aim and intent of the
analysis phase, focusing on the components of the risk assessment process rather than on
stressors. Include a figure and perhaps a table showing types of exposure and effects relevant
to increasing levels of biological organization. Also include a hypothetical deterministic
example that at some point in the section includes the issue of uncertainty; the example could
be carried through subsequent sections of the document. Later in the Analysis section, using
illustrative examples where possible, note aspects of the analysis phase that are unique to each
major type of stressor.
3-5
-------
Guidance
The Analysis section of the Draft Guidelines rarely provides specific guidance on complex
issues.
RECOMMENDATION 14. Provide criteria for choosing between methods and approaches
and, where possible, note preferred methods and approaches and conditions for their use.
Also note that preferences sometimes depend on the level of sophistication (tier) of the
ecological risk assessment.
Hierarchies
Multispecies testing is insufficient to analyze community- or ecosystem-level effects.
Considering a large area or scale does not necessarily mean that ecosystem or landscape issues
are being addressed.
RECOMMENDATION 15. Discuss more fitlfy what is meant by "community-lever and
describe attributes applicable of the ecosystem and landscape levels of organization.
Dose-Response Relationships
In contrast to the Draft Guidelines as a whole, which assume an elementary level of
expertise, the level of treatment of dose-response relationships varies from one subsection to
another in the Analysis section; information is elementary in some places and much more
detailed in others. If the document is intended to be a guidelines document rather than a
"Framework IP document, EPA should provide more information on dose-response relationships
related to nonchemical stressors (e.g», the introduction of exotic species or removal of a
disturbance to which a system is adapted) to raise these subsections to the level of other sections.
RECOMMENDATION 16. Improve the consistency with which dose-response relationships
(and other topics) are addressed in the various subsections of the Analysis section of the
Draft Guidelines.
3-6
-------
Natural Disturbances
Ecological risk assessments encompass both anthropogenic and "natural" disturbances, but
the Analysis section does not address both with equal clarity. Background disturbance regimes
and the effects of physical stressors relative to these regimes need to be discussed much more
clearly and explicitly. The intensity, frequency, and magnitude of disturbances need to be clearly
delineated as factors to consider in an ecological risk assessment. A short separate
section—possibly written with the help of experts outside the Agency—would be useful.
RECOMMENDATION 17. Provide a dear, explicit treatment of background disturbance
regimes, noting the effects of physical stressors. Discuss the intensity, frequency, and
magnitude of natural (and possibly chemical) disturbances against which to measure human
effects. Explicitly comment on whether disturbances smaller than background are of concern.
Biological and Physical Stressors
The sections on biological and physical stressors are somewhat superficial compared to
the section on chemical stressors.
RECOMMENDATION 18. Expand the discussion of biological and physical stressors,
relying in part on the Issue Papers by Simberloff and Alexander and by Sheehan and Loucks.
RECOMMENDATION 19. Focus on biological and physical stressors over which EPA has
jurisdiction (Le., those with which EPA has experience).
Multiple Stressors
A fuller discussion of multiple stressors would be useful.
RECOMMENDATION 20. Divide the section on multiple stressors into subsections on
chemical mixtures, habitat alteration, mixed types of stressors (biological, physical, chemical),
and receptor-driven assessments. Note the different approaches and methods used for each.
3-7
-------
RECOMMENDATION 21. Develop an Issue Paper on receptor-driven assessments. Review
easting case studies (e.g., striped bass in the Hudson River) to gamer common themes and
lessons learned.
Biological Indices
Guidelines on biological indices are inadequate and written in a tone suggesting a
negative bias against these indices. EPA appears to have dismissed biological indices without
adequately exploring the available literature.
RECOMMENDATION 22. Review appropriate literature and revise the discussion of
biological indices accordingly, noting the utility as well as potential problems of these indices.
Indicator Species
To be useful, indicator species must have relevant characteristics that are relatively closely
congruent with those of the species they are believed to indicate.
RECOMMENDATION 23. State that indicator species should be used as a response
measure only after the characteristics believed to be indicated are clearly stated, documented,
and futty justified.
DQOs and Terminology
The Draft Guidelines appear to confuse the terms "DQO" and "quality assurance/quality
control (QA/QC)," which are not synonymous. Data quality concerns more than the procedures
by which data are collected. In this light, should line 6 on page 58 refer to DQOs or QA/QC?
The Draft Guidelines also appear to confuse the terms "data" and "information." In line 5 on
page 56, for example, "data" should be replaced with "information," since model outputs are not
really data.
3-8
-------
In addition to addressing these terminology questions, the discussion of DQOs should
emphasize the need for flexibility to avoid exclusion of data that might be useful for decision-
making. For example, data obtained through Good Laboratory Practices (GLP) might not be
useful for decision-making; these data should not be used at the expense of other, more useful
data. To sort out this issue, EPA should provide criteria for selecting and using
data/information.
RECOMMENDATION 24. Clarify the real differences between DQOs and QA/QC and
check the use of these terms throughout the Draft Guidelines. Provide criteria for including
data/information in a risk assessment; in so doing, note that useful data/information should
not be excluded and that GLP might be too rigorous in some cases.
Meta-Analysis
Although meta-analysis is a developing field, this method of analysis should be addressed
in the Guidelines, particularly given EPA's goal of producing a living document" that is
sufficiently broad to avoid becoming obsolete quickly.
RECOMMENDATION 25. Consider discussing meta-analysis and other emerging methods
of analysis; indicate areas of science that need further development.
Modifiers of Exposure
The figure on page 69 fails to mention modifiers of exposure, as does the text. A clear
discussion of how exposure can be modified is required.
RECOMMENDATION 26. Add a short discussion of exposure modifiers, explaining the
relationship between dose levels, toxicokinetics, bioaccumulation, and factors that affect these
exposure parameters relative to food chain issues (where modifying factors are particularly
important). Clarify how the environmental chemistry and fate of chemicals can modify the
exposure potential of some chemicals.
3-9
-------
Monte Carlo and Uncertainty Factors
The Analysis section would benefit from more guidance on the use of Monte Carlo
(which is already commonly used), more guidance on the use of safety factors, and more
contextual information about uncertainty factors. Uncertainty factors, which actually are
extrapolation factors (also called application factors), are considered by many to be arbitrary.
Since uncertainty factors have not been revisited since their inception, the Agency should do so
by retrospectively analyzing existing data bases (e.g., to examine the relationship between acute-
chronic ratios and valued ecosystem components). Potential resources for this effort include
Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA) data bases, work performed by the
Europeans, and the pyrethroid data sets.
RECOMMENDATION 27. Improve the discussion of uncertainty factors by including
references, rationales for their use, advantages and disadvantages of their use, criteria for
when and when not to use them, appropriate terminology, and so on. Provide resources to
revisit, update, and improve the case for uncertainty factors.
Statistical Expressions
Rather than trying to definitively establish cause-and-effect relationships, confidence
limits around point estimates might be more appropriate than other statistical expressions. Use
of confidence limits would minimize or eliminate inappropriate statistical inferences.
RECOMMENDATION 28. Explain the potential for misuse and misrepresentation of
statistical anafyses. -Encourage readers of the Draft Guidelines to express statistical inferences
as point estimates with measures of precision when possible.
3-10
-------
Uncertainty
The Draft Guidelines provide a great deal of useful discussion on sources of uncertainty,
but they provide little guidance on how to incorporate uncertainty in stressor-respdnse profiles
(e.g., ECa, ± 95% confidence limits) and exposure profiles (e.g., confidence limits, skewness,
kurtosis, underlying distribution, importance of considering sampling design).
RECOMMENDATION 29. Provide guidance on how to incorporate uncertainty in stressor-
response and exposure profiles.
Monitoring and Validation
When possible, risk assessments should include and be followed by "reality checks."
Reality checks could take the form of monitoring and/or appropriate and realistic validation.
RECOMMENDATION 30. Emphasize the importance of monitoring as a tool for estimating
exposure and existing impacts, for evaluating predictions, and for monitoring recovery.
Roles of Decision-Makers and Interested Parties
The Draft Guidelines effectively explain the roles of decision-makers and interested
parties in the problem formulation and risk management phases of ecological risk assessments. .
These groups should also be involved hi—or at least consulted during—the analysis phase.
RECOMMENDATION 31. Emphasize the continuing interaction between risk assessors and
interested parties, decision-makers, and scientific experts. Consider developing terms of
reference for each of these groups.
3-11
-------
Window to the Literature .. - .
Although the Draft Guidelines document is not (and should not be) a textbook, citing key
references (e.g., seminal papers, review papers) to provide a window to the literature for readers
who need to understand issues in greater depth would substantially improve the Analysis section
of the document.
RECOMMENDATION 32. For all major issues in the Analysis section of the Draft
Guidelines, cite review articles (not just Issue Papers), seminal papers, and existing guideline
documents from other jurisdictions.
Training and Certification
This issue was discussed in the context of the inadequacy of many present-day ecological
risk assessments. Because many groups will be using the Guidelines, some training would be
very useful. EPA also might wish to consider establishing certification procedures to ensure
adequate training of those involved in ecological risk assessment.
RISK CHARACTERIZATION
Richard Kimerle, PhJX
Senior Fellow
Monsanto Company
If goals addressing all interested parties' concerns are clearly articulated at the beginning
of an ecological risk assessment, and if all aspects of the plan generated by the problem
formulation step are carefully considered, the overall assessment is likely to be successful.,
Linking all components of the assessment—problem formulation, analysis, and risk
characterization—is critically important to ensuring that the assessment fulfills the established
goals. As the final step in the technical and scientific portion of ecological risk assessment, risk
characterization serves this purpose. All data outputs on exposure and stressor responses from
the analysis phase—as well as considerations such as elements of uncertainty, assumptions,
limitations, scientific judgment, ecological relevance, and the concerns of all interested
3-12
-------
parties—are integrated into a judgment about the probability of an impact (adverse or beneficial)
occurring in some component of an ecosystem or in an entire ecosystem. During workshop
discussions, reviewers recognized the importance of risk characterization and offered the
following suggestions for improving this section of the Draft Guidelines.
Congruence with Analysis Phase
The Analysis and Risk Characterization sections of the Draft Guidelines should follow a
similar organization to provide a logical flow from analysis to risk characterization. Indeed, the
Risk Characterization section should be, but is not yet, a powerful section that builds on the
Analysis section. Improving the connectivity between analysis and risk characterization,
especially with regard to exposure and response profiles, would be useful. Similarly, although the
Risk Characterization section presents many of the "tools" to be used hi characterizing risk, it
does not provide sufficient detail on the use of specific tools. Carrying examples through all
sections of the Draft Guidelines to illustrate the pros and cons and the utility and non-utility of
various tools under various circumstances would be useful.
RECOMMENDATION 33. Through an example carried through the Problem Formulation,
Analysis, and Risk Characterization sections of the Draft Guidelines, elaborate on how "tools"
or methods of integrating assessment results can be used in characterizing risk and
formulating risk decisions.
Ecological Significance
The Draft Guidelines' discussion of ecological significance appears to confuse ecological
consequences and human values. These two factors should be distinguished, and their roles in
determining ecological significance explained. Moreover, EPA's interpretation of ecological
significance (i.e., whether the magnitude and probability of observed effects are ecologically
relevant) is unclear. EPA needs to clarify its interpretation of ecological significance and note
that it is appropriate to consider an effect's magnitude in evaluating ecological significance.
3-13
-------
The discussion of ecological significance belongs in the Problem Formulation section
rather than the Risk Characterization section. The Problem Formulation section should clearly
define the ecological relevance of various types of effects relative to their magnitude. It should
also address the question of whether assessment and measurement endpoints are important to
determining ecological significance and emphasize the importance of knowing "where one is
going." For example, establishing the level of change that can be detected (i.e., determining the
power of the test) is important to ensure that a significant effect can be determined. If such
decision criteria are not established at the beginning of an assessment, interpreting results will be
difficult. As with other aspects of problem formulation, stakeholders should be involved in
setting these types of decision criteria.
RECOMMENDATION 34. Clarify the meaning of ecological significance, distinguishing
between ecological consequences and human values and their roles in determining ecological
significance. When discussing ecological significance, stress the likelihood and consequences
of effects.
RECOMMENDATION 35. State that all parties should agree on ecological significance
issues (Le., "if this is found, it means this and this should be done...") a priori during problem
formulation rather than setting these issues aside until risk characterization.
Analysis Versus Risk Characterization Methods
The discussion of simulations as methods of analysis beginning on page 118 belongs in
the Analysis section, since risk characterization should not involve selecting methods of analysis.
The use, possible misuse, advantages, and disadvantages of various risk characterization methods
(e.g., from quotients to probabilistic methods), on the other hand, should be discussed in more
detail. For example, the discussion of uncertainty analysis for the risk estimate (section 4.2.4)
should be expanded to discuss in more detail the advantages and limitations of and best practices
for various techniques for estimating uncertainty (see also related recommendation to develop
guidance on communicating uncertainty under the heading "Risk Communication" below).
Similarly, at least part of section 4.2.3.4 (Causality) might be more appropriate in the
Risk Problem Formulation and/or Analysis sections. The Risk Characterization section cannot
and should not review all methods of determining causality, but should focus on (and explain
3-14
-------
more clearly) the principles involved in determining causality, stressing the difficulty of
establishing causality and the advantages of multiple lines of evidence. Case studies (e.g., herring
gulls and organochlorines, acid rain examples) might be useful to illustrate these principles.
RECOMMENDATION 36. Move methods of analysis to the Analysis section so that the
Risk Characterization section focuses on risk characterization methods and issues. Provide
more detailed guidance on the use, possible misuse, advantages, and disadvantages of various
risk characterization methods.
RECOMMENDATION 37. Discuss and expand on the advantages and limitations of and
best practices for various techniques for estimating uncertainty (from quotients to probabilistic
methods).
RECOMMENDATION 38. Consider moving part of the discussion of causality to earlier
sections of the Draft Guidelines. In the Risk Characterization section, focus on principles
involved in determining causality, stressing the difficulty of establishing causality and the
advantages of multiple lines of evidence; if possible, use case studies to illustrate these points.
Multiple Stressors and Total Risk
Like the Analysis section, the Risk Characterization section provides little guidance on
how to address multiple stressors and estimate the total risk associated with these stressors.
Endpoints for total or aggregate risk do exist, and EPA should review the literature in this area
to provide more guidance in the document.
RECOMMENDATION 39. Discuss available population models, community simulation
models, and other state-of-the-art techniques for estimating total risk. Describe the
advantages and disadvantages of these techniques (including their utility in describing
ecological consequences and estimating total risk) and provide a window into the literature
on these techniques.
RECOMMENDATION 40. Stress the importance of identifying and addressing all key
stressors and identify diagnostic tools for doing so.
3-15
-------
Recovery
Determination of recovery time, currently discussed only briefly, needs to be clearly
articulated in both the Analysis and the Risk Characterization sections. Recovery, like risk, is
best addressed in a probabilistic framework. Discussions of recovery should note that one cannot
return a system to exactly pre-perturbation conditions, that recovery from one impact might be
followed by another impact (thus, consideration must be given to sustainability through and
beyond the end of anthropogenic insults), that restoration is part of recovery, and that
"something" of value must be retained. EPA might be able to draw from National Trustee
language on ecological services provided by resources and on situations when substitution of
resources is acceptable.
RECOMMENDATION 41. Clearly articulate issues related to recovery time, including the
issue of whether a return to pre-perturbation conditions is possible and the relationship
between multiple insults over time, persistence, and recovery. Consider combining the
discussion of restoration with that of recovery in section 5.3.2.3.
Natural Variability
Reviewers questioned whether anthropogenic impacts should be discussed in the section
on natural variability, but decided that this is unnecessary given previous discussions (assuming
EPA implements the reviewers' other recommendations).
Weight of Evidence
EPA should expand the section on weight of evidence (page 121), and perhaps add
hypothetical examples, to elucidate what "weight of evidence" means and how to use weight of
evidence. A document on this topic prepared by a volunteer work group in Massachusetts
(details available from Nancy Bettinger) might be a useful resource.
Weight of evidence can involve determining how well endpoints are characterized in an
assessment and how useful they are for final problem-solving; this, in turn, sometimes points to
3-16
-------
new measurement endpoints for an assessment endpoint. Similarly, weight of evidence can be
used to compensate for deficiencies and to estimate risk in the face of conflicting information (a
topic that should receive more attention in the Risk Characterization section). Thus, weight of
evidence can be defined as a systematic, rigorous method of developing a risk assessment
through logical interim decision points to achieve a final decision. Evaluating the weight of the
evidence results in either a decision on the magnitude of the risk (which in turn results in a
management decision) or a decision to iterate to the next tier.
RECOMMENDATION 42. Expand section 5.3.1 to explain the meaning, reasoning involved
in, and use of weight of evidence. If possible, provide examples (cf. previous recommendation
for an example, possibly hypothetical, carried through all major sections of the Draft
Guidelines).
RECOMMENDATION 43. In this or a separate subsection, provide more guidance on how
to deal with conflicting lines of evidence.
Optimum Use of Information
Interplay between risk analysis and data acquisition (e.g., via tiers) helps ensure optimum
use of available assessment data in characterizing risk and in determining what is needed if a
decision is made to proceed to the next tier.
RECOMMENDATION 44. In part through text and in part through an example carried
through all sections of the Draft Guidelines, provide guidance on how to exploit all of the
data generated by the analysis phase; describe both effective use of data in risk
characterization and in determining data needs for a subsequent assessment iteration/tier.
Roles of Interested Parties in Risk Characterization
The Draft Guidelines appear to suggest that interested parties should be involved mainly
in the problem formulation phase of ecological risk assessment. Interested parties play a key
role in risk characterization as well. Their involvement in risk characterization helps ensure that
assessment results are expressed in a way that will be meaningful at the management phase.
3-17
-------
RECOMMENDATION 45. Discuss the roles of risk assessors, experts, decision-makers, and-
interested parties in risk characterization. Note that this step should be appropriately inclusive
to ensure that assessment results are expressed in a way that is useful for decision-making; the
appropriate level of inckisivity depends on the type and goal of the assessment (being too
inclusive might be unnecessary to meet the Agency's needs or might stall the process).
Monitoring •
Monitoring, especially as it relates to recovery, should be discussed in a separate section.
RECOMMENDATION 46. In a separate subsection in the Risk Characterization section,
provide more guidance on monitoring, paying special attention to the need to develop data
over time and the use of monitoring to determine the effectiveness or impact of ecological risk
management decisions.
Risk Communication
Audiences of risk communication include nontechnical individuals and groups. Reaching
these audiences should include communication and education during problem formulation, not
just preparation of readable documents after completion of the assessment. Communicating
uncertainty during and after the assessment is particularly challenging and should be a topic of
additional guidance.
RECOMMENDATION 47. Note that risk communication is an integral part of problem
formulation, include this topic in the text box on planning, and emphasize this topic more in
the Risk Characterization section. Spelling out objectives and elements of good
communication, addressing risk perception, and providing a window to the available literature
would be useful.
RECOMMENDATION 48. Develop guidance on how to communicate uncertainty to a
variety of audiences. Cover such issues as putting uncertainty into context, explaining
quantifiable as well as qualitative uncertainty, and discussing sources of high uncertainty.
3-18
-------
End of Section
Currently, section 5 ends "flat." This could be corrected by adding a summary that
addresses the following topics: use of the Draft Guidelines to simplify what is often perceived as
a hopelessly complex challenge, the role of scientific judgment, integration of uncertainty issues
discussed throughout the document, and preparation of the overall ecological risk assessment
report (providing a generic framework or outline would be useful). Scientific judgment and
values especially need to be highlighted, and a separate section on reporting might be useful.
The bullets on pages 127-128 should be integrated and presented earlier in the text or in a text
box earlier in the document.
RECOMMENDATION 49. Develop a new section (section 5.4) to provide closure for the
Risk Characterization section and a logical lead-in to section 6. Address at least the
following topics (and perhaps others): simplifying ecological uncertainty, scientific judgment
and values, and the final assessment report.
3-19
-------
-------
SECTION FOUR
OBSERVER COMMENTS
At the end of each workshop day, time was allotted for workshop observers to comment
on the guidelines and workshop deliberations. These comments are summarized below.
DAY 1 OBSERVER COMMENTS
Elizabeth Kelly, Los Alamos National Laboratory
This observer commented on the definition of ecological risk assessment in the Draft
Guidelines. Noting that some workshop reviewers and observers had struggled with whether an
effects-initiated assessment is truly a risk assessment and if so how it differs from a stressor-
initiated assessment, the observer suggested that a deficiency in the definition might account for
the confusion. Dlustrating her point mathematically, the observer compared what she considered
a traditional definition of risk to the Draft Guidelines' definition of ecological risk:
Traditional definition of risk:
R = P(Hazard n Consequence)
Risk equals the probability of a hazard and the consequence of the hazard. In the case
of ecological risk, this translates to:
R = P(SnE n AE)
Risk equals the probability of the stressor and the exposure causing the effect (i.e., the
hazard) and the adverse effect (i.e., the consequence) associated with the stressor and
exposure.
Draft Guidelines definition of ecological risk
ER = P(AE/SnE)
Ecological risk equals the probability of an adverse effect given an exposure resulting
from a stressor. (The slash in the equation means "given" or "conditioned on".)
4-1
-------
Given these equations, traditionally defined risk (R) equals EPA's ecological (ER) risk-
times the probability of exposure given the stressor, times the probability of the stressor:
R =P(SnEDAE)
» P(AE/SnE) x P(SnE)
= ERxP(SnE)
Thus, traditionally defined risk and ecological risk are the same (R = ER) only when the
probability of the stressor is one and the probability of exposure given the stressor is one. In an
ecological risk assessment, the former might be true (i.e., the existence of the stressor might be
known), but the latter might not be true. The observer contended that EPA's definition of ,
ecological risk ignores the dependence of exposure on the stressor — the probability of observing
a particular exposure given the stressor — and that this might cause problems in assessments '
involving comparisons.
To illustrate, the observer discussed the implications of the absence of the probability of
exposure term for an assessment in which it is known that there is an effect, that multiple
sources of exposure exist, and that each source is (or could be) associated with a different level
of risk. Mathematically, the risk associated with source i is the probability of source i and the
exposure given the observed adverse effect, times the probability of the adverse effect:
Rj = P(S£nE n AE)
= P(S;nE/AE) x P(AE)
Since the effect is known to exist, the probability of the effect is one, and that term can
be dropped:
Ri = P^DE/AE) x P(AE)
= P(S;nE/AE) x 1
= P(SiDE/AE)
4-2
-------
Equating this to EPA's definition of ecological risk as shown above produces the
following:
= ERixP(E/Si)xP(Si)
Assuming that the assessment examines stressors that have been observed, the probability
of the stressors is one; but the probability of the exposure given the stressor remains unknown.
As a result, traditionally defined risk and ecological risk are not the same — they differ by the
probability of exposure given the stressor. It is unlikely that exposure will be the same for
different stressors. Thus, using the more limited definition of ecological risk in the guidelines
could lead risk assessors to miss an important component of risk. The observer suggested that
EPA reconsider the definition of ecological risk, or at least point out how and why it differs from
the traditional definition of risk.
The observer's comments sparked a number of questions. To help clarify the discussion,
one reviewer suggested that the observer's comments reflect differences in how ecological risk
assessment and "traditional" or engineering risk assessment have evolved. He suggested that
analysts in both fields do the same things, but that this has not necessarily been captured in the
definition of ecological risk. In ecological risk assessment, he said, risk assessors focus on the
probability of an adverse event and assess the consequences of the event in a separate step.
Thus, the observer's concern is not addressed in the ecological risk definition, but it is addressed
hi the risk assessment process as a whole. The observer and reviewer agreed that risk assessors
probably address the issue in practice, if not in equations or definitions.
Michael Harrass, Amoco Corporation
This observer recommended that EPA:
Examine products developed by other organizations to supply details on methods and
approaches mentioned in EPA's Draft Guidelines. In particular, several American
Society for Testing and Materials (ASTM) groups have developed or are
developing consensus technical documents on identifying assessment endpoints,
identifying potentially impacted sites, performing risk-based corrective actions, and
4-3
-------
so on. It might be productive to consider using these documents to support
EPA's Draft Guidelines—and to promote harmonization of methods and
approaches.
Consider what sites should be assessed and what conditions make a site suitable for
ecological risk assessment. Should an ecological risk assessment be conducted for
every site? Can meaningful ecological risk assessments be conducted for service
stations? For operating wellheads? What are the minimum conditions needed to
conduct a sound, meaningful assessment? How should assessment activities be
scaled in different situations?
Require that risk managers take the time to participate in the problem formulation
step if EPA decides that this step is important. The Draft Guidelines should reflect
the reality that risk managers' time is often very limited.
Note the need to train people who conduct ecological risk assessments, since these
individuals are usually engineers or human health risk assessors rather than
ecologists.
Distinguish between ecological risk assessment activities and other types of activities.
For example, investigating a site known to be impacted might be a diagnostic
activity rather than an ecological risk assessment. Similarly, verifying the presence
of a stressor at levels sufficient to produce the observed consequences might not
be a risk activity.
DAY 2 OBSERVER COMMENTS
Alan Robin, Water Environmental Foundation
This observer urged EPA to produce a practical product that:
• Provides (or aids in generating) numerical standards, management practices, or
mechanisms that can be written into regulations to reduce the impact of regulated
materials on the environment.
• Recognizes the beneficial as well as damaging impacts of regulated materials—and
notes that the resulting overall picture should be compared to that associated with
alternative materials. This information is important to risk managers in crafting
regulations.
• Addresses the issue of ecological significance. For example, putting biosolids on
degraded arid range land produces an impact: greater crop growth and
replacement of animal populations adapted to sparse conditions with populations
that do well in lush soil. Is this a positive impact, negative impact, or no impact?
4-4
-------
Addresses terrestrial and marine systems.
Reflects the real-world data needs of different programs. The sewage program, for
example, needs actual field information rather than laboratory or experimental
data. This should be reflected in a discussion of hierarchy of data and weight of
evidence.
Daniel Michael, Neptune and Company
This observer noted that workshop reviewers and observers commented on the lack of
specific guidance in the Draft Guidelines, the need to reorganize the document, and the need to
expand on the discussion of the DQO approach. He suggested that these perceived deficiencies
result from an effort to address all levels of decision-making at once. Thus, a reader consulting
the document from a particular point of view (e.g., that of needing to support a particular level
of decision-making) would not know how to proceed. This problem could be resolved, he
suggested, by dividing the Draft Guidelines into sections based on the three levels of decision-
making:
• Site-specific decisions (e.g., whether to remediate a site, what limits to set for a
specific discharge at the site).
• Programmatic decisions (e.g., at what level to set a threshold, whether to register
a pesticide, how to deal with sewage sludge).
• Policy decisions (e.g., whether to regulate a new chemical or industry, whether to
introduce a new species).
The observer suggested starting with the site-specific level, providing relatively specific
guidance at this level. Subsequent sections on higher-level decisions would provide broader and
broader guidance. Each section could include guidance on how to implement the ecological risk
assessment process and how to present the results of the process, both of which depend on the
level of decision-making being supported.
The observer commented that dividing the Draft Guidelines in this manner also would
facilitate resolution of other issues raised during the workshop. The question of tiering, for
example, becomes straightforward in the context of a specific decision level. Tiering has been
4-5 ^
-------
well described at the site-specific level, where each tier is intended to reduce the level of
uncertainty and the level of conservatism used in estimates to avoid wasting public dollars.
Tiering takes on different meanings at the programmatic level and at the policy level.
One reviewer suggested that some ecological risk assessments might not fall easily into
any of the three decision levels. He asked, for example, how the observer would categorize a
watershed assessment. The observer replied that he would ask what kind of decision is to be
made. Most likely, the assessment would be aimed at supporting a site-specific
decision—whether to remediate various sources affecting the watershed. He went on to state
that, having worked with nearly every EPA program, it is his experience that every EPA decision
can be categorized into one of the three levels listed. A reviewer added that the approach
suggested would have the additional benefit of underscoring the concept of risk assessment as a
decision-making tool.
The observer concluded by suggesting that the expanded discussion of DQOs
recommended by some reviewers during the workshop might not be warranted because the
document specifically excludes data collection from the ecological risk assessment process. He
contended that the DQO issue would be clearer if the document's discussion of drawing people
into the process distinguished between who needs to be involved in assessments supporting
different decision levels.
Unidentified Observer
This observer agreed with Daniel Michael that distinguishing between assessments for
different decision levels may have some utility. He noted that when the U.S. Food and Drug
Administration, for example, reviews food and drug product applications, it is being called upon
to make what are clearly nationwide, programwide decisions. A watershed assessment, he
suggested, would be a site-specific assessment because the watershed has a specific "address."
Distinguishing between these two types of assessments (nationwide versus site-specific) is useful
because they involve different approaches to assessment and decision-making.
4-6
-------
Suzanne Marcy, National Center for Environmental Assessment, ORD, EPA
Suzanne Marcy concurred that there may be some utility in presenting the information
this way, but that reorganizing the document accordingly might be difficult. Anne Fairbrother, a
reviewer, said that she favors the current organization because it reflects the fact that the core of
ecological risk assessment lies in defining the question being asked and that the rest of the
assessment process is driven by that question. The observer responded that he is not
recommending that the three phases of risk assessment (problem formulation, analysis, risk
characterization) be removed; rather, he recommends that issues related to each of the three
decision levels be defined and explained in each phase.
4-7
-------
-------
SECTION FIVE
NEXT STEPS
Prior to the work group sessions, William Wood, Director of EPA's Risk Assessment
Forum, addressed a series of questions that might influence how reviewers and observers view
the Draft Guidelines and this workshop. His remarks, and reactions to them, are summarized
below.
WHERE DOES EPA GO FROM HERE?
EPA is following a very deliberate course in developing the Draft Guidelines. Having
concluded early on that no one document can cover all aspects of ecological risk assessment,
EPA is developing a series of guidance documents. The process of developing these guidance
documents began with the development and publication of a number of reports: Framework for
Ecological Risk Assessment, several case studies, and several issue papers. The Guidelines expand
upon and will replace the Framework Report, EPA expects that the Guidelines will in turn form
the basis for a series of followup activities and documents on areas needing additional guidance
development. This workshop is intended to produce recommendations on what guidance can be
provided now and what must be addressed later.
Following this workshop, the Risk Assessment Forum (RAF) will consider all
recommendations offered and revise the Guidelines accordingly. The RAF will then seek both
internal and interagency review of the Guidelines. The RAF will revise the document again and
submit the revised document as proposed guidelines to the Federal Register for public review and
comment. At the same time (probably in the late Spring), the RAF will submit the proposed
guidelines to the SAB for review. The RAF will then work with the public's and SAB's
comments to revise the document. EPA hopes to publish final guidelines by the end of calendar
year 1996 or early in calendar year 1997.
5-1
-------
WHO IS THE AUDIENCE FOR THIS GUIDELINES?
In general, EPA guidelines contain two types of guidance:
• State-of-the-science guidance, consisting of information with which most risk
assessors would agree. The goals of this guidance are to document the Agency's
position on the topics addressed and to make the information understandable to
people inside and outside the Agency.
• Science policy guidance, consisting of EPA's policy positions on key issues,
especially where methodology or data gaps exist. These positions are not
necessarily positions that other agencies or private groups would take. Thus,
science policy guidance in EPA guidelines are primarily aimed at Agency risk
assessors. EPA hopes that this guidance is reasonable enough that others might
adopt it, but the Agency does not presume to impose it on others.
The Draft Guidelines being reviewed at this workshop are intended to provide these two
types of guidance.
WHAT ARE EPA'S GOALS IN TERMS OF GUIDANCE?
In providing state-of-the-science guidance and science policy guidance, EPA aims to use a
range of guidance language based on EPA's "guidance pyramid" (see figure 1) in which the
majority of guidance language pertains to the use of judgment; somewhat less language describes
case-by-case issues, guideposts, and preferences; and relatively little sets forth specific rules.
In their premeeting comments, many reviewers indicated that the Draft Guidelines should
provide more "guidance," meaning preferences and rules. EPA recognizes that the document
might rely too heavily on guidance language pertaining to the use of judgment rather than
preferences, and the Agency hopes that the workshop yields suggestions for improving the
balance between the various types of guidance. At the same time, EPA recognizes that the
guideline must be a living document—one that is not overly prescriptive and, thereby, becomes
dated.
5-2
-------
c
-------
WHAT IS THE RELATIONSHIP BETWEEN THESE GUIDELINES AND EXISTING
PROGRAM GUIDANCE OR PROGRAM GUIDANCE THAT MIGHT BE DEVELOPED IN
THE FUTURE?
Publication of the guidelines for ecological risk assessment is expected to spur
development of program-specific guidance materials. Compared to Agencywide guidelines,
program-specific guidance generally places greater emphasis on preferences and rules. This is
appropriate because programs have particular applications in mind, and they know what types of
data are typically available. In addition, programs can update program-specific guidance
materials more frequently than can the Agency as a whole.
DISCUSSION
Observing that the premeeting comments were quite lengthy, one reviewer remarked that
addressing the comments would be a monumental task and wondered how EPA plans to
accomplish the task. William Wood agreed that the task is monumental, but said that EPA
hopes that the workshop produces consensus recommendations to guide the revision process. He
also noted that EPA will seek further input from within and outside the Agency. A reviewer
asked if this will ensure some consistency between program offices. William Wood responded
that the programs have reviewed predecessor documents and that, although debate continues, the
Agency is moving toward concordance. Another reviewer asked whether EPA would seek input
from the European Union, commenting that international harmonization is a worthy goal.
William Wood replied that EPA has shared drafts of the documents and will continue to do so.
During the course of the workshop, reviewers offered additional suggestions related to
EPA's planned next steps:
• The reviewers encourage EPA to move forward with the guidelines as quickly as is
reasonable.
» Developing issue papers on complex or poorly understood topics (e.g., multiple
stressor interactions) would be useful, but should not delay revision and
publication of the Guidelines.
5-4
-------
Similarly, providing resources to revisit, update, and improve the case for
uncertainty factors would be useful, but should not delay work on the Guidelines.
Funding technology transfer of the Guidelines would represent a wise investment
of Agency resources.
5-5
-------
-------
APPENDIX A
REVIEWER LIST
-------
-------
v>EPA
United States
Environmental Protection Agency
Risk Assessment Forum
Workshop on the Draft Ecological
Risk Assessment Guidelines
Holiday Inn - Georgetown
Washington, DC
December 6-7, 1995
Reviewer List
Lawrence Barnthouse
Principal Health Scientist
McLaren Hart/ChemRisk
109 Jefferson Avenue - Suite D
Oak Ridge, tN 37830
423-482-8978
Fax: 423-482-9473
E-mail: oakridge
6ffice@mclrnhrt.uucp.netcofn.com
Steven Bartell
Hifhwood Farm
P.O. Box 688
Wartburg, TN 37887
615-346-7406
John Bascietto
Environmental Protection Specialist
U.S. Department of Energy
1000 Independence Avenue
(EH-413)
Washington, DC 20585
202-586-7917
Fax: 202-586-3915
E-mail: john.bascietto@hq.doe.gov
Nancy Bettinger
Environmental Analyst
Massachusetts Department of
Environmental Protection/
Office of Research and Standards
I Winter Street - Third Floor
Boston, MA 02108
617-556-1159
Fax:617-556-1049
E-mail: nbettinger@state.ma.us
Joanna Burger
Professor
Environmental & Occupational
Health Sciences Institute
Department of Biological Sciences
Rutgers University
Piscataway, NJ 08855-1059
908-445-4318
Fax: 908-445-5870
E-mail: burger@biologx.rutgers.edu
Peter Chapman
EVS Environment Consultants
195 Pemberton Avenue
North Vancouver, BC V7P 2R4
Canada
604-986-4331
Fax: 604-662-8548
E-mail: chapmanp@pl.apfnet.org
James Clark
Senior Staff
Environmental Toxicologist
Mettiers Road (CN-2350)
Exxon Biomedical Sciences, Inc.
East Milstone, NJ 08875
908-873-6039
Fax: 908-873-6009
William Cooper
Institute for
Environmental Toxicology
C-231 HoldenHall
Michigan State University
East Lansing, Ml 48824
517-353-6469
Fax:517-355-4603
Peter deFur
Senior Scientist
Environmental Defense Fund
1875 Connecticut Avenue
Suite 1016
Washington, DC 20009
202-387-0070 Ext: 14
Fax: 202-234-6049
E-mail: peterd@edf.org
James Donald
Senior Toxicologist
Office of Environmental Health
Hazard Assessment
California Environmental Protection
Agency
601 North Seventh Street
P.O. Box 942732 (MS-241)
Sacramento, CA 94234-7320
916-445-5652
Fax: 916-327-1097
E-mail: hwl.jdonald@hwl.cahwnet.gov
Charles Eirkson, III
Toxicologist
U.S. Food and Drug Administration
Environmental Safety Branch
7500 Standish Place (HFV-152)
Rockville, MD 20855
301-594-1683
Fax: 301-594-2297
E-mail: eirkson@al .fda.cvm.gov
i Printed on Recycled Paper
A-l
-------
Anne Fairbrother
Senior Wildlife Ecotoxicologist
Ecological Planning and
Toxicology, Inc.
SOIOSWHoutStreet
Corvallis, OR 97333-9541
541-752-3707
Fax:541-753-9010
E-mail: falrbroa@aol.com
Alyce Fritz
CRC Branch Chief
Coastal Resource Coordination
NOAA/HAZMAT (N/ORCA-32)
7600 Sand Point Way, NE
Bin CI5700
Seattle, WA 98115
206-526-6305
Fax: 206-526-6941
E-mail: alyce_fritz@hazmat.noaa.gov
Robert Goldstein
Manager, Environmental
Risk Analysis
Electric Power Research Institute
P.O. Box 10412
Palo Alto, CA 94303
415-855-2593
Fax:415-855-1069
E-mail: rogoldst@eprinet.epri.com
Lawrence Harris
Professor, Department
of Wildlife Ecology and Conservation
Program in Landscape Ecology
University of Florida
303 Newins-Ziegler Hall
Gainesville, FL 32611
904-392-4851
Fax: 904-392-6984
Richard Kimerle
Senior Fellow
Monsanto Company
800 North Lindbergh Boulevard
(U4E)
St Louis, MO 63167
314-694-3286
Fax:314-694-1531
E-mail: raklme@ccmall.monsanto.com
'Timothy Kubiak
U.S. Fish and Wildlife Service
Division of Environmental
Contaminants (ARLSQ-330)
•4401 North Fairfax Drive
Arlington, VA 22203
703-358-2148
Fax: 703-358-1800
Lyman McDonald
President/Senior Biometrician
Western Ecosystems Technology, Inc.
2003 Central Avenue
Cheyenne, WY 82001
307-634-1756
Fax: 307-637-6981
E-mail: lymanmcd@csn.org
Dwayne Moore
Senior Evaluator
Commercial Chemicals
Evaluation Branch
Environment Canada
Place Vincent Massey Building
14th Floor
351 St. Joseph Boulevard
Hull, Quebec KIAOH3
Canada
819-953-1664
Fax: 819-953-4936
'Gerald Niemi
Director, Center for Water
and the Environment
Professor, Department of Biology
Natural Resources Research Institute
University of Minnesota
5013 Miller Trunk Highway
Duluth, MN 55811
218-720-4270
Fax: 218-720-9412
E-mail: gnlemi@sage.nrri.umn.edu
Richard Orr
Senior Entomologist
Animal and Plant Health
Inspection Service
U.S. Department of Agriculture
4700 River Road
Unit H7-Room3-C-38
Riverdale, MD 20737-1221
301-734-8939
Fax: 301-734-5899
E-mail: rorr@aphis.usda.gov
Kevin Reinert
Section Manager
Ecotoxicology & Environmental
Risk Assessment
Rohm and Haas Company
727 Norristown Road
P.O. Box 904
Spring House, PA 19477-0904
215-641-7479
Fax: 215-619-1621
E-mail: rstkhr@rohmhaas.com
William Smith
Professor of .Forest Biology
Greeley Laboratory
Yale University
370 Prospect Street
New Haven, CT 06511
203-432-5149
Fax: 203-432-3929
E-mail: whsmith@minerva.cis.yale.edu
Ralph Stahl, jr.
Senior Consulting Associate
DuPont Company
1007 Market Street (B-12212)
Wilmington, DE 19898
302-773-0611
Fax: 302-773-1889
E-mail: stahlrg@al .csoc.umc.dupont.com
Randall Wentsej
Chief Scientist
U.S. Army
ATTN: SCBRD-RTL
Aberdeen Proving Ground,
MD 21010
410-671-2036
Fax: 410-671-2081
E-mail: rswentse@cbdcom.apgea.army.mil
A-2
-------
APPENDIX B
CHARGE TO REVIEWERS
-------
-------
CHARGE TO REVIEWERS
FOR ERA'S ECOLOGICAL RISK ASSESSMENT GUIDELINES
EPA's Risk Assessment Forum first
published Agency-wide guidelines for human
health risk assessment in 1986 and has been
working to develop ecological risk assessment
guidelines since 1989. As discussed in the
Foreword and Introduction to the guidelines,
EPA has approached guidelines development
in a step-wise fashion, with each subsequent
step drawing upon prior products and
experience. Consistent with the
recommendations of previous peer reviewers
and the Science Advisory Board, the
guidelines are organized by the major phases
of the ecological risk assessment process:
problem formulation, analysis, and risk
characterization. The goal of the guidelines is
to improve the quality and consistency of
EPA's ecological risk assessments.
This charge is divided into two parts.
The first part asks for your assistance in
identifying and highlighting important guidance
principles in the document. The second part
asks specific questions concerning particular
portions of the guidelines. Please focus on
those Part II questions that you feel are most
important and that best suit your expertise, and
feel free to comment on other issues that may
not be addressed by the questions. Your
comments, along with those of other reviewers
and recommendations from the workshop, will
be considered in revising the guidelines.
EPA's Risk Assessment Guidelines
Guidelines are developed by tie Htsk
Assessment Forum, with oversight from EPA's
Science Policy Council. The guidelines are
intended to describe the scientific principles and
approaches for risk assessment and establish
-EPA's science policy on new or controversial
issues. They are a product of consensus- *
building and peer review'processes.
Guidelines are intended, to:'
•• - 'improve the quality of EPA's risk
assessments; ' /:
• • promote Agency-wide consistency; and
• inform the scientific community and the
public. , ^ %
Guidelines are not:
-• ' rule books. Guidelines are intended
primarily for use within EPA.
• cookbooks. Guidelines do not provide
detailed "how tos". Rather, they address
major issues of concern.
* iext books. Guidelines do not contain
extensive background material for novice
readers., - '.,,',-
* program-specific guidance. It is left to
each EPA program (e.g., Superfund,
1 pesticides) to adapt Agency-wide
guidelines to their own needs.
PARTI. GENERAL PRINCIPLES
For each phase of the ecological risk assessment process, the ecological risk assessment
guidelines discuss the general options available and the various considerations, strengths, and
limitations that the risk assessor should keep in mind when evaluating these options. The National
Research Council has recommended that EPA's human health risk assessment guidelines indicate
the basic assumptions underlying risk assessment approaches, the default positions for these
approaches, and the circumstances under which deviations from the default positions are
peimissible. Please answer the following questions, and keep them in mind as you review the Part
II questions.
• Specifically, how can the ecological risk assessment guidelines be improved to more clearly
indicate underlying assumptions and default positions?
• What changes would you recommend in each section of the document?
B-l
-------
PART II. SPECIFIC QUESTIONS
Guidelines Balance
These questions concern crosscutting issues that involve all parts of the ecological risk
assessment guidelines.
1. Considering both the present state of the science and present and future Agency needs,
how well are the guidelines balanced regarding the range of stressors, levels of biological
organization, ecosystem types, and spatial/temporal scales? Specifically, what would you
emphasize or de-emphasize?
2. What would you suggest to improve the use of case illustrations in the guidelines? How
useful is Appendix A in illustrating a range of applications of the risk assessment process?
3. Some Agency reviewers of the guidelines have suggested that more examples of terrestrial
assessments and field approaches (e.g., bioassessment techniques) should be used.
Specifically, what, if anything, should be added?
4. Areas of uncertainty are summarized in the problem formulation (section 1.5), analysis
(section 3.7), and risk characterization (section 5.2.4). How useful is this approach in
providing guidance on uncertainty issues?
Introduction and Scope
The introductory portion of the guidelines (section 1) describes background, scope, and the
intended audience for the guidelines. Section 1 also discusses the importance of ecological risk
assessment for environmental decision-making, contrasts these guidelines with the previously
published Framework Report, and clarifies terminology issues.
5. How could the Introduction be modified to more clearly communicate the scope and
content of the guideline?
6. Terminology, especially related to endpoints and exposure, has always been controversial.
What changes, if any, would you recommend in guidelines terminology1?
7. The overall framework figure for the ecological risk process has been retained, although
some changes have been made to the diagrams for problem formulation, analysis, and risk
characterization. What further modifications, if any, are required?
Risk Manager Interactions
Although the primary focus of these guidelines is risk assessment, not risk management,
appropriate risk assessor-risk manager interactions are critical to the success of ecological risk
assessments.' Sections 2 and 6 of the guidelines highlight these interactions at the beginning and
end of the ecological risk assessment process.
8. Please comment on the description of the risk manager's role at the initiation of an
ecological risk assessment and on the principles for selecting management goals.
9. What additional points, if any, should be covered about relating ecological information to
risk management decisions after the completion of an ecological risk assessment?
B-2
-------
Problem Formulation
During problem formulation (section's), the purpose for the assessment is articulated, the
scope and boundaries are defined, the assessment endpoints and conceptual model are
developed, and the plan for analyzing and characterizing risk is developed.
10. How useful is the categorization of assessments as either stressor- or source-initiated,
effects-initiated, and ecologipal value-initiated?
11. Please comment on the discussion of assessment endpoints and their relationship to
management goals and "measures."
12. What, if anything, would you add to the discussion of risk hypotheses and conceptual
models to give the reader a clearer understanding of their nature and content?
Analysisf ~
The analysis phase (section 4) includes technical evaluation of data on potential effects of
and exposure to stressor(s) identified during problem formulation. Primary activities include
exposure and ecological response analysis. The outputs of the analysis phase are summarized in
exposure and stressor-response profiles. Section 4 is organized by stressor type: chemical,
physical, biological, and multiple stressors.
13. How would you change this section, if at all, to improve the balance in the discussion of the
different stressor types?
14. Are there additional points or principles that should be emphasized for either chemical,
physical, or biological stressors?
15. What additional principles would you suggest, if any, for the analysis of multiple stressors?
Risk Characterization
Within risk characterization (section 5), key elements include estimating risk, evaluating
ecological significance, and determining the weight of evidence. Major uncertainties, assumptions,
and limitations of the assessment are summarized.
16. What additional principles should be highlighted in the discussion of risk estimation
techniques?
17. The guidelines propose four criteria for ecological significance. How should this list of
criteria be modified, if at all? What additional guidance, if any, might be added to the
discussion of these criteria?
B-3
-------
-------
APPENDIX C
WORKSHOP AGENDA
-------
-------
vvEPA
United States
Environmental Protection Agency
Risk Assessment Forum
Workshop on the Draft Ecological
Risk Assessment Guidelines
Holiday Inn - Georgetown
Washington, DC
December 6-7, 1995
Agenda
Workshop
Chair
William Smith
Yale University
New Haven, CT
WEDNESDAY, DECEMBER 6, 1995
7:30AM Registration/Check-ln
PLENARY SESSION
8:30AM Welcome
William Wood, U.S. Environmental Protection Agency (EPA),
Office of Research and Development (ORD)
8:35AM Introductory Remarks
Robert Huggett, EPA, ORD
8:45AM Workshop Background, Objectives and Format,
and Overview of Guidelines Issues
Workshop Chair: William Smith
William Wood, EPA, ORD
9:05AM Discussion
9:35AM BREAK
9:50AM Problem Formulation Highlights
Suzanne Marcy, EPA, Office of Water
10:OOAM Problem Formulation Issues
William Smith
i Printed on Recycled Paper
C-l
-------
WEDNESDAY, DECEMBERS, 1995 (continued)
CONCURRENT WORK GROUP SESSIONS (2)
10:1 SAM Problem Formulation Discussion
II:35AM LUNCH
I2:45PM Problem Formulation Discussion (continued)
2:45PM BREAK
PLENARY SESSION
3:OOPM Problem Formulation Discussion: Comments and Recommendations to EPA
Richard Kimerle, Monsanto Company; Dwayne Moore, Environment Canada; and
William Smith
4:OOPM Analysis Highlights
Susan Norton, EPA, Office of Research and Development
4:15PM Analysis Issues
Dwayne Moore
4:30PM Observer Comments
5:OOPM ADJOURN
THURSDAY, DECEMBER 7, 1995
CONCURRENT WORK GROUP SESSIONS (2)
8:OOAM Analysis Discussion
IO:30AM BREAK
PLENARY SESSION
IO:45AM Analysis Discussion: Comments and Recommendations to EPA
Richard Kimerle, Dwayne Moore, and William Smith
II:45AM LUNCH
I2:45PM Risk Characterization Highlights
Patricia Cirone, EPA, Region 10
I :OOPM Risk Characterization Issues
Richard Kimerle
CONCURRENT WORK GROUP SESSIONS (2)
1:15PM. Risk Characterization Discussion
2-.45PM BREAK
3:OOPM Risk Characterization Discussion (continued)
G-2
-------
THURSDAY, DECEMBER?, 1995 (continued)
PLENARY SESSION
4:OOPM Risk Characterization Discussion: Comments and Recommendations to EPA
Richard Kimerle, Dwayne Moore, and William Smith
5:OOPM Observer Comments
5:30PM Workshop Chair's Closing Remarks
5:45PM ADJOURN
C-3
-------
-------
APPENDIX D
PREMEETING COMMENTS
-------
-------
Workshop on the Draft Ecological
Risk Assessment Guidelines
PREMEETING COMMENTS
December 1995
&EPA
Risk Assessment Forum
U.S. Environmental Protection Agency
Washington, DC
-------
-------
TAB LE OF CONTENTS
Lawrence Barnthouse I
Steven Bartell 7
John Bascietto 15
Nancy Bettinger 27
Joanna Burger 33
Peter Chapman 43
James Clark . . . ; 59
William Cooper 73
Peter deFur 79
James Donald . 95
Charles Eirkson, III 107
Anne Fairbrother 113
Alyce Fritz . 137
Robert Goldstein 151
Lawrence Harris 161
Richard Kimerle 171
Timothy Kubiak , 195
Lyman McDonald 201
Dwayne Moore 209
Gerald Niemi 225
Richard Orr 235
Kevin Reinert 239
William Smith . 255
Ralph Stahl, Jr 269
Randall Wentsel 287
-------
-------
Lawrence Barnthouse
-------
Lawrence W. Barnthouse
Comments on Ecological Risk Assessment Guidelines
General Principles
For the most part, this document is simply an incremental elaboration of the Framework for
Ecological Risk Assessment According to the Charge to Reviewers, guidelines are intended to
"promote Agency-wide consistency and to "inform the scientific community and the public." In
their present form, the guidelines are much too vague to perform either function effectively. To
be effective, the guidelines should:
• provide explicit, detailed guidance on issues expected to be common to most or all
agency programs. Examples of such issues might include the use of phased or tiered
assessments; the use of formal scoping procedures such as the Data Quality Objectives
Process; a preference for populations/ecosystems as assessment endpoints (except when
protected species are involved); establishment of a hierarchy of preferred data types
(QSAR to life- cycle chronic tests); a preference for using weight-of-evidence
procedures rather than the most conservative line of evidence;...
• clearly identify issues that differ among programs and for which the program offices
should develop their own. One easy distinction is between predictive (PMNs, pesticide
registration, NPDES permit), retrospective (RCRA, CERCLA, FIFRA special review)
and nonregulatory (watershed management) programs. ;
Ecological risk assessment guidelines should strive for the same level of specificity the agency's
health risk assessment guidelines. Otherwise, almost any assessment that can be massaged into
the format of the Framework can qualify as a valid ecological risk assessment.
Specific Questions
Guidelines balance
I'm not overly concerned with whether this document provides a complete listing of the
stressors, levels of organization, etc. that are or may be relevant to ecological risk assessments. I
-2-
-------
Lawrence W. Barnthouse
think it's more important to identify the types of stressors, endpoints, and scales associated with
the different types of assessments of interest to EPA. Clearly, assessments that involve
persistent, bioaccumulative chemicals require a longer time-scale and more complex endpoints
than do assessments of acutely toxic but nonpersistent chemicals. Watershed-level assessments
almost by definition require large spatial scales and consideration of nonchemical stressors. A
thoughtful discussion of the types of scientific information needed for different types of
assessments would add a great deal of value to the guidelines.
Unfortunately, almost all of the case studies predate the Framework, and for this reason
they can do little to illuminate the current guidelines. Adding additional examples of the same
type won't help much. What would be useful would be a discussion of how the case study
assessments could have been improved by adherence to risk assessment practices that are defined
in the guidelines.
The taxonomy of uncertainty described hi section 1.5 is very useful. I would, however,
recommend either revising or eliminating the column of "examples" included in the tables.
Some of these are quite confusing. Definitions of the various sources are already provided (they
could be highlighted in a table or sidebar); eliminating that column would leave space for more
details concerning the strategy for dealing for each source - some of these are pretty perfunctory.
Introduction and scope
The discussion of scope and content should be more specific. As noted above, the
current scope and content differ only marginally from the Framework document and don't
provide much, if any, concrete guidance. Interminable arguments over terminology have been a
pathological affliction in the ecological risk assessment community and have significantly
impeded progress. I choose not to add to this problem. The Framework modifications are minor;
any additional changes would be irrelevant. Breaking out the assessment plan from the
conceptual model is a good idea. This material has to go somewhere; shoehorning it into the
conceptual model (as was done in the Conceptual Model Issue Paper) is clumsy.
Risk Manager interactions
-------
Lawrence W. Barnthouse
The guidelines would benefit from a much more thorough discussion of just who/what is a" risk
manager" and about the whole topic of the role of managers and the public hi risk assessment.
This needs more than a sidebar. There aren't many situations where the "risk manager" is a
single person, even within the agency. The Data Quality Objectives Process (EPA QA/G-4)
suggests that representatives of the risk manager should be directly involved in the development
of assessment endpoints and decision criteria. I'll admit that the health risk assessment guidance
is also very weak on this point. The only attempt I know of to determine how EPA makes
decisions concerning ecological risk is the 1994 report on "Managing Ecological Risks at EPA"
(EPA/600R-94/183). This report, which isn't even cited in the guidelines1, provides a number of
concrete recommendations concerning risk manager/risk assessor
Problem formulation
The categorization of assessment types is useful. It could be more aggressively
exploited by identifying specific agency decisions that fall into each category and describing any
associated issues that should be addressed hi program-specific guidance for those decisions.
There's nothing new to say about assessment endpoints and associated measures. It's time to
stop splitting terminological hairs. More explicit examples of risk hypotheses (as in the
conceptual model issue paper) should be provided; the focus again should be on distinguishing
between issues common to all agency assessments and issues that are program-specific.
Analysis
The section on chemical stressors is much too lengthy. Subsections 4.2.2.3 (Estimating
Chemical Exposure), 4.2.3.1 {Estimating Primary Effects), 4.2.3.2 (Extrapolations') and 4.2.3.4
(Secondary Effects') could be reduced to a couple of paragraphs each. These subsections simply
survey available methods and provide no judgements about which should be used in a given
circumstance. Since the program-specific guidance-writers will have to consult the original
literature (which will have changed substantially by the time guidance is written)., it seems
pointless to include an overview of methods in these guidelines. Physical disturbances, on the
other hand, are treated in a highly perfunctory manner. The primary and secondary effect
dichotomy is ill-defined and confusing. More often than not, many ecosystem characteristics are
affected virtually simultaneously by physical disturbances. The only example discussed is the
-------
Lawrence W. Barnthouse
FORFLO modeling case study. The scientific literature on ecological effects of physical
disturbance is large and diverse; a thorough review is needed to support the preparation of
guidelines. The section on biological introductions is somewhat more thorough,, but still
provides relatively little material for prospective guidance-writers. The Simberloff/Alexander
issue paper the USD A reports are much more comprehensive. For both physical disturbances
and biological introductions, the main emphasis in these guidelines appears to have been to show
how these stressors fit into the Framework. This is not sufficient. Again (am I repeating
myself?), the guidelines should clearly identify the scientific issues that should be addressed in
program-specific guidance for each of the agency's major program areas.
All of my comments on physical disturbances and biological introductions are equally
applicable to multiple stressors. In the case of multiple stressors, however, the scientific
literature is scattered, confusing, and not always reliable. My opinion is that the term should be
reserved for combinations of qualitatively different stressors (e.g., chemicals + habitat
disturbance) and should not be applied to chemical mixtures. Multiple stressors are routinely
encountered in retrospective risk assessments and are the rule rather than the exception for all
watershed-level assessments. In hindsight, this should have been an issue-paper topic. An
expert review such as the ongoing ILSI/EPA effort would be extremely helpful; without such a
review the agency can provide little or no substantive guidance.
Risk characterization
This section identifies most of the key issues, but in some respects lags behind the
current state-of-the-science. Uses of the "quotient method" and "physical models and field
surveys" are often substantially more sophisticated than is implied in the guidelines. This is
certainly true in many CERCLA-driven ecological risk assessments. The theoretical limitations
discussed in the issue papers and in the guidelines are recognized by practitioners. Procedures
for screening-level assessments and for weight-of-evidence-based risk characterization intended
to account for these limitations are now widely applied. As with the rest of the document, it is
essential in the risk characterization section that the scientific issues associated with each type of
assessment problem be identified. Simulation models, for example, are (or could be) an
extremely useful risk characterization method in predictive assessments. They could also be
-------
Lawrence W. Barnthouse
used in some retrospective pesticide risk assessments (e.g., projecting changes in raptor
population characteristics due to changes in pesticide use patterns). Such models are much less
commonly used in retrospective assessments of contaminated soil or water. On the other hand,
criteria for inferring causation are critical for retrospective assessments but rarely applicable to
predictive assessments.
The discussion of recovery in section 5.3.2.3 is somewhat vague and confused. A wide
variety of factors influence the recovery of ecosystems from disturbance; there is no point in
attempting to list or discuss them here. This section should clearly identify two critical aspects
of recovery: the potential for recovery (i.e., the reversibility of the predicted or observed change)
and the rate of recovery. Nothing further is needed. Otherwise, the four significance criteria
discussed in section 5.3.2 provides a reasonable foundation for guidance development.
Overall, I would say that the draft guidelines report is a good start but is still much more
a revision of the Framework report than a bridge between the Framework and program-specific
guidance.
-------
Steven Bartell
-------
S.M. Bartell
PART I. GENERAL PRINCIPLES
The draft guidelines represent an important first step towards developing the necessary Agency
guidance towards promoting scientifically credible and consistent ecological risk assessments.
The authors should be congratulated concerning their efforts, yet much work remains to provide
the final document. The following comments are offered towards that end.
There are several issues and general principles emphasized in the draft document. The draft
rightly emphasizes the importance of discussions between the risk manager and the risk
assessors prior to the formulation and design of the overall assessment. The draft also
emphasizes the importance of the problem formulation phase in determining the final quality and
effectiveness of the risk assessment. Flexibility in addressing chemical, physical, and biological
stressors is a major theme in the draft guidelines. Importantly, risk assessment is presented as
an iterative process that provide information for decision making. Finally, the issue of
uncertainty, in all its forms, is addressed repeatedly in the draft as a central component of the
risk assessment process.
UNDERLYING ASSUMPTIONS AND DEFAULT POSITIONS
The draft does not really specify the assumptions underlying various methods and approaches
for assessing ecological risk. The document, in its attempt to provide for flexibility for specific
Agency programs, is quite general in its guidance. In order to become more specific to the point
of identifying key assumptions and default positions, the guidelines would have to assume more
of the character of a "how to" manual. Specific tables of stressors, endpoints, data sources,
methods for exposure and effects assessment, and methods for characterizing ecological risk
would have to be described in sufficient detail (including example calculations) to facilitate their
use in specific assessments. The result would be an immense document that would necessarily
remove some of flexibility advocated as necessary in the current draft.
RECOMMENDED CHANGES
Specific comments will be provided in Section II. However, some general remarks may prove
useful. One, there should be better balance among the sections - the document as it stands is
unnecessarily long. Many issues (e.g., uncertainty) are repeated across sections. The few
pages devoted to risk characterization provide minimal guidance at best. Two, references to the
issues documents for important detail to support contentions made in the various sections does
little to serve the reader, who might not have ready access to these documents. (Also, it's not
-------
S.M. Barte«
quite clear what is meant by changing or modifying the issues paper's contents to meet the
Agency's needs. Three, each of the sections would benefit from a clear statement of the
purpose of the material presented. In attempting to be general and flexible in guiding ecological
•risk assessments, much of the writing come across as unsuitably vague. A general approach to
guidance will prove only generally useful. Unfortunately, nothing happens in general. Four, the
material presented as case studies in Appendix A should be reviewed carefully to make sure the
presentations are (1) consistent with the overall ecological risk assessment framework, and (2)
fairly even in their detail of presentation for each of the sections in each case study.
PART II. SPECIFIC QUESTIONS
Guidelines Balance
1. It's difficult to anticipate future Agency needs. However, considering the present state of the
science, the draft guidelines attempt to provide fair treatment of physical, chemical, and
biological stressors. As reflected in the years of work preceding the guideline draft, much
emphasis is placed on chemical stressors. The sophistication in treatment of the various
stressor categories is certainly biased towards toxic chemicals. Given the effort in
developing "the cube" by Harwell, Gentile, et al., the minimal presentation and discussion of
different levels of organization and ecosystem types in the draft is somewhat perplexing.
Perhaps the cube" might be incorporated into the guidelines to demonstrate the
comprehensive nature of previous discussions of organization, scale, ecosystem type, and
ecological complexity.
2. The use of case illustrations would be improved by replacing the general discussion in the
text boxes with some actual example calculations, i.e., do some risk assessment, at least as
examples. Don't be afraid to show some methods and applications in relation to the case
studies. As suggested previously, the case studies developed in Appendix A would benefit
from better balanced, more detailed, and quantitative presentations.
3. Perhaps examples of risks posed by insect pest outbreaks (e.g., spruce budworm, gypsy
moths), habitat loss and alteration resulting from changes in land use, or risk and recovery
from fire (forests vs. prairies) might serve as useful terrestrial examples.
4. Put all the discussion of uncertainty in a single section, perhaps the risk characterization
section. This would remove much of the redundancy regarding uncertainty from the current
draft.
-------
S.M. Bartell
Introduction and Scope
5. It's not clear that the history and background discussions are particularly helpful to the
reader. The treatments are too brief to accurately reflect the years of effort represented by
the draft. The discussion of risk and decision making does not make the connection between
the quantitative nature of risk assessment and its potential for integration with quantitative
methods for decision analysis. Such a combination might provide a powerful decision
support tool for decision making. More attention could be given to brief description of each
component of the process, e.g., what the reader will encounter in the sections to follow.
6. The terminology issue has not been resolved. As much jargon as possible should be
removed from the document. It only lends a pseudo-intellectual air to the process and
serves no real purpose other than to confuse and obfuscate. The entire process distills into
identifying what is at risk, what are the odds, and what are the consequences. In the least,
the terminology selected for presentation here should be removed to the glossary.
7. The modifications to the figure appear reasonable and useful. Perhaps remove "relevant"
from the data boxes? Also, I'd argue for including the Planning box inside the'Problem
Formulation box. I'd also drop the iteration figure.
Risk Manager Interactions
8. The description is suitably general and makes difficult any real criticism or review. The
collections of questions in box 2-2 seems rather arbitrary. It's also not clear why the text box
concerning sustainability is presented; there are certainly other management goals (e.g.,
complying with environmental laws). Why is the planning outcome not a plan, instead of a
summary of decision (p. 23)? Perhaps the focus should be on the nature of risk manager -
risk assessor interaction in designing the assessment, rather than on the manager and
principles for selecting management goals. Either get much more specific and
comprehensive (maybe identify actual positions of decision making power and the kinds of
decisions entrusted to their care) or greatly abbreviate this section.
9. It might prove useful to mention that one strength of quantitative risk assessment lies in
being able to exploit (i.e., analyze) the various uncertainties associated with the assessment
to identify the critically needed information for refining the assessment in successive
iterations until a decision can be made and justified.
10
-------
S.M. Bartell
Problem Formulation
10. The categorization is not particularly useful. What is the value added to problem formulation
by recognizing such categories? They are sometimes confusing, for example, is an effects-
initiated assessment really a source-initiated assessment, one step removed? Isn't is merely
a matter of focus? The value-drive assessment is also vexing - i.e., whose values?
11. The discussion of assessment endpoints follows fairly straight from the supporting
documents (e.g., the Framework). However, the endpoint definitions (e.g., p. 35: "explicit
expressions..." and "... the valued attributes..") are too vague to be of real use to the reader
- maybe omit them and get right to the "focus of a risk assessment should be..."
12. "Risk hypotheses" should be removed from the discussion of the conceptual model. The
reader might confuse risk assessment with statistical hypothesis testing. Risk assessment
might involve hypothesis testing, but does not require it. The "hypotheses" stated in text box
3-22 are not hypotheses in the strict sense. These statements are really sequences of
deduction, inference, or speculation. The discussion of the conceptual models is, otherwise,
pretty consistent with previous deliberations (e.g., the framework and supporting technical
papers). Figures 3-2 and 3-3 are not good examples of conceptual models (i.e., they are
ecosystem model flow charts); figure 3-4 is much better (it relates stressorto exposure and
possible effects).
Analysis
13. Before discussing stressor types, consider removing the section on uncertainty from the
analysis phase and constructing a single section on uncertainty regarding all phases of the
assessment. Similarly, the discussion of sensitivity analysis should be moved to the risk
characterization section. To balance the discussion of the different stressors, it might be
useful to begin with the necessary components of the exposure profile - then develop
separate examples using the chemical, physical, and biological stressors. Shorten the
discussion of what exposure is and the dimensions of exposure. Note the contradiction
regarding the use of biomarkers on p. 69 and p. 70. Also, simply introduce more examples
of physical and biological stressors to balance the section.
14. Perhaps even more emphasis should be placed on determining the relevant spatial-temporal
scale for the stressor (physical, chemical, or biological), as well as for the ecological effect of
interest (e.g., different levels of organization). The juxtaposition of these scales should assist
in developing meaningful assessment endpoints, conceptual models, and corresponding
scales in assessing exposure. In essence, such considerations should lead to appropriately
scaled (e.g., minimize variance) stress-response functions for use in risk characterization.
11
-------
S.M. Bartell
More comprehensive coverage of different levels of ecological effects could be provided.
Additional tables of the kinds of effects that have been used as endpoints in assessments
would be helpful in guiding the reader. Perhaps the multispecies assays and field
experiments sections could be moved to risk characterization. The section on causality also
needs to be developed in more detail or removed.
15. No comment.
Risk Characterization
The risk characterization section seems not to reflect the effort that entered into previous
discussions and workshops on this topic. Given the important integrating role of this phase of
the overall assessment process, more attention should be given to quantitative concepts and
measures of risk (i.e., the quotient does not characterize risk), risk estimation, and addressing
uncertainty - with perhaps less attention on weight-of-evidence discussion.
16. Evaluation of risk principles in the draft guidelines depends in part on the interpretation of
risk assessment and risk estimation. The draft is a suitable start if one adopts a broad
definition of risk to include qualitative risk assessment. If one insists on reserving risk
assessment to refer specifically to quantitative assessments that explicitly address
uncertainty (e.g., a probabilistic framework), then much of the section on risk
characterization, particularly risk estimation, needs considerable revision. The first of the
principles offered on p. 114 is an unsubstantiated generalization (e.g.,"... a simple approach
... is more credible than a complex assessment...'). If the approach is simple, yet insufficient
(i.e., quotients), the accuracy of the data is a moot point. The second and third principles
don't really seem to be principles of risk characterization, as much as they are advice and
cautions regarding problem solving in general. Given the importance of risk estimation in
this phase of the assessment, perhaps more of the information included in Wiegert and
Bartell (1994) should be included directly (e.g., maybe in text-boxes ortables), ratherthan by
reference. Note that the first bullet under the quotient discussion seems to contradict the
advantages outlined in the preceding paragraph. On p. 118, what is meant by the sentence,
"Simulations can be applied to regression analysis to determine cause and effect relations."
On p. 119, replace the term "risk hypotheses" with assessment endpoints? Also, why would
exclusions in the assessment be addressed in the characterization phase instead of the
problem formulation phase?
17. The current draft identifies the following criteria for ecological significance: nature and
intensity; scale; recovery; and natural variability and disturbance. It might be worthwhile to
12
-------
S.M. BarteH
note that an ecological response can also be scaled in relation to the overall life history
characteristics for species (population) level impacts. The dynamic nature of systems,
mentioned under the recovery section, also refers of course to natural variability and
disturbance. It is important to emphasize that recovery is best assessed in probabilistic
terms; recovery complements risk in ecological risk assessment.
Miscellaneous Comments
Time and resource constraints preclude a detailed, point-by-point review of the draft document.
Nonetheless, several observations and comments might assist in revising the draft guidelines
document:
1. The guidelines should be written under the assumption that the reader will not be intimately
familiar with the history and development of ERA process that led to these draft guidelines.
Therefore, ecological, toxicological, and regulatory jargon should be kept to an absolute
minimum (i.e., see the first bullet under clarity, p. 127).
2. The guidelines should provide a clear picture of the process of ecological risk assessment -
emphasizing the purpose and produces) of each phase. The current draft loses the train of
thought as the result of introducing a plethora of issues, considerations, caveats, strengths,
and limitations, etc. Diverse ecological and toxicological phenomena, observations, and
generalizations seem selected and interspersed throughout the sections without any common
thread or theme. The net result does not engender so much a feeling of generality and
flexibility, as it does confusion and bewilderment! It might be more effective to select one
or two very different assessment problems (i.e., case studies) and "walk them through" the
assessment process. The same set of important issues can be raised, but with a constant
frame of reference for those who might not be so broadly trained and experienced as those
who have participated in the long evolution leading to the draft.
3. Replace most of the general text boxes with more specific examples and calculations based
on a small set of case studies (see above). This information should provide more detail for
the interested and technical reader instead of being merely additional text that could have
just as well been incorporated into the main body of each section.
4. Some of the discussion is so general as to be essentially truisms that provide minimal
guidance, either in concept or method, to potential risk assessors and managers (e.g., p. 42:
" The assessment of multiple stressors..." and p. 43: the best assessment endpoints...").
5. Is the conceptual model really a Verbal description" (p. 44)? Documentation is offered
elsewhere as an important aspect of the overall risk assessment process - why verbal here?
13
-------
S.M. Bartell
Also, the statement, "hypotheses are assumptions..." (p. 44) poses real confusion to risk
assessors and is a good reason to omit the use of "risk hypotheses" throughout (see also
text-box 3-12). Hypotheses are not assumptions!
6. The modifications made to the original risk assessment diagram appear beneficial. Yet I fail
to understand the need to point out the modifications to those who might have never seen
the original.
7. Put the uncertainty table (e.g., Table 3-1, et seq.) in one place, one time. The examples are
not particularly compelling or enlightening. Just focus on the nature and source of
uncertainties and provide some clear-cut examples (maybe even some basic statistics) in
relation to a few well developed case studies, as mentioned above.
8. Remove Figure 4-4. Omit lines 13-22 on p, 69.
9. Remove text-box 4-5.
10. Remove first sentence on line 13, p. 74.
11. What is the difference between adjunct material and examples? (lines 18-23, p. 75).
12. Leslie models do not extrapolate effects from individuals to populations (text-box 4-7).
13. The secondary effects section needs more basic conceptual development (p. 84). Much of
the following section seems more appropriate to risk characterization than analysis of
secondary effects.
14. The stressor-response profile should be discussed in greater detail, particularly from the
viewpoint of biology, ecology, and endpoint selection (p. 86).
15. Why are only "clean" sediments a physical stressor (p. 88)?
16. In Table 4-3, why use "extirpation"? Do you really mean local extinction?
17. On p. 90, lines 21-26 would make a good lead-in to section 4.3.2.
18. Why put the bottomland hardwood information in text-box 4-9 if it is better presented in the
Appendix?
19. Delete lines 1-2 on p. 93; they are redundant.
20. The "delphic" approach is mentioned several times (e.g., p. 97), but is not well described. It's
also not in the Appendix B glossary. Maybe provide a text-box description and example?
21. The selection of key terms seems rather arbitrary. Several of the definitions (e.g., for
community, ecosystem, ecological component) are too superficial to be helpful.
14
-------
John Bascietto
15
-------
John J. Basciettb
REVIEW OF THE U.S. ENVIRONMENTAL PROTECTION AGENCY'S DRAFT
PROPOSED GUIDELINES FOR ECOLOGICAL RISK ASSESSMENT, EXTERNAL
REVIEW DRAFT, OCTOBER, 1995.
REVIEWER: JOHN J. BASCIETTO
U.S. DEPARTMENT OF ENERGY
WASHINGTON, D.C.
The U.S. EPA's Risk Assessment Forum staff should be commended on drafting this next
phase of EPA's effort to produce ecological risk assessment guidelines. While the guidelines
development effort has spanned many years (1987-present), it has been carried out under
sometimes difficult circumstances, and always with insufficient budgetary and manpower re-
sources.
\
\
The following comments are offered in the spirit of constructive criticism, with the best
intentions. I hope these comments help to make the ecological risk assessment guidelines
successful, and useful to a wide variety of stakeholders.
PART 1. General Comments
EPA has provided several disclaimers concerning the scope and intended use of the
guidelines: that they are intended "primarily" for use by EPA staff and EPA .programs; that
they are not a rule book, nor a cook book or a textbook; that they address only general
principles, and do not contain extensive background materials for novice readers. EPA
should realize, however, that like it or not, these guidelines will instantly become the de facto
standard for ecological, risk assessments performed in this country, and perhaps abroad. This
should have been evident from the enormously positive reaction to EPA's 1992 report "A
Framework for Ecological Risk Assessment," and its almost universal adoption.
16
-------
John J. Bascietto
It is clear that these guidelines do not constitute a manual on "how to do" ecological risk
assessment. An experienced risk assessor will see that the results of approximately twenty
years of negotiated ecological risk assessment science policy is reflected in the equivocal tone
of much the current draft. The discussions on the underlying science of ecotoxicology, the
vast array of science policy choices and the discussions on how to handle uncertainty, all with
a special emphasis on the interface between risk assessor and risk manager, do indeed reflect
the lessons learned from twenty years of operating without the benefit of the guidelines.
The EPA has characterized these draft guidelines as "not a textbook." I'm not sure I
agree. The document is certainly pedantic enough, and in this reviewer's opinion, could indeed
serve as manual for an advanced undergraduate or graduate student taking a beginning course
in ecological risk assessment. The current draft does a good job of describing the underlaying
principles used by many ecological risk assessors, particularly if they are working in a
regulatory context (I was struck with the comprehensive nature of the discussions), but it also
disappoints, in the sense that the document barely lives up to its title - it-presents precious
few guidelines, per se. The draft serves to expand on the 1992 "Framework" document in
great detail. The authors' stated intention to do just that was indeed fulfilled.
Nonetheless, with a few welcome exceptions, I found little in the way of ecological risk
assessment "guidelines" that could be of immediate use to experienced assessors who are
engaged, with EPA regional and state counterparts, in the difficult questions of assessment
design, endpoint selection, sampling and analysis, and hypothesis testing. Ecological risk
,•
assessors in the U.S. Department of Energy could certainly benefit from these proposed
guidelines, but of more immediate use, DOE would like to see much more concrete, consistent
and specific information from EPA about what is expected, what is required, what is
acceptable, and what is not acceptable in risk assessment documentation. This is where the
guidelines are most needed, and they need to apply equally to the regulators as well as the
regulated community.
17
-------
John J. Bascietto
I should probably state my own bias, preferences and expectations for the current
manuscript. In this reviewer's opinion, a "guideline" clearly spells out limits or boundary
conditions. I expected.that these guidelines would integrate and synthesize information from
the two earlier volumes of published cases studies of ecological risk assessments, and the 1994
report, "Ecological Risk Assessment Issue Papers."
A guideline should not be an inflexible rule, but should .provide a discrete method and
context for moving through a systematic process (such as risk assessment), even as new data
and hypotheses regarding the process are being formulated. Speaking as one who has devel-
oped EPA ecological risk assessments without the benefit of guidelines, this reviewer is
concerned that the lack of even qualitative ecological risk assessment guidelines could be (and
has been) used as an excuse to dismiss significant and serious investments of time and effort,
out of hand, as mere speculation.
At this point in time, given the enormous resources at stake in many environmental
compliance programs and with pressures from legislators for more reliance on risk-based
decision making, there is clearly a need for EPA to issue unambiguous statements of what is
acceptable and what is not, in a risk assessment. Default positions are needed to help risk
assessors move through controversial areas of science policy. Such policy statements should
be clearly labeled as such. Default positions should be rebuttable by proponents of an
alternative hypothesis, provided that they offer new, problem-specific, or stressor-specific data.
Particular risk assessment problems could be addressed in specific guidelines (e.g., a
population risk assessment for acute toxicity of pesticides to birds; or a community risk
assessment for a freshwater stream receiving releases of uncontrolled hazardous wastes, taking
into account acute, subacute and chronic effects). I suggest that specific guidelines could be
prepared for or by EPA Program and Regional scientists, and reviewed and finalized by the
Risk Assessment Forum.
As a follow up to the Framework, EPA published two volumes of case studies of specific
18
-------
John J. Bascietto
ecological risk assessments. This presented an opportunity to write the guidelines for the
types of risk assessment problems illustrated by the case studies. EPA instead chose to use
the case studies to see if the Framework was appropriate for these types of exercises. Some
parties suggested that the case studies could also be used as interim guidelines until the "real"
guidelines could be produced.
It is not my intention to discourage further use of this draft report. It is a very good
.catalogue of potential issues that one will run across in the course of one's employment as an
ecological risk assessor. EPA has now certainly done more than enough on the "process."
Perhaps EPA could consider the current draft as a preamble for the next iteration, which
should be supplemented with some specific guidelines, as I've indicated. For example, the
conceptual model section of the current draft was very helpful, but should have been
integrated, more closely with the risk hypothesis formulation section, which itself needed
further development. The detailed discussion integrating these two sections could have been
the subject of a guideline, and focused on a particular type of ecological risk assessment.
EPA should consider providing more detailed information on default positions when
appropriate, and should address three specific questions in any guideline: what is in bounds?,
what is out of bounds?, and when is it permissible to step outside the bounds of a presumptive
(default) approach? EPA should integrate the Agency's science policy decisions (based on the
issues raised by the "Issue Papers" report) into the next interation of the draft guidelines.
Initial efforts could center on dose-response methodologies and modeling, risk hypothesis
formulation, endpoint .selection, hypothesis testing, use of computer codes, risk estimation
methodologies, and uncertainty analysis.
19
-------
John J. Basciett'o
PART II. Specific Questions and Comments on Sections
Guidelines Balance
This draft attempts to cover too many bases, in ibis reviewer's opinion. Considering the
present state of the science and the needs of the Agency and its regulated community, EPA
should consider postponing development of the biological stressor guidelines at this time.
EPA should instead focus on population, community and ecosystem effects from chemical and
physical stressors. t
There appeared to be a uneasy balance in the treatment of levels of biological
organization. There is not enough specific discussion of individuals, populations and the
community level issues. EPA should, and' has begun to explore the ecosystem level issues (to
the extent that they are really ecosystem issues and not really community level issues).
The temporal/spatial .scaling issues had better balance than the biological organization
issues. This is acceptable, since the latter are usually much more important to actually
performing an the assessment, once the endpoints have been selected.
(I've covered the remaining questions for "Guidelines Balance" in the body of my Section-
by-Section comments).
Section 1 - Introduction.
The purpose of this discussion was not at all clear to me. In addition to needing an editor,
the ideas should flow like a good story, from beginning, to middle to end. It is confusing for
EPA to cite specific definitions for terms from previous EPA documents, implying that they
are obsolete, without also stating or referencing the current, official EPA definition of that
term. For example, I could not tell what the current', official EPA definition of an ecological
risk assessment is!
20
-------
John .J. Bascietto
It is particularly confusing to split the perfectly good term "measurement endpoint" into
three new terms, each with an only a marginally different definition from the original, given
the inclusive context of the term "measurement endpoint" in the Framework report. It seems
to me that very little value was added by the splitting the term into "measures of exposure",
"measures of effect" and "measures of ecosystem characteristics." The point is that you either
can measure this thing, or you cannot! I took this as a warning sign that EPA may be getting
too focused on the "process" of risk assessment, and perhaps not focused enough on the
substance.
Section 2 - Planning ...Assessor and Risk Manager.
The stated purpose for the assessment is too narrow. It not only includes bringing assessors
and managers together, but also some very important 'stakeholders, and the public in general.
Recognize that "planning" is not a step in the assesment, per se. It is only distinct from the
risk assessment process on the Framework report's diagram for purposes of clarity of
communication. In reality, the process and the players should form a team, and continually
"plan" for contingencies (see my comments on data quality objectives below). Recognize also,
that it is very difficult to overcome stakeholders' perceptions that "government policy" and risk
"managers" bias the scientific analysis. Such perceptions can only be overcome by constant,
conscientious efforts to involve the stakeholder in a meaningful way, clear communication of
"the rules of the road" (i.e., the guidelines), combined with a rigorous adherence to good
laboratory and field practices, plus a thorough independent review process.
I recommend adding a clear and rigorous process of Data Quality Objectives (DQO)
development to the process guidelines. Both the risk assessor and the risk manager need to
participate in.setting DQOs, as well as the other stakeholders. The planning section tries to
get at this notion, but seemed to develop only a cursory discussion of DQOs. The guidelines
should spell out a formal DQO process.
DQO's are designed to answer the following questions:
21
-------
•Tohn J-. Basciett'o
o What is the problem we are trying to solve? - (management goal)
o What decision are we going to make ? (action alternatives?)
o What data is needed ? (assessment and measurement endpoints)
o What are the study boudaries (conceptual model, spatial and temporal
considerations)
o What rules will be used to make the decision ? (risk hypothesis, statistical
limits design)
o What are limits on the decision error? (probability of Type I and Type II
errors)
o What is optimum study design? (trade off between precision of the decision, v. cost of
the sampling and assessment design)
Ideally, the guidelines would specify DQO's for each type of risk assessment. However,
for certain ecological risks, DQOs development may be very difficult. For example, it may
not be possible to quantify decision error limits if EPA is unwilling to specify which endpoints
should be used when multiple endpoints could be used; or if acceptable reductions could not
be specified for a population risk assessment.
Risk assessors should not allow risk managers and political officials to unfairly put the
burden of a developing a "magic bullet," (also known as "risk-based decision making") on the
assessor. Government agencies and responsible public officials need to know that if they are
unwilling, or unable to pay for needed systematic and defensible data collection and risk
analysis efforts, they should not expect then* decision-making to be "risk-based." The DQO
process is a very effective way to establish a systematic approach to risk assessment and risk
management, and prevents the "raising of expectations" of public officials and the public,
beyond our ability to deliver defensible results.
Section 3. Problem Formulation Phase.
While I very much enjoyed the discussions on the initiation of risk assessments (i.e.,
22
-------
John J. Bascietto
source-, effects- and values-initiated), I was left wondering how this information helps me to
develop an ecological risk assessment, and why EPA spent so much time describing it. It
seems, to me that one ends up in the same place, regardless of how one got there: conceptual
models and risk hypotheses need to be developed, endpoints selected, assessment methods
agreed upon, data collected, massaged and described.
Based on the text box and other written descriptions of the case studies, I could think of
several ways in which a so-called "effects-initiated" assessment of granular carbofuran could
have been construed to be a "stressor-initiated" or "values-initiated" assessment. Likewise,
the Spotted Owl assessment (ostensibly an "effects-initiated" assessment) could be construed
as a "values-initiated" assessment. In fact, regardless of what the case study says, the actual
Special Review of granular Carbofuran was supported by a "stressor-initiated assessment"1, not
an "effects-initiated" assessment. The time taken to split intellectual hairs would have been
better spent developing formal DQOs.
I very much appreciated the discussions on Management Goals and the cascading of goals
to objectives, assessment endpoints and measurement endpoints. These are real-life risk
assessment issues which deserve this kind of detailed reflection. Please continue to refine and
expand this aspect of the guidelines. These issues should be clearly hashed out by risk
managers in planning sessions with risk assessors, but are not necessarily obvious to everyone
involved, including the risk managers. They are prerequisites to DQO planning.
I'm not-sure whether there is a great deal of value-added information in the continuing
efforts to refine the Framework report's process diagram. The diagram (and the process)
probably does not need expanding and refining. This does not actually help risk assessors do
their job. Again, spend more time developing ecological DQOs.
•"•The granular Carbofuran case study published by EPA and cited in these draft proposed guidelines has
apparently been rewritten at least three times subsequent to the original risk assessment used by. the OPP staff to
initiate the Special Review. The original was a stressor-initiated risk assessment, which met the toxicity-based
regulatory criteria to initiate a Special Review of a pesticide, at the time the review was initiated (1985).
23
-------
John J. Bascietto
I very much enjoyed the discussions on "Ecosystem Potentially at Risk." This and related
sections need considerable refinement and editorializing (there is an inappropriate emphasis on
geographic location, Tather than ecosystems as "systems." One point however: the current
revision of the Framework process diagram, Problem Formulation Phase (Fig. 3-1) makes it
appear that every assessment should have an ecosystem component. That interpretation of the
figure is probably not what EPA intended, nevertheless, its confusing.
I was quite surprised to find (in Sec. 3.4.1.3, "Susceptibility to the,Stressor") a familiar
discussion of the appropriateness of using particular species as assessment endpoints. An
example is used in which turkey and deer could be selected as assessment endpoints because
they are good surrogates .for other species in terms of sensitivity to a particular contaminant.
This scenario very closely parallels an actual assessment endpoint issue at a DOE site in the
Southeastern U.S. The dispute unfortunately reached an disproportionate level in my opinion,
with both sides taking strong opposing positions. This is why I find it rather disturbing to
read about the issue here. I hope EPA is not trying to make a point in these guidelines. This
is not the place to win an argument.
Furthermore, I did not see that the 'turkey and deer' example well illustrates the point being
made - i.e., 'should deer and turkey be included as assessment endpoints because they well
represent the sensitivity of ungulates and birds to a chemical contaminant?' The example,
which indicates that deer and turkey may not be good assessment endpoints because they may
not live at the site, fails to consider the very real possibility of off-site exposures to these
critters, which clearly would have been a concern for DOE and the other. Trustees. In the
DOE case, sensitive endpoints of relevance to both the Trustees and EPA's programmatic
goals should have been allowed to be pursued.
Section 4. Analysis Phase
There are many good issues raised in this section, and much good information that will be
of use to assessors.. This discussion clearly comes the closest to providing "guidelines."
24
-------
John J. Basciettb
Rather than hash but each opportunity in the section when EPA should have gone the extra
mile to a guideline, I suggest that the reviewers and Agency personnel may want to discuss
this effort at the workshop. Generally, the discussions here were well thought out, clearly
presented and well supported. There is ample bulleted and tabulated information here, and
also available, in the literature, for EPA to begin to make the tough science policy choices that
will be needed for the guidelines. EPA should make these choices in the next iteration.
I refer the authors to a new (1995) book, by Dr. Michael C. Newman, Savannah River
Ecology Laboratory, U. of Georgia, entitled Quantitative Methods in Aquatic Ecotoxicologv
(Lewis Publishers), which contains an excellent discussion of the calculation, use and
interpretation of biological indices, particularly useful for community level assessments.. These
measures deserve another look by EPA.
Section 5. Risk Characterization
Although I agree that the discussion of the hazard quotient has to be presented in this
section, EPA, may want to consider more clearly defining the hazard quotient in terms how it
"" / ' •
differences from other measures of risk, per se. Ideally, one should estimate the magnitude of
the effect and the probability of that magnitude materializing in order to assert "risk." Since
the hazard quotient does not do this, EPA may to reconsider such statements as "the higher the
quotient the higher the risk."
In the uncertainty analysis section (5.2.4), EPA should include an explicit guideline for the
risk manager to indicate precisely how the uncertainty was interpreted and how the
interpretation effected the decision;
Some of the bulleted information requested to be discussed in the Risk Description (Sec.
5.3), especially with respect to data sources and conceptual models, would have already been
discussed in the previous phases of the risk assessment, and therefore may not need to be
repeated here.
25
-------
John J. Bascietto
Does EPA really think it is necessary to discuss "Human Error" (Table 5-1), or that it will
be a significant contributor to the uncertainty? Can't a good editor mitigate some of the human
error?
Section S..3.2.1, Nature and Intensity - What are we going to do with this-information?
In the discussion of ecological significance, it should be pointed out (but not in a
guideline) that "change" in nature is constant and not necessarily "bad." There is change in
the Darwinian sense, and then there is the kind of change that we notice. Even rapid change
might not be "bad" if we all agree that we like the results. However, even "good" change
may not be acceptable from certain stakeholders' point of view.
The elimination (or reduction in numbers) of one or more species opens niches to other,
perhaps exotic species. Exotics are probably not desirable as a general proposition. I suppose
it depends on your management goal. In any case, the real the trick will be carrying these
types of issues and predictions through the risk assessment. Maintenance of suitable,
preferably native habitats, and a reservoir for reintroductions are likely to be more important
than avoiding certain population reductions.
The EPA suggests that "the evaluation of-recovery is made more difficult because
ecosystems are dynamic and will not return exactly to a preexisting state." While this may be
true, I'm not sure that we need to worry about it. To the extent that this concern reflects a
Management Goal (i.e., restoration of preexisting state) it's probably doomed to failure
anyway. The more significant ecological" value should focus on restoration of ecological
services and the their implications for biological resources. This is the concept ("restoration of
natural resource services") used by natural resource trustees to restore (not necessarily
replicate) important structures and basic functions to a system, in lieu returning a resource, per
se to its pristine state.
26
-------
Nancy Bettinger
27
-------
The Draft Pro/pawl Guidelbu* fi>r> EaJagitol Rial A^nm^t
-------
measurement error arises from variation in the characteristic of interest Further,
the table suggests constructing probability distributions as a strategy for addressing
measurement error. This presentation confuses measurement error with variability
in soil concentrations.
The Guidelines should provide a more extensive di§cus§i«?a of the implication* of
different types of uncertainty, For example, risk'as»sesora and risk managers should
recognize that, in many cases, risk assessment is done tem*3§, of lack of knowledge,
not just in spite of it, Thus, tempering risk estimates with descriptions of uncertainty
due to lack of knowledge may be circular logic. The risk managers should be grren
a f\ill appraisal of all types of uncertainty associated with the as*££*m*at, Ibut
different types of uncertainty may have very different meaategs for risk management
decisions.
Problem Formulation
Question 10. The categorization of assessments appears to be very useful. It is an
improvement on the predictive/retrospective categorizsti«» scheme becaust it
emphasizes consideration of the available information, It provides a basis for applying
general principles consistently to a variety of assessment problems, while encouraging
the risk assessor to apply a process that best fits the type of assessment in question.
Question 11. The discussion of assessment endpoints and their relation to
management goals is clear, but an expansion of Idxe guidases on applying the endpoint
selection criteria would be helpful. Specifically, the Guidelines should indi
-------
potential assessment endpoints, the Guidelines should indicate which, criteria should
be given more weight when selecting assessment endpoints,
The discussion of relationships between assessment endpoints and measures of tfftct
should he expanded to include guidance on comparing the utility of various measures
of effect* for evaluating an assessment endpoint. The strength of association betwoen
the measurement result and the assessment endpoint, anticipated data quality, imd
study design issues such as spatial representativeness should be addressed in detail
The Massachusetts Weight of Evidence Workgroup has identified measuremtnt
endpoint attributes that determine how well the measurement estimates or predict*
the effect defined by the assessment endpoint (Draft Report - A Weight-of-Evideioce
Approach for Evaluating Ecological Risks), Although the attributes were initwilty
developed to evaluate weight of evidence in risk characterization, the Workgroup itlso
concluded that the same attributes were equally valuable for selecting measurement
endpoints (or measures of effects) in problem formulation. These attributes may be
useful as a basis for EPA guidance on developing criteria for selecting measurement!
of effect.
Question 12. The guidelines should describe in more detail the relationship between
hypotheses and assessment endpoints and measures of effects. They are discussed
separately in the text, even though they are closely linked.
Analysii
Question 14. In Section 4.8.2.1 under Chemical Stressors - Estimating Primary
Effects, the texts states that in hazardous waste sites, field studies are often an
integral part of ecological risk assessments. The Draft goes on list a number of
reasons for using field studies in risk assessments, but only briefly mentions the
Ctmsmntt MWIttid by Naacy Button*. Hmirtumtt Mg/OBo vttmtitr&ui lt»a
-------
importance of appropriate reference sites and statistical power. In our brief
experience, the lack of availability of appropriate referenda sites and study design
deficiencies associated with low statistical power (sm§U sample size, for example) often
lead to field study results that are less than compelling, These drawback!, howwwr,
seldom seem to discourage investigators from conducting field studies and placing
relatively high weight on the results. Guidance on deciding when a field study may
not be appropriate would be helpful, as would examples to illustrate the effect of hiigh
variation and tow sample size on statistical power.
Riak Characterization
Question 16. In the section on Weight of Evidence, the relevance of the evidence, to
the assessment endpoint is mentioned very briefly, This discussion should be
expanded substantially. The Massachusetts Weight of Evidence Workgroup !aas
recently produced a draft report on the results of the workgroup's efforts, which raay
be useful as a starting point for expanding EPA's weight of evidence guidelines the Draft Guidelines indicate that localized contemplation
may be significant if the impacts translate into landscape or ecosystem-level effects.
In several other places, the Draft Guidelines indicates in several places (Section 3,3.2
title and first paragraph) that ecosystem effects should be the focus of most ecological
risk assessments. The implication is that assessment endpoints should represent
ecosystem level effects, or that the results of any asgsesmsnt should b« extrapolated
to the ecosystem level However, risk assessments are often conducted for stresiiors
_/ mm_i__n inniiPiiii !•._" .__-^~.-:—--- -.-:=.•"-•-•••-^uilurjf.^"-~~'WmiU v •-r-n"" n « i \tm—
CeoMttt* NlfeRlttel iyNuuy BrtttnflW, MiMMkuMlti MP/Oflk» of StttMife Mi Studs*
31
-------
that are likely to affect \sn\y a small area and ar« unlikely to affect an ecosystem. It
seems a waste of resources to focus an assessment of localised effects and ttten
temper any positive results by a judgement that those effects are unlikely to translate
to ecosystem level effects. If the agency considers effects «*g*/Om«» of Swank u4 ataadwb
32
-------
Joanna Burger
33
-------
JOANNA BURGER
JQANNABURGER
GENERAL PRINCIPLES
GUIDELINES BALANCE
1. The guidelines are generally well-balanced. However, I
feels that there is still an overemphasis on chemical
stressors, and on individual and population parameters rather
than community or ecosystem-level interactions. I would add
(rather than delete) these aspects.
2. My copy did not have Appendix A, nor did it have the
discussion papers. My other concern is that most of the
references are to the discussion papers, rather than to the
original papers. This leads to too many chemical examples,
rather than a balance. It might be useful to have the
committee come up with some other references that use more
biological and physical stressors.
3. I agree, the document as it now stands relies too heavily
on aquatic systems, which in the past, have largely taken a
single-species approach. Thus ecosystem approaches are given
less time, as are terrestrial ones.
4. It is useful to have the uncertainly issues outlined in
each section. However, it might also be useful to pull them
together in an Appendix so the assessor could get a complete
picture of the problem.
INTRODUCTION AND SCOPE
5. More emphasis should be placed on the importance of
examining ecological risks to ecosystems and ecosystem
function and to mixtures. Managers and the public will be
interested in this aspect. Quality assurance-considerations
should be listed up front. The importance of evaluating
whether the risk assessments were accurate should be included.
We cannot get to a predictive stage if we do not evaluate our
34
-------
JOANNA BURGER
past risk assessments.
6. I feel that assessment and measurement endpoints should be
clearly defined up front, in this section. They should both
be defined, and have clear examples. This is especially
important for managers and for the public. Although you refer
to sources for definition, these are very important points,
and the reader will find it easier if they are here.
To the definition of disturbance, Page 15, you might add
processes. Processes are important endpoints for ecological
disturbance. To the first sentence in section 2.3: I think
that ecological value should be defined, perhaps in a box with
some examples.
7. None
Other additions and suggestions to introduction and scope
a. page 2; sometimes new data might be required, and this
should be incorporated in the process.
b. page 3, under 1.4': it might be useful to add "evaluating
competing risks" to the initial sentence.
page 3, last sentence: it might be useful to refer to dose-
response here, since many risk assessors will want to see just
how this fits in even though a decision was made not to use
it. We still must deal with the real world, and should be
able to talk to human risk assessors
Page 3-4: in the bullets, I feel we should add a phase for
evaluating the efficacy of the risk assessments (after Burger
1994)
Page 8: figure should be enlarged slightly for readability.
Page 9 (top): it might be useful to have a box that
delineates some of the landscape-level effects you are talking
about since this may be the phase most unfamiliar to risk
assessors.
Page 12: Need to add something about data quality
assurance. '
35
-------
JOANNA BURGER
Page 13: first bullet: may need to give an example or two.
RISK MANAGER INTERACTIONS
8. The most important thing that is missing from the
discussion is stakeholder participation. Although it is
implied that the manager will have this in mind, and there are
several references to the public, I think this needs more
inclusion.
In this same vain, stakeholders should be clearly defined
and enumerated. For example, stakeholders other than those
living in the immediate vicinity should be considered. For
example, when considering logging in the northwest, the
American public outside of the immediate area was very
interested in the outcome, and had a stake since these were
public lands (in many cases). This might well need a box.
9. Environmental justice considerations should receive more
attention.
Additional comments on this section:
Page 17: I would consider adding stakeholder to box 2-1.
Stakeholder is absent from most of this section; and this will
become increasingly important over the years.
Page 19: Top sentence; I might add commercial uses to the
public perspective.
Page 19: In addition to risk managers and risk assessors
coming to agreement on goals and scope, they should also agree
on the timing of the process.
Page 20: to box 2-2, the question: How soon will recovery
occur should be added.
Page 21: Box 2-3: I am still uncomfortable with
sustainability and its definition.
Page 21: to the first sentence under 2.4, I might add
endangered species as an example of implementation of
management decisions. It would be good to have a non-chemical
36
-------
JOANNA BURGER
stressor here.
Page 25: To the data acquisition phase on the right of the
diagram I would add quality assurance instead of verification
because it includes more factors.
PROBLEM FORMULATION
10. Overall I think these terms are confusing as presented,
or at least will be to managers when presented with them. The
distinctions are excellent, they just need more examples.
I do not agree that value-initiated assessments are
conducted at the landscape scale. Many of our value ones are
based on endangered species, or appealing ones that people are
concerned about. ,
11. In discussing assessment endpoints, it would be useful to
distinguish between "indicator species" and assessment
endpoints - perhaps a box would do nicely.
12. I would add a couple of more models that show conceptual
models for real systems., Burger 1995; A risk assessment for
lead in birds (J. Toxicol. and Environ. Health) has one for
lead that might help for some risk assessors. Another one or
two of different types would be good.
Specific comments
Page 27: It might be a good idea to give some physical or
biological examples in addition to the chemical (middle of
page)
Page 27: Box 3-2 is less clear than some of the others.
Page 28: May need to expand the bullet to give some examples
It might be useful to have a box for both physical and
biological stressors.
Page 28: 3.2.3 It might be helpful to add endangered or
appealing species to the values.
Page 30: Need to expand here on the stakeholder
participation, since this is critical to value-initiated
37
-------
JOANNA BURGER
goals.
Page 31: I might also include a species level box (such as
for Spotted Owl, one managers will relate to).
Page 33: Add half life of chemicals
to box 3.5
Page 34: Might want to add endangered species or those of
special concern to box 3-6.
Page 37: Need to mention indicator species, since this seems
like what you are discussing in many ways.
Page 38: the first sentence is confusing.
The last sentence of line 20 is not a complete sentence.
Page 39: ecological entity needs to be defined,,
Page 40: Assessment endpoints are not management goals: this
needs a bit more expansion. It is an important point.
Page 42: Box 3-10 makes little sense as it stands. It needs
to be more readable on its own.
Page 44: Risk hypotheses need more boxes;the one given is
good. More would help
Page 44: add "ecological relationships " to the list of
bullets at the bottom
Page 46: Need to have a box to show relationships of more
conceptual models. Burger 1995 for a "Risk assessment for
Lead in Birds" (J.of Toxicol and Environ. Health) has one
that applies to the real world that might be good.
ANALYSIS
13. I think the balance is fine, might have more biological
stressors since this aspect has been largely ignored by
ecological risk assessors.
14. I think there needs to be a discussion about dose-
response information from the laboratory, what kinds of dose-
response data there are from the field, and how to interpret
them. Also, how to deal with the lack of such data.
I also think it would be good to get some examples from
38
-------
JOANNA BURGER
the main ecological literature, instead of only using
Barnthouse and Bartell examples. This will vary the interest,
and make it more relevant for other readers. I can provide
some if you wish.
Multispecies approaches (page 79) are NOT measuring
community level effects. They may indicate problems at
different trophic levels, but community level effects can only
be measured by doing so.
15. None, it is fine.
Additional comments for this section
Page 58, middle of page: I would add quality assurance to the
carrots.
Page 59: I think you need further definition of secondary
effects (line 18) . For many risk assessors, this will be a
new thought
- line 26-8: Might need to add life stages. That is, some
organisms (such as frogs) live their life in very different
habitats (aquatic vs terrestrial) and these should be
included.
Page 61: (line 22). I wonder if error in selection of
appropriate endpoints should be mentioned.
Page 62: Table 4-1: Under Variability: I would add life
stages and allometry.
Page 63: Considerations of power should be done before any new
field data are collected.
Page 67. line 1: might add or stressors that interact.
Line 5: not only concentration, but species or form of
contaminant.
I did not'find Box 4-4 useful on its own.
Page 68: line 1-3. Might be useful here to give an example.
For example, the presence of selenium decreases the uptake of
mercury.
Line 19: Might need a box to define "reference"
39
-------
JOANNA BURGER
environment. This is a term that is not equally used in all
communities. Yet it will be used more in the future.
Page 70: line 4: may need to be compared across systems as
well.
line 10-12: May need a further explanation of the
relationship between dose and bioassays. In my experience,
there are not enough studies to examine these relationships.
Page 71: Need to add some discussion concerning
toxicodynamics, and the difficulty of determining dose from
tissue levels in different tissues over time.
Line 16: Need to add life stages and life history
information
Line 20-22: Need to add secondary behavioral effects.
Page 72: Box 4-5 is less useful than it might be.
Page 73: top: Might need to discuss thresholds here briefly;
and have a box explaining how they relate to risk assessment.
DDT and eggshell thinning might be a good example.
Line 24: Might want to summarize the usual ones.
Page 74: line 11-14: May want to include some reference to
dose-response data.
Line 29: Might want to add "and how well they represent
time, space and life stages variations"
Page 79: Line 21: I disagree that multispecies assays evaluate
community level effects. They can be evaluated only by
measuring community level effects (there are such measures,
such as productivity, nutrient cycling, species composition)
Page 80: TOP: Need to test the assumptions of extrapolations
as well.
Page 83: line 14: and in different species, should be added.
Line 27: Can find good studies to reference here (see
Burger and Gochfeld 1994, 1995 - call if you wish reference).
These studies conducted dosed studies in the field with wild
birds.
Page 84: line 10: need references to, secondary effects for
40
-------
JOANNA BURGER
general reader.
Page 86: line 27: add habitat stages (since some may be in
water, others on land).
Page 89: You might add Beach Nourishment as another
disturbance to give an oceanic one.
Page 93:line l:you might add daily variations.
Line 20: You need to define and give a reference for the
Habitat Suitability Index work.
Page 94: Analysis of Biological Introduction: I wonder if
Extinction should- be added here. We might want to do risk
assessments for how extinctions would affect particular
ecosystems.
Page 96: The secondary effects listed in table 4-4 should be
more specific to be meaningful
Page 101: line 12: I wonder if you want to add "what are the
effects of natural competitors?"
Page 102: lines 1-8: I wonder if you want to add landscape-
scale features?
Page 104: Line 5: The delphic approach may not be known to
all readers, perhaps a box is needed. "
Page 106: line 29: again, I would add synergism (such as the
effect of selenium on mercury uptake).
RISK CHARACTERIZATION
16. I would suggest adding Meta-analysis. It will become
increasingly important in risk assessments (both human and
ecological).
17. This was not clearly spelled out; and deserves a box of
its own. It should be highlighted, not buried in the
guidelines.
Add i t i ona1 comments:
Page 114: Line 16-18: Again, quality assurance should be
added here.
41
-------
JOANNA BURGER
Page 115: lines 1-15: I wonder if the cost of being wrong
with a particular analysis or characterization should be added
in some way.
bottom: Sublethal effects should be added more clearly.
Page 116: Another bullet should be added that refers to
bioaccumulation and species differences.
Page 119: bottom. Again, I wonder whether the potential
ecological costs for being wrong should be added.
Page 121:line 7: Power tests should be added here.
Under weight of evidence: I wonder if a discussion of
metanalysis should be added here.
Page 123: Under scale: some other landscape features should
be added like corridor, pattern, etc.
Page 125: It might be useful to do a box on recovery
42
-------
Peter Chapman
43
-------
ENVIRONMENT
CONSULTANTS
REVIEW COMMENTS:
EPA'S ECOLOGICAL RISK ASSESSMENT GUIDELINES
Prepared (November 1995) by:
Peter M. Chapman, Kathy Godtfredsen, and Andrea La Tier
EVS Environment Consultants
General, then sequential comments are provided. These directly or indirectly address the specific
questions asked of reviewers.
Genera] Comments
1 The document is very good, though there are some improvements which could be made as
noted below. We commend the authors for their efforts. In particular., past risk assessments
have often contained vague and ambiguous assessment endpoints. The document provides a
clear explanation of the definition and objectives of assessment endpoints. Text box 3-9
(page 41) is particularly helpful in that it provides examples of correctly worded assessment
endpoints. However, we are concerned that this document not replace past guidance
documents which actually guide the risk assessor through the process. The present document
is a compilation of expert reports which focus on specific considerations and examples and
covers such a wide range of topics that only vague statements can be made regarding
chemical-based ERAs, which will probably be the major focus (correctly or incorrectly) of
users of this document.
2 There is a general avoidance of discussion related to issues which would improve
standardization of the risk assessment process. For instance, there is no discussion of the
application of uncertainty or species sensitivity factors for extrapolating between taxa used
in food web modelling. EPA Region 8 has developed guidance for applying uncertainty
factors, which may provide a basis for sound risk assessment decisions and help to ensure that
risk assessors are applying these factors in a uniform manner.
3 Selecting the most appropriate toxicity reference values from the literature is critical to
accurately estimating risk to ecological receptor species using a food web modelling
approach. Basic criteria which should be followed when selecting appropriate values should
be outlined in the document. Trophic transfer of contaminants is a significant component of
both screening level (predictive) and higher level risk assessments. More information should
be provided on a food web modelling approach to exposure and effect assessments,
44
-------
ENVIRONMENT
CONSULTANTS
particularly for terrestrial receptor species. The discussion of field sampling techniques can
be tied into the food web model approach so that the appropriate data types can be collected.
A discussion of the criteria used to select toxicity reference values and EPA approaches to
applying uncertainty factors for interspecies extrapolations could be included.
4 There is a need to hilite more in the report as detailed in the sequential comments below. The
document contains many useful concepts and ideas which should receive greater emphasis.
As it presently stands, however, the document is somewhat fragemented and lengthy, which
makes it difficult to read in one sitting and difficult to use as a reference document.
5 Some of the text boxes are very good, some are not, and some are useless. Under sequential
comments below suggestions are made as to which text boxes to remove and what text boxes
to add. Ensure all text boxes are refered to in text, are relevant, and appear where they
should in the document.
6 A Glossary should be provided at the beginning of the document. Such a Glossary should
build upon but not be restricted to Appendix B. For instance, Appendix B does not: (1)
clearly indicate changes since the 1992 document; (2) define all key terms (a few examples
are "weight of evidence", "delphic"; lots of other terms, noted below, need definition). Both
need to be done. Specifically, with regard to (1), in the definition note in italics any changes.
7 To avoid confusion (there will be lots) it is important to provide, up-front, a "Summary of
Changes to 1992 Framework Document" with reference to specific sections where details are
provided.
8 The case illustrations are all noted to be flawed and, in at least one case, incomplete. Why
not provide one or two generic illustrations which are as close as possible to the ideal?
9 Text discussion related to maintaining the status quo (e.g., sustainability) is confused and
inappropriate given the reality that one cannot return exactly to pre-disturbance conditions
(noted towards end of document).
10 Particular emphasis must be placed on defining "weight of evidence". In particular, what the
significance of conflicting lines of evidence is needs to be addressed. In other words, some
sort of guidance is required though it is recognized that flexibility must be maintained.
11 The most common form of analysis is by Monte Carlo. While not espousing this method
(which is 50 years old in 1996), a good discussion of how to use Monte Carlo analysis in risk
45
-------
ENVIRONMENT
CONSULTANTS
assessment would be most useful. We don't believe such explanation/guidance exists
anywhere?
Sequential Comments
1 Page 2, hilite first sentence.
2 Text box 1-1, page 2: The information contained in this text box is more suitable for the text
(Section 1.2). Instead, most of the information in the first two text paragraphs should be put
into a text box, as this is more important to hilite than case illustrations, particularly when it
is noted that these illustrations "all have strengths and limitations that cannot be fully explored
in this document." This latter statement detracts from the utility of these illustrations. Why
not provide generic illustrations built upon these that are less flawed?
3 Page 3, lines 27-31; page 4, lines 1-9: This information should be hilited in a text box.
4 Page 6, line 3: Weight of evidence, used here and throughout the document, needs to be
defined. A text box would be very useful as well as a Glossary definition.
5 Text box 1-2 is good, but should appear on the next page adjacent to Figure 1-3 (it is
presently separated by a page). Figure 1-2 is not useful; delete. Specific problems with this
figure include the lack of a decision point determining whether one makes a decision or just
"rolls along". Instead, provide a text box on the iterative nature of ERA using information
beginning on page 6, line 6.
6 Page 9, lines 9-12: Should be hilited by a text box.
7 Page 9, line 16: "survival, growth, and reproduction of individuals" is not much clearer than
"healthy populations". How much mortality allowed? How much growth and reproduction
needed?
8 Page 11, lines 15-21: Good.
9 Page 11, line 30: The term "measurement endpoint" is widely in use and finally understood.
Switching to "measurement of effect" is disruptive and does not improve conceptual
understanding.
10 Page 12, line 9: Data acquisition is part of the risk assessment process. Text Box 1-3: The
46
-------
ENVIRONMENT
CONSULTANTS
new terms "measures of exposure" and "measures of ecosystem and receptor characteristics"
are awkward and do not improve conceptual understanding. However, if they are to be used,
they should be introduced in the main text also to receive the deserved emphasis. At present
these terms are not introduced until page 53.
11 Page 13, line 22: A recommendation is required on terminology that is being used
synonymously (i.e., either recommend relative risk assessment, or comparative risk
assessment) to standardize terminology. Line 29: Insert (1992) after Leibowitz et al.
12 Page 14, line 23: The italicized definition of stressor should be included as the word stressor
to maintain a consistent format. Ditto for the term exposure on page 15, line 6. Line 31: The
term stressor should be used instead of agent (to be consistent with Text Box 1-5).
13 Page 15: Delete this text box and put information in text; this is not worth hiliting. We
suggest, under Section 1.6.3, if you insist on using both terms, defining both agent and
stressor. Similarly, page 16 delete this text box and simply define and discuss stress regime
irt text. Remember, you are not trying to have text boxes spread nicely around in the text; you
are trying to hilite what is key.
14 Page 17, first sentence: hilite. Line 6, are you really referring to Figure 1-2 or do you mean
1-3? More importantly, all parties in the ERA process should be involved in the planning
stage including all regulatory parties. The use of the term "planning" seems rather arbitrary.
Planning is just part of the problem formulation process and discussions should occur at
frequent intervals during the entire process to keep all parties apprised and involved. Text
box 2-1: include a reference to text box 1-4 and/or Section 2.2 re what risk managers do.
15 Page 18, lines 2-6: Hilite. Maybe a text box?
16 Page 19, lines 1-2 and 9-10: Hilite these sentences, they are important. Also hilite sentence
on line 25. End of Section 2.2: Clarify whether risk managers and assessors work jointly from
the beginning or not. There are pros and cons to both possibilities.
17 Page 20, text box 2-2: Fifth bullet re uncertainty under Risk Managers should also occur
under Risk Assessors.
18 Page 21, line 2: The term "place-based" is awkward. Text box 2-3: We do not agree that
sustainability (i.e., keeping the status quo) is intuitively the right approach. Lack of
sustainability (death of dinosaurs, development of mammals and hominids) is a reality that is
47
-------
ENVIRONMENT
CONSULTANTS
not always negative. This box needs to be rethought, especially in the light that you can never
return to pre-disturbance conditions after remediation (mentioned later in the text).
19 Page 21, line 24: Hilite this sentence.
20 Page 22, text box: Good.
21 Page 23, line 2: Hilite. However, re lines 10-11 note that infinite resources do not equal a
lack of uncertainty. Section 2.6: This section should discuss the tiered approach in general.
For example, screening versus phase I or II type assessments should be discussed with
guidance criteria for determining the degree of rigour necessary for an ERA and how best to
proceed.
22 Page 24: Hilite first sentence under 3.0 and first sentence under 3.1. Sentence covering lines
7-9 should be a text box. However, line 9: The third shortcoming identified under case studies
"failure to identify important risks" seems ambiguous. The problem formulation stage of an
ERA involves: (1) compiling all available information about the site (e.g., industrial processes,
waste disposal methods, historical reports of environmental effects, ecological communities
at risk); (2) determining the extent of contamination using available data; (3) selecting
chemicals of potential concern; (4) developing a conceptual site model; (5) selecting
ecological receptor species; and, (6) developing assessment and measurement endpoints. Risk
identification is addressed in the analysis phase, specifically in the exposure assessment when
receptor species exposure point concentrations are presented. Perhaps this third point is to
be related to one of the components of the problem formulation identified in Figure 3-1 on
page 25? Bullet the three numbered points in Section 3.1 to hilite these. Text box 3-1: Direct
links need to be clearly made between measurement and assessment endpoings, and
emphasized in the problem formulation. Care should be taken not to separate them too much
(i.e., identify measurement endpoints in the analysis plan without stating relevance with
respect to assessment endpoints).
23 Page 25, Figure 3-1: Under source and stressor characteristics, list these characteristics
instead of repeating the words sources and stressors:
24 Page 26: Second sentence is not a sentence. Lines 10-13 should be a text box, and refer
specifically to Section 3.7 at end. Line 13: Actual section numbers should be listed instead of
phases like "in the final section". This will facilitate organizational understanding. Hilite
sentence on line 21. Section 3.2 is awkward and needs focus.
48
-------
ENVIRONMENT
CONSULTANTS
25 Page 27, line 17: Omit "of'. Section 3.2.1: Divide into two sections (i.e., separate stressor-
and source-initiated). Text Box 3-2: The example is too general to be specific to aquatic
environments. Line 3): The conceptual site model does not show the relationship between the
stressor and the assessment endpoint. The model displays the contaminant transport
pathways and, therefore, the exposure pathways to the ecological receptor species (e.g.,
inhalation, ingestion re aquatic species, dermal exposure). To quote the September 28, 1994
review draft: "assessment endpoints are explicit expressions of the actual environmental
values that are to be protected". The conceptual site model does not take into account the
assessment endpoints (e.g., reproduction and growth of small mammals) identified in the
problem formulation, but only the potential for exposure resulting in the assessment endpoint
being evaluated.
26 Page 28: line 8 - is constituency the same as stakeholders? Line 10: The sentence "ecological
receptors that originally showed the adverse effect are frequently selected as the assessment
endpoint" is confusing. Ecological receptors are not selected as assessment endpoints in and
of themselves. An assessment endpoint represents a measurable ecological characteristic
(e.g., diversity of the benthic community) of the receptor species. Assessment endpoints are
developed to direct the evaluation of potential adverse effects of a stressor on the ecological
receptors. In most cases receptor species themselves are not the values to be protected.
More likely assessment endpoints would, for instance, be: a reduction in the receptor species
population (i.e., survival of birds), or community diversity (e.g., reduction in benthic
macroinvertebrate community diversity). Lines 16-17: Biomarkers, depending on which and
how used, are not always reliable indicators of exposure. Lines 29-30: Hilite this sentence.
Line 31: omit "a vista in a national park"; surely this is trivial?
27 Page 30: hilite sentences on lines 23-24 and 27-28. A lot of this text refers to the need for
balance; distill this into a text box. A general comment: there is here and elsewhere a lot of
implicit emphasis on the negatives of anthropogenic inputs. Be careful with what appears to
be an implicit bias; some anthropogenic inputs are beneficial, others are neutral.
28 Page 31, text box 3-4: Good. Second sentence of first bullet: shouldn't that read "could also
have been based on", not "can also be based on"?
29 Page 32, line 5: "Actual, inferred, or estimated" data should be defined. Lines 27-28: Hilite
sentence. Also, an important point that cannot be overlooked is that one must look beyond
artificial or other boundaries to check re outside impacts or influences (past, present or
future).
49
-------
ENVIRONMENT
CONSULTANTS
30 Page 33, text box: Define stochastic, chaotic. What non-natural stressor event is lunar?
Consider adding: Is there any historical evidence of adverse ecological effects? Examples
might include: federal or state agency incident reports offish kills or waterfowl mortality.
Fourth bullet: substitute recruitment for reproduction.
31 Page 34, text box: add sediment after soil otherwise this sounds primarily terrestrial. Delete
mention of QSARs relative to ecological effects information. This does not make sense, or
are we missing something?
32 Page 35: hilite sentences on line 10 and lines 19-20. Line 12: The statement beginning "Their
relevance to assessment..." is circular. Line 14: The statement should read "whether there are
measurable characteristics" and not "whether they are measurable characteristics" because it
is the measurement endpoints that must be measurable (and linked to the assessment
endpoints). Lines 28-31 and lines 1-3 on page 36 should be a text box.
33 Page 36: The text box is too obvious to hilite; put into text. Delete lines 6-14. Begin Section
3.4.1.1 with paragraph starting on line 26. Second paragraph of this section is present second
paragraph starting on line 15, but is presently in need of rewording. Line 24: Omit "in
national parks" after clean air. We need such everywhere.
34 Page 37: Hilite lines 9-11 (replace "can be very useful" on line 9 with "are vital"). Reword
lines 14-16 as follows and hilite: "Ecologically relevant endpoints sustain the natural structue
and function of an ecosystem." First paragraph under Section 3 A. 1.2 needs more discussion
of the keystone concept, which now includes groupings of species and also includes bacteria,
i.e., it is not just predators which are really charismatic species in most cases, not keystones.
KGUte sentence on lines 26-28.
35 Page 38, line 4: Sensitivity is not a relative characteristic (i.e., all sensitivities are not
compared to a standard sensitivity). Lines 16-17: But note that adult bivalves after energetic
depletion due to spawning are the most sensitive life-stage for those organisms. Line 20:
during what? Lines 26-27: what is "birds require unobstructed views where they roost"?
Sounds anthropogenic, though we gather much later on that this may be a requirement of
whooping cranes? In any case it is both unclear and confusing.
36 Page 39: Hilite sentences on lines 1-2 and 18-21. Text box is very good. Commas needed on
lines 29 and 30. Lines 2-6: This is too simplistic an example; more cogent discussion of
laboratory versus field is required. Section 3.4.2 should go first in Section 3.4.
50
-------
ENVIRONMENT
CONSULTANTS
37 Page 40: comma needed on line 3. Hilite sentence on line 6.
38 Page 41: What is meant by the Specific under New Chemical?
39 Page 42: Line 13, second word should be "individual". Text box is too vague early on. For
.^Instance, if endpoint is too vague, what should it be? What is the midges example? What is
turkey and deer example? In comparison, the last two bullets are clear.
40 Page 43: Hilite lines 7-8.
41 Page 44: Hilite sentence on lines 4-6. Line 4: replace the word "verbal" with "written".
42 Page 44, line 14 and page 45, line 13: Similar statements are repeated. This type of repetition
occurs regularly throughout the document. Other examples include: page 44, line 8 and page
46, line 12; page 80, line 13 and page 82, line 22.
43 Page 45: Hilite sentence on lines 11-12.
44 Page 46: Hilite sentence on lines 12-13. Line 19, explain stressor-response curves.
45 Page 47: Quotes which continue on page 50 appear to include material which could not have
been verbatim, for instance the figure numbers (lines 7, 19) and Appendix A, Case A-l (line
23). Also, Odum, 1971 (line 24) is not in the References. -
46 Page 50, line 2: Diagrams such as Figure 3-3 are complex and not readily comprehensive;
they should not be included in generic guidance. Lines 4-17: Too much exerpted text; more
generic guidance is needed. Generalizations (e.g., lines 18-20) should be given before specific
examples, so as to orient the reader. Line 21: on, not up.
47 Page 51, Figure 3-4b: Delete one of the arrows pointing from "ingestion of particles by birds"
to "death".
48 Page 52: Line 4, into, not on. Hilite sentences on lines 11-14 and 17-18. Line 18: reference
the Framework Report. Section 3.6: The inclusion of an analysis plan as defined is a welcome
addition to the process.
49 Page 53: Hilite sentence on lines 7-9. Line 11: Assessment endpoints include responses of
receptors. Line 17: Assessment endpoint is used interchangeably with ecological receptor
51
-------
ENVIRONMENT
CONSULTANTS
throughout a substantial portion of the document. It would be clearer to distinguish between
the two since the assessment endpoint incorporates both the ecological receptor species itself,
and the receptor species attribute that is being assessed (page 40, line 4). Statements such
as "well defined assessment endpoints reduce uncertainty hi a risk assessment" (page 40, line
27) also clearly indicate that the assessment endpoint is more than the ecological receptor
species. Line 24: importance, not important.
50 Page 54: line 26: The six main areas should be listed and discussed in the text. The
uncertainty section should not be broken into so many different sections throughout the
document (Section 1.4, 3.7, 4.1.2, 5.2.4...) Tables 3-1, 4-1, etc. seem repetitious. Hilite
sentence on lines 27-29.
51 Page 55: First example is not "as clear as possible" as previously noted. Table 3-1: It appears
that two examples are given for variability. This is confusing to the reader since all other
categories have just one example.
52 Page 56: Hilite lines 4-7.
53 Page 58: Lines 4-23 should be captured in a text box.
54 Page 59, line 25: Introduction of the section organization comes too late to be useful - instead
it confuses the reader at that point who then questions what it is they have been reading.
Another example is page 75, line 3.
55 Page 61: Line 5, examples, not samples. Hilite sentence on lines 9-10. Section 4.1.1. should
not be a section; it is too short and says too little.
56 Page 63: All periods (end of sentences, between section numbers, in e.g.) are missing from
line 12 to the end of the page. Line 25: communicate, not communication.
57 Page 64: Hilite lines 28-30.
58 Page 66, text box 4-2: delete; put this information into the legend for figure 4-3. Also, delete
text box 4-3. Text boxes should be hilites that are key to reader comprehension. This is not
and adds little to the text. In fact, it is also confusing. Also, neither of these text boxes is
refered to in the text. Hilite sentence on lines 16-18.
59 Page 67: Hilite sentence on lines 9-11. Line 10 would be clearer if it read: "...and may include
52
-------
ENVIRONMENT
CONSULTANTS
measurements of the concentrations of contaminants in the air, soil..." This text box is usefbl.
60 Page 68, line 1: The "speciation" of chemicals has more meaning than the "nature" of
chemicals as used in this sentence. Line 19, seventh word should be "from". Hilite sentence
on line 27.
61 Page 69, lines 16-21: Should this be in quotes?
62 Page 71, lines 4-8: Agreed, hence previous comments re biomarkers.
63 Page 72, line 11: Sentence is difficult to understand.
64 Page 73, line 10: More guidance is required regarding spatial scales and averaging techniques.
Line 24: text should accompany each point. Section 4.3.3.4 was particularly helpful with
respect to designing an exposure profile. It prompts the risk assessor to consider both the
receptor species' natural history characteristics and abiotic factors that influence exposure.
It should be stressed that local or regional receptor characteristics (e.g., home range size)
should be considered over general literature information whenever possible.
65 Page 74, line 13: The ecological response analysis, as defined, is far more than simply a
number crunching analysis and requires a careful examination of the type and usefulness of
toxicity data available.
66 Page 76, line 17: "...such as..."
67 Page 77, first (incomplete) paragraph: Good. Line 12: Discussion should first be on
interpreting and using available data rather than skipping to QSAR when chemical-specific
data are not available. At the very least the section on Single-Species Assays, "the most
common method", should be discussed first.
68 Page 78, line 17: What does "q.v." (used twice on that line) mean? Line 24:
Recommendations should be made.
69 Page 79, line 15: An example of data which may not support regression analysis should be
provided. Line 28: Laboratory testing with field media should not be defined as a field
experiment since the purpose of the field experiment is to assess in situ conditions which are
disrupted when transferred to a laboratory setting and manipulated. The Triad should be
discussed in this section.
53
-------
ENVIRONMENT
CONSULTANTS
70 Page 80: Line 18, toxicants not toxicant. Lines 20-21: Omit "In the previous example". Line
29: This document should make recommendations.
71 Page 81: Text box 4-7 is good. Ensure Leslie Matrix is defined. However, section on
Extrapolations Between Respones covers too many topics to be user-friendly.
72 Page 82: Line 15, the 1982 Gentile et al. reference is dated; why not use more recent
example(s)? Line 17: data were, not data was (data are the plural of datum).
73 Page 83, line 17: Guidance should be provided on other useful factors to regress.
74 Page 84: Line 5, "...geographic area..." Line 16, insert "to" after close parentheses. Lines 20
and 24: Physical models, ecosystem models and microcosm models should be better defined
with examples. Line 22, rewrite: "Costs and time are certainly factors..."
75 Page 85, Section 4.2.3.4: This section does not make sense here. The reader is lost at this
point as to the overall organization and point of Section 4.
76 Page 86, line 24: "...contained in..." Line 25: additional information that would be useful
should be listed in the text.
77 Page 88, lines 8-9: A verb is needed to mesh with the format of the other bullets. Line 29:
If characterization of exposure and effects is discussed somewhere in the document, these
sections should be referenced.
78 Page 90: Hilite sentence on lines 9-10. Text box 4-8: Omit first "were"; this box is not very
useful. Ditto for text box 4-9 on page 92.
79 Page 91: Can't read the figure; hopefully this will not be the case for the final.
80 Page 93: Text box 4-10 is good but is not referenced in the text. All text boxes need to be
referenced in the text so their usefulness is clear.
81 Page 94, Section 4.4: Analysis of Biological Introductions is an awkward title. Introduction
of Biological Stressors would be much clearer.
82 Page 95: Text box 4-11 is good. However, line 6: differences are not given in the text box
54
-------
ENVIRONMENT
CONSULTANTS
as stated in the text. Line 22: The use of "for instance" does not make sense here.
83 Page 97, line 18: Define fault trees.
84 Page 99, lines 5-6: Explain more re the Chestnut Blight. What are the lessons? Maybe such
were provided earlier, but the reader cannot be expected to remember them. At least provide
a reference to where hi the text to find this information. Line 19: The risk assessor will not
know if a worse case estimate is being made if little information is available. Line 22: This
discussion should start more generically and then move to discuss genetically engineered
organisms.
85 Page 101: Text box 4-13 is not useful and is not referenced in the text. Omit.
86 Page 102: Text box 4-15 is not useful. Instead, make lines 4-10 into a text box. Line 29:
"Primary" effects should be stated for clarity, not "direct" effects.
87 Page 103: Text box 4-16 is useless, delete (note APHIS is not defined). Line 9: is, not in.
Line 13: delete "of the". Hilite sentence on lines 23-25. Line 27: Define "qualitative
estimates" of what.
88 Page 104: Line 5, define delphic approach. Line 6: "...also by..." Text box 4-17 is useless;
delete.
89 Page 106, line 7: Response addition should be discussed in more detail or omitted. Line 21,
second in should be is. Lines 24 to 3 on next page should be a text box. But, line 26: it is not
clear what is meant by "exposure measurements for mixture components". Is this a reference
to the form of the individual chemicals (e.g., arsenate as opposed to arsenite), to determine
potential bioavailability of the mixture?
90 Pagel07: Text box 4-18 is not refered to in text, is not the only way of doing this, suffers
from problems not elaborated here, and should not be the subject of a text box. Delete this
text box. Instead, make a text box of lines 21-27.
91 Page 108, line 12: This sentence implies that the physical stressors in the Detenbeck study
were independent and had additive effects. Line 21: If sediment chemistry and toxicity test
results are available, they are more useful than simply comprising part of the problem
formulation phase.
55
-------
ENVIRONMENT
CONSULTANTS
92 Page 109: Text box 4-19 is good. Line 26: It is not clear why "indices that include both
exposure and effect measures should be avoided."
93 Page 110, line 13: We cannot [yet] generally link biomarkers to effects. Sentence is
confusing; rephrase. Section 4.5.4: This section should be combined 'with the other section
on causality.
94 Page 111: Text box 5-1 is good. However, it is not referenced in the text (as many are not)
and contains important information which should receive greater attention and inclusion in
the main text. The distinction between risk estimation and risk description is not clear.
Where does the uncertainty analysis go? Hilite sentence on lines 5-7.
95 Page 113: Hilite sentence on lines 3-4. Line 30: "...exposed to a..." Lines 7-27 and Text Box
5-2 are out of place and seem repetitious - were they discussed elsewhere?
96 Page 114: Lines 1-20 should be a text box. Line 11: Approaches such as population and
ecosystem modelling should be described in greater detail to point out the kind of output
expected. Line 22: omit "are". Line 31: The HQ equation should be given.
97 Page 115, line 1: Most risk assessors use the NOAEL instead of the LD50 in the HQ
approach.
98 Page 116: Explain why subtle short term effects may be missed if results are only tabulated
on the final day. This is from text box and is presently unclear. The four days should be
consecutive unless additional time-response information is available. The 28-d results should
also be applied to consecutive exposure. Line 13: The use of distributions does not alleviate
all of the pitfalls identified in Section 5.2.2.1. Line 29: Distributions are still useful when
evaluating a single species since many times a wide array of toxicological data is available.
99 Page 117: Lines 7-16 should be a text box. Line 27: semicolon, not comma. Section 5.2.2.3:
The use of the term physical models for field testing is misleading.
100 Page 118, line 2: Something is missing on this line or else omit "types". Line 23: Delete this
sentence. Simulations are commonly used for many reasons.
101 Page 119: The following should be text boxes: lines 7-15, and lines 27 to 6 on page 121.
Hilite sentence on lines 18-20. Sections 5.2.4 and 5.3: The proposed format for risk
estimation, uncertainty analysis, and risk description is not clear. Line 17 discusses a risk
56
-------
ENVIRONMENT
CONSULTANTS
summary which is assumed to be the risk description.
102 Page 121: Make a text box of lines 15 to 7 on page 122. -Section 5.3.1: ASTM is coming out
with guidance on ranking weight of evidence. This information, presented at the Second
SETAC World Congress in Vancouver this November, should be included in this section.
103 Page 122, line 9: Much more is needed re when different lines of evidence are in apparent
disagreement (e.g., see lines 27-28 on page A-19). Line 25: Hilite this sentence.
104 Page 123: Line 7, "...aspects of an..." Also, why the periods at the ends of the headings on
lines 12 and 19?
105 Page 124, line 4: Something wrong here. Line 13: The temporal scale should range from
seconds to centuries, not minutes to days.
106 Page 125: Hilite sentence on lines 2-3. Make a text box of lines 27 to 2 on page 126.
107 Page 126: Text box 5-4 does not illustrate well the importance of understanding natural
disturbances. Such a text box would be useful; this is not it. Also, hilite sentence on lines 20-
24.
108 Page 127: Hilite lines 3-4. Line 6: This sentence is not clear and should be shortened, or the
information should be presented in two sentences. Make a text box of lines 20 to 14 on page
128.
109 Page 128, line 12: Consistency characteristics should be presented with bullets as well. Last
sentence: This is an important point which is also controversial and which needs much more
discussion and justification.
110 Section 7, References (pages 129-144) only spot checked.
Ill Page A-3, line 3: Connor here, Conner on page A-21.
112 Page A-13, line 16: from a, not froma.
113 Page A-21: Where is rest of Harris et al. (1994) reference?
114 Page A-22: U.S,EPA (1993, 1994) cited in Appendix A?
57-
-------
ENVIRONMENT
CONSULTANTS
115 Appendk B: No References but many citations. Page B-2, line 10: omit second (lower case)
"a". Page B-3, line 24: needs close parentheses; line 25, needs a comma before "(3)".
58
-------
James Clark
59
-------
Comments by J.R. Clark
COMMENTS PERTAINING TO ISSUES THROUGHOUT THE DOCUMENT
1) Document quality and responsiveness to outside input is very high.
This- document incorporates many of the discussion points of the May 3
Colloquium held in Washington, DC, to discuss stakeholder perceptions of
ecological risk assessment guidance. The authors are commended for being
responsive to that input and accommodating many of the suggestions. This
document is well-written, the ideas and topics are clearly presented and logically
organized. The examples and appendices provide useful insight into the t
application of the guidelines and concepts important to the agency.
2) Greater detail in guiding principles for conducting risk assessments is needed.
These guidelines must link the need to have flexibility in how ecological
risk assessments are conducted and the necessary consistency in ecological risk
assessment products. This can be achieved only by providing greater detail hi the
criteria by which decisions are to be made when in the middle of the risk
assessment. An overly prescriptive guidance document will not be useful to
resolve the myriad of environmental issues the agency must address. However,
this document lacks sufficient detail in the guiding principles of ecological risk
assessment to achieve consistency in products developed by following this draft.
The reader needs more specific direction on where to go for basic information,
how to recognize good data from bad, what to search for when looking for
surrogates, or reference sites. The text cites a number of useful references, but
does not provide the detailed guidance necessary to achieve consistency in
products. Some of these issues are discussed more fully below.
3) References to use of common sense and good judgement require more
60
-------
Comments by J.R. Clark
elaboration to form a useful guidance document.
In numerous places, the text rightly refers to areas where common sense
and good judgement are required to make an assessment regarding the role of
some factor or data in the risk assessment process. However, it is the function of
a guidance document to lay out the basis to define the application of good
judgement or common sense. These areas in the text should be enhanced to set
up criteria by which decisions can be made regarding the direction the risk
assessment is to take. Instead of limiting the discussion to "use common sense",
the document should direct the reader's actions (provide a list of do's and don'ts,
such as go with documented or peer reviewed information only, seek direction
from government agencies, make a decision on your own and document the
rationale, etc). Such an elaboration is necessary in (at least) the following areas:
S 4.2.3.2, P 80, L 27; S 4.2.3.2, P 82, L 29-30; S 4.2.3.3, P 85, L 8-11; S 5.2.1, P
114, L 16-19; S 5.3.2, P 123, L 1-10).
4) Data quality/data usability issues need to be addressed in a detailed, cohesive
manner. "
Throughout the text there are limited references to literature regarding
appropriate use of data, but the guidance lacks a focused and comprehensive
discussion of the attributes of good data, how to compare old and new data, data
generated by different methods, etc. The text includes references tor various EPA
documents regarding DQOs and quality assurance issues, but appropriate data use
in ecological risk assessment is much broader than merely generating quality
analytical data. In particular, the guidance document should have a separate
section that addresses the issue of how to decide to use or exclude data or data
sets based on data quality, data usability, and data relevancy issues. These issues
should be discussed in a separate section, then referenced as necessary to
pertinent areas in the following sections (S 3.6, P 54, L 12; S 4.1, P 58, L 6-9; S
61
-------
Comments by J.R. Clark
4.3.2, P 74, L 27-29; S 4.3.2.1, P 76, L 22-27; S 4.2.3.2, P 82, L 3; S 4.2.3.4. P 85, L
25-26; S 5.2.1, P 114, L 2-4; S 5.2.2.2, P 116, L 28).
5) A new section on selection and use of surrogates (species, endpoints,
indicators) is needed.
The guidance document lacks an adequate overview and detailed guidance
on what attributes are to be considered when selecting surrogates for receptors or
assessment endpoints. A detailed discussion regarding a search for surrogates
with similar life histories, exposure pathways, toxicological sensitivities, predator-
prey interactions, etc., is needed to guide risk assessors in selection of surrogates.
This pertains to discussions presented in S 3.4.1.3, P 39, L 1-6; S 4.1.2, P 63, L 9-
13).
SECTION-SPECIFIC COMMENTS
Section 1.2. The perceived audience for this document should be broadened
(P3, Ll-4).
Although the agency needs a guidance document for its risk assessors, the
regulatory aspect of agency functions require that all stakeholders must
understand the basis for agency actions. This document will become the
cornerstone for regulatory agencies, as well as business, industry,
contractors, and landowners. The agency should recognize this
responsibility and create a document that targets a much broader audience.
Section 1.4. Risk assessment is only a small part of decision-making.
The text should recognize that there are numerous factors to decision-
making, and risk assessment is only one (although important) component.
A diagram (enclosed) taken from an OTA document shows five other
factors affecting a regulatory decision. This diagram helps risk assessors
62
-------
legislative/
political
factors
Elements in risk management. Modified from the
U.S. Congress. OfScs of Technology Assessment [25].
63
-------
Comments by J.R. Clark
understand where their contribution may fit into a decision.
Section 1.5. Figure 1-3 and accompanying text are very useful.
This diagram and the text describing the activities in each box are so much
more useful than the overall framework diagram. By breaking out inputs,
actions and deliverables, the figure offers much greater, insight on activities
and expectations associated with each step in the risk assessment process.
Section 1.6.1. Need greater consistency in use of "adverse" to describe effects.
The text in this section needs greater consistency in the use of the term
adverse, as it relates to the types-of ecological effects that a risk assessment
will address. If in the final analysis, as indicated in line 14 of page 3, any
change (positive or negative) must be assessed, then the definition can
focus on ecological changes alone. In my opinion, the guidance should
focus on significant adverse effects, not any change. The terminology in
this section is not consistent with discussions in Section 1.6.3, where the
terms adverse response (P 14, L 25) and disturbance (P 15, L 28) are used.
Section 1.6.2. Don't confuse hazard quotients and risk quotients (P 14, L 6-9).
The discussion in this section may confuse readers as to the value of risk
quotients versus hazard quotients, since the reference to the SETAC
document includes an exposure and toxicity term. The text here should
elaborate on the differences between risk quotients and hazard quotients.
Risk quotients are mentioned extensively in Section 5.2.2.
Section 1.5. Address Tiered risk assessment process in more detail.
The iterative nature of a risk assessment is discussed in sufficient detail
here, providing examples of how data and analyses become more
64
-------
Comments by J.R. dark
complicated as necessary. Others have used a tiered approach, designed
around data availability or costs. Screening-level tiers might use only
available toxicity data, yet may have several iterations as investigators
search for more published data.
This document should have a whole section dedicated to tiered approaches
in ecological risk assessment. It should address issues such as how to
organize a tiered approach to a risk assessment, setting up decision criteria
at each level (perhaps with multiple iterations within each tier). Taking
the reader through a discussion of how a screening level assessment is
organized and executed, a process for formulating a decision on why and
how to proceed to the next tier, when and where site-specific risk
assessments are useful, how and when to collect more field or lab data, and
other aspects of tiered approaches as necessary to demonstrate cost-
effective, efficient approaches to risk assessment. The costs in time, data,
and effort at each level could be discussed and a rationale for how tiered
assessments are used in de,cision-making would be useful. Currently, this
document implies, perhaps implicitly, that a risk assessment is always a full
blown, data intensive, lengthy and costly endeavor to be undertaken for
every significant environmental decision. Ecological risk assessment is not
practiced in that fashion today, nor does it need to evolve to that level for
every decision.
Sections 2.0 and 2.1. Broaden the discussions concerning, and the definition of,
risk managers.
The discussion on page 17 regarding who is a risk manager (along with Box
2-1) should be broadened to include risk managers in the private sector of
business and industry who manufacture products and operate facilities.
65
-------
Comments by J.R. Clark
Likewise, the discussion of planning objectives should be broadened to
include all stakeholders (business, industry, land owners, etc.).
Section 2.2 Quality job on delineating roles.
This section provides a good delineation of the roles for risk assessors and
risk managers. These responsibilities hold for those in the private sector
involved in risk assessment/risk management as well.
Sections 2.3 and 5.2.2.1. Costs and benefits should be addressed.
These sections should include references to cost/benefit or other economic
factors that must be considered in risk assessments and discussion of the
goals for risk management. Examples used in these sections omit
reference to economic factors, which are drivers in nearly every case.
Many management decisions come from Tier I studies which are low cost
tools, for low cost issues.
Section 2.3. Sustainability should be considered in a management goal (P 21).
In Box 2-3, 'the discussion on sustainabiliry is too narrowly focused.
Sustaining populations and communities of ecological resources is indeed
the goal of an ecological risk assessment. \Vften expressed in this manner,
sustainabiliry does meet the criteria for an assessment endpoint.
Section 2 overall. Good job on discussion of planning and interactions between
risk managers and risk assessors.
Overall, Section 2 provides a solid explanation of the roles and activities of
the risk' assessor and risk manager as a risk assessment is initiated. This is
a well written chapter, the examples are clear, the roles and expected
outcomes of interactions are delineated. There are good examples of
66
-------
Comments by J.R. Clark
government activities, responsibilities, and topics of dialogue!
Section 3.2.2. Beef up text on causality (P 28, L 15-20).
This discussion points out three criteria (bullets) for linking exposure and
effects. The text should emphasize that all three conditions must be met in
order to confidently assign causality to an observed ecological change.
Section 33.1. Effects driven assessments should not be witch hunts
(P 32, L 10-18).
When working with generalized conceptual models, any type of change
often is linked to suspected chemical stressors, leading to a witch hunt for
the offending toxicant. The text here should reflect the conditions for
causality outlined on page 28 (L 16-20), or at least refer the reader back to
those ideas. In addition, the text in this section needs to reinforce the
need for balanced, scientifically-based, problem formulation.
Section 3.3.2. Management should not define ecosystems.
The text implies that issues other than ecology define ecosystems. Perhaps
the intent is to say that risk managers define the area/geography/locale of
concern, then ecologists can discern the relevant ecological boundaries.
Most ecological receptors do not obey risk management directives (the
same might be said of risk assessors, too).
Section 3.3.3. Link discussion of effects more closely to exposures (P 34).
Effects to be considered are those resulting from some form of direct or
indirect exposure. The discussion should focus the reader only on effects
, that can be linked to exposures. Text in Box 3-6 does not address
exposure pathway analyses, which is crucial hi assessing potential for
67
-------
Comments by J.R. Clark
effects. This text should be tied into the discussion of selecting what to
protect (Section 3.4.1), focusing on exposed populations that lead to
adverse ecological effects.
Section 3.4.1.2 Discussion of ecological relevance is very useful (P 37).
The text in this section provides the reader with valuable guidance on what
to take into consideration when selecting endpoints that are ecologically
relevant.
Section 4.2 Add section addressing issues regarding background conditions and
use of reference sites. .
This guidance document should include specific recommendations
regarding how to select or define appropriate reference sites and
background conditions. Ecosystems which have adapted to baseline or
background exposure conditions should be the basis of site-specific risk
assessments. This document should explain, in detail, the procedures for
obtaining appropriate information regarding background conditions and
exposure levels. An expanded text should incorporate areas briefly
discussed on pages 67, L 1-6; P 68, L 21-24; P 107, Box 4-18.
Section 4.2.2.3. Use of biomarkers of exposure should be discussed more fully.
The guidance document should provide a list of attributes that make
biomarkers reliable and useful. Currently, a number of useful and useless
biomarkers are being incorporated into risk assessments (and other
ecological activities), which confuses their role and taints their overall
utility. The text should lay out the expectations of what makes a useful
biomarker, the documentation and validation necessary to support it, and
the role it can play in risk assessments.
68
-------
Comments by J.R. Clark
Sections 4.2.3.2 and 4.2.3.3. Data quality issues with modeling should be
emphasized.
In addition to an overall discussion of data quality and usability (my
comment #4), the sections discussing applications of models should have a
paragraph providing guidance regarding use of quality data in modeling
attempts. Use of poor quality input data destroys the credibility and
usefulness of an otherwise excellent modeling tool.
Section 4.2.3.4. Reference to use of anecdotal information requires accompanying
cautions.
Effects-driven assessments must be based on sound science. Quantitative
information must be a priority need when conducting a risk assessment on
this basis. The text does not go far enough in providing caution regarding
use of anecdotal or incomplete information. The guidance document
should emphasize that when confronted with a "wide array of data",
scientific evidence should be heavily weighted and decisions to utilize other
information should be openly documented and justified.
Section 4.3.2. Spatial scale of disturbance must be considered (P 92, L 15-27).
Discussions regarding loss of habitat and physical disturbance must be put
into perspective of the spatial and temporal scale of the disturbance/loss.
This concept is captured elsewhere in the document (4.3.4), but should also
be introduced here. This text comes across as an absolute statement, and
is incorrect without some modification. Discussions of relative risk and
incremental risk could be added to this section as well.
Section 4.5. Differentiate multiple stresses and multiple risks.
The presence of multiple stressors is indeed a difficult task to assess, and
69
-------
Comments by J.R. Clark
the guidelines provide useful explanations for how to address these.
However, the guidance document should recognize that there may be a
difference between the impact of multiple stressors on a single (or limited
number of receptors), and the risks posed by the presence of these
stressors. Multiple adverse impacts to a receptor may pose a single
ecological risk, if the ecological risk hinges on,the,growth, reproduction
and survival of that receptor. Cumulative impacts from multiple stressors
can pose a single risk. Summing the effects of multiple chemicals, as
discussed on P 106, does not necessarily result in summing the individual
ecological risks. The guidance document should elaborate on this concept.
Section 5.2.1. Introduction needs tighter link to subsequent text.
The introduction to this section provides an overly simplistic assessment to
a complex problem. The subsequent text (all of section 5.2) provides an
acceptable description of how quantitative and qualitative approaches are
used, and how they differ. However, this introductory material provides
absolute statements (e.g., "Qualitative assessments rely solely on
knowledge") which would be more accurate when accompanied by
modifying and limiting adjectives and adverbs ("Some qualitative
assessments", or "may rely solely on", or "rely heavily on"). This
introductory paragraph should more accurately reflect the numerous
combinations discussed in subsequent sections.
Section 5.2.2.1. Quotient method is more useful than text implies.
The text provides insight into development and application of single
quotients in a very simplistic context. There are opportunities to compute
various quotients based on other types of toxicity data, or for other effects
measures. The quotient is not a sophisticated risk assessment tool, but it
70
-------
Comments by J.R. Clark
can be a very efficient and cost effective screening level risk assessment for
decision-making. The guidelines should provide a broader, and open
discussion of the uses, an perhaps misuses, of quotients in risk assessments.
This tool has may useful purposes, and is worthy of greater attention.
Section 5.23. Distinguish between sensitivity analysis of a model and sensitivity
assessment of the ecosystem (P 118, L 24-25).
The text in this section provides a good overview of the role of modeling in
ecological risk assessment. However, the reader should be reminded that
there may be significant differences between the model of the real world
and the real world itself. Therefore, sensitivity analysis and data
prioritization based on a model may not necessarily reflect the real world
conditions. Important parameters identified through empirical studies
should be prioritized over important parameters identified through
modeling runs.
Section 5.3.2.3 Is the statement regarding logging impacts on recovery true?
(P 125, L 17-18).
Is there some context where the statement proclaiming continuous logging
of old-growth forests will eliminate forest ecosystems is true. Harvesting at
what rate, using what techniques? With replanting, selective harvesting, or
other management techniques, some aspect of a forest ecosystem could be
sustained. Physical alterations, as described in subsequent text, could lead
to a loss off of forest ecosystems, but this case seems different than the
impacts of continuous logging.
Section 6. The focus of the risk communication text is too narrow
(P 128, L 16-19).
71
-------
Comments by J.R. Clark
This short discussion regarding risk communication is a topic worthy of its
own section in this document. Risk communication is a very important
process throughout the entire process of risk assessment. It begins with
discussions between the risk manager and risk assessor. Good risk
communication must occur among the multidisciplinary team conducting
the risk assessment. Communication is important at the end of the risk
assessment, not only when dealing with the public, but with all stakeholders
and risk managers.
SIMPLE EDITORIAL COMMENTS
P 43, L 1-6. This text is redundant with P 42, L 9-23.
P 53, L 24. Change "important" to importance.
P 82, L 17. Replace "data was" with data were.
72
-------
William Cooper
73
-------
Review of EPA Draft Guidelines
for Ecological Risk Assessment
by Bill cooper
Generalt The document is well written and covers the subject
very well. The distinction between Effects - Initiated
and Ecological Value Initiated is one of scale, not one
of substance. The selection of assessment points are
common to both. If you want to maintain 21 clear
distinction between these two, you need more work on
this - pages 28-30.
Specifics,;
Page 3, line 2.
line 4.
lines 23-26.
Page 4, line 20.
Page 6, line 3.
Page 9, line 28.
Page 18, line 4.
Page 22, line 26.
Page 28, line 11.
"risk assessors at EPA" - With the shift of
power to the states, this is no longer true
(Florida).
"ecology and ecological risk" - This is not
true with many risk assessors in EPA. This
is why you needed so many outside
consultants.
These are all chemical insults.
base.
.Broaden your
"preferable alternatives" - One still needs a
relative risk analysis to determine
"preferable".
"weight of evidence" *• The quality of the
evidence is more important than the
magnitude. '
"carefully designed experiment" - How do you
carefully design an experiment for something
you don't even know exists? How do you
define a null hypothesis? .--,- ; <
The focus is primarily a management function.
The extent and complexity will be dictated by
the science for any given level of acceptable
uncertainty. You might expand the definition
of roles. . , ,
"community, court-ordered or legally
mandated" - Why would only the knowledge of
the origin of request help determine the
maximum uncertainty?
"adverse effect - assessment endpoints" -
(Page 29, line 24 - early selection of
assessment endpoints). These are the same.
One is empirical and the other is
74
-------
anticipatory.
line 20. This is presumed guilt by association. Be
careful here.
Page 32, lines 12-15. Again, these only differ by observation
versus anticipation of endpoints. If
one observed effects that were
considered not important, one would ntft
initiate a risk assessment.
Page 33, Test Box 3-5. "absence of such influences" - The
absence of the stabilization of water
levels in the Everglades (stabilization
is the stress) is not stochastic or
chaotic, but a return to the normal
yearly cyclicity in water levels.
Page 38, line 4. "one individual" - I can't think of an
ecological situation where one individual
response makes a difference. Ecologically,
individuals are expendable.
line 14. "small home ranges" - No! Metapopulations
are multiple small patches coupled by
frequent migration.
line 20. "during" - What comes next?
line 30. "spatial extent" - also location - the
contextual component of landscape ecology.
Page 39, lines 1-6. Most test organisms for laboratory
dose/response calibrations for sensitivity
don't match anywhere near this well -
"ungulates to deer".
Page 41, Text Box 3-9. Carbofuran - The assessment endpoint
should not be individual bird survival.
It should be a population
characteristic. See pages 81 and 82 of
this report.
Page 43, lines 1-6. This is redundant to the same statement on
page 42, line 19+.
Page 44, line 31* . "what initiated...known and unknown" - This
is not clear. The knowns and unknowns
included a lot more than stressor
identifications. Stressor and effects are
not as difficult to hypothesize as getting
all the detailed data on transformations,
fates, potency, etc.
75
-------
lines 5 and 8. "affected ecological entity" and
"undesirable changes observed" are the
same. The value of this distinction
between effects and values is not
obvious.
Page 47, Test Box 3-12. "Effects and value" - They both have
the same logic, only the scale is
different.
Page 59, line 9. "assessment endpoint identifies" - No. It is
the other way around. The level of
biological organization and the type of
effect identifies the assessment endpoints.
Page 63, line 3. "random variation" - Variability is a genetic
property of the biological system. Based on
the magnitude of this variation, a different
sample size is needed to achieve a given
level of discrimination. This is not
"error". Errors would involved bias sampling
(non-random), poor analytical techniques
(level of detection), etc.
Page 84, line 16. to large scale.
Page 85, line 18. "other 95 percent" - This is very incorrect.
The 0.5 level is the probability of falsely
rejecting a true hypothesis. Ho's are always
negative statements of causality. The Type
II error of incorrectly accepting a false
hypothesis still exists. It is not correct
to say that 95% of the time the causality is
proven.
Page 106, line 24+ Steve Safe's TEQs are obvious by their lack
of presence. Why?
Page 114, line 22. remove "are".
Page 115, line 14. "data are limited" - This is where it is most
risky to use the single value quotient
method. Even the Gftl mandates a 28-day
chronic.
Page 118, line 2. "of type kinds" - makes no sense.
Page 122, line 8. "greatly increases" - only if the biases and
uncertainties are independent.
Page 124, lines 4 & 5. Something is wrong here.
Page 125, line 2. Populations and communities also do not
retain the original condition. The new
76
-------
ecosystem constraints prohibit this.
line 17. "eventually eliminate" - Not necessarily. It
might select for "weedy" tree species.
Page 128. One should also identify easily observed response
• - characteristics to verify the impact of the risk
management.
77
-------
-------
Peter deFur
79
-------
P.L.deFur
Draft Proposed Guidelines
for
Ecological Risk Assessment
Comments by Dr. Peter deFur, Environmental Defense Fund
General Comments
1.1 commend EPA staff for creating the guidance and putting a great deal of
information together. The guidance shows that EPA has given a great deal of
consideration to incorporating new developments and concepts into the way in
which ecological risk assessments are practiced.
2. In no small part because the field of ecological risk assessment is changing and
developing, further concepts and practices need to be incorporated into the
guidance document. The conceptual basis for risk assessment guidelines needs
enrichment and expansion even farther beyond the initial steps of chemical-specific
risk assessments. While the present draft does not rely solely on chemical, source
and end-point specific assessments, the text needs to more fully develop the issues
beyond such narrow foci.
3. The challenge for EPA in preparing such guidelines is in adding the basics of
health risk assessment along with the diversity of ecological endpoints and the
mathematical modeling that has come to be used in ecological RA. Furthermore,
the practice of RA is changing rapidly in all areas, especially in ecological areas.
EPA will need to update this guidance in order to keep the readers and users of
the guidance up to date. EPA should indicate this developing nature of the
guidance in the document, and, if possible, suggest the areas where changes might
80
-------
PJLdeFur
and might not be expected.
4. The guidance document would benefit from a more thorough discussion of
processes that are not analytical. There is some reference to data collection, but
not to other areas, in which process is determinative. Numerous other areas in
which deliberative and collaborative activities are required or highly desirable
should be identified and the process summarized.
5. There is not enough discussion of public participation and participation of
interested and affected parties. This guidance should not propose or imply that
ecological risk assessment is conducted without full and complete involvement of
interested and affected parties. This involvement encompasses much more than
public hearings and notices. Substantive and meaningful involvement requires
membership on advisory and review committees, participation in-planning, and
even joint assessment and analysis, in appropriate cases.
6. There are too many indefinite antecedents to the "this's" throughout the
document. .
7. Section 3 must include describing the context for the risk assessment. The
assessment process must begin with a determination of the context - physical,
chemical, ecological, social, political, economic, etc.
8. The scope of ecological risk assessments has to be intentionally determined and
set in the beginning. Given EPA's current efforts to use the watershed for more
and more activities, ecological risk assessments should have a default assumption
that the watershed will be determined for each assessment, unless compelling
reasons exist to the contrary.
9. The guidance still needs more language on non-chemical situations.
10. This reviewer disagrees with the premise that risk characterization is
81
-------
P.L.deFur
primarily or exclusively risk estimation and a qualitative description of risk. The
statements about all the excluded aspects are not in agreement with many
commonly accepted and practiced areas of risk characterization. IF EPA is going
to maintain the assertion that rick characterization is only the quantitative
estimation and the subjective description accompanying the estimate,, then EPA
will have to defend to position that human health and ecological risk assessment
are different procedures.
11. The guidance document needs to provide more explanation of process,
particularly deliberation and consultation.
12. There is an underlying assumption that the reader (and the process for which
the document is being used) is familiar with computer models applicable to these
cases. There is also an assumption that the reader can USE the models. I doubt
the accuracy of either assumption.
82
-------
P.L.deFur
Comments in/Response to Reviewers'Charge
•• • ' • . ' • -;,:..( •... .-. . ' -- ' • - ' • : -
General Principals
Assumptions and defaults
The guidance could identify assumptions and defaults in a single separate section
or in labelled subsections of the document. In any case, it is probably prudent to
indicate the assumptions in explicit and separate statements.
Balance . . .
1. There are a few areas that need more explanation. First, more discussion of
non-chemical stressors is needed. The document could and should point out that
any change to an ecosystem can be considered a stressor. Next, all stressors come
in combinations in the real situations that assessor face. The legal or regulatory
conditions may limit the consideration given to some or all of the stressors, or may
offer other barriers, but the guidance should acknowledge to the readers that all
can be stressors. There is less use of terrestrial than aquatic systems. EPA should
consider adding some more terrestrial examples in the text, not necessarily as new
case studies.
2. No major recommendations on the use of case studies. EPA may want to
consider where additional use could be made of the ones in the Appendix.
3. In adding cases or examples of terrestrial ecosystems, EPA may want to
consider those related to land use cases and events. Logging and wetland filling
are all too well discussed, so site selection for a facility, or transportation activities
would be common and applicable examples.
4. I found the uncertainty approach quite useful, particularly in the manner in
which it was carried through the document. I do think the document should refer
83
-------
PX.deFur
the reader to additional reading in key areas, this being one of the more important
and significant ones.
Introduction and Scope
5. The scope and content are clearly stated in the Introduction, but as written, all
the people who now rely on such guidance are excluded from the intended
audience. I suggest EPA explicitly identify the longer list of people who are "
expected to be using this document. I also believe that EPA wants state agency
regulatory staff to use this, and wants consultants and private analysts to at least
consult the document. Therefore, EPA should say so explicitly. If some legal
barrier prohibits this language, then say that.
6. Terminology changes should be clearly identified up front, and then when used.
Too much repetition of terminology changes is not possible.
7. Further modifications to the figure are required in language only. Not all of the
figures throughout the document have sufficient legend explanation and could not
stand alone as they should. In the framework figures, the language needs to
indicate multiple assessors, multiple managers, and interested and affected parties
as participants in the process. The risk characterization box needs to be inclusive
of more factors, primarily in the Risk Description that could be multiple circles.
EPA may consider adding iteration steps along the left side, indicating that there
is constant flow of information and refining of analysis as the process goes forward.
Risk Manager interactions
8. My specific comments include more details on this point. To sum here, the risk
managers have to, be involved to set the decision-making stage and inform the
analysis from start to finish. The document is overly ambitious in expecting the
goals to all be spelled out, BUT should identify such as a goal. Risk assessments
may not be able to wait until management goals have been agreed to be all the
84
-------
P.L.deFur
parties (toxic chemical controls for the Chesapeake Bay watershed, or Mississippi
River delta).
9. The relating information to the manager should not be expressed as though it
were a simply linear one-way flow with a single manager and assessor. The entire
process should be recognized as more inclusive and more iterative. Therefore, the
step to which this question refers should be a "confirming" step in which the
management team and the assessment team agree that a logical stopping point has
been reached. ,
Problem Formulation
10. The approach of the "driver" seems quite pragmatic and usable in its present
form. Some specific comments are found in the later comments.
11. It is hard for reviewers who are familiar with the assessment/measurement end
point topic to objectively answer this question. I found some places where the
distinction was unclear, and others where it was perfectly clear. I have tried to
identify these in the detailed comments. I suggest a read from someone who is not
so familiar with the issue and make text refinement and improvements.
12. I think the hypothesis component is fine. I suggest using other and more
diverse examples for the demonstration of conceptual models. It is later in the
analysis where the use of models is more difficult.
Analysis
13. The balance is in the direction of chemicals. EPA could expand the other parts
to include some additional discussion of the same type that is found in the
chemical section.
14. The chemical section should note combinations with non-chemical stressors.
85
-------
P.L.deFur
Guidance on dealing with these is especially needed in state level actions.
Multiple chemical stressors and time and space differences can and should be
noted specifically here and in the later section on multiple stressors.
15. There is an additional "principal in multiple stressors : the unknown and the
unexpected should be expected as the norm in these cases. Multiple stressors tend
to be usual, but more complicated in the more difficult situations that have less
predictability. As with biological stressors, it is the unanticipated outcome that
will undo the system at risk.
Bisk Characterization
16. The document needs to include reference to risks that are downstream,
indirect, delayed or multigenerational. Impacts that are generated in non-adjacent
trophic levels, the next generation, or in another medium (air v water v soil) must
be explicitly considered. The document should point this fact out and give some
guidance on considering such non-immediate and indirect effects.
The document should say that quantitative risk estimation is an extension of
quantitative risk analysis.
17. Ecological significance is manifested in two additional areas: predictability and
generational/evolutionary nature of the impact. On the first point, the
predictability is not the same as uncertainty as used in the present guidance,
though it could be added. Predictability refers to whether or not outcomes have
been measured or otherwise assessed, if there is a pattern and whether or not the
type and nature of actual outcome can be anticipated from the known inputs.
Many biological stressors have low predictability.
The evolutionary/generational scale of effects is an important one that must be
considered. It is now apparent that some chemicals alter the reproductive
86
-------
P.L.deFur
capabilities of entire populations over multiple generations. Changing the species
composition of ecosystems can and will often provoke much wider and longer
change that are evolutionary in nature. These should be included.
Other -- This section should not state that risk characterization excludes social,
political, economic, etc considerations, as in the last paragraph. The concept of
characterizing risk includes a great range of perspectives and expertise, ranging
from the statistical to the cultural. Loss of bald eagles may cause little ecological
harm, no economic cost and an undetermined social cost. Yet, the bald eagles will
be protected and saved for a great variety of critically important reasons. The risk
characterization described here would exclude these areas from the
characterization.
SPECIFIC COMMENTS:
[4:7] State specifically, limit with unknown endpoints
[4:ft] But are there any terminology changes that we should note?
[10:29-30] Do not expect human errors can be eliminated entirely, especially
as a regular matter of course.
[12:6-18] Paragraph is too fragmented.
[12:20] Typo: Figure noted should be 1-3, NOT 1-2.
[13:15-16] "and help .. . potential." This is too narrow, only one option. Rather
CRA is useful in understanding the landscape of issues -- risks,
stressors, and existing future concerns.
[13:19] "Relative Risk" category should be separate.
[13:23] Which is recommended?
[13:29] "impact" Is this ecological impact?
[14:21] Ballast water release
[14:28] Or any unknown outcome
[16:box 1-6] "(more clearly described ..." Clarify.
[17:1] Change to: "2. PLANNING: DISCUSSION BETWEEN RISK
87
-------
P.L.deFur
ASSESSORS AND RISK MANAGERS"
[17:6] Typo: Figure noted should be 1-3, NOT 1-2.
[17:22] Inform the analysts and the analytical process
[17:box 2-1] The text needs to conform to usage of term "general public.
[18:2] This seems ambitious. In the context of a requirement, e.g. statutory, setting
the management goal at the outset is fine. But in cases of Greenfield, new activities, de novo
or one time, the context way does not exist. For example, even now, what is the management
goal for land use in the Chesapeake Bay, for toxic chemicals in Terrebone Benataria, or living
resources in the Mojave Desert? I suggest rephrasing this to agreeing on the management
context, goal, if possible.
[18:9] "Planning is initiated . . ." Not really. How about in urban settings where
planning has been ongoing for years?
[18:19] after "local resource managers " insert: "interested and affected parties"
[18:23] add "IN PLANNING" to end of title
[19:1-8] Narrow use of the term risk manager in entire paragraph. The decisions, time
frame, severity, and nature of consequences.
[19:11-13] Here too, the public lAP's may have vital information.
[19:23] This place hi the process is where values and perceptions initially arise — that
should be the case. Intended use and value of the resource should be clearly
articulated as part of planning context.
[20:1-19] These assumptions still need to be articulated e.g. this habitat is to remain
intact and undisturbed.
[20:box 2-2] "Questions for Risk ManagMENT" and "Questions for Risk AssessMENT"
[21:27-28] Multiple activities, stressors impacts
[22:3] extent and complexity?
[22:19] Delete "The"
[22:20] Change to plural: "managers" and "assessors"
[22:29-30] Delete "the" and change to plural "managers" and "assessors"
[23:19] Add here that this is another iterative process and includes multiple intereste,
perspectives, expertise.
[24:17] "from human activities" Do not discount the impacts of natural, events, mostly
disasters.
88
-------
P.L.deFur
[24:box 3-1] "contains several changes" ADD "from the framework"
[27:32] These assessments are not without value issues, especially on the part of the
lAP's that may hold or rank different values.
[28:16-20] Give other examples in text list decline in harvested resource — e.g. crabs, or
z. mussels
[29:22] "present special challenges to ..." Expand — different conditions, especially
of process and interaction with the lAP's who may wish to be over the
shoulder of the "risk assessor"
[29:24] "early selection of assessment" Not so! It is the point above {see 29:22] and
thelAP's.
[30:13,16] ??
[30:27] Where are the values and importance?
[32:25] "the ecosystem" [ADD] and components"
[32:32] Must work out differences between ecological and political boundaries.
[33:box 3-5] bullet 5: Spatial
[35:6] text box 3-6: on this page
[36:all] Less political controversy, especially among the public and lAP's. The RA
may be admittedly easier to use with LCV's or commercial SP's, but those are
issues of practicality.
[36:8-14] Reword: Reality should be stated as such without value judgements.
[36:27] "to them or" INSERT even "find them annoying"
[36:box 3-7] Final sentence: state as an assumption
[37:9-11] Public involvement at the earliest stages is necessary. The manner and mode
of involvement will be determined by the situation. Representation may be the
best method in one case; public hearings may be required by law.
[37:17] even if only within a limited area.
[37:22-23] refer to past experience with Eco RA.
[38:8-9] Toxicity testing for ambient physical and chemical conditions also.
[38:13] Why Bamthouse?
[38:18] Include here indirect exposures that are separated by time, space, or severation
i.e. downstream sediment accumulation of high k-ow chemicals may kill fish
fry later because the chemical is passed to eggs from the female.
89
-------
P.L.deFur
[39:25-26] Change order to: "Two elements are required to operationally define an
assessment endpoint."
[40:17-18] This is where lAP's are needed.
[40:21] Reword for the positive. Change "can" to "must" and change "inappropriate"
to "appropriate."
[40:26] Suggest a way of fixing this error-reproducing population of indigenous - .
species?
[43:1-6] Why repeated almost verbatim?
[43:22] And what guidance for cases unknown, not well studied, etc.?
[43:23-30] Again, this has to be broadened beyond two folks talking.
[45:2] "The risk assessor ..." Comment: but may not be fully characterized.
[46:12] "are assumptions ..." Comment: They should not be assumptions. They
should be testable statements of relationships and predictions.
[47:box 3-12] Stressor-initiated: This is a testable hypothesis, not an assumption.
[50:26-28] Should at least reference and cite an example for information and refer to
another source for greater detail and explanation, e.g. Chesapeake Bay nutrient
model, Commencement Bay, etc.
[59:11-17] Are we referring here to interactions and relationships mat influence the
stressor, the target, the response, etc.? I am not sure of the intent.
[63:15] ADD comma after "rationale"
[63:21] ADD comma after "results"
[63:25] ADD period after "research"
[63:27] ADD period after "analysis"
[63:29] ADD period after "communicated"
[63:30-31] This is a problem! More guidance needed or something! Is Mpnte Carlo even
considered uncertainty analysis? I don't think so.
[64:11] "much of the..." ADD "activities intentionally termed" and REMOVE
"activity"
[65:chart] Not clean The sequence and flow do not seem right. Put explanatory text in
the figure at the bottom next to "Analysis example." ;
[67:10] change "may" to "should usually"
[67:11] ADD "and" after "water" ,
90
-------
P.L.deFur
[67:17] Change "They" to "Measurements(?)M
[67:19] The specifics of sampling designs, procedures, and analysis
[67:28] Hg v. MeHg
[68:9] Typo: "transforming it the"
[68:11] ADD "primary" before "production"
[68:12] ADD "if algal population crashes" after "decomposition"
[68:18] Where? High flow is the concern for some pollutants for erosion, etc.
[69:29] Other techniques may be appropriate or different assumptions may apply.
Depending on the source of nature e.g. national as in acid-rain.
[71:14] "exposure .. ."ADD "during the period of anoxia"
[71:17] After "individual" ADD "or species and distribute the exposure among
individuals or species."
[72:28-29] Multigenerational exposures must also be considered and critical periods of the
exposure, e.g. fetal, lactational
[73:31] Perhaps this is the place to flag life stage specific exposure.
[74:4-7] Pulse, sine, background, mixtures
[74:13] Also dose-response?
[75:16-22] Not clear — especially how this is number crunching
[75:31] Multi or trans generational e.g. endocrine, immune
[79: 3-4] these lines should refer to other EPA documents that are relevant, i.e. water quality
criteria
[79:10] this paragraph should refer to the intrinsic variability in systems, especially biological
ones in nature
[80: 26] make some note about data for invertebrates, i.e. shrimp, crabs, worms
[81: 1] this language needs caution as the population genetics of many species is not well
known or is poorly understood
[81: 26-30] more explanation is needed for this paragraph to help the reader; provide specific
information tihtat is needed to extrapolate or when such is not possible
[83: 8] there is an abundant physiological literature dating back many decades on allometric
scaling. One text on the topic has been written by K. Schmidt-Nielsen, and numerous reviews
by C. R. Taylor.
[84: 3-4] include reference to climatic and geological factors
91
-------
P.L.deFur
[84: 16] "to" is missing
[85: 1] refer to the current model for Chesapeake Bay modeling on the basis of trophic
analysis
[86: 15] note that this is equivalent to dose response
[88: 14] include clear cutting forests as example
[88: 20] refer to other agency's reports ~ ACOE, HRC, USDA, etc
[90: 13] what is identifiable outfall?
[93: 15] spell out the measurement and assessment end points here to help the reader
[93: 22] By this .point I am convinced that I will have to use computer models to complete any
analysis, without knowing why that is so, what good it will do, if I am correct in the
assumption and which ones to use.
[93: 28-29] I don't understand the terms
[95: 1] refer to reports by OTA and J. Carlton
[95: 3] add "in contrast to the case of POTW discharges, thermal discharges, etc"
[95:the box]: What about unanticipated outcomes of biological stressors?
[95: 17] The fact is that the "hydrilla" in the Potomac was 5 species of submerged aquatic
vegetation; please correct for accuracy
[99:15-8] expand the explanation here
[99:20] elaborate on the details
[99: 22] Open this paragraph with a more general statement
[100:119-24]. This section mixes topics so that it is confusing. Separate the topics into different
paragraphs.
[101: 5] what about temperature here?
[101: 26] I would refer to an authority on species dispersal at this point
[103: 17] make the assessor plural
[105: 2] This section needs discussion of multiple categories of stressors.
[105: 26] the following section is too chemical dependent
[108: 23] see review by R. Diaz on hypoxia in aquatic systems
[109: 2-3] I disagree that this is a problem formulation only consideration
[109: box] — another error is the exclusion of observational data.
[Ill] Generally, the Risk Characterization should summarize all the components and
processes that have gone into the assessment/analysis. RCh includes consideration of process
92
-------
P.L.deFur
as well as substance, the design and plan, the participation and review by interested and
affected parties. An analysis that simply crunches the numbers will not be able to adequately
characterize risks.
[115: 28] it is not clear what the "toxicity measurement" is.
[126: 129] I suggest referring to the overfishing and the risk assessments that did/did not deal
with that going back as far as 20 years.
93
-------
-------
James Donald
95
-------
EPA's ECOLOGICAL RISK ASSESSMENT GUIDELINES QUESTIONS
The following comments represent the collective opinions of staff in the Ecotoxicology
Unit of the California Environmental Protection Agency's Office of Environmental Health
Hazard Assessment. These staff are:
James M. Donald, Ph.D. (Senior Toxicologist)
Regina M. Donohoe, Ph.D. (Associate Toxicologist)
Karin E. Ricker, Ph.D. (Senior Environmental Research Scientist)
Julie T. Yamamoto, Ph.D. (Associate Toxicologist)
GENERAL PRINCIPLES
• Specifically, how can the ecological risk assessment guidelines be improved to more
clearly indicate underlying assumptions and default positions?
These draft guidelines present a good attempt at reviewing the status and general
advantages and disadvantages of current methodologies for ecological risk assessment. They are,
however, heavily oriented towards descriptions and discussions of general options available to
the risk assessor, as is explicitly recognized hi the opening discussion of the scope and intended
audience of the document. A similar point is made in relation to the importance of scientific
judgment in successfully completing an ecological risk assessment. This is all well and good, as
far as it goes. The major flaw in the existing draft "guidelines" is that they provide little hi the
way of actual guidance as to the optimal or preferred methods for conducting or reviewing •
ecological risk assessments. The discussion provided of the strengths and limitations of alternate
ecological risk assessment approaches is useful, but some indication of appropriate criteria for
selecting preferred, or at least acceptable, methodologies and approaches under specific types of
risk assessment would be of much greater value both to the regulated and regulatory
communities, in that it would greatly improve the consistency and appropriateness of ecological
risk assessments hi general. This is not inconsistent with the stated intent of avoiding a
requirement that certain procedures always be followed.
Specific improvements that could be made include:
Explaining rather than just mentioning methods/procedures etc. (e.g., give reference for
Habitat Suitability Index; explain how and why multivariate statistical techniques might provide
alternative approaches when measuring effects of multiple stressors; explain why the success of a
risk assessment depends on the quality of communication early in planning)
Explaining more unusual technical terms/jargon, such'as delphic; tree fault, etc: (these
could be included hi Appendix B).
Avoiding examples that are not well introduced, out of context, or where the point is not
immediately clear. Some of the examples used were confusing to the reader. Some style and
organizational changes would greatly improve the clarity of the guidelines. Due to the
96
-------
considerable size of the document, efforts to facilitate the reader's ability to focus on summary
points is critical. Currently, four basic types of summary information (important principles,
checklists, examples and supplementary text) are highlighted hi a variety of inconsistent ways
(bulleted list, italicized list, italicized headings with text, text boxes and plain text) which
distracts the reader and makes it difficult to comprehend the relative importance of the points.
Each type of information could be presented in a uniform manner that would represent the
hierarchy of their relative importance.
• What changes would you recommend in each section of the document?
Most of the comments on individual sections are incorporated into the responses to
subsequent questions. A couple of specific suggestions are:
Provide a more lengthy discussion of the analysis plan in the Problem Formulation
section. If it is not appropriate to discuss specific methodologies, then provide guidance on
limitations/advantages of different types of approaches (e.g., field data versus lab studies versus
literature search/model results) for purposes of analysis.
In the Analysis Phase section, provide discussion of monitoring. This is an important
mechanism for the iterative process, for verifying predictions (reducing uncertainty), and for
evaluating recoverability of ecological values due to mitigating actions.
Guidelines Balance
1. Considering both the present state of the science and present and future Agency needs,
how well are the guidelines balanced regarding the range of stressors, levels of
biological organization, ecosystem types, and spatial/temporal scales? Specifically,
what would you emphasize or de-emphasize?
The range of stressors covered by the guidelines is comprehensive, but the treatment of
stressors such as physical stressors and biological introductions, which may be relatively
unfamiliar to some users of the guidelines, could be expanded and made more detailed.
Similarly, dealing with multiple stressors may be the most complex and difficult challenge facing
ecological risk assessors. The discussion of current approaches, all of which have obvious
limitations, would be greatly unproved by inclusion of discussion on potential future
improvements. If no improvements are foreseen, this important issue should also be discussed.
With regard to levels of biological organization, individual to community effects were
considered, but could be discussed at more length with respect to relationships and extrapolations
between levels, and other issues/problems associated with assessing risk at each level. It would
also have been useful to have included discussion of sensitive/fragile ecosystems besides
wetlands (e.g., desert, prairie).
(See response to question #13 for further discussion of sections on stressor types)
97
-------
2. What would you suggest to improve the use. of case illustrations in the guidelines: How
useful is Appendix A in illustrating a range of applications of the risk assessment
process?
In general, the case illustrations included in the text were helpful and showed that risk
assessment can be applied to a wide variety of environmental problems. Additional general
information on the risk assessor/risk manager interface would enhance the risk assessor's ability
to consider various approaches to this interaction. Specifically, the cases studies do not
uniformly address the process and outcomes of the discussions between the risk assessor and
managers during the planning and post-assessment phases. While it is stated that details of the
cases are intentionally limited to avoid prescriptive use, the potential benefits of providing
detailed information to the assessor may outweigh the feared misuse. Ecological risk assessment
is an evolving process and consideration of past practices will help assessors develop and refine
available methods. As with the USEPA case study review (1993), outlining the general strengths
and limitations of the case assessment may assist hi preventing widespread adoption of imperfect
methods. In some cases where the information is spread and imbedded throughout the main body
of text, the example is difficult to "digest". Perhaps for some sections, the case illustration
should be presented hi a more cohesive and comprehensive way. This would also make the main
text easier to read and follow - hi some areas the examples seem to cloud the concept that they
are meant to clarify.
Appendix A is a good overall summary of the selected case studies and illustrates the
common elements as well as the range of risk assessment applications, but would have been
more useful overall had it provided more details of the examples provided. It would be useful to
have an evaluation of how well these risk assessments fit hi with the proposed U.S. EPA
guidelines or if the result would be very different if the EPA guidelines had been used to conduct
the risk assessment (e.g., a conceptual model was developed only for the Waquoit Bay Estuary
risk assessment). A critique of each risk assessment would also be useful.
3. Some Agency reviewers of the guidelines have suggested that more examples of
terrestrial assessments and field approaches (e.g., bioassessment techniques) should be
used. Specifically, what, if anything, should be added?
It is unlikely that any six case studies would encompass the universe of ecological risk
assessments. Additional case studies to represent other ecosystems (desert or semi-arid regions),
cumulative impacts of multiple stressors (overgrazing or urban/agricultural runoff) and
assessment endpoints (survival of endangered species) would be useful.
Addressing exposure hi a terrestrial situation is very different and in some ways more
complex than in the aquatic situation. Methods/approaches for exposure characterization would
therefore be useful, such as discussion of the use of exposure factors, extrapolation from
laboratory rodent studies to relevant terrestrial wildlife, and treatment of multiple exposure
pathways (oral, inhalation, dermal). Mention of commonly used terrestrial tests may be beyond
the scope of the document.
98
-------
4. Areas of uncertainty are summarized in the problem formulation (section 1.5), analysis
(section 3.7), and risk characterization (section 5.2.4). How useful is this approach in
providing guidance on uncertainty issues? „
Summarizing areas of uncertainty and providing guidance on how to appropriately deal
with that uncertainty are not necessarily the same thing. As noted above, actual guidance on
appropriate methodologies, or criteria for selecting appropriate methodologies and/or rejecting
inappropriate methodologies, would be of great value. Overall these sections are useful, but
expression of uncertainty is a critical feature of ecological risk assessment and is worthy of more
extensive treatment. These sections would be improved by inclusion of more guidance on what
types of uncertainty are qualitatively or quantitatively evaluated and on how to go about this
evaluation. Perhaps there should also be a discussion of how to define unacceptable levels of
uncertainty.
The list of uncertainty sources for each phase of the assessment is a worthwhile
classification scheme. While this is a rapidly developing area, a distillation of developed
approaches and required improvements would promote a more consistent and rigorous
quantitation of uncertainty hi assessments. Approaches towards integrating all the sources of
uncertainty to estimate or rank the overall assessment uncertainty should also be discussed.
Under Section 1.5., the reader might benefit more from an abbreviated version of the uncertainty
issues. The amount of detail provided (bullets) distracts somewhat from the focus of this
section, i.e., an overview of the risk assessment process. Just listing the types Of uncertainties at
this point would be fine. Under Section 3.7. the level of discussion is appropriate, although a
link to section 5.2.4 could be provided here.
In Section 5.2.4., it might be helpful to summarize how Table 5-1 and Table 3-1 differ,
i.e., what stays the same and where changes/progress have been made during the risk assessment
process. (It might also be useful to change Table 5-1 to include information from Table 3-1 with
an additional column of the issues that are part of the Risk Characterization Phase Strategies.
This way, a side-by side comparison would be easier).
Introduction and Scope
5. How could the Introduction be modified to more clearly communicate the scope and
content of the guideline?
The intended audience, scope and content are clearly defined in the introduction.
However, it is questionable whether the scope and content fully meet the needs of the risk
assessor. It is stated that techniques, methods and models are beyond the scope of the guidelines
but this is exactly the type of information the risk assessor needs to conduct the assessment.
While the data acquisition, verification and monitoring steps are outlined in the paradigm, they
remain a vague, black box. More detailed descriptions of the available data acquisition/analysis
methods and general evaluations of their strengths and limitations are strongly suggested. Rather
than being prescriptive, these descriptions would instead serve to illustrate that ecological risk
assessment methods are continually evolving. Further discussion would be extremely helpful
99
-------
and applicable to many program's needs. This opinion was also voiced in the recent series of
public workshops sponsored by Cal/EP A. Workshop participants consistently requested more
detailed guidance be developed in the areas of extrapolation factors, uncertainty expression,
trophic transfer factors, reference site selection and mixture evaluation. At the very least,
annotated bibliographies of available methods should be included as Appendices or
Supplements. Since the field of ecological risk assessment is in its infancy, it is probable that
any set of guidelines will become outdated and require revision. The focus of the document
should be to facilitate information transfer in order to assist in continued improvements in
method.
Several terms used and statements made were not entirely clear or understandable, e.g.,.
"In a larger sense, experience with many risk assessments can help verify the usefulness of the
overall ecological risk assessment process" (p. 12, line 29). Similarly, the definitions and usages
of "stressor" and '-'receptor" were not fully understandable to all readers. Expansion on the
statement that "primary effects may become stressors" (page 14, line 29) would be helpful. It
would also be helpful to put the use of ecological risk assessments into some regulatory context,
in addition to being presented as a general problem-solving approach. Perhaps a table of federal
mandates that require ecological risk assessments or for which ecological risk assessments are
commonly carried out could be included, as well as examples of the regulatory decisions in
which ecological risk assessments play a large role.
6. Terminology, especially related to endpoints and exposure, has always been
controversial. What changes, if any, would you recommend in guidelines terminology?
While the general definition of exposure (page 15, line 19) fits all stressor types in a
literal way, it may be open to misinterpretation. For example, chemicals must contact the
receptor not just co-occur. To avoid confusion, two operational definitions for chemical and
physical/biological stressors may be helpful.
Addition of the phrase "a process with three phases" to the definition on page 13 (line 8-
9) adds needed detail to the definition of ecological risk assessment presented on page 1 1 (line 6-
Additional terms that would benefit from a formal definition include: risk hypotheses,
risk management, problem formulation, analysis phase.
7. The overall framework figure for the ecological risk process has been retained,
although some changes have been made to the diagrams for problem formulation,
analysis and risk characterization. What further modifications, if any, are required?
The planning phase could also be considered to have a distinct output or plan (page 23).
This could be diagrammed in a separate circle, similar to other phases.
100
-------
Risk Manager Interactions
8. Please comment on the description of the risk manager's role at the initiation of an
ecological risk assessment and on the principles for selecting management goals.
As noted in text box 2-1, the risk manager can be a single person or represent groups of
interested parties. This important mformation on who can be a risk manager should be expanded
upon and incorporated into the text; the text box then could serve to provide specific examples or
summarize what was said in the text.
The statement on page 19 of the document that the level of certainty needed should be
determined by the risk manager was rather confusing. It was not clear whether "certainly" was
used quantitatively or qualitatively, or if the ability of the risk assessor to guarantee that level of
certainty would determine whether or not the risk assessment was conducted. Also, the apparent
predetermination of the required level of certainty by the risk manager seemed questionable - it
may be more appropriate to have the risk assessor work with the risk manager in arriving at that
determination.
Determining the amount of resources available for a risk assessment hi advance of
implementing the assessment can certainly help to keep the risk assessment focused and prevent
escalation of costs. Also, the risk assessor needs should be involved in determining the amount of
resources he or she needs to properly conduct the assessment. Obviously, the amounts of
resources allocated may greatly influence the quality of the risk assessment. Determining what
resources are needed should be an evolving discussion between the risk manager and risk
assessor.
With regard to selection of management goals, the relationship of "management goals" to
"legal endpoints", or "protection goals" should be clarified since, in some instances, the law
spells out what is to be protected or maintained.
Increasingly, the risk manager is called upon to" develop stakeholder involvement to
ensure that the decisions arrived at have considered the range -of viewpoints. It is recognized that
some regulatory programs have greater requirements for developing stakeholder involvement
than others, and that it is not appropriate to include in this guidance directions for risk managers
to develop stakeholder involvement. Nevertheless, a generalized comment to the risk manager
on the need to consider stakeholder involvement would seem to be appropriate.
9. What additional points, if any, should be covered about relating ecological mformation
to risk management decisions after the completion of an ecological risk assessment?
Clarify what is meant by "status of peer review explained" (page 128, line 11). This is
the first mention of peer review. If internal/external review is advocated, details of the
recommended purpose and extent should be clarified. Perhaps this should be mentioned earlier
101
-------
in the problem formulation phase where it should be noted that a team of experts will be required
to design the assessment and review the outcome.
Other discussions that might help to communicate ecological risk to managers include
reiteration and further explanations of the ecosystem at risk (e.g., driving forces, functions and
foodwebs of interest) and stressor behavior in the ecosystem.
Also, the terotns "clear", "transparent", "reasonable", and "consistent" would be better
understood with more definition, especially with respect to what each achieves in conveying risk
assessment information, and why it is important to the process (e.g., to some, "clear" and
"transparent" may have the same or very similar meaning).
10. How useful is the categorization of assessments as either stressor- or. source-initiated,
effects-initiated and ecological value-initiated?
This is potentially useful in conceptualizing and organizing the information to be
obtained hi an ecological risk assessment, and may help to point out that the sequence of events
in the process is related to the type of information available when the assessment is initiated.
However, the bulleted text boxes and use of examples throughout the text do not illustrate the
sequence differences for the three classes in a clear, concise manner. Three separate flow
diagrams might better serve as examples to illustrate the differences, followed by explanatory
text in Section 3.2.
11. Please comment on the discussion of assessment endpoints and their relationship to
management goals and "measures."
The link between these three components would be better established by adding a
"measures" column to Text box 3-9. It would also be useful to expand on stakeholder
involvement (Section 3.4.1.1). The document defines three criteria that need to be considered
when selecting endpoints: policy goals and societal values; ecological relevance; susceptibility to
stressor. While ecological relevance and susceptibility to stressors are discussed, no guidance on
how to determine societal values is provided. We recognize that specific risk management
guidance in this area is difficult and beyond the scope .of this document. It is briefly mentioned
that "public meetings during the initial stages can be useful...", but no further discussion or
procedure on how to elicit stakeholder involvement is provided. It would be useful to at least
briefly summarize the more commonly used techniques to gather this type of information as well
as discuss the importance of stakeholder involvement. At a minimum, appropriate references
should be included.
12. What, if anything, would you add to the discussion of risk hypotheses and conceptual
models to give the reader a clearer understanding of their nature and content?
The conceptual plan examples do not reflect all the recommended attributes of a
conceptual plan. Are mere better examples or would addition of a hypothetical conceptual plan
serve to illustrate the ideal situation? The section on conceptual models could be better
102
-------
organized. For example, risk hypotheses are described twice; once in the introductory
paragraphs of the section and once under the risk hypotheses heading. Also the example of a
simple conceptual model (single stressor/receptor situation) is given twice (page 44 and 45). The
relationship of the conceptual plan to the analysis plan, and how these processes differ, is not
explicit. The role of the manager/management goals in risk hypothesis and conceptual plan
formation and evaluation is not clear. While data collection and analysis has been stated to be
outside of the scope of this document, these issues perhaps should be addressed at least briefly.
For example, describe the general types of data used in conceptual plan development, and how it
is collated and assessed hi terms of applicability and confidence.
Analysis
13. How would you change this section, if at all, to improve the balance in the discussion of
the different stressortypes?
The statement that "a detailed treatment of data acquisition and model development is
beyond the scope of these guidelines" (page 58) is not explained, and appears to arbitrarily
exclude an issue of major importance to the regulated and regulatory communities. This section
could be more cohesive if the information on the different stressor types were organized in a
uniform fashion. For example, the headings used for the chemical stressor section could become
general headings under which all three types of stressors were discussed. This may make
analysis of different stressors more uniform and comparable. In particular for physical and
biological stressors, some mention of habitat quality evaluation methods would be helpful.
Additional examples for the biological stressor section would be instructive, particularly
retrospective assessments. One potential example would.be increased disease incidence in fish
and wildlife due to pathogen exposure from domestic animals, humans (sewage effluent) or
introduced species. The emphasis put on host-pathogen interactions leaves the reader with the
impression that the majority of biological stressors are microbes, particularly plant pathogens and
genetically engineered organisms. Perhaps expand the section with examples of other biological
stressors that have been encountered in the past (e.g., Zebra mussel).
Other examples of primary and secondary effects scenarios in different ecosystems for
different physical stressors would be helpful. The section could also benefit from comparison
between physical and chemical stressors (some physical stressors could be more similar to
chemical stressors than others). Table 4-3 could be expanded to include additional physical
stressors such as UV, radioactivity, noise. Some discussion of how the focus of risk assessments
for evaluating effects of different physical stressors may vary would be useful (e.g., UV radiation
vs. the building of a dam).
For both chemical and physical stressors, summary text or a bibliography of key
principles, methods and case studies would be useful. It was also not clear why the discussion on
causality (Hill's Factors) and selected points on extrapolation were placed in the chemical
stressor section when they have relevance to biological and physical stressors.
103
-------
In addition to interactions between multiple chemical stressors, it would be helpful to
address interactions between different types of stressors (i.e., combinations of chemical, physical,
and biological stressors). For example, how would the presence of a physical stressors (e.g.,
severely fragmented habitat) influence risk from exposure to a toxic chemical? ,
The statement on page 81 that uncertainty factors are empirically based is unsupported,
and the fact that they are presented in multiples of 10 suggests that there is at least a degree of
arbitrariness in their selection. References to support this statement should be provided. Also
Figure 4-4 on page 69 appears to represent only spatial or temporal intersection, but not both. It
is so poorly representative of the concept as to be misleading.
14. Are there additional points or principles that should be emphasized for either chemical,
physical, or biological stressors?
Criteria for establishing or evaluating reference or background conditions should be
addressed. For chemical stressors, discussion of bioaccumulation and trophic 'transfer factors
would be useful.
15. What additional principles would you suggest, if any, for the analysis of multiple
stressors?
In section 4.5.2 the only potential interaction discussed is additivity. For completeness,
the potential for synergism and antagonism in chemical mixtures should be mentioned. Further
discussion of potential approaches for integrating risk from the three classes of stressors over
space and time (cumulative impact assessment) would be beneficial.
Establishment of baseline or reference points for all measures should be discussed.
The focus is primarily on multiple chemical stressors in this section, or multiple stressors of the
same type. More strategies on how to isolate specific stressor effects against a background of
many other stressors, and how to combine effects of different types of stressors would be helpful,
although difficult to formulate. Also, what about the use of models for assessing complex
exposure and effects relationships?
Risk Characterization
16. What additional principles should be highlighted in the discussion of risk estimation
techniques?
Text summarizing the strengths and weaknesses of existing methods for risk
characterization would be preferred to referencing the reader to numerous citations. Also, the
statement on page 114 that any method should be critically evaluated before being applied to a
risk assessment is of great importance, but no guidance on how to evaluate such a method is
offered. This is particularly important for new methods, the use of which is encouraged when the
situation warrants. How is a risk assessor supposed to determine if use of a new technique is
warranted?
104
-------
17. The guidelines propose four criteria for ecological significance. How should this list of
criteria be modified, if at all? What additional guidance, if any, might be added to the
discussion of these criteria?
It should be emphasized that one's understanding of the ecological relevance of a finding
is dependent upon the degree of understanding of an ecosystem. If a given ecosystem has long-
term monitoring data available, more refined estimates of ecological significance will be
possible, compared to a relatively unstudied ecosystem. This type of uncertainty/bias should be
emphasized in the introductory section 5.3.2. Another criterion might be the actual nature of the
effect (not addressed although there is a subheading called "Nature and Intensity" in this section).,
that is, how important the effect is with respect to the ecosystem mtegrity/function. For example,
is the effect directly on a keystone species or on a species that is nonessential to the functional
aspects of the system. This topic is also related to that of ecosystem recoverability.
It may also be worth considering whether it is appropriate to expand the definition of ;
recovery to "partial or full return of a population, community, the ecosystem, or services
provided by a natural resource". (The incorporation of "services provided by a natural resource"
is a definition from NO AA's proposed rule for NRDA under OP A).
105
-------
Typographical Errors
pg 16, text box 1-6. Item (1) is incomplete, in that the sentence reads "both chemical stressors..."
and does not go on to identify the other consideration. Also, parentheses are opened but not
closed in the same sentence (text omitted?)
pg 11, line 21; change "ignoring" to "ignore"
pg 12, line 20; should refer to figure 1.3, not 1.2
pg 13, line 4; delete "in"
pg 32, line 24; change "affects*5 to "effects"
pg 43; text box 3-11 not referenced in section text and lacks explanatory text (missing?)
pg 48; the diagram of contaminant transport processes shows an arrow from zooplankton to
piscivorous fish, but no arrow from zooplankton to forage fish. Was this arrow transposed?
pg 50, line 21; change "up" to "upon"
pg 53, line 24; exchange "important" with "importance"
pg 63, line 14; no periods between section numbers
pg 69, line 24; heading has no section number
pg 71, line 26; heading has no section number
pg 75, line 16; insert "for" between "guidance" and "preparing"
pg 84, line 16; insert "to" before "large scale"
pg 103, line 13; delete "of the"
pg 105, line 11; change "give ecosystem" to "given ecosystem"
pg 113, line 30; insert "a" before "variety"
pg 114, line 22; change "are rely on" to "rely on" ,
pg 118, line 2; change "of type kinds" to "of kinds"
pg 124, line 4; truncated sentence
A-l 0, line 5; change "Note the these" to "Note that these"
A-13, Ime 22; change "marts" to "mats"
106
-------
Charles Eirkson, III
107
-------
C. Eirkson
Comments on Draft Proposed Guidelines for Ecological Risk
Assessment
October 1995 EPA/630/R-95/002
1. I had trouble tracking all of the questions that were presented in the
charge to reviewers. I have simply listed my comments below. Additionally,
my comments are biased toward stressor-initiated actions. I hope this will be
acceptable and that the comments provide some value.
2. Overall the document is very well written and easy to follow and read. Its
concurrent publication with new texts in ecological risk assessment (ERA) is
beneficial.
3. There is an underlying assumption that we can accurately predict potential
risks to the environment through ecological risk assessment . The document
could be improved by providing citations to cases where predictions were
made and data exist to show that the prediction was reasonably correct. In the
same vain, the document could be improved by expressing the need to do
monitoring and compare the monitoring with the predictions. This is only
way, I can see to turn the perceived art of ecological assessment into a
predictive science.
4. On page 7, there is discussion of the iterative nature of ecological risk
assessment. Although an iterative process is academically and scientifically
desirable and should be held as the ideal, it presents a concept that may not be
acceptable to some. In a regulatory atmosphere, under which much
ecological risk assessment is conducted, unrestricted iterative processes imply
the potential for unrestricted expenditure of time and resources. This is not
usually acceptable to the regulated industry who desires and is somewhat
entitled to know some specifics on the rules of the process and what is
expected. To accomplish this, it is appropriate to encourage tier testing in at
least the earlier (hazard assessment portions of an assessment). Early tiers
could be well-defined and specific cutoff criteria (e.g., safety factors,
assessment factors) could be suggested. As the tiers progress and the
108
-------
C. Eirkson
likelihood of risk increases the process would be less defined in some respects
but come to rely more heavily on probabilities of occurrences.
5. The use of the term stressor as defined should be continued.
6. The concept of sustainability should be emphasized more. The assessment
endpoints that are important for sustainability should be included. This is
important because nearly every system should be able to continue to provide
the value it was created or modified to perform, e.g., an agricultural field
should be expected to produce crops without decline.
The exception should be that the value is purposefully modified, e.g., change
an agricultural valley into a water reservoir, a forest into an agricultural
field.
7. The need to educate risk managers and decision makers should be
emphasized in the document. One of the primary obstacles to performing ERA
is the lack of knowledge of individuals with regard to ecosystem theories. If a
decision maker or manager does not understand the ecological and
toxicological concepts associated with assessment endpoints, measurement
endpoints, and management goals, then the ERA may serve only a cursory role
in a decision (see also comment 15 below).
8. On page 39, its is expressed that if a species is unlikely to be exposed it is an
inappropriate assessment endpoint. This is true but the results from effects on
an unexposed species can still serve as a measure endpoint. This was not clear
to me.
9. On page 43, the discussion that if the response of an assessment endpoint
cannot be directly measured, or be predicted from a measure of responses by
surrogate or similar entities, it cannot be assesses, causes me some
bewilderment. How can we ever hope to validate a prediction if we can never
measure the affect on an assessment endpoint.. It is very important that we
begin to recognize a need to systematically monitor for predicted outcomes.
Without some monitoring, ERA will always be perceived as an art form rather
than a science.
109
-------
G. Eirkson
10. I liked the relatively clear cut concept of assessment endpoints and
measurement endpoints. Although it is sometime difficult to accept that
measurement endpoints actually reflect an assessment endpoint, the concept is
relatively straight forward and I think helps to communicate the process by
clearly distinguishing between the two. "Measure of effect" is not as concise
as measurement endpoint. "Measure of effect" could be interpreted or
misinterpreted to be the effect on the assessment endpoint which is not
measured in the prediction. On page 53, could the "measure of" items be
categorized under measurement endpoints.
11. (On page 18 and elsewhere - uncertainties) Expressions .of uncertainty
imply that the risk assessor can validly express uncertainties associated with a
prediction. If an expression of uncertainty is valid and not just a
mathematical extrapolation, we need to expressly include the validating
information in the document Is there field data that can be used that verifies
predictions? For example, on page 83, lines 26 and 27, implies that data exist to
prove predictions, a citation of the proof is needed.
12. A case study for a stressor-initiated action which includes monitoring or
other data that verifies the predication would improve the document.
13. The sections on uncertainty are useful, primarily in emphasizing the
importance of uncertainty and the need to express uncertainty in the
assessment. A case study that demonstrates the use of uncertainty would be
useful maybe in conjunction with item 13, above.
14. The introduction section should strive to be honest about the capability of
successfully predicting environmental impacts especially associated with
stressor-initiated actions.
15. The role of the risk manager in determining the assessment endpoints is
critical. My main concern as already mention above is that a risk manager or
decision maker may have little to no knowledge of environmental science,
ecology, chemistry, etc., and make decisions heavily weighted heavily relative
to available resources (personnel and money). Routinely ecological
assessment is not well supported and the less the support the more likely an
110
-------
C. Eirfcson
inadequate assessment will be done. This is a significant catch, e.g., the fewer
the resources the poorer the assessment, the poorer the assessment the fewer
the resources. As this document indicates, ERA is a complex resource intensive
(specialized) process.
16. Categorizing assessments as stressor- or source-initiated, etc. is very
useful. It clearly distinguishes that different ERAs require different concepts,
information, and conclusions.
17. Management goals and measures can be biased relative to office resources
(see item 15 above).
18. In section 5.3.2.1, the "nature" part of "nature and intensity" is not
discussed.
Ill
-------
C. Eirkson
Editorial items:
Page 32, line 13 identity
Page 38 line 20, ending sentence with during
Page 60 and other, many periods are missing
page 68, line 9, it the inorganic
page 84, line 16, add to
112
-------
Anne Fairbrother
113
-------
FAIRBROTHER
1, General Principles
1.1 Underlying assumptions and default positions
This document provides a general guide to ecological risk assessment. Essentially, it
provides a review of the principles laid out in the Framework for Ecological Risk
Assessment and embellished in the subsequent Issue Papers. It does not provide any
new information or guiding principles.
In some sections, the underlying assumptions about a particular aspect of a risk
assessment are given but not always. Furthermore, default positions are never clearly
stated. For example, when estimating wildlife exposure, one underlying assumption
might be that only herbivores consume soil and that the default position is to assume a
2% ingestion rate. Such details are never provided.
Another example is provided on page 81 lines 17-19. This paragraph states that
"empirically derived uncertainty factors have been used...[that] range from 10 to 100.
Use of these factors is contingent on the assessor's knowledge about the chemical and
class to which it belongs". First, this is a very superficial treatment of a complex and
very controversial subject. Secondly, no guidance is given on how to select an
uncertainty factor beyond a knowledge that all assessors are assumed to have. There
have been studies conducted both within and outside the Agency regarding the use of
uncertainty factors and an examination of the empirical basis for which one to select.
Selection depends not only on the chemical of concern but even more on the receptors
of concern. Aquatic organisms and terrestrial organisms differ in the amount of
information available about chemical effects, mechanisms of 'action, etc. and so would
have different uncertainty factors applied. Furthermore, the use of uncertainty factors
is influenced by the requirements of a particular risk assessment in regards to how
precise (and accurate) the assessment needs to be. In summary, this paragraph states
the obvious but provides no guidance for how to approach a controversial issue.
Still another example is on page 79 lines 10-12 where the document states that "some
investigators...have proposed using regression analysis as an alternative approach to
114
-------
FAIRBROTHER
hypothesis testing." The reader is left wondering which approach EPA recommends
should be used. If both are useful, what criteria should be used to select which one to
apply to a particular situation?
Although the guidelines are not intended to be a "cookbook", some guidance should be
provided for how to select among the multiple options presented. For example, in
section 3.3.2 Considerations of the Ecosystem Potentially at Risk several generic
statements are made. "Ecological factors determine how to translate these
[management] goals into ecologically relevant boundaries". How should this be done?
The two examples that follow are both based on hydrological "forcing functions". If
management goals are based on maintenance of populations, should metapopulations
be considered (i.e., source /sink populations) or only local populations? Should the
ecosystem boundaries be defined as impermeable or porous boundaries (i.e.,
immigration, emigration, nutrient input/output, etc.)? In other words, there is a myriad of
ecological questions to consider when discussing the boundaries of a particular risk
assessment. The guidelines indicate that this is the case but provide no guidance on
how to address the problem. Granted, this is a complex question but so are most of
the components of an ecological risk assessment. These "guidelines" highlight this
complexity but provide little "guidance" for how to work through the issues involved.
The Issues Papers provide a good in-depth discussion of many of the issues. What,
then, is the value-added of these synoptic guidelines?
2. Guidelines Balance
2.1 Stressors, biological organization, etc.
In general, the guidelines strive for balance in terms of stressors, discussing chemical,
physical, and biological stressors. It is obvious, however, that the authors are most
familiar with chemical stressors and least familiar with biological stressors. Section 4.4
(biological stressors) refers to the use of Professional Judgment much more frequently
than for the other stressors. The statement is. made... on line 6-7 page 97 that
115
-------
FAIRBROTHER
quantitative exposure and response profiles are not generally attainable at this time for
biological stressors. This is not true for pathogenic organisms where LDSOs, LCSOs,
etc. can be generated as easily as they are for chemical stressors. Introduction of
exotic species that compete for food and habitat (i.e., are not pathogens) does, indeed,
create a problem in terms of quantification of dose-response relationships. However,
this categorical statement that quantitative methods are not available for all biological
stressors and a relatively superficial discussion of transmission factors, host
susceptibility, and available models (e.g., Bailey's book The mathematical theory of
infectious diseases and its applications (2nd edition) 1975 published by Charles Griffin
& Co., LTD) suggests that the authors are outside their comfort zone in this area.
The emphasis that EPA has on chemical stressors (of necessity, given the legislative
mandates,that the Agency works under), is reflected in the six case studies. One
concerns a biological stressor, one concerns a physical stressor, three are related to
chemical stressors and one is unclear about what the stressors are. However, given
that the preponderance of risk assessments (both within EPA and outside the Agency)
are chemical driven, this balance is appropriate.
Spatial/temporal effects are not dealt with in much detail. They are mentioned, such as
on page 33 where the Florida Everglades are used as an example of the need to take a
landscape perspective in some cases, but guidance is not provided on how to
determine what scales are/are not appropriate (see above comments). While it is easy
to make the statement that populations and communities are more relevant ecologically
than are individuals, there is no guidance given on how to develop exposure/response
profiles at these levels of ecological organization. Should distributions of exposure and
response profiles be used in a Monte Carlo-type simulation to determine a hazard
quotient for a heterogeneous population? When is it appropriate to assume "worst-
case" exposure and response scenarios and develop point estimates for population
assessments? What types of predictive population models are more or less
appropriate for particular types of risk assessments? Landscape level risk
assessments are not dealt with at all (other than the above example). How could GIS
be integrated into the process to describe heterogeneous exposure patterns over a
116
-------
FAIRBROJHER
landscape? How should habitat variability (size, shape, and distribution of patches) be
integrated into an exposure assessment? Large scale risk assessments (i.e., greater
than 10 square miles) currently are dealt with by breaking the area into smaller
"operable units". Is this the proper approach? If so, how should "operable units" be
defined? If not, what alternative approaches are available? While references are
made to other reviews (for example, John Emlen's review of population models), the
guidelines provide little guidance in how to conduct risk assessments at these higher
levels of ecological organization.
2.2 Case studies
The case studies in Appendix A illustrate a physical example, a biological stressor, one
"site with presumably multiple stressors, a retrospective pesticide assessment, a
prospective toxic substance assessment, a Superfund site, and a site with presumably
multiple stressors. Missing from this mix is an example of how to conduct a future
assessment of a new pesticidal product (either chemical or biological) — i.e., how Office
of Pesticide Programs does business. Also missing is an example of how Office of
Water set criteria for water and sediments that are protective of various forms of
aquatic life and/or terrestrial components of the aquatic food chajn.
The following additions would strengthen the utility of the examples:
• description of who the risk managers were
• identification of the stakeholders and when they were/were not involved in the
process
• an explicit description of the conceptual model, including flow diagrams indicating
exposure pathways
• a more precise description of the level of ecological organization of concern (e.g.,
"effects on wildlife" is too broad), the spatial scale, and the time frame of concern.
• a listing of uncertainty factors applied to the exposure and/or response analysis
117
-------
FAIRBROTHER
• a statement of whether a particular analysis approach would be considered
sufficient/appropriate and what might have been done better.
* For example, the bottomland forest wetlands case study used "published
literature concerning the habitats and the wildlife they support." The
assessment concluded that the "analysis was applicable only for the
selected species, and effects on other wildlife must be evaluated in a
qualitative way based on professional judgment." Was this case an
exception in having sufficient literature to provide habitat information on
species of particular concern or is this generally the case? Did the species
for which the data were available adequately predict "ecological effects"?
* The Baird & McGuire Superfund Site case study, for example, evaluated
only direct toxic effects. Is this sufficient for a Superfund evaluation? How
were Data Quality Objectives (DQOs) determined for exposure analysis?
This case study could provide a lot more detail on the Superfund process.
2.3 More examples?
As indicated above, there are some EPA applications that were not illustrated by
specific case studies. However, rather than adding a long list of case studies
(particularly since EPA has previously published two volumes of Ecological Risk
Assessment Case Studies), I would prefer to see more detail provided in the six case
studies already presented. Having said that, I think that the Waquoit Bay Estuary case
study should be replaced. Given how incomplete the case is, it does not help provide
insight into how such an assessment should be done correctly or incorrectly.
Furthermore, I urge the authors to use the case studies to show how something should
NOT be done (or was done insufficiently) as well as examples of how an assessment
should be done.
118
-------
2.4 Uncertainty issues
It is appropriate to discuss uncertainty issues in several places in the document as such
issues must be addressed more than once during the risk assessment process. There
are times, however, where more precise language would be helpful. For example,
using "increased precision" rather than "reduced uncertainty". Section 1.5 does not
really discuss uncertainty; it is addressed much more thoroughly in section 1.6.1 on
pages 9-10 and in section 4.1 on pages 61-64. I like the way the discussion has been
woven through the entire document.
3. Introduction and Scope
3.1 Modifications
The introduction and scope are well laid out but I have concerns about particular points
that were made. First, I do not think that these guidelines will replace the widely-used
Framework Report as suggested on page 2. The Framework was really a seminal
piece of work that significantly changed and focused the thought process behind an
ecological risk assessment. It is well-written and short enough to be a usable working
document. This guidelines document appears to embellish on the Framework by
extracting information from the Issue Papers, but it probably will not (and should not)
replace the Framework Report.
Apparently some underlying assumptions were made about the qualifications of an
ecological risk assessor that will be reading this document. The top of page 3 says that
the assessor will have a "basic understanding of ecology and ecological risk
assessment principles." What does this mean? That the assessor has taken an
introductory Ecology course at the undergraduate college level and read one text on
ecological risk assessment? What about a rudimentary knowledge of toxicology or
parasitology/pathology/virology? Similarly, at least a gross level of understanding of
environmental chemistry is necessary for someone who will conduct a chemical-based
risk assessment. If you are going to assume a knowledge base, you need to be more
explicit, particularly since "professional judgment" is so heavily relied upon in the
119
-------
FAIRBROTHER
guidelines (it may be instructive to do a word count and see how many times
"professional judgment" is used).
3.2 Terminology
These guidelines are backing away from the use of."Measurement Endpoints" in favor
of "measures of effect", "measures of exposure", etc.. Personally, I liked the two terms
"Assessment Endpoinf and "Measurement Endpoint" as I felt that they adequately
characterized the difference between the questions being asked (the Assessment
Endpoints) and what was measured in order to answer those questions (Measurement
Endpoints). However, I see no problem with breaking "Measurement Endpoint" into
more specific terms. It appears to me that the greater confusion occurs in the definition
of "Assessment Endpoinf and how this differs from "Management Goals". This is
discussed further in my comments related to Risk Manager Interactions. Perhaps the
term "Assessment Endpoinf should be dropped in favor of "Management Goals and
Objectives".
3.3 Framework
No further modifications are required. I particularly like Figure 1-2 showing the iterative
nature of the risk assessment process.
4. Risk Manager Interactions
4.1 Risk Managers' role
I disagree with the statement on the bottom of page 2 that a discussion of the use of
the ecological risk information in the risk management process is beyond the scope of
these guidelines. I recommend that section 1.4 on Ecological Risk Assessment and
Environmental Decision-Making and Section 2 on Planning: Discussion between the
Risk Assessor and Risk Manager be expanded to explicitly discuss how to incorporate
stakeholder involvement into the risk assessment process. This needs to be integrated
into the risk assessment process at several points, from the beginning until the end.
120
-------
FAJRBROTHER
See attached Figure for an illustration of this. By including stakeholders in the dialogue
from the beginning of the process, greater acceptance of the results of the assessment
will be achieved. While I recognize that EPA has the final decision on policy, including
risk management decisions, allowing stakeholder input to the process of endpoint
selection, information identification and analysis, and risk interpretation will result in less
public dissension and greater political support.
Text Box 2-1 defines who are risk managers. One very important category is left out:
Industry. Many companies perform ecological risk assessments in their planning
process. Life Cycle" Assessment (LCA) is an example of this. Companies that own
land frequently need to assess what the ecological consequences might be of a
proposed action (i.e., will they run afoul of any environmental protection laws?). There
are many reasons why such assessments should be done and it is inappropriate to
assume that all environmental risk managers work for public agencies.
On page 19, the word "resources" is used in a confusing manner. On this page it
generally is intended to refer to monetary resources, but in an ecological context the
word "resources" has a much different meaning. Therefore, I suggest substituting or
inserting the word "monetary" where possible.
Page 19 lines 9-15 that suggests that the risk assessor provides insight to appropriate
scales, values, and focus for the ecological risk assessment. I believe that it is the role
of the risk manager to define these attributes as they are part of the question that is
being asked. The risk assessor can help the risk manager (and associated
stakeholders) to clearly define these implicit attributes through a dialogue based on a
generalized conceptual model of the system at risk. For example, the risk manager
may not be aware that a desired fishery is dependent upon the maintenance of a
particular benthic community. The risk assessor could make these types of interactions
clear so the manager better understands the system dynamics. However, the
Assessment Endpoints should not be defined by the risk assessor, but rather by those
who must make the management decision and/or have other reasons for involvement
in assessing potential risk. Perhaps this is what is intended by the discussion in lines 1-
8 on page 19, but I am left with the overall impression that the risk assessor
121
-------
FAIRBROTHER
participates too heavily in defining the questions and that the risk manager (and
stakeholders) do not participate enough.
In Text Box 2-2, the following questions should be addressed by Risk Managers, not
Risk Assessors:
• what is the scale of the risk assessment?
• what are the critical ecological endpoints, and ecosystem receptors?
• what is the nature of the problem?
Page 21 discusses that significant interactions among diverse groups are required for
setting management goals for "places". This is an excellent discussion and should be
expanded further to include all ecological risk assessments. This discussion could be
expanded to provide guidance on employment of expert facilitators for public meetings
and goal setting; methods for consensus building (e.g., Delphi methods, Vital Issues
Panels, etc.); methods of information elicitation (e.g., structured interviews or
questionnaires). This portion of the risk assessment should not be done by a biologist,
but rather by a social scientist trained in these methods.
4.2 Risk communication
Additional points about relating ecological information to risk management decisions
after the completion of an ecological risk assessment include:
• an admonishment that the ecological risk assessment is only one input to the risk
manager's decision-making process. Human health concerns, economics, social
and religious factors are some of the other inputs to the decision process
• a recognition that the risk manager may not always select the alternative leading to
lowest environmental risk. However, if the risk manager makes such a decision on
the basis of complete, accurate, and precise information (within the constraints
imposed on the risk assessment), then the risk assessor has done a proper job.
122
-------
FA/RBROTHER
The success of the risk assessment should hot be judged on the basis of which
option was selected.
• a discussion of risk/benefit analysis, risk perception versus calculated risk, and
expanded discussion of cumulative risk and comparative risk.
5. Problem Formulation
5.1 Assessment categorization
The categorization of assessments as stressor/source-initiated, effects-initiated, and
ecological value-intimated is quite useful. It helps to categorize the questions being
asked.
I have .argued for years that an effects-initiated "risk assessment" is not a risk
assessment. Risk, by definition, is the probability of an adverse effect occurring in the
future. An effects-initiated assessment has a 100% probability that an effect occurred;
therefore, there is no risk. The purpose of this type of assessment is to diagnose the
cause of the observed effect, not to predict if a particular action will (or will not) have an
undesirable consequence. For example, when you are sick and go to the doctor for
help, you don't ask him/her to do a risk assessment of your condition. Rather, you ask
him/her to diagnose the cause (and hopefully propose a cure). The effects-driven
situation is analogous. Here, a deductive reasoning method is used where all potential
causes are listed (a "differential diagnosis") and then systematically eliminated until one
remains with a causal linkage to the observed ill-effect. Recommendations are then
made on how to break the stressor-receptor link and so remediate the problem. This is
fundamentally different from a predictive risk assessment where potential future effects
of one or more stressors on one or more endpoints are evaluated. Therefore, I will
continue to argue that effects-driven assessments (e.g., Superfund site evaluations) be
given another appellation (e.g., diagnostic ecology) and not labeled" risk" assessment.
123
-------
FAIRBROTHER
5.2 Assessment endpoints
Assessment endpoints should have a direct relationship to management goals as they
are the formal articulation of the ecological components that are of concern to the risk
manager and related stakeholders. As mentioned above, I see nothing wrong with
keeping the Framework terminology of "assessment" and "measurement" endpoints.
The statement on page 43 line 17-18 that "the best assessment endpoints are those for
which there are well-developed test methods, field measurement techniques, and
predictive models" is true but must not be taken too literally. The best assessment
endpoints have these attributes but also are direct and succinct representations of the
ecological concerns of the resource managers and stakeholders. If they are only those
endpoints that are easily measured, then we will end up defining our questions by what
we can measure, rather than defining our techniques by what questions we are asking.
5.3 Risk hypotheses and conceptual models
The Conceptual Model discussion (Section 3.5) was very good. I do not, however,
understand how there can be more than one conceptual model for a particular
ecosystem (top of page 46). This discussion may result from confusion caused by an
effects-driven assessment, where there are multiple hypotheses about what has
caused the observed effect (i.e., differential diagnosis). Once again, this refers to my
argument about why an effects-driven assessment is not a "risk" assessment.
Figure 3-4: Why are there two arrows from the box Ingest/on of Particles by Birds to
the Death box? A box labeled Ingestion of Invertebrates is missing and should be
placed just below the Ingestion of Particles by Birds box. Similarly, the Ingestion box in
part a) should include soil invertebrates as well as particles.
The Analysis Plan (section 3.6) would benefit from a discussion about the usefulness
(or not) of phrasing questions as null hypotheses. This is the method employed in
scientific investigations but there is considerable argument about its applicability in
ecological risk assessment. If the hypothesis is phrased as a "null", the burden of proof
is then to prove that there is a probable risk. If it is phrased as a "positive", then the
124
-------
FAIRBROTHER
burden of proof is to show that there is no probable risk. Innocent until proven guilty or
guilty until proven innocent?
Page 53 lines 4-5 states that the analysis plan includes a clear description of
assumptions made during the development of hypotheses and models. This is an
excellent comment that cannot be overstated.
Human health risk assessments provide significant guidance on selection of Data
Quality Objectives (DQOs). This discussion is missing in this document. Is there some
reason why it is not applicable to ecological risk assessment?
6. Analysis
6.1 Balance of strcssor types
Much more space is given to chemical stressors than to any of the other stressor types.
This, of course, reflects EPA's mission and the state of knowledge -in regards to
applications of ecological risk assessments. The section on biological stressors
certainly can be expanded as a great deal is known about conducting monitoring,
surveillance, and assessments of insect pests, viral pathogens (particularly
arboviruses), arid bacteria (e.g., Lyme disease). Less is known about genetically
engineered organisms but EPA already has published guidelines on how to conduct
ecological risk assessments of GEMs. There is a lot of descriptive a posteriori
assessments of larger introduced species (e.g., kudzu, starlings, or zebra mussels), but
our ability to conduct detailed risk assessments is most limited in this area.
Nevertheless, many of the same predictive modeling techniques used for assessing
potential risk of chemical receptors could be used to assess risk of biological stressors
(particularly community models). Therefore, this section could be expanded.
125
-------
FAIRBROTHER
6.2 Additional points or principles
6.2.1 chemical stressors
Page 68 line 22 introduces the concept of a "reference system." This concept needs to
be developed further and guidance given on whether or not reference systems are
appropriate. It is alluded to again on line 5 page 80 but is still not discussed in any
detail. If reference systems are appropriate, how are they selected (i.e., what are the
criteria used)? How many reference systems are needed; is one sufficient? How does
the use of a reference system incorporate biological variability and the changing nature
of ecosystems? If the purpose of a reference site is to determine if an area of concern
is different from the reference area, then a single site is enough. However, if the
purpose of reference systems is to determine if the area of concern differs from the
norm, then a sufficient number of reference sites must be characterized in order to
determine the natural variability of the system. Guidance is needed on how to
determine the need for a reference condition or site for various kinds of ecological risk
assessments.
Page 70 lines 24-27 state that tissue residue information is difficult to interpret without
information on the chemical's distribution and metabolism and knowledge of the
organism's behavior. The real difficulty in interpretation of tissue residue data is
knowledge of the dose-response relationship. What effect is associated with a liver
concentration of 32 ppm of some chemical? The literature is replete with information
about tissue residues but is depauperate in regards to dose-response relationships.
The last sentence of the first paragraph on page 71 says that "These [biomarker]
measurements, however, can provide valuable confirmatory information that exposure
has occurred." What is being confirmed? The beginning portion of the paragraph
stated that interpretation of tissue residue data is difficult. Are the biomarkers
confirming the tissue residue data? This is not clear.
Page 72 (lines 29-30) to page 73 (lines 1-5) discusses use of averaging concentration
and contact rates to develop an estimate of exposure over time. Another approach
would be to use distributions of concentration and contact rates in a Monte Carlo
126
-------
FAJRBROTHER
approach. The guidelines suggest using toxicodynamic models in an "extreme case".
Guidance is needed to define an "extreme" case, more discussion of toxicodynamic
modeling would be useful, and an appreciation of Monte Carlo analysis and its potential
for application would be helpful
The next paragraph on spatial heterogeneity would benefit from a discussion about the
applicability of GIS-based techniques - concatenation of exposure estimates within
individual polygons distributed throughout a heterogeneous landscape. A statement
such as is given in lines 10-12 that "A general solution to the problem of incorporating
pattern into ecological assessments has yet to be developed; this issue is normally
addressed on a case-by-case basis" skirts the issues and does not provide any
guidance in this area.
Similarly, the exposure profile section (4.2.2.4) says that the exposure profile may be
expressed as a point estimate or distribution. It would be helpful if guidance were
provided for when a point estimate would be better than a distribution or vice versa.
The subsection on "summarize the most important exposure pathways" could use some
embellishment (e.g., is an "important pathway" determined by being the largest
contributor to the total loading to a receptor?).
Line 27 on page 73 says there are 3 general dimension so exposure: intensity, time,
and space. I believe time should be subdivided to duration and periodicity, resulting in
4 exposure dimensions.
Estimation of primary effects (section 4.2.3.1, page 75) discusses toxicological effects,
not ecological effects. While all chemicals cause ecological responses only as a result
of toxicological effects on individuals, I believe this should be clearly stated so that the
ecological focus of the risk assessment does not become lost in the toxicological
details. Indeed, the sentence in lines 28-30 that sates that "...the characterization of
ecological effects for chemicals has concentrated on evaluation of effects that are
readily observed" contributes to this confusion as the document then goes on to state
that these effects include mortality, growth, and reproductions all of which are biological
responses. Also, throughout the document, growth is always included on par with
127
-------
FAIRBROTHER
mortality and reproduction. While growth is an extremely important parameter in fish
and is directly related to longevity and reproductive performance, it is of much less
importance in birds and mammals that cease to grow once they reach adult size.
Lines 6 - 20 on page 76 discusses environmental factors that alter bioavailability of
compounds (such as metals). This discussion should be in the Exposure section not in
the Effects section.
Lines 24-25 assert that a risk assessor needs to be familiar with test protocols and
Good Laboratory Practices (GLPs) in order to avoid being fooled by anomalous data.
While 'I agree that familiarity with the test procedures, assumptions, and shortcomings
is essential, I am not as certain that GLP familiarity is needed. GLPs are procedures
designed to enhance accountability and record-keeping; they have no bearing on
whether the study was correctly designed, only if it was carried out according to the
written protocol. •
The section on Structure-Activity Relationships (page 77) should be caviated to say that
they are most applicable to aquatic systems and may or may not be useful for terrestrial
systems. Guidance could be given for what type of ecosystem is amenable to this form
of estimation (e.g., freshwater only? salt water as well? terrestrial?).
Page 79 lines 8-15 discusses the uses and limitations of hypothesis testing. However,
the reader is left wondering what EPA's guidance is on when to use regression analysis
or when a hypothesis testing approach would be more appropriate. As these are
meant to be guidelines, some indication of the criteria to use for methods selection
should be provided.
Page 79 lines 26-30 discusses field experiments and includes laboratory testing with
media collected from the field. I do not categorize such bioassays as field studies.
They are laboratory studies under well-defined (and generally static) conditions that
happen to employ media collected from the environment. A discussion of bioassays
may warrant a separate section or may be included in the previous sections on single-
species or multispecies assays.
128
-------
FAJRBROTHER
Page 81 lines 1-.11 discusses extrapolations between responses, in particular the
application of the acute-to-chronic ratio and MATC. The last sentence indicates that
"although developed for aquatic organisms, the approach could be used for terrestrial
organisms as well". However, this has not been tested and I am not sure what the
empirical basis is for making such a statement at this point. Given that this is a
guidance document, I would suggest the use of caution in suggesting that the approach
be extrapolated to terrestrial systems in a wholesale manner.
As stated previously, the paragraph on uncertainty factors provides no guidance on
how they should be applied (lines 17-19 page 81).
Pages 82-83 discuss extrapolations between taxa. On page 82 line 24, the statement
is made that "in lieu of actual testing, the only practical approach is to attempt to
extrapolate...". When should "actual testing" be done? Should it always be done but is
limited due to monetary constraints (or endangered species problems)? The top of
page 83 suggests the use of allometric scaling factors for interspecific extrapolations.
However, this does not take into account many toxicodynamic differences that
influence chemical uptake and elimination. For example, a moose is not merely a large
mouse as a moose is a ruminant and a mouse is not. This difference in gut physiology
profoundly affects metals uptake and, consequently, sensitivity to dietary
concentrations of metals. AHometry provides an incorrect estimation in this instance.
Furthermore, Pierre Mineau (Head of Pesticides Registration, Environment Canada)
recently did an analysis of allometric scaling factors for birds exposed to
organophosphorus, carbamate, organochlorine, or anticoagulant pesticides and
determined that almost all the values were greater than one (about 1.4), rather than
0.68-0.75 as determined from mouse-to-elephant mammalian extrapolations.
Therefore, I suggest caution in recommending the wholesale adoption of allometry for
interspecies extrapolations. This is stated in lines 17-18 on page 83, but it is not clear
by what is meant that "allometry adjustments should be used only as one part in the
overall process of estimating interspecies differences." What are the other parts? How
much weight should be given to allometric estimation versus other approaches?
129
-------
FAIRBROTHER
Page 85 section 4.2.3.4 on Causality states that in effects-driven assessments the
assessor should present evidence that a particular stressor could have caused the
observed effects. Generally in diagnostics, potential causes are ruled out, not in, and
those that remain are considered plausible explanations. This returns to my argument
that an effects-driven assessment is a diagnostic riddle, not a risk assessment.
6.2.2 physical stressors
The introduction to this section (page 88) states that physical .stressors include the
"exploitation and harvesting of resources". "Exploitation" is a value term and should not
be used in this context.
Line 19-20 on page 88 should include industry as a partner with EPA and other
agencies in evaluating issues related to physical stressors.
Page 93 lines 19-21 recommend the. use of Habitat Suitability Indices (HSIs) developed
by the U.S. Fish and Wildlife Service as particularly useful for evaluating impacts of
disturbances on wildlife species. However, there are limitations that should be
discussed. Foremost is the fact that HSIs have been developed for a relatively limited
number of species, almost all of which are found in the eastern U.S. Secondly, HSIs
are very simplistic models and may be most appropriate for screening-level risk
assessments.
6.2.3 biological stressors
Guidance is given on page 104 line 17-18 to take a delphic approach coupled with a
knowledge of past case histories when determining risk from biological stressors. This
approach really is no different than what is frequently done in chemical risk
assessment, although it is not usually identified as such. Furthermore, there are two
inherent properties of biological stressors that are. sufficiently different from chemical
stressors to warrant a more detailed discussion. These are reproduction and dispersal.
Guidance should be given to pay particular attention to reducing the uncertainty in
these two aspects of the assessment. Theories of extinction biology could be invoked
130
-------
FAIRBROTHER
to provide methods for determining the minimum viable population size. Finally, the
concept that an introduced species may have become a valued part of an ecosystem
needs to be discussed, not only in light of new biological stressors but also in regards
to interactions with other agents. These concepts are all discussed in Simbertoff and
Alexander 's issue paper and I see no reason why they should not be treated with as
much detail as is done for chemical stressors.
6.3 Multiple stressors
The primary shortcoming of this section is that it discusses additivity of stressors but
does not address antagonism or synergism.
Page 109 line 29-30 ends the .section with the intriguing statement that "multivariate
statistical techniques might provide alternative approaches." They most certainly
would. What techniques are available? What guidance can be provided for when to
use which one(s)? For example, when is principal components analysis better (or
worse) than classification analysis? A whole section should be devoted to this issue as
it will significantly help in providing guidance for how to look at effects of multiple
stressors on multiple receptors.
The last sentence of this section (lines 13-15 page 110) says that the risk assessor
may have to rely on weight of evidence analysis to evaluate the causal factors involved.
What type of evidence? How is each piece weighted relative to other pieces of
evidence? Is this "professional judgment" or is it based on empirical information? This
sentence, as written, does not provide guidance and only makes a vague statement
about the need to consider all available information.
131
-------
FAIRBROTHER
7. Risk Characterization
7.1 Risk estimation techniques
Overall, this section was quite good. The bullets on page 114 clearly stated that the
accuracy of the data is as important as the methods used (i.e., garbage in = garbage
out) and that a range of risks is preferable to a single-point estimate. Excellent points.
The paragraph from line 30 on page 114 to line 3 on page 115 needs some
clarification. First, the word "benchmark" needs to be defined for ecological risk
assessment as- it gets used in many ways, all of which are different from the human
health definition. Second, the quotient method does not always adjust the risk quotient
by an uncertainty factor as stated here. Third, a statement that "the higher the quotient
the higher the risk" needs to be made with some caution, as a risk quotient of 100 does
not mean that a risk is 10 times greater than a risk quotient of 10. There is a lot of
information imbedded in these numbers so they do not necessarily scale in a linear
fashion (if at all). Lastly, the actual relationship of the risk quotient to the ecological risk
is not understood and should not assume to take any particular function. An increase
in the risk quotient may not necessarily mean an increased ecological risk (e.g., if an
LC50 is used in determining the quotient, an exposure value that is 100 times greater
than the LC50 may not confer 10 times the risk of an exposure value of 10 times the
LC50 as many environmental factors may alter the exposure-response relationship).
The sections on point estimates (section 5.2.2.1) and distributions for exposure and
effects (section 5.2.2.2) are very good and provide some good guidance of when one is
used and another is not.
Section 5.2.3 on simulations should have the following cautions added on line 17 page
118:
However, simulation models are only as good as the assumptions on
which they are based. They should be treated as hypothetical
representations of reality until appropriately tested with empirical data.
132
-------
FAIRBROTHER
7.2 Ecological significance criteria
As a genera! comment, I assert that "ecological significance" is not an inherent property
of any ecosystem but rather is a human, anthropocentric construct. One component of
an ecosystem is no more "significant" than another unless "significance" is defined in
terms of maintaining the general structure of that ecosystem (i.e., the desire to maintain
a wetland as such and not turn it into an upland). Having said that, however, I would
generally agree with the statement on line 23-24 page 122 that ecological significance
is defined as "changes in an ecological component that are important in terms of the
structure and function of the ecosystem." (note that the grammar of this sentence as
originally written is incorrect as the word "its" refers to the ecological component, not to
the ecosystem at risk).
Otherwise, I think that this discussion of ecological significance reads well although the
"four criteria" for ecological significance are not clearly stated (bullet statements would
help). I assume this is referring to:
• nature and intensity
• scale (spatial and temporal)
• recovery potential
• natural variability and disturbances
I have not identified any other criteria to add to this list.
8. Summary comments
This document raises many options for the issues associated with ecological risk
assessments but does not provide guidance on how to select among the options. In
many cases, good review papers, books, and other documents are available to provide
in-depth analyses of when to use what approach. For example, there are many
writings dealing with the application of multivariate statistics to biological questions. At
the very least, these references should be provided. Merely raising the issues for
133
-------
FAIRBROTHER
consideration does not provide any more guidance than currently is available as most
of these issues (and others) are discussed in the published Issue Papers.
These guidelines should not be a cookbook that states which option should be used.
Rather, they should provide the criteria for how to choose among options. What is
considered appropriate and state-of-the-art today is likely to change in the future as our
thinking about ecological processes, structure, and function evolves along with our
knowledge of stress ecology. New techniques for measurement of exposure and
effects will be developed that may change how we approach the Analysis phase of the
risk assessment. These guidelines provide the opportunity for laying the groundwork to
direct innovative thinking in ways that will result in ecological risk assessments correctly
tailored to the questions being asked.
134
-------
135
-------
-------
Alyce Fritz
137
-------
Alyce T. Fritz
206/526-6305
Comments on Draft Ecological Risk Assessment Guidelines
Parti. General Principles
The important contributions of this document include its thorough explanation of
management goals, assessment endpoints, the new "measures" terms, and the emphasis
on careful development of a conceptual model and analysis plan. However, there are
inconsistencies in the use of terms (perhaps due to the inherent difficulties of multiple
authorship) and a lack of emphasis or consideration of the potential effects of the
stressor. The specific comments below detail potential sources of confusion and offer
suggestions to clarify the text.
The use of text boxes and full-page tables in this document is generally very effective.
However, these graphics occasionally distract the reader's attention rather than reinforce
an underlying principle. For example in the Analysis phase, the excerpts from case
studies (particularly for Physical and Biological Stressors) should be interwoven in the
context of the text to reinforce the major points. The text boxes sometimes appear as
disconnected items; the information enclosed should be able to stand alone or be
carefully referenced to the issue in the text. The text boxes in the previous Framework
Document are very effective in highlighting steps or issues that need to be addressed to
ensure a technically adequate ecological risk assessment for any discipline or EPA
program. The Draft Ecological Pvisk Assessment Guidelines document further outlines
several series of detailed considerations for steps in the ecological risk process. The
Guidelines provide specific questions for risk assessors to be addressed,when planning
and implementing ecological risk assessments for different types of stressors. These
series of questions or recommendations could be highlighted in text boxes (as noted
below in specific comments). Ideally, a selected group of these text boxes and tables
could be excerpted for an executive summary or as a guide to pages to be used for
"quick reference cards" for ecologists and risk assessors in the field.
The last chapters should reiterate that this document provides more detailed guidelines
to improve quality and consistency of the ecological risk assessments within EPA and
for other agencies. It should be emphasized that there are numerous benefits in using
these guidelines for advancing the current state-of-the-art and for promoting the
exchange of information and lessons learned. The Risk Characterization chapter (5) now
ends rather flatly with the knotty problem of fishery population dynamics and an issue
138
-------
Alyce T. Fritz
206/526-6305
of natural variability that should be addressed in conceptual model development and
the analysis phase. The reader could be left with the idea that the difficulties of dealing
with complex ecological interactions may make it impossible to determine ecological
risk and management alternatives. The Risk Characterization chapter should conclude
with a summary of the process with an emphasis on how professional judgment is
necessary to estimate and describe risk, particularly in terms of ecological relevance.
This section should end with a positive discussion of how, given the multiple sources of
uncertainties and inherent ecological variability, ecological risk assessment for different
types of stressors can be conducted, improved, and communicated in a credible fashion
to risk managers. Many lessons could be learned and shared, particularly if guidelines
are widely used and ecological risk is communicated effectively as described in Chapter
6.
Problem Formulation
Question 10. The description of effects-initiated assessments in the text box example is
not convincingly different from the stressor-initiated example except that it includes an
ecological effects component. If effects are observed, causes may be explored, but then
this appears to become a stressor-driven assessment. The examples both use pesticides
and thus appear to fit under the stressor-initiated (text box 3-3 is a carbofuran
assessment) and effects-initiated assessments. Using a pesticide for both examples may
cause confusion because we automatically associate predictive assessments with this
type of stressor. It might help to emphasize the assumption in the text box for the
stressor-initiated example that an adverse ecological effect has not occurred, or at least
has not been detected/measured. The goal is to predict risk in order to prevent adverse
effects. Source-driven assessments (i.e., point-source discharges) may not share the
same characteristics if releases are potentially causing adverse effects. Perhaps these
should be a separate category. The text boxes provide useful examples of the different
components of problem formulation and analysis phase. The terms "predictive" and
"retrospective" also may be useful here in explaining the different approaches. This
section could distinguish stressor-initiated or source-initiated (predictive assessments
such as for pesticides or point sources) from effects-initiated (which can be both
predictive and retrospective, particularly for Superfund), and very complex, ecological-
value-initiated assessments (where assessment endpoints need to be tied closely to well-
defined ecological values and management goals). The discussion of differences could
clarify the terms "stressor," "source," and "effects-initiated." Such a distinction could be
139
-------
Alyce T. Fritz
206/526-6305
beneficial to risk assessors who are developing management goals, assessment
endpoints, and conceptual models. However, these assessment categories are not
carried through the whole document. Although this is not a fatal flaw, it is unclear how
critical these approaches are to developing analysis plan, analysis phase, and risk
characterization. This section represents an opportunity not only to discuss differences
in approaches but also to stress that all risk assessments should be as site-specific as
possible, using field-investigation data where feasible.
Question 11. As mentioned in General Principles, the discussion of assessment
endpoints, management goals, and "measures" is a particularly valuable part of the
guidelines. Specific comments are offered below to help clarify the discussion of
conceptual models and assessment endpoints.
Pages 6_and 7, Figures 1-2 . Iterative nature of ecological risk assessment.: Iterations in
the risk process are necessary not only to reduce uncertainty but also to address different
types of risk. For example, once baseline risk is established in Superfund, another
iteration of the process must occur in consultation with the risk manager. In this
iteration, they evaluate the risks associated with various cleanup alternatives (e.g.,
predominantly physical stressors rather than chemical stressors). They must consider
rates of recovery under different cleanup scenarios (disturbance) and evaluation of risk
reduction. The measure of effects then ideally will not only be used to estimate risk but
to document ecological recovery or effectiveness of risk management and remedies.
Page 24,_Section 3.1. line 22: Conceptual models should describe relationships between
stressors and assessment endpoints.
Page 24, text box 3-1: Conceptual models should describe pathways for contaminants at
the site to reach receptors AND the mode of toxic action (how the contaminants will
affect receptors). The outcome of the conceptual model should be assessment endpoints.
Although some preliminary assessment endpoints might be proposed before the
conceptual model is completed, final assessment endpoints cannot be identified without
a conceptual model.
Page 24. Figure 3-1: I would add the extent and magnitude of contamination to the
source and stressor characteristics. The arrow between the assessment endpoints and
140
-------
Alyce T. Fritz
206/526-6305
conceptual model should go both ways (see above comment on the text box). In the
analysis plan circle, the concept of "system/habitat" is unclear.
Page 26, third paragraph: The discussion of uncertainty is unclear. Much of the
uncertainty in the problem f ormulatioh phase is due to missing data. These data gaps
can be filled in the analysis phase to reduce uncertainty.
Page 28, line 16: This requires that a stressor be identified, turning the assessment into a
stressor-initiated assessment.
Page 28, lines 25 and 26: Assessment of reduction of breeding birds from logging or
spread of a parasite that kills trees are both stressor-initiated assessments (logging and
parasites being the stressors).
Page 29, line 7: The term "damage" should be changed to avoid confusion with natural
resource damage assessment.
Page 3L lines 4 and 5: The "sources" and "stressors" in these examples are confused.
Oil (or PAHs) is the stressor in line 4; spills and tar pits are the sources. Similarly, air
pollutants and sediments are the stressors in the next line; automobiles and non-point
sources are the sources.
Page 32^ line 15 and line 19: These sentences should be clarified to be consistent with
section 3.4.2, which indicates that assessment endpoints are not necessarily equivalent to
receptors.
Page 33, text box 3-5: "Type" should indicate chemical, physical, OR biological.
Page 34, line 24: "affects" should be changed to "effects."
Page 35. section 3.4: This section should be consistent with section 3.4.2, which indicates
that the terms "assessment endpoints" and "receptors" are not interchangeable (for
example, page 35 lines 19-20; page 37 line 23; page 39 line 1; page 39 line 14; and page 45
lines 4-6). This first paragraph indicates that assessment endpoints should be chosen
before conceptual models are developed. This may result in a risk assessment that
141
-------
Alyce T. Fritz
206/526-6305
ignores risk to sensitive receptors. For example, bird reproduction may be chosen as the
assessment endpoint at a wetland site containing PCBs. A conceptual model may be
developed in support of reproductive effects to birds when, in fact, aquatic organisms
may also be at risk.
page 35, line 26: The term "significantly" is unclear. This is also a problem on page 37
line 18.
Page 37. line 10: "Selecting effective assessment endpoints" should be changed to
"identifying potential assessment endpoints." Although it is necessary to ascertain what
the public views as an important resource to protect, the actual assessment endpoints
should not be selected at public meetings. It is essential that professional judgment and
ecological risk assessors be used to make those decisions so that components or
receptors in ecosystems are not ignored because of lack of expertise in ecological
sciences. The Suter reference implies that wetlands would not have been figured into
assessment endpoints 30 years ago because of negative public perception. This is
particularly disturbing given that ecologists and biologists were very aware of their
value and critical functions when wetlands filling was legal. We have a responsibility to
the public to protect the environment and, in turn, to solicit their input to help prioritize
ecological values. However, this responsibility should not prevent us from using
appropriate endpoints. We need to communicate how we are going to conduct good
science or scientific evaluations so that the public can understand why seemingly
unattractive or insignificant creatures (e.g., midges) are important in a trophodynamic
structure that ultimately supports widely valued fish and wildlife.
Page 37, last paragraph: If assessment endpoints are not ecologically relevant, the
results may predict risk but money will certainly have been wasted identifying risk to
irrelevant ecosystem components. Are there any examples where an irrelevant
ecosystem has been assessed? Sometimes an assessment endpoint is chosen involving a
receptor that is not actually affected by the stressor (rather than the truly sensitive
receptor) and risk is potentially underestimated. Although this would be inappropriate
for the particular ecological risk assessment, the ecosystem component is not irrelevant.
This is why professional judgment and interaction of qualified ecologists are necessary
during all phases of problem formulation and risk evaluation to avoid these pitfalls.
142
-------
Alyce T. Fritz
206/526-6305
Section 3.4.2: This is an excellent section that will be a major contribution to future risk
assessments. It should be clear that it is possible to have more than one assessment
endpoint for any assessment.
Page 40, last paragraph: Definition of assessment endpoints: The sentence discussing
"the boundaries of an assessment to 'genetic exchange in regional populations' of an
organism rather than 'genetic exchange in local populations'" is confusing. Either
phrase helps define the assessment endpoint dearly. The difference is in the scale,
which has implications for the level of variability to be addressed in the analysis phase.
The examples in Text Box 3-10 do not reflect the idea of "Common problems
encountered in selecting assessment endpoints." Highlighting examples is a good idea
but the "Ecological resource is better as a measurement endpoint (e.g., midges
example)" and the third bullet with (e.g., turkey and deer example) is confusing. For
the midges example the suggestion was to use salmon, which is also an ecological
resource, albeit more "charismatic" or appreciated. Of course, NOAA appreciates both
the midges and the salmon as ecologically significant.
Page 43. first paragraph: This paragraph should be deleted because it is appears earlier
in the report.
Page 43. lines 17-19: The statement "In many applications, the best assessment
endpoints are those for which there are ..." is used to reiterate three qualities necessary
to provide good information to evaluate the endpoint. However, I disagree that all three
conditions must be met. If effective tests and field measurement techniques are
available, then predictive models are unnecessary. Perhaps, instead of using a direct
quote, these ideas could be captured or reorganized in a sentence reflecting the authors'
opinions.
Page 44: The bullets should include the fact that conceptual models should include the
mode of action of the stressor.
Page 44, line 30: In addition to the three basic elements, pathways and effects must also
be identified.
143
-------
Alyce T. Fritz
206/526-6305
Page 45. line 1: This line should be changed to read "identify where hypotheses and
data must be generated."
Page 45. first full paragraph: This description of the conceptual model does not include
potential effects of the stressor (conceptual models must include more than exposure
pathways).
Page 45. text box 3-12: "PMN" should be spelled out. The "effects-initiated" example is
really a carbofuran (stressor-initiated) example. The hypotheses in the text box should
be simplified. For example: "Reproduction (number of viable offspring) of birds
exposed to chemical A will be reduced when compared to that of unexposed birds."
Since these risk hypotheses are to be tested using measurements, they should actually be
phrased as "null hypotheses."
Pag6j46, Section 3.5.2: Although flow diagrams are a useful component of conceptual
models, the diagrams themselves do not comprise the model. At best, the flow diagrams
can only illustrate some of the relationships. Typical examples include food-web
relationships and contaminant-migration pathways. Representations of adverse effects
are usually not shown in diagrams. This section should make it clear that the diagrams
cannot stand alone. The text beginning on page 47 could be reworded or condensed so
that the main points of Barnthouse and Brown (1994) are presented without the long
description of Odum's diagram, which does not in the end serve as a good example.
Page 47, lines 11-12: Figure 3-2 does not show effects.
Page 50, line 3: Figure 3-3 is an example of a figure that is not readily comprehensible to
non-experts; however, it does not stand alone to make that point in this document. It is
not helpful as a positive example for the diagrammatic part of a conceptual model.
Perhaps this figure could be replaced with an example of complex food web diagrams
frequently submitted as (or in lieu of) conceptual models. The relevant portions of this
diagram, including the important exposure pathways for the stressor and sensitive
receptors (e.g., potentially adversely effected receptors used as assessment endpoints)
could be highlighted to demonstrate the ideal for a conceptual model.
144
-------
Atyce T. Fritz
2O6/526-6305
Page 50, line 7: The definition of assessment endpoint as used in Barnthouse and Brown
is apparently inconsistent with section 3.4.2.
Page 50. line 10: It is not clear why pesticide application rates are the only relevant
environmental fate data for the example.
Page 52, line 22: It seems misleading to suggest that some risk hypotheses should be
dropped due to feasibility concerns. Risk hypotheses regarding the assessment
endpoints must be tested if the assessment is to evaluate risk. However, the actual
methods used to test title hypotheses can vary due to feasibility concerns.
Page 52. lines 29-30: I would like to see the use of data from other locations and species
discouraged where it is feasible to collect site-specific and species-specific data.
Page 53: I would present the measures in a different order. I think it makes sense to
present the measures of ecosystem and receptor characteristics first, followed by
exposure measures and effects measures.
Page 53, line 15: Measures of exposure should also include determining how a stressor
changes in the environment (for example, methylation rate for mercury).
Page 54, Section 3.6: The discussion of the analysis plan should be clarified to indicate
that objectives for the plan must be identified first. Specific measurements that will be
made should be described in sufficient detail to document how the measurements will
meet the objectives and address assessment endpoints. The plan should indicate very
clearly how the results will be interpreted. These questions should come before
descriptions of heterogeneity, data gaps, and QA/QC concerns.
Page 55, Table 3-1: These strategies appear to have limited ability to reduce
uncertainties. A better way to reduce uncertainty would be to decide up front what
constitutes significant risk. For example, one strategy to deal with variability is to
determine what will define a "significant" response. This will allow appropriate sample
sizes to be determined. Some of these uncertainties can be reduced by collecting more
information. For example, one strategy that deals with extrapolation uncertainty is to
conduct tests with the species you are concerned about.
145
-------
Alyce T. Fritz
206/526-6305
Analysis :
Questions 13,14, and 15.
The balance is appropriate and, at this time, I have no major recommendations for
additional principles. I would not characterize the ecological-effect analysis as simply as
a "number crunching exercise" (page 74), particularly when dealing with data from field
investigations. (This reflects my bias for site-specific assessments using field assessments
and toxiciry testing and my emphasis on using professional judgment in interpreting
results.)
The biological steessor type discussion could be improved with a clear definition of the
delphic method and a more explicit example of a fault tree. In addition, the text boxes
in this part of the chapter, along with others for stressor types, are not particularly
useful. Many of the points should be integrated into the text to be effective examples.
On the other hand, several sections contained valuable series of points, issues, and/or
questions that could be effectively highlighted in text boxes: a) Exposure Profile—
uncertainty associated with exposure estimates (pp. 72-74); b) Stressor-Response Profile
(pp. 87-81); c) Physical stressor—list of things to summarize or describe (p. 93); and
d) Multiple stressors-^-issues to address (p. 106).
Risk Characterization
The discussions of risk estimation and weight of evidence are excellent!
The end of the chapter dealing with ecological relevance seems to confuse risk
management and risk assessment. For example, determining the potential for recovery
is not part of the baseline or primary risk assessment. Recovery could be discussed as to
how it affects determining magnitude of risk rather than as a tool for comparing risk due
to other stressors or cases in other systems. When the existing or primary risk has been
characterized, additional questions regarding the risk of alternatives and potential rates
of recovery are usually raised during the risk management phase. The risk of alternative
cleanup or management actions can be determined as another iteration of ecological risk
assessment (as discussed earlier in response to question 11). For example, in
determining the risk of no-action alternatives or various remedial alternatives/
management actions used to reduce existing risk, the potential for, and predicted rates
of, natural recovery become important. In the Superfund process, this usually involves
a predictive evaluation of risk due to physical stressors (altering contaminated habitat)
146
-------
Alyce T. Fritz
206/526-6305
that considers potential recovery under different management scenarios or remedial
actions. However, the potential for natural recovery does not contribute to determining
baseline risk.
Page 122. Section 5.3.2: The beginning of this section provides important information
that should feed back into the design of the analysis phase. These issues should be
addressed up front in problem formulation. Ecologically significant effects, determined
through appropriate measures of effect (if the risk assessment has been designed
properly), should be detected with statistical significance. It makes sense that
professional judgment concerning issues of ecological significance could override
determinations of statistical significance in a situation where uncertainty and
susceptibility are high. However, it is not clear that this was the intended meaning here.
I am concerned about the potential for ambiguity: it would be unfortunate if the
statements regarding statistical significance could be construed to mean that the result of
measures of effects and risk estimation could be easily discounted, i.e., the risk is not
"significant enough."
Concerns with Ecological Relevance - Recovery and Natural Variability
Page 123, Section 5.3.2: Some indication of ecological significance can provide risk
managers with a framework for comparison to other cases. However, the intent of the
risk assessment should not be to compare the risk of one situation to other risks (for
example, comparing the risk at a hazardous waste site to the risk of filling wetlands in a
watershed).
Page 123. section 5.3.2.2: This section seems to confuse risk management and risk
assessment. The scale of the risk is important in risk-management decisions. If the risk
is over a small area or time, perhaps a reasonable management decision would be to take
no action at all.
Page 125. second paragraph: Recovery rates as described for the different communities
used as examples are too short. Specifically, fisheries could take several decades to
recover or more likely, species composition may shift completely to a distribution that
no longer supports other important species or commercial fisheries. Forests and
associated productivity may not recover even after 50 or 100 years. Additional factors
147
-------
Alyce T. Fritz
206/526-6305
to be considered include the mode of reproduction (k-selected versus r-selected) and
population growth, along with the availability of adequate stock.
It is appropriate to evaluate recovery of a system from adverse effects of original stressor
versus recovery under various remedial alternatives; this iteration of risk assessment
should be built into the analysis plan at the problem formulation stage (as mentioned
before) in consultation with the risk manager. Most of-the recovery discussion is
appropriate but the last paragraph on page 126 containing a Superfund example is
disturbing. In the Superfund process, we frequently agonize over the concept of
needing to weigh and balance the existing risks of chemical contamination in sensitive
habitats versus .the risk of remediation destroying habitat, particularly certain wetlands
that are unlikely to recover important ecological functions. In fact, we use this as an
example for why site-specific risk assessment is so important to determine significant
risk and ecological functions to avoid unnecessary cleanup of habitat to meet numerical
criteria. However, we very infrequently find, in situations where significant risk exists,
that the contaminants are likely to detoxify in a time frame that will not cause years of
irreversible harm to ecological resources.
In wetlands, heavy metals may not be readily bioavailable, the contaminant source may
be easily stabilized or hot spots removed to prevent migration, and sedimentation rates
may enhance natural recovery. However, usually measured or significant risk is due to
persistent toxic contaminants (e.g., PCBs, Hg, PAHs, and dioxin) particularly those that
continue to migrate out of the wetland downstream or into the food chain. My concern
is by concluding the recovery section with an example that assumes relatively quick
recovery from effects of toxic chemicals could lead to the interpretation that if natural
recovery is possible the significance of existing risk can be discounted as well as the
need to take actions to reduce ecological risk. I agree with the need to consider the risks
of remediation, but the particular phrasing of the example and implication of mixtures
as problems and "detoxifying chemicals" potentially leads to a conclusion that perhaps
is not intended.
Page 126, Section 5.3.2.4: Determining natural variability is part of the analysis phase
and seems out of place in the risk characterization discussion. If no effect can be
discerned from natural variability, then there is no apparent risk. However, analyses
should be designed with sufficient power to distinguish an ecologically significant effect
148
-------
Alyce T. Fritz
206/526-6305
from natural variability. Taking into account natural stressors and how extremes in
natural phenomena could increase risk can be an important part of a risk description.
The discussions of natural variability, recovery, and ecological significance raise issues
here that should be addressed earlier in the analysis planning phase. A well-conceived
analysis plan (achieved using these guidelines!!!) should address these issues to help
minimize ambiguity in risk characterization.
Finally, NOAA very much appreciates the opportunity to participate in this review
process. This document will be a very valuable tool and reference. I am delighted with
many of the detailed recommendations and examples. Although many of the detailed
comments are devoted to the Problem Formulation chapter, this section provides
excellent clarification of the process and its components. The change and/or additions
to terminology should be embraced by risk assessors and the description of four stressor
types worked fairly well in the Analysis Phase. I am concerned about the potential for
the possible misinterpretation of certain statements in the risk characterization section
that may lead to descriptions of ecological risk as insignificant or that may lead to
conclusions that no actions are necessary to reduce risk. I would re-emphasize mat
observed adverse effects equals risk; ecological risk assessments should be site-specific
where possible! However, most of these concerns can be easily addressed with some
clarification in the discussions or slight recrafting of language. Again, I would like to
thank the EPA Ecological Risk Assessment Forum and the authors' outstanding efforts
to obtain and meld varied input to produce a well-written and useful document.
149
-------
-------
Robert Goldstein
151
-------
Robert A. Goldstein
REVIEW OF DRAFT GUIDELINES FOR ECOLOGICAL RISK ASSESSMENT
GENERAL PRINCIPLES
I believe that the report could be more specific as to what exactly is an ecological risk
assessment and what are its motivating factors. How does an ecological risk assessment
differ from an ecological impact or effects assessment? What is new or unique about
ecological risk assessment? What are the advantages or strengths of an ecological risk
assessment versus traditional ecological effects or impact assessments? What does the
term "risk" imply in the context of an ecological risk assessment? Some people interpret
the term to signify an adverse occurrence while others take it to signify uncertainty.
These alternative interpretations should be discussed.
The guidelines define ecological risk assessment as being limited to adverse effects. This
is too limiting. The approach is equally valuable for neutral and beneficial effects such as
outcomes of mitigation actions. Furthermore, as the text points out, the categorizing of
effects as adverse depends on the value judgments of the stakeholders. These value
judgments can be highly diverse and vary with time.
The text does a very good job of describing the interconnection between risk assessment
and management. These two activities are so closely intertwined that it is totally
artificial and incorrect to claim as it is done in the text that they are separate. Although
one can draw a schematic and stick these activities into two separate boxes, in practice,
they are inseparable. The objectivity of the assessment is always at risk of compromise.
Calling assessment and management distinct in no way reduces that risk. Environmental
management decisions are influenced by scientific, engineering, economic, social and
political factors. It is reasonable to expect that decisions will not be based on
optimization with respect to a single factor type. In order for the ecological risk assessor
to be effective, he cannot separate himself from those who bring the nonscientific factors
to the decision process. He must be able to articulate what the ecological trade-offs are
with respect to alternative management options.
The text does an excellent job in discussing the relationship between risk assessor and
manager. It needs to do likewise for the relationship between risk assessor and
152
-------
Robert A, Goldstein
traditional environmental scientist. The text frequently refers to risk assessment as
scientific, which it is, implying that the risk assessor and the traditional environmental
scientist (e.g., ecologist, physiologist, limnologist, oceanographer, geologist, atmospheric
scientist, etc.) belong to the same fraternity, which they don't. The risk analyst provides
a bridge between the environmental science community and the stakeholder
community. The risk analyst modifies through extrapolation and simplification the data
and theory developed by environmental scientists so that it is amenable to the risk
analysis of alternative management options. In many instances, the techniques of
uncertainty analysis and. extrapolation used by the ecological risk analyst are not part of
the background and toolbox of the traditional environmental scientist. Unless the risk
analyst justifies his approach to the traditional environmental scientist, there is a risk
that the traditional environmental scientist will oppose the analysis and undermine the
ecological risk assessment. Hence the interaction between risk analyst and
environmental scientist is as critical as the relationship with the stakeholders.
The text needs to identify what are its most significant points. I assume that the
executive summary will be used to do this. Within the main body of the text perhaps
bulleting or boldfacing could be used to identify key concepts.
To me, it is the analysis of the sources of uncertainty and how they influence the
ultimate assessment that is the essential characteristic of risk analysis. Most of the text,
although generally well written and insightful, pertains to ecological analysis in general
and is not specific to the risk element. I do not necessarily object to the inclusion of this
material, but I would like to see more substantive discussion on uncertainty analysis. A
major point of contention in risk analysis is the application of reduced form models. This
subject is never mentioned, neither is uncertainty propagation. It is pointed out at least
twice that simplification can add uncertainty; however, it is equally true that increasing
complexity can increase uncertainty. The uncertainty in a calculated entity is a complex
function of the uncertainty of the algorithm used to do the calculation and the input
data, which includes values for initial conditions, algorithm parameters, boundary
conditions and forcing functions. A motivation for increasing the complexity of an
algorithm is to increase its accuracy; however, this also tends to increase input data
requirements. Hence, it is possible to decrease the uncertainty in the algorithm while at
the same time increasing the uncertainty in the input so as to produce a net increase in
the uncertainty in the calculated output.
153
-------
Robert A. Goldstein
QUESTION 1
I find that ecology is underemphasized relative to toxicology and that risk (uncertainty)
analysis is underemphasized in general. Ecology is not sufficiently underemphasized to
be of concern, but the guidelines should be beefed up with respect to how sources of
uncertainty are estimated, how uncertainty is propagated in fhe analysis, and how
uncertainty is characterized in the completed assessment. Consultants with expertise in
uncertainty analysis in contrast to simply toxicological or ecological analysis should be
hired to provide the additional material. Possibilities are Virginia Dale (ORNL) and Bob
Gardner (Appalachian State).
More emphasis should be placed on multiple stressors. In the future, this will become
the dominant assessment situation. Historically, environmental policy and technical
analysis have for the most part treated stresses individually. This is because society's
concern was concentrated on only a few stresses. As society has become more concerned
about ecological impacts (in contrast to just human health impacts) and has identified a
significantly greater number of stresses, the technical community has come to realize
that most receptors of concern exist in an environment where they are exposed to
multiple interacting stresses, anthropogenic and natural. Analytic approaches that work
well for single stresses, do not necessarily work well for multiple stresses; hence, the
need to provide additional emphasis on the latter subject.
QUESTION 2
I thought that the document's authors, in general, did an excellent job in clearly
integrating the illustrations into the text, hence I find Appendix A somewhat
superfluous. I found the illustrations to significantly enrich and enliven the text by
providing concrete, real-world examples of presented concepts. However, I think it
creates confusion to refer to these illustrations as "case illustrations"; since as the text
explains, they are not meant to be used as examples of either what to or not to do.
The weakest of the illustrations was Illustration 5, Assessing Risk of a New Chemical
under the Toxic Substances Control Act. Because of the illustration's vagueness as to the
exact chemical and its source, the illustration tended to be somewhat confusing. Cannot
more specifics be added without identifying the chemical, which I assume is the reason
154
-------
Robert A. Goldstein
for the vagueness. If not, can another example be chosen where more specifics can be
given?
If Appendix A is retained, the descriptions of Uncertainties for each of the illustrations
need to be beefed up. At present, they are too perfunctory. This relates to my answer to
the previous question where I identified a general need for more depth and insights in
the guideline's treatment of uncertainty.
QUESTIONS
I did not note such a deficiency and have no recommendations.
QUESTION 4
The approach is okay but the information needs to be expanded to included more depth
and insight. Please refer to my answer to question 1.
QUESTIONS
Please see comments under general principles. The Introduction needs to clarify what
makes an assessment a "risk assessment" and how risk analysis differs from traditional
environmental science. The insistence that risk assessment and risk management can be
separated is incorrect and misleading. The limitation of risk assessment to adverse
effects is unwarranted.
The discussion (page 4, lines 8-15) of when a risk assessment may not be required
provides no guidance. Issues concerning habitat protection and endangered species are
certainly amenable to ecological risk assessment. Furthermore, professional judgment is
a valid risk assessment tool as presented in the Chilean log illustration.
QUESTION 6
I thought that the approach taken in the text tended towards being pedantic. The
concepts are common sense and can be explained well in one or a few paragraphs. The
concern about and need for concise, precise definitions seems unwarranted and
155
-------
Robert A. Goldstein
probably is the source rather than the solution to the controversy referred to in the
question. As stated before, I would not limit risk assessment to adverse effects.
QUESTION 7
I think that the framework is okay. t
QUESTIONS
I think that this is done well.
QUESTIONS-
I find this question confusing, since it is my impression that the text does not include
any discussion of the interaction between assessor and manager (stakeholders) after the
completion of the assessment This is an important topic and should be included.lt falls
into the category of risk communication which has also been left out of the document.
QUESTION10
It is okay.
QUESTION11
I thought there was a tendency to be pedantic. It was not clear what material had specific
applicability to risk assessment versus assessment in general.
It should be pointed out that policy goals and/or societal values often do not form a self
consistent set.;e.g., recreational and flood control goals for impoundments. There
should be a discussion of methods to help resolve this type of situation.
QUESTION 12
Again, what is the specific relevance to risk assessment versus assessment in general?
156
-------
: Robert A. Goldstein
QUESTION13
As stated in the answer to question 1, more attention should be given to multiple
stresses. In addition it is preferable to separate exposure analysis into source arid
pathway analysis; since, source and pathway calculations usually draw on different
data sets and models, involve different groups of technical experts, have distinct
uncertainties, and are the focus of different control and mitigation options.
QUESTION 14
On page 83, lines 26-27, it is stated that it is reasonable to assume, when there is no data
to the contrary and exposure profiles are similar, that laboratory effects represent field
effects. No backup support or citations are given for this statement which I disagree with
since it ignores the possibility of environmental mediation as well as secondary jnd
indirect effects. The stated assumption is not even conservative, since secondary effects
can be more adverse than primary effects. ' •
On page 84, lines 5-6, it is stated that extrapolations across geographic regions require
coordination between exposure and effects analyses. What is needed is a mechanistic
receptor model.
On page 84, on the discussion of cosms one might want to mention open-top field
chambers as a more complex cosm. It would also seem appropriate to mention here
ecosystem experiments; e.g., addition of nutrients or acids to lakes, modification of
precipitation or canopy throughfall, field fumigation, etc. : : ,.
On page 84, lines 21-22, it is stated that as the cosm increases in complexity, variability
among the control and treated replicates can mask effects. The reader should be
cautioned not to infer from this statement that the increase in variability is an artifact of
the cosm design. Variability is a property of the real system. The purpose of this report is
to guide the reader on how to deal with the variability.
._'••,- - ! S
On page 84, the discussion of secondary effects abruptly switches from cosms to
ecosystem models, with no explanation that the ecosystem models, unlike the cosms, are
mathematical models.
157
-------
Robert A. Goldstein
The statement on page 85, line 15-17 is confusing. Predictive risk assessments that utilize
field data also rely on hypothesis testing, regression analysis and other statistical
techniques. What is meant by the phrase," but not absolutely prove causality"? An
hypothesis can be a causal hypothesis. The causality section is written from a
toxicological perspective. It needs ecological perspective.
Figure 4-6 has a stressor one and three, but no two? The figure also tends to be confusing
because there is a general flow of information from right to left instead of left to right.
Page 95, lines 15-16, the statement that opinions about whether a biological stressor is
beneficial or deleterious can differ provides additional support as to why ecological risk
assessment should not be limited to adverse effects.
QUESTION 15
Traditionally, environmental policy and regulations have been source oriented. As
society in general and the environmental community in particular become more
sophisticated with respect to identification and analysis of environmental issues,
identify an increasingly larger number of potential stressors and their sources, and
become more concerned about ecological effects, there will be a greater recognition of
multistress issues. In an environment where one is dealing largely with multistress
issues, it is advantageous to switch one's focal point from source to receptor, since the
receptor is the natural integrator of the multiple stresses. One can calculate the total risk
for a receptor of concern and decide whether it is acceptable. If it is not acceptable, one
can analyze the ecological and cost effectiveness of alternative management strategies to
reduce the risk. As the text states on page 107, lines 8-9, "The most accurate predictions
will result from a sound understanding of system structure and function." Thus a major
thrust of the ecological assessment community should be to develop and verify
mechanistic model of ecological receptors at levels of temporal and spatial resolution
that are of concern. The level of detail required by these models is a function of how the
various sources of uncertainty affect the uncertainty of the final assessment. This is one
of the most important facets of ecological risk analysis and it is not at all treated in the
guidelines.
158
-------
Robert A. Goldstein
QUESTION 16
As I have already noted, there should be some guidance as to how risk (uncertainty) is
treated mathematically within the context of a risk assessment.
I have two comments about Fig. 5-1. There should be a two-way arrow connecting the
data box on the extreme right with the problem formulation-analysis box on top. There
should be a one-way arrow leading down from risk management to another box called
management decision to signify that ultimately the process should culminate in a
decision. The process cannot go on indefinitely without some decision. Of course, the
process can be restarted at some later time leading to a modification of the original
decision.
QUESTION 17
The fourth criteria mixes two different subjects which should be separated. One is to
what extent can an effect be recognized against a background of natural variability. The
second is to what degree do natural disturbances interact with anthropogenic stressors
to alter risk.
159
-------
-------
Lawrence Harris
161
-------
L. Harris
I. IDENTIFYING AND HIGHLIGHTING IMPORTANT GUIDANCE PRINCIPLES
A. INTRODUCTION.
1. The following 'general principles' are very laudable:
can*! eliminate all risk
we must operate under much uncertainty, and that
we must explicitly evaluate uncertainty. :and that
both magnitude and likelihood of risk are important, and
we must work toward well-defined endpoints or measures of effect
and that there will be increasing
interactions between assessors and managers. Figure 1.1 is a good
representation of the overall process. I also give high marks to figures
1.2 and 1.3 (with some provisos).
2. The following 'general principles' are conspicuous by their absence and need to
be dealt with. I hope that we deal with them in the workshop.
a. The introduction as illustrated by Figure 1,1,and examples is strongly
biased toward single-factor effects and neither explicates nor implies
much importance to simultaneous multiple stressors. I conclude that
this is a serious issue (even though it is treated considerably in the
analysis section). Most of the real-world examples that I am
aware of (e.g., the Everglades) are infinitely more complex than can
be handled by this simplistic portrayal. Therefore, even though I fic.d
the progress made in Risk Assessment to date very laudable, I thhiik
it very important to note that this greatly over-simplified material iin
the first 50pp. is only a prelude .to the much more complex material
dealt with yet another 50pp further, along (i.e. around p 100).
b. Equally, if not more, disturbing is the lack of explicit acknowledgement
that interaction effects of multiple, simultaneous stressors and/or
ambient conditions are very real, but also very difficult and discussion
is deferred to the end of the analysis section (where it is weak).
It seems to me that both the wording and the illustrations lucidly
portray how to handle single-factor stressors, one at a time, but both
the text and the figures fail to either implicate or explicate the
importance of interaction effects.
c. It is laudable to explicate that both magnitude and likelihood of effect
must be evaluated. But equally important is the notion of reversibility
vs. irreversibility. Destruction of a sawgrass or Spartina marsh may
be lamentable, but it is reversible. Extirpation of the Dusky Seaside
Sparrow is not reversible. I suggest that reversibility be added to the
list and that magnitude, likelihood, and reversibility must stand as
a trio. Reversibility is probably the most important of the three.
162
-------
L. Harris
d. Regarding unclear communication (p.£,1.14).
Yes, dropping down a scale to simple demographic parameters allows
the full power of the Cartesian (reductionist) approach to science as
opposed to those nasty qualitative terms such as health or healthy.
But this occurs at great cost too. In almost every year during the last
decade there has been more births than necessary deaths of the
Florida panther; the Florida panther population is demograpbically
exemplary. But do any occur in Everglades National Park, no!!! Is
the Florida panther population 'healthy', no!!!
Simple demographic parameters niiss the point regarding a majority
of the endangered species on earth!!!! What is the population
viability???
.e. Regarding the issue of Default and non-default, I conclude that this needs
clarification, I do not know what an agency default for a model
structure is. Lets try to clarify this.
f. Regarding definition of ECOLOGICAL RISK ASSESSMENT. This is;
a seductively cogent and clear definition. It is good. But it's so
simple and beguiling as to lead to great problems later on. Should
not this definition be made explicitly more complex so as to reflect
the complexity of the real world. For example, should we not strongly
imply or explicitly state that there are a least a dozen simultaneous
and interactive stressors at work in destroying the Everglades as a
viable system????
If this were a document on health hazards of lead-based paint I would
not be as concerned about the apparent single-factor, no-interaction
orientation that comes across to me to this point in the document. But
big systems are going to the happy hunting grounds because of very
complex interactions of multiple, simultaneous stressors, not because
of identifiable single-factor effects.
Inasmuch as the document does treat some significantly complex
material later on, it seems quaint that this first 50pp comes across as
so overly simplistic.
g. Definitions of ENDPOINT/MEASURE OF EFFECT and SOURCE are
good.
h. Definition of STJRESSOR is, in my view, too simplistic and naive.
For example, it explicates only nouns, not verbs. It neither implies
nor explicates that rising sea level, urbanization, sprawl,
habitat fragmentation, and lack of burning were interactive forces
that caused extermination of the Dusky Seaside sparrow. It neither
calls for nor openly allows that free-ranging cattle (overgrazing),
lack of burning due to Smokey-The-Bear, and forestation led to the
extinction of Bachraan's warbler. I believe it is the interaction effects
of multiple, simultaneous stressors are the stressors we must get at.
163
-------
L. Harris.
Inasmuch as the last part of the analysis section treats some much more
complex situations and scenarios, I wonder if the complexity of the
real issues should not be signalled up here in the early sections.
The two cases just cited are among many that I tried to elucidate in
a 1988 paper THE NATURE OF CUMULATIVE IMPACTS ON
BIOTIC DIVERSITY OF WETLAND VERTEBRATES, Environ.
Manage. 12:675-693.
i. Definition of EXPOSURE seems too narrow for practical application to
many of the problems that I am aware of. It is the lack of burning
that threatens scores of species in the Southeastern Coastal Plain.
Exposure to fire is necessary. How do we say that 'exposure* to no
fire (very common) and to fire prevention (not uncommon) is the
stressor. This is dealt with in some considerable degree toward the
end of the document and so it is puzzling that these simple, categorical
statements would ring so loud and clear in the early sections.
j. Definition of 'Disturbance' seems too narrow and jaded. Fire is not a
"disturbance" in the Southeastern Coastal Plain, it is a necessary fact
of life. To call these necessary fijres a 'disturbance' is not consistent
with what we know about southeastern ecology. It would be helpful
to find a few better words here. The exact same case can be made
for flooding in the bottomland hardwood forests. We need floods in
order to maintain any semblance of a natural system. How do we
describe "exposure" to a no-flood regime?
(I am now aware that this is treated somewhat toward the end of the
paper, still, at this early stage it Seems to come across as a red flag).
k. Text box 1-6. STRESS REGIME.
I understand what is being said very well. Still, tthe phrase come:;
closer to describing the native environmental dynamics that are needed
in the Southeastern Coastal Plain than what we are replacing it with.
B. PLANNING
1. The following points or principles are worthy of note
p.17,1 3-4
Purpose of R.A. to provide scientific information about risks
p.17,113-14
By using scientific info, the R.A. ensures (?) attention to all important
ecological concerns
p.17,118-22
Keeping the planning distinct from the R.A. helps keep politics separate from
science.
p. 21,11-6
The agency increasingly emphasizes "place-based" or "community-based"
management. This is good but begs a critical question.
164
-------
L. Harris
P.21,128-30
R.A. involving multiple stressors. This could profit from some
expansion of the idea and perhaps examples.
2. The following points are cause for concern:
Use of scientific Information in a Risk Assessment will not ensure that
attention is paid to all critical ecological concerns. It is my experience that
contracts such as R.A.'s are farmed out fo consulting firms that see it as a
job to execute with minimum fuss and mass. Rarely are top-of-the-line
thinkers and scientists involved in the planning and problem formulation
stage. Thus, the R.A. becomes little more than a voluminous statement of
the obvious with little or no probing of the more far-reaching questions.
While "place-based" or "community-based" management sounds good on the;
face, it is increasingly out of step with reality. Take our National Parks and!
biodiversity preserves as an example. iOO years ago, it would have been
adequate to do 'place-based* assessment of impacts on Yellowstone or the
Everglades, With today's knowledge about landscape and region-wide
phenomena at work the 'place-based* approach seems almost like an
oxymoron. Regional Landscape analysis is the key to relevant analysis in
growth states such as occur in the sunbelt. Even the putative 'advanced*
ecosystem management is already out-of-date because of the rapidity of
developments in the physical environment and in the science of landscape
ecology.
p.21,1.27-30. I believe that this section on region or watershed analyses that
involve multiple stressors and their interactions need much more emphasis
if ecological risk assessment is to be relevant to big systems such as the
Everglades.
C. PROBLEM FORMULATION
1. Positive points or principles worthy of note
a. p. 24,17-9
the statement here about consistent shortcomings is good and important
b. p. 26,121-24
the statement here about why/how ERA's are initiated is good
c. p. 28,1 29-31, and p.2$>,1.1-6
the statement about complexity and landscape scale are exemplary. More
of this is needed. As the science of ecology continues to grow, Landscape
ecology continues to play a greater and greater role. This is demanded and
forced by the increcising isolation of ecosystems and thus the increasing
irrelevance of ecosystem analysis without a landscape contextual analysis.
165
-------
L. Harris
-------
I/. Harris
temperature, but the effects will not be manifest (and there is »o way we
can know) until 25 years from now (because they are so long-lived) ????
g. p.38,1 29-31
The business of degraded habitats badly needs to be considered at the
landscape scale because of issues such as configuration, adjacency, etc.
h, p.43,1. 7-23
I know that this is logical and nice, it is also problematic. The implications
are horrendous. Can or should we soften this. Frank Egler made a
famous statement one tune that said: 'the computer has encouraged the
belief that the only things that count are those that can be counted"
i. p.44,1. 8 hypotheses
I would prefer to state that they are falsifiable predictions rather than
assumptions.
D. THE ANALYSIS PHASE
1. Positive points and principles
a. p. 59,1.4n7
It is good to see this discussion of natural driving forces, which I noted
above, are not necessarily stressors. .1 believe it is important to hammer
this point home that driving forces (the pulse paradigm) are indeed
commonplace and that they are not disturbances, and that the real
disturbance comes into play when humans or our structures prevent or
mitigate the 'natural' pulses.
b. p. 60, fig 4-2
This figure goes a long way to allay anxieties that the Ecol. Risk Assessment
process can handle some complexity, I would advocate inclusion of an even
more complex one. Although I do not have one in mind, I believe we can
develop one based on cumulative impacts that I am well aware of.
c. p. 63,1.21-24
This is a good example but should be fleshed out a little more to drive
home whatever point is being made.
d. p. 71,1.19-24
This is a good point about interaction effects but even it is simplistic
e. p. 88,1.17-20
This is an important paragraph. Could/should we expand on the SAB
conclusion that habitat alteration ranked so highly and could/should
we develop an example or even a box to drive this point home. I think
it is very important and is presently being slighted.
f. p. 90,121-27
This is an important point to make, but do we really want to stick to the
word disturbance when all of the speci.es and communities are well-adaptE d
to these pulses??
167
-------
L. Harris
g. p. 92,1. 20-23
Its good and reassuring to see that we are finally getting around to
secondary effects. May want to amplify.
h. p. 93,1. 5-11
This is a good point/principle, in my judgement it should be
expanded/elaborated/amplified.
j. p. 93, text box
Good to include the indicator species approach but this is problematic
k. p.96, table 4-4
The use of Hydrilla as an example really shows that values do encroach
on science etc. It is highly debatable that Hydrilla increases "habitat",
yeah, but at what cost in other 'habitats'. I recommend we drop this
example, it is polemical to say the least.
What about including a really complex example of secondary, tertiary
and/or quaternary effects in here???
1. p. 99,1.22-30
It is good to bring this point of one-sided science out. Should we amplify
the now obscure notion that this whole scenario most change in the future.
m. p. 103, L 4-5
Simberloff and Alexander make a gopd point here.
n. p.105,14-10
This is a very important paragraph, as far as I am concerned it must be
emphasized more, perhaps expanded. !The problems that I had with the
first 50 pp of the document is that they did not imply or state that ecol. risk
assess would/could get into the complexities of the real world.
o. p. 105, L 13-15
I believe this is so important as to beg for expansion
p. p. 106,1. 5-7
re: assume lack of interactions. This i's probably a true statement, but it
is lamentable and ecologists should not let it pass. Something must be done
about this.
p. p. 108,1. 4-13
It is encouraging that progress is befog made at the levels of higher
complexity and larger scales and we should encourage it. To my mind,
this is what is called for. Perhaps we should develop a box and say
something about present limitations.
168
-------
L. Harris
2. Points of concern or problems
a. p. 58,1 30-31
evaluating stressor individually is indeed a major problem and it is one
thai should concern all ecologists and resource managers with regard to
future ERA's.
b. p.88,121-24
As I have mentioned repeatedly, the word "disturb" and the phrase
'disturbance regime9 are problematic (in my view). A disturbance regime
is hardly a disturbance regime if everything that lives there is highly
adapted to it. Events such as the Yellowstone fire of 1988 were disturbing
to humans, it crushed our illusions, but it was hardly a disturbance to the
Yellowstone system.
c. p. 93, text box 4-10
The business of "indicator species", especially the ones cited here, is very
tricky. There is a very rich early 20th century literature on indicator
species and one of the principles drawn from all that work is that
indicators had to be stenotypes and not eurytypes. In today's applications,
the indicator species are almost always chosen because they are generalises
(eurytypes). It is for this reason that use of species such as the gray
squirrel and downy woodpecker are almost laughable. What they indicate
is that managers can make all sorts of errors and still maintain their
favorite small galaxy of "indicators" that things are o.k.
d. p. 95,1. 15-20
We must give serious thought to this description of Hydrilla's effects; 1
submit that we are too ignorant to say much about the pluses and
minuses. Surely, the increase of largemouth bass does not compensate for
all the other negatives, or does it??
e. p. 96, table 4-4
remove HydriUa
f. p. 97,1.18-20
need to explain and or elaborate on "fault trees"
g. p. 103,1. 9-14
This seems like a good example but the point being made is not made well.
Moreover, there needs to be a caveat about the effects of outbreeding
depression.
h. p. 104, overall
Do we not need an overall conclusion, statement about the generalities of
exotic species problems????
169
-------
L. Harris
i. p. 105,I.12-15.
I conclude that our ability to conduct good ERA's is much more limited
by the nonpaititipatiou of good minds, .inadequate dollars and motivation)
and the fact that they are usually done at too small a scale.
j. p. 105-107
This, section on predicting effects of multiple stressors almost immediately
falls back to the 'assumption of additivity' and other simplistic examples.
I would like to see the cases involving a big system such as the Everglades
emphasized, and then describe our inabilities against that light.
170
-------
Richard Kimerle
171
-------
Richard Kimerie
Nov. 15, 1995
Premeeting Comments For Workshop On "Draft Ecological Risk Assessment
Guidelines"
I. General Comments Reflecting My Initial General Reaction To The Draft
Guidelines
A. The Draft Guidelines are an enhancement of the old Framework document
addressing many of the deficiencies. It is a real improvement. Everything that has been done
and included in the Guidelines is quite good and very difficult to find significant faults. I can
see that it will be useful guidance in conducting all stages of an ecological risk assessment
helping the risk assessor and manager understand what needs to be considered in an
ecological risk assessment and why it is important. However, this draft document, which was
stated to "....replace the widely-used Framework Report" still does not adequately address
many of the important suggestions made in the past "peer review" meetings over the past
couple of years. I would suggest additional consideration of the many excellent comments
made at the May 3rd Colloquium, and other meeting like the "Issues Papers Workshop".
Several of the common concerns brought up in the May 3rd review have been covered quite
well, (i. e. the needed discussion off the risk assessor and manager). Many of the others
concerns expressed are mentioned in the new Guidelines but it is as if they were only given
"lip-service" to satisfy the participating reviewers by including their points (i. e., tiers and the
iterative process are only discussed in a superficial manner; more helpful guidance on
probabilistic approaches to risk characterization; non-chemical stressors; and terrestrial
environment risk assessments). If some of these issues were handled in a more
comprehensive manner it would add to the useful of the guidance. In other words the breadth
of coverage is reasonable and could be improved but the depth of coverage in some of the
key areas could use significant improvement.
B. The inclusion of the section on "Planning: Discussion Between The Risk Assessor
And Risk Manager" provides much needed guidance on this issue that was not adequately
addressed in the previous Framework Report. However, the statement that "economic, legal,
political, or social implications (I add societal and the concept of availability of resources ) of
the risk assessment results is beyond the scope of these guidelines" leaves the risk managers
and assessors without much needed guidance. I believe that many persons feel that these
172
-------
difficult issues cannot be ignored and must be viewed as part of the total ecological risk
process.
C. I think I understand why the EPA has developed these guidelines for almost
exclusive use by its internal technically qualified professional risk assessors and managers.
However, I wonder if the entire Guideline process and end product would be more valuable if
more consideration were given to the needs of other potential users at local, state, regional,
international levels and by industries and businesses that must make ecological risk
assessment decisions on a regular daily basis? What will be the next step to be taken by the
EPA to enhance the benefits of this general ecological risk assessment guidance at the user
level, including your own EPA program offices?
D. I was somewhat disappointed because of a lack in details and need for
embellishment of many technical guidance points with more specific help in the form of
discussion of references or in some cases at least references to turn to for more help. Again
this may reflect the stated intent that this document is primarily for use by highly qualified
professionals within the EPA. Recognizing that these guidelines will likely be around for
numerous years, and we are all going through changes in the way we go about our business,
would it not be an opportune time to provide a bit more generic and useful detail than to leave
it out and assume that the expertise is available to get the job done right? What will the
technology transfer program be to make the Ecologcial Risk Assessment Guidelines as useful
as possible to users both within the EPA and in outside industries and businesses?
E. The case history used as examples were very helpful but they seemed a bit
superficial and contrived at times in order to satisfy earlier stated needs for including specific
items and covering more issues (i.e., stressors other than chemical, non-aquatic examples,
use of QSARs, etc.).
F. I liked the use of shadow boxes to emphasize key important points.
G. I did not see a need to change the terminology on measurement and assessment
endpoints.
173
-------
Response To Questions We Were Asked To Address
Part I. General Principles
* How can ERA Guidelines be improved to more clearly indicate assumptions and
default positions? The biggest assumption that will detract from the intended success of the
Guidelines is that the not enough detail is provided in most of the sections. It is all good
guidance but depends very heavily on the expertise involved in carrying out the risk
assessment.
* Changes in each section? Look for places where additional details would better
assure a quality assessment.
174
-------
AIHC Perspective On Ecological Risk Assessment
Framework And Guideline Development
Presented By Richard Kimerle
American Industrial Health Council QAIHC)
Ecological Risk Assessment Subcommittee
EPA Public Colloquium On
Ecological Risk Assessment
Guideline Development
May 3,1995
Alexandria, VA
175
-------
A1HC Perspective On Ecological Risk Assessment Framework And Guidelines
Slide 1 '
The American Industrial Health Council (AIHC) thanks the EPA Risk Assessment
Forum for this opportunity to speak at this important Colloquium on the Ecological Risk
Assessment Framework and Guideline Development process. I am Richard Kimerle from
Monsanto Company and am vice chairperson of the Ecological Risk Assessment
Subcommittee of AIHC, which is chaired by Dr. Charles Pittinger of Proctor and Gamble. A
group of about 10 members of the subcommittee met for a couple of days, reviewed
numerous materials on ecological risk assessment, and prepared this presentation.
Slide 2
AIHC welcomes this opportunity to participate with the EPA in these early stages of
development of Ecological Risk Assessment (ERA) Guidelines. The industrial segment of the
regulated community, as represented through AIHC, is interested in helping to assure that the
Guidelines are technically sound as is possible, practical so we can all use them,
appropriately protective of our natural ecosystems, and flexible enough to be of value to all
users. Although the Guideline process may be just beginning, AIHC is well aware of the
excellent progress made over the past couple of years by the Risk Assessment Forum in the
development of the Framework, Issue Papers, and Case Histories. The series of "Purple"
books have become a standard reference on the desks of ecological risk assessors. We look
forward to the next book on Guidelines.
Slide 3
We perceive that there is a clear benefit when governmental regulatory agencies work
with industry in a partnership. There are many examples of this type of cooperation producing
quality products that are more widely accepted by all the stake holders in regulatory guidance
through compliance. I personally have had the privilege participating in many of these joint
regulatory programs going back into the 1970s: ie. Algal Assay Procedure (AAP); Acute
Toxicity Testing Methods For Invertebrates, Fishes, and Amphibians; developing ASTM
consensus methods; Whole Effluent Toxicity (WET) testing; and numerous SETAC Pellston
workshops.
Slide 4
We in industry believe we have considerable experience in ecological risk assessment.
We would like to maximize the utilization of that expertise in the development of the
forthcoming ERA Guidelines. Ecological scientists in most major industries have been
practicing ecological risk assessment for many years in guiding internal decisions for product
development, manufacturing site emission issues, hazardous waste site assessments and in
working with regulatory agencies all over the world. We are prepared right now to commit
176
-------
whatever resources that are necessary to assure that future developments in ERA are as
technical should as is possible.
Slides 5
One of the purposes of this Colloquium was to review the appropriateness of the ERA
Framework. AIHC considers the Framework to be a sound foundation which can provide a
useful and consistent approach for ecological programs across all EPA program and offices
and other federal and state agencies needing to address ecological issues. It also provides a
useful basis upon which to build future Guidelines. The technical components of problem
formulation, characterization of exposure and effects, and then ecological risk characterization,
when property used, are a proven useful tool for judging ecological risks.
Slide 6
However, AIHC believes that the overall usefulness of future ERA guidelines will be
improved by implementing several changes in the Framework and suggested
recommendations for ERA guideline development:
1. Provide additional guidance on the nature of the policy issues that need to be dealt
with in the initial and final discussions between the risk assessors and risk managers. To
date, the EPA has appeared to focus its attention on technical issues inside the technical
boundaries of risk assessment. This simply does not meet the needs of society, the regulated
community, or "Mother Nature" herself.
2. There needs to be a clearer incorporation of the concept of "tiers" in the
Framework. We can not afford to collect all the data we would all like to have before making
a risk management decision. We must learn to rely not upon the size of the data base but on
the margin of safety between exposure and effects concentrations. We need a stronger
foundation on the use of an iterative process that would assist us in collecting only the data
needed to resolve the adequacy of the margin of safety.
3. Address several technical and programmatic issues.
Slide 7 & 8
The usefulness of the Framework itself, and future ERA Guidelines, will be enhanced
if more guidance is provided on what policy issues need to be discussed before and after
conducting the technical part of an ecological risk assessment, initial planning discussions
need to consider the following points in Problem Formulation:
I. Policy Considerations.
Existing laws and regulations.
International implications.
Corporate Stewardship policies.
Societal values and concerns.
II. Scaling the risk assessment problem.
Valuation of the ecosystem at risk.
Protection of which species and to what extent.
The magnitude, duration, and scale of risk.
Balance in consideration of negative impacts and benefits.
177
-------
Ml. Constraints.
What is a reasonable amount of money to spend?
How much of our private and public personnel resources should be used?
What are the timing and duration consideration?
During and after risk characterization, risk assessors and managers need to discuss
risk policy issues. Some of the points that need to be considered in the Frame work include
the following:
I. Was the right problem worked on or should the problem be re-formulated?
II. Was the characterization adequate to achieve the goal of the assessment?
ill. Is there a need to conduct another "tier" of data collection and risk characterization
in order to better resolve technical issues in the risk assessment?
IV. Does the risk assessment meet the need of the risk manager?
V. Before making the final risk decision are then any constraints, values, or policies
that should be considered?
Slide 9 & 10
Special consideration should be given by the Risk Assessment Forum to alter the
Framework document, and also include in the Guidelines, more explicit information on the use
of "tiers". There are numerous technical issues which should be part of educating the ERA
assessors and managers on potential components of data collected in the tiers from the more
simple to complex studies on exposure and effects. Topics include: use of QSARs; role of
single species tests; modeling to measuring to monitoring; role of micro and mesocosms;
primary, secondary, and tertiary effects; and field ecosystem studies. Guidance is also
needed on when the existing exposure and effects data bases are inadequate to make a risk
decision, that is the pass-fail-need more data criteria. Their needs to be an explanation on
the role of assessment factors in transposing data from simplifying assumptions of early tiers
to assessment endpoints desired in advanced tiers. Dealing with the uncertainty is of
paramount importance. For risk characterization there is a need to balance the appropriate
use of the more simplistic quotient approach with the growing field of probabilistic risk
assessment. The latter offers more opportunities for making risk management decisions.
Slide 11
There are a number of technical and programmatic issues that if addressed now, will
enhance the effectiveness of the Guidelines and risk assessment process in the future. In the
Problem Formulation phase, it is exceedingly important to fully understand the ecological
significance of the stressors so that risk assessors and managers focus on developing the
right conceptual model and attainable goals. Uncertainty is one of the major technical
challenges that must be dealt with more effectively than it has to date. Too often, ecological
scientists are viewed as not being able to deal with all the uncertainties of their "complex"
ecosystems. We need a system that supports understanding, communicating, and managing
data uncertain. This will prevent "uncertainty" from becoming the driving force that leads risk
assessors and managers on useless and costly "witch hunts". An emphasis has traditionally
be placed on chemical stressors and point source discharges and thus many of the "tools" of
ecological risk assessment don't adequately address the non-chemjcal and nonpoint source
stressors. Through the forth coming process to develop ERA Guidelines an opportunity exist
to correct this situation. Consistency or harmonization of the ERA methods within the EPA,
178
-------
across governmental agencies, and at the international level is a recognized need. In order
for this to occur there needs to be opportunities for multi-stakeholder involvement in the
Guideline development process and in the peer review on an ongoing basis. Ultimate
acceptance and widespread use of the ERA Guidelines will be assured if they are technically
sound, practical, flexible, and have been communicated in a specifically planned technology
transfer program.
Slide 12
Many of the principles and concepts of the Ecological Risk Assessment Framework
are contained in EPA's Framework publication Figure 1 page 4 (EPA 1992). The AIHC has
modified this figure to present some of their ideas for improvement. First, the flow as depicted
by the arrows has been altered significantly reflecting what we perceive as a significant
philosophical change. First there is no separation by a heavy line of technical ecological risk
assessment issues from initial or final policy issues. Starting with the initial planning
discussion between the risk assessor and manager, in which policy issues are more
comprehensive and spelled out, there is a one way flow of information into the Problem
Formulation Phase, down into the Analysis Phase and into Risk Characterization. After Risk
Characterization, the risk assessor and risk manager review the results. They can either
decide to make a "final" risk management decision or decide that there needs to be an
iteration back to Problem Formulation or collection of additional exposure and/or effects data.
Results from this iteration are once again considered after Risk Characterization in a second
discussion between the risk assessor and manager. They can once again decide to iterate
back up for more information or make a "final" risk management decision. Notice that there
are no back and forth arrows. Instead, arrows better reflect the iterative process, going in one
direction, through tiers of data collection after discussions between the assessor and
manager. This modified figure more accurately reflects state-of-the-science practices.
Slide 13
The AIHC Ecological Risk Assessment Subcommittee feels confident that the window
of opportunity exist now to develop ecological risk assessment Guidelines that will be
scientifically sound, practical, and prove very useful under a wide variety of program needs.
We are committed to making the expertise of AIHC member companies available to the EPA
to assist in whatever manner would be most appropriate to attain our mutual goal.
179
-------
^
(/)
£
^ —
re
o
"5)
o
o
o
LJJ
C
O
0)
m>
j^ji
o
o
0.
0)
ac
"0
[O
•^
o
•o
c
<
^
mewor
re
•••
u.
HM
c
juium Or
ssessmeni
5 <<
o ^
u S
S 'eS
'S o
s •*•<
£ OD
O
•< /o
^ {^
-M
=
£>
S
•ft
O
elineDevel
T5
_ ^^
• ^^
^3
O
1/5
O\
y^ ^
C\
i-H
r\
m
>»
«
^
>
lexandria.
<-
'ss
fi
S
0
U
i-3
-M
"3
o
Industrial II
AIHC
S3
C3
O
"S .
s
•^
u
1
nt Subcom
isk Assessme
fr
*~+
C«
*5i)
^^
^^j
f \
I
1
a)
GL W
O
-------
V) CD
O ~~
O
O
o
h- Q- "3
.= O 3
i *- o
Jj =
0
8 «
£ w
o
) o
o
CD ro
5.S
w0-
•"~ +J
— CO O G)
a. LU Q
••= e
o w •—
«
S ft
r- ^
I H
% g
i-^ • H
S e
o fe
^v -^^^
« a
^
Q«
H»5
^
QC
Ass
^
/^ ^
& »rJ
«
0>
SS
« A
5/5
a>
U
5/5
^ ^>
^> C
& ^
C^ Q^
^
a C
« ^ «
^
« 5
S ^
& o
0^ ^
>> *PN
Jj ^»>
sefu
ework,
ii
S^ HH O
^ KJ ^M
>>.• CL,
r«
cs
181
-------
(/>
0)
3
•o
C
C
o j2
3
O)
O
EJ
J*= c ^
05-0)
C P O)
(0 -rj
0 ^ O
•-*
o
j=
V5
"5
o
A* S—>
C3 ^
bX) S*
^Sr a
182
-------
o
h-
O
o
l-
0
3
0)
(0
I
!/)
3
•a
—
•*j
ogical Risk Assessmen
o
o
LJJ
o
EH
S3
O
•rporate Commitments And Act
e
u
Environmental Stewardship
^
n
W)
0g
*»2
P
a
o
0>
•1— <
-
^^5
^^J
c
^
u
^
terface Routinely With Regulate
C
.?»
•g
^g
O
U
=
rs
QC
|
feJD
O
. -
r\
=
W)
fl
Q
^
immitment Of Resources -Time.
>
w
U
And Personnel
/-*
urce Of Useful And Relevant D'<
o
183
-------
«
(V
"(5
o
"5>
o
o
o
LLJ
'o
CO
Q.
£
3
^
o
o
E:
'c
4S
o
Q.
£
"""•*
essment Guidelines Is
ndustry.
C/) """*
(/) O
< H
Cuts Across All Regulations
+
WnJ
Often Becomes Enforceable ]
*
S3
O
QC CS
*? ^
Regulations And Standai
Impacts Competitiveness At
*
And International Level
^
CJ
Affects Allocation Of Resour
184
-------
o
•= c
> 3
m £
o"i o
(0 -o 5? CO
2 <
O ^
^•^
o o _ o
uj ^ 2 +*
w ® '5 -5
Q. II O
UJ LL.
Data Acquisition: Verification and Monitoring
OO Jr ^^
e
a?
.
-o c '
.
Q O <0 CO <— '
CDE £
O)
O j£
0) CO
^5-sr
® T3 42
0 =3
c « to
o »- o>
18-
Sg
8 «
tn
c
o
o
c
CO
c
CO
sk
185
-------
O)
CO
CO
0
£
0
E
0
E
0
Q,
"O
£
V)
0
0
Q
com 5 =
0
CO
0
CD 0
£ £
CO
0
0)
"O Q "O
^ 3
2 o°
£ 1 O
E ® u_
M W
2 c
H "" Ul O
(0 O
CQ
N
0 CO
0
0
m
0 CO "^
•— /I* ^
O) 05
"U
£
0
E= £
o
5 "5
CD
0 £
0
o
o
0
3
s-*
OS
Gu
O '"^
43 d
H °
O
O)
s
o
3
C<
^ S fl
d ft
C3
T3
i
QC
o>
S3
C/5
T3
d
S J^ o
C3
d
o
S 3
00
-
O O
c/5
C/5
3J
GO
C/5
H^J
00'
-
;O
•^ O'-tf
o^
o
CJ
W)
e
fl
C3
^
CO W
8 ^
£ S... H
a 2
•••< *^
H —
o>
>
era
53
35;
5/J
186
-------
•D
tf)
O
tf) c-
5 S co
u 5- w
G) «: O
CQ ^ W
W
O "D
(f)
ion
— O .±
05
CO a)
0 0
0
CD CD CL
cussion Bet
d Managers
s
s
sues
g Dis
ors an
Is
c
C3
W2 -^
co rv>
> s § .= -s g
.^ wJ U OT 03 -5 > £
o
oo
.
« s -s .s 53
S.ffl c S E- :OH
°
CJ
187
-------
13
CD
CO £ ^
CO o CO
•3 ?T^ **""T
^^^M m ^^^^r
f \ ^^L ' ^^^_
ffj ^HB XI 4
^^ Cl J
• ••M »•• ^»y
Qfl3 *c
^,,1 i^^
^2^* «^*^
® rn
fl[| ^^^ V^*
•u -- •— o-
"5 Q£ 15 co
OT5 "5.
0)5=-.°-
liat Policy Issu
r Risk Assesso
iring And After
laracterization
«^. ^O ••* «"ta«
^ CQ Q O
,*»
4>
w c
Q ,f ^ -
AMU _ . ertiic >onir
^ ^
c v v
c
£
i
15 C e 5
WO | |
m ^8 •= _ •;
J P _U If
~
a
• C — T 2
2 o g ° ui
S =
C iC O
1 i < 1 I
a> 2 "™ •" 1 » | "
o g S 18
"S OL |
fii
A
\£
= 2 S » •=
•1 P s , II
* = * T2 = C ^
3 | < s 5 5 -
£ ^ j: * E
o « « « .*-
o 2 g... ._
^
*•»
^
j
«
> w
O v
~ <
J
)
3
3
*
•^
X
^
iscussion On Assessment And 1
uring Or At Completion Of Ris
Q Q
... l^^l
a monuonnt [^ «J™
t
1
_i_
•° • fl
V^ B 1= i
^>'g Jl- 1
^^ *- o -j " g3
JJ 03 g S "
1 ' I i-t **! '
^v^ O « 8 i
i i*j> ^ ° 2
EC
*
*
^' N
^ \
» x \
\
s
A . ..'
N'
\
\
s
\
\ , .
• \
£ H ^ 8
N ^
* *>^ * *"^
i_ i r~
tj-i 5
0 =
-B ^~" *co -a ^
S j§ ^ J ' ^ '
«3 ^ ' c ^ 2
S 1*5 ! 1
^ is P- s.
e- S § S -8 -
V- rt ^ H-C « . OT .
1 1 11 ! 11
! 1 H| i tf
1 1 P, 1? it.
|. £ ^1| <£ 1|
5 j= l^i 2! aj .
•— " t^ ^r?co .„ "^ *-' "^
co co £. ^- « g^'«^
c3 cc ^"oco O .52 ' ^ .52
^ ^' 43
PQ •< Q C^ !> C^
-1
188
-------
o
«^
o
<
T3
C
0)
E
delines Should Recom
• MM
3
0
»
o
"£
.2
cr
o
<
CO
CO
o
o
(0
o
••Ml
Q.
<
•D
^^
, ,
' *••
O
£
(/)
o
(/)
C/5
jo
t5
c
0)
2
^^^^
X
LU
•
c
"o
Q
o
CO
m
i
C/>
E
c
O
^^^_
0)
a:
't/T
•HM
O
^^^^_
^^^^B
LU
T5
C
r\i
Simple To Complex (Exposun
- QSAR
*
- Single Species Tests
0)
c
"£Z
0
«
O
s
- Modeling To Measuring To
- Micro / Mesocosms
CO
o
sS-
LU
- Primary, Secondary, Tertia
- Field Ecosystems
.2
<5 «.
^^^ "*^
o *S
^ CO
Clear "Pass / Fail / More Data'
- Assessment Factors Are U
+• .
ff\
(fj
'o
o
CD ®
— •;•••
<
CO
c
o
"co
'o
0
Increased Efficiency And Qua
4>
189
-------
o
o^
pI-D •O
CD "U
CO
0
N
co
(-j «_> ~ W
H
C
53 o
S3
53
to
J o < | f '- | .•« C -E
/—\ *^- !—« f'S i-) r* O "-— <^"") d
O ••r -—-4 *^_ •^>. — -^ /*•!_.! dZ
•__, r— —* ^ «^% . £/} rj w C! ^-
^ 's o "§ i ? o _ ~ .2' •§)
<^^ C-c^i c. ^ w^ t>> ?^
i—1 *! ^ yr™ f-> 3 co G <— , x
H O ta ca .5 O< < E- < £
Discussion
Between The
Risk Assessor
and
Risk Manager
(Planning)
anagement
190
-------
"go
"re ®
'? =^
c re w
^ ___ , ^ ^^ (
nsideration Of Certain Tech
Dgrammatic Issues Will Enh
:ectiveness Of Ecological Ri
sessments In The Future.
o c q= w
O Q. LU <
»-
s
u
•o
"3
•o
o
Ecological Significance/ Conceptual
Goals
^
Uncertainty Needs To Be Described
- Transparency
- Assumptions
- Defaults
^.
.0
o c
Jr .2
CS 0 03
« S ,£f
^Vfc^ ^v • p**^
^^: • ^^^J
i^^ ^^^\ ^^^^
^ "S &
° S e3
5fJ Q OPM
c^ ,*: o
Inclusion Of More Probabilistic Asse
More Attention To Physical And Biol
Opportunity For Flexibility, Site Spe
Consistency / Harmonization Needed
- Within EPA
^ ' + 4. ^
2
C
"3
c
0
«3
- Across Agencies
- Internationally
Communication / Technology Transf
Opportunities For Peer Review
* +
->
191
-------
c/>
0
O
C
03 *->
JC C
O ®
•&£ Q.
O O
^ «••—
> 0
G) >
la
"O =
10 2
0 3
8®
3 C
^ jt^L '""" •""
\fi _.
V) CO
— o
CO
CO
(0
E
15
"5
"o
u
HI
C
tc •
Dblem Formul
••
^f
ol
A
•
'
•"
| ANALYSIS
c
o
•4->
03
fft
i-
u
£
N UJ
!»*
2
C8
O
u
"o
u
UJ
acterization
of
xoosure
L.
(0
T
c
o
**••
S _ o o fe
g o£g o>B
E w C gj -o c -
I UlS|l
"o
Q.
III
•
c
o
^> N
V ^
isk Character
i
t±
•
1^-
b.
05
(0
f? 0> CO
Assessment and 1
Discussion Betwe
Risk Assessor and Ri
(Results)
4-»
C
0)
r
Risk Manage
Decision
:
192
-------
<^
^ 0
CO
CD
H
*-» O
C I—
3J2
^ CO
o *
°UJ
J2
0)
Q)
u. r
O °
X
<0
c
3
>»
= O
CO H
£ (D
= >
c S
C.2Q.
O O —
£(/> =
W 0) >
O
E
O
O
O
CO
03
O
>>
c
CO
v? +•• CO
15-^
S i— t3
"co w ^
O c °-
• ••• ••••
WK ««"• «^
^-Q)^
O
LL
"5
*G)
LU
O
HI
193
-------
-------
Timothy Kubiak
195
-------
Timothy Kubiak
PREWORKSHOP COMMENTS
EPAs DRAFT ECOLOGICAL RISK ASSESSMENT GUIDELINES
On a general first read, I was impressed with the overall content, scope and organization of the
draft ecological risk assessment guidelines (guidelines). As someone who has been involved
with articulating risks and effects of multiple limiting factors on our natural resources as well as
recognizing the difficulty thereof, I commend the authors and cooperators for the effort that has
been expended to date and the overall result of containing the potential to write more than can be
absorbed by practicing risk assessors hi their daily duties. The complicated workings of natural
systems, species associations and then- relationships to various limiting factors will always be
difficult because our scientific knowledge is imperfect and the process of modeling risk and
documenting effects (given this lack of knowledge and properly described as uncertainty in these
guidelines) is sobering to all professionals tackling such endeavors. Regardless, the structure is
well-developed. What is needed is a further iteration of examples and qualifications of the
textual presentation so as to begin to operationalize the paradigm without compromising
flexibility and innovative thought by those participating in such activities. I would hope that
EPA will endeavor to update these guidelines at frequent intervels once a decision is made to
finalize initial deployment.
Questions 2 & 3: A schedule for extensive use of technical guidance documents to allow for
communication across the scientific community that will reduce uncertainty to the use of the
process and promote new and better ways of articulating risk as well as documenting causality
with the weight of evidence approach.
196
-------
Timothy Kubiak
Question 8: The introduction needs to stress the need for risk assessors and managers to take a
multidisciplinary approach to the risk assessment process. The nature of the environmental
sciences today requires the use of many discreet, but related; disciplines. Risk assessors and
managers need to make a conscious effort to facilitate acquisition of the needed talent and assure
that the process is supported by the expertise that is needed to drive a credible, timely and
scientifically defensible process.
Question 9: The risk assessment guidelines do include the process of using cause -effect
epidemiological criteria but the guidelines do a relatively poor job of providing guidance on how
they are to be used and factored into the decision-making process. The question that needs to be
answered for use of this information once it is generated is: What do I as a risk
manager/assessor have to ignore as scientific understanding of cause-effect information and what
remains uncertain with regard to the understanding of risk and as well as cause-effect? These
questions are most important when undertaking a retrospective assessment of existing problems,
but many of EPAs actions, within its varied programs, deal with these problems.
There is a need for the manager and assessor to have a visible reminder to acknowledge the
bidirectional feedback loop from operational risk assessment and the research needs and research
input into the process. The diagrams could be improved by having a double pointed arrow
between research and the risk assessment process. Language should be added that uncertainty
and the resulting needed research should not be used as a shield to make informed management
decisions given that there is always going to be some degree of uncertainty. Further, the risk
manager needs to be informed by operational specialists of the need to identify secondary"
management actions or issues that come about because of the initial risk management decision.
For instance, not fully remediating a degraded aquatic resource to "criteria or standards" may
result in the secondary decision to propose and adopt temporary variances and/or losses of a
beneficial use. Intragency action is clearly the intent of the guidelines but these decisions can
197
-------
Timothy Kubiak
have effects on other agencies and related statutory responsibilities. There are not many
"actions" for which there are not "reactions". There is a need to fully define the scope of the
risk management decision that very often goes beyond the immediate primary decision.
Question 10: Categorization of the types of assessments is valid. It would be my wish to have
sufficient guidance developed for/by EPA programs that routine use of the process is not held
hostage by lack of technical support guidance.
Question 13: I believe EPA needs to consider a more important issue relative to stressors as a
definition and relative to adverse effects assessment. Consider using Liebigs Law of the
Minimum, as the overarching synonym for stressor. This law states that if all but one of the
substances necessary for growth of a plant are present in appropriate concentrations, the
inadequacy of the one that has too low a concentration will prevent growth. This law can be
used to address excessive concentrations as well. Indeed, the law is the basis for all stressors
identified in the guidelines and regardless of whether the limiting factor exposure is negative or
positive. Two things are important here, 1). the identification of limiting factors relative to
management issue and 2). appropriate rank-ordering of the major limiting factors (chemical,
biological or physical) for effective management actions to be undertaken, if action is the
selected option.
EPA also needs to more fully discuss the interrelationships among and between stressors/limiting
factors. For instance, if a chemical insult is remediated absent the knowledge that the habitat is
physically or biologically incapable of providing for minimum sustainability for restoring a
target species, both limiting factors would have to be addressed and acted upon for management
actions to succeed. Many wildlife disease outbreaks are mediated by host organisms, some dead,
that are caused by physical and biological inteactions. Chemically- induced immune suppression
which leads to disease is also a well established risk although less is known about actual cause
198
-------
Timothy Kubiak
and effect.
Page 86 of the guidelines concludes the brief discussion of causality. This discussion is
particulary weak and contains no citations from which the authors can draw examples of the
cause-effects linkage process which shows how the analysis of these criteria is performed.
Examples are the two cause-effects linkages meetings on Great Lakes contaminants effects and
resulting analyses that are published in the full volume of the Journal of Toxicology and
Environmental Health (Vol.33, No 4,1991) and the partial volume of the Journal of Great Lakes
Research (Vol 19, No. 9, 1993). Since the discussion ends on the issue of importance of
causality in retropsective analyses, it would seem prudent to include these references since they
collectively depict a range of "certainty" with respect to causality and would serve to provide the
background for seeing how difficult generation and assessment of causality information really
is. This should not be viewed as diminishing the need for inclusion of this concept into the
guidelines, but rather to help focus attention on the difficulty as well as the utility of assessing
and demonstrating causality to various degrees.
199
-------
-------
Lyman McDonald
201
-------
Review of
DRAFT PROPOSED GUIDELINES
FOR
ECOLOGICAL RISK ASSESSMENT
Prepared for the
Risk Assessment Forum
U.S. Environmental Protection Agency
Washington, DC
by
Lyman L. McDonald
WEST, Inc.
Cheyenne, Wyoming
November 14, 1995
PARTI. GENERAL PRINCIPLES
• How can the ecological risk assessment guidelines be improved to more clearly
indicate underlying assumptions and default positions?
o I recommend that a distinction be made between evaluation of risk of a
present stressor(s) (e.g., the Baird & McGuire Superfund Site) and the risk
of a future stressor (e.g., the New Chemical case study). The evaluation of
risk of present stressor(s) allows study designs and data collection similar to
those used in "impact evaluation" due to the release of hazardous substances
under the Comprehensive Environmental Response, Compensation and
Liability Act. The U.S. Department of the Interior has promulgated
regulations for the performance of natural resource damage assessments and
those regulations should be referenced in this document on risk assessment.
• A discussion should be given of the value of monitoring after risk assessment and risk
management decisions are made.
" Monitoring for occurrence of ecological effects provides a reality check on the
predictive capability of risk assessment. For example, if the levee is
constructed in the Bottomland Forest Wetland case study, are the predictions
of loss of wildlife values realized. Are there fewer birds being killed by
insecticides now that carbofuran is not being used?
o Monitoring allows for development of data bases which will be of value should
reanalysis of risk assessment be necessary.
202
-------
Lyman L. McDonald, 14 November 1995
PART H. SPECIFIC QUESTIONS
Guidelines Balance
2. • I found the case studies to be very useful and recommend that they be
included in the Introduction Section. At least, the Foreword and Executive
Summary should include a recommendation that the appendix containing the
case studies be read first if the reader is not familiar with the jargon of the
text.
• Specific comments
« Clarify differences between "assessment endpoints" and "measurement
endpoints"; if there are any, hi lines 25-4, pages 11-12. Indicate which
endpoints are in each category in the case studies if appropriate. See
lines 22-26, p. A-9.
o Re-emphasize in the case studies that choice of a model is itself a
subjective (qualitative) decision and quantitative results depend on the
professional judgement that the model is appropriate. In particular,
see lines 12-16, page A-5. .
o Stress future monitoring of ecological effects of management decisions
as a means of advancing predictive ability of risk assessment.
• o In the carbofuran case study, Ecological Effects and Endpoint
Selection (lines 17-22, p. A-6) emphasize the population level
ecological effects on birds. Later in the case study, population level
effects seem to be the target, not the death of individual animals.
Include discussion of uncertainties associated with population level
effects (lines 25-30, p. A-7).
o Give a map or figure showing the location of Waquoit Bay (p. A-17).
° Eliminate most of the acronyms in case studies, particularly in the
New Chemical case study. There are 20 acronyms on p. A-16.
o The New Chemical case study is understandable on pages A-16 to A-
18, but is not presented very well in the text of Section 4, The
Analysis Phase.
4. Guidance on uncertainty issues?
203
-------
Lyman L. McDonald, 14 November 1995
• Some additional work is necessary to clarity the relationships of "statistical
inferences" and "uncertainty" as used in this guidance document.
o Statistical inference; for example, in the form of a confidence interval
on a ratio or EQo, is one of the sources of uncertainty.
o Statistical inferences are to the experimental protocol or to a study
area (with data collected by a specific protocol).
o Uncertainty in risk assessment includes questions concerning whether
the experimental protocol or data from a study area are appropriate
for evaluating risk in the situation under investigation.
o Statistical inferences; with the associated uncertainty concerning
applicability of the protocols, are but one source of information for
further deductive professional judgements in risk assessment. Final
risk characterization will always be deductive, regardless of statistical
inferences or model predictions.
o I recommend that the sections which mention statistical tests of
hypothesis and power analysis be rewritten to emphasize use of
confidence intervals for summary of statistical inferences. Present the
inference in the form of a confidence interval when possible. Most of
the problems associated with misuse of statistical inferences go away.
5. Modification of the Introduction?
• Move the case studies to the introduction,
6. Terminology?
• Expand section 1.6. Definitions and terminology.
o Emphasize the definitions in section 1.6.1. as was done in section 1.6.2. For
example, ecological risk assessment in line 5, p. 11, should be in a bullet.
Also, retrospective risk assessments, assessment endpoints, measurement
endpoints, etc. should be in separate bullets (Lines 10-16, p. 35 should be
repeated with the definition of assessment endpoints) etc risk
characterization, risk estimation, risk description, conceptual model,
transparency of the risk assessment, Ecologically relevant endpoints, exposure
profiles, stressor-response profiles, group boundaries, outer boundary, effects
characterization, intensity of stressor, synoptic index,
204
-------
Lyman L. McDonald, 14 November 1995
o My copy does not have Appendix B - Key Terms, but there are many terms
which need definitions. For example: variability should be contrasted to
uncertainty in a statistical inference,
° Some words are not in my pocket Webster's dictionary. For example,
connotes and delphic (lines 21-22, p. 13), overarching (line 3, p. 29), and
colocated (line 9, p. A-19). Try to simplify the text when possible.
8. Risk manager's role?
• If possible, a member of the risk manager's staff should be on the risk assessment
team.
Problem Formulation
11. Your intent with "assessment endpoints" and "measures" was not clear to me on first
reading and I may not understand now. I suggest that the case studies be rewritten to
separate the assessment endpoints from the measures under different bullets. In the
Bottomland Forest Wetlands case study, assessment endpoints included forest community
structure and habitat value to wildlife as measured by the HSI and FORFLO modes.
Measures included tree species present, abundance, canopy cover,...?
12. Risk hypotheses and conceptual models?
• Rewrite the case studies to include the risk hypotheses and conceptual models. The
carbofuran, importation of logs, New Chemical, and Baird & McGuire Superfund case
studies do not appear to mention risk hypotheses or conceptual models in separate
bullets. Risk hypotheses were implied in the Waquoit Bay case study, but none were
given.
Analysis
13. Changes to the Analysis section?
• First, the section does not flow well with the preceding and following sections.
Terminology is not always consistent and it appears that the section may have been
written by a different team without sufficient editing to tie the sections together. For
example, in lines 9-10, p. 63, should the terms "assessment endpoint" and "measures"
be used rather than "attribute of interest" and "attribute being investigated"?
• The New Chemical case study is not effectively introduced and appeared naive on first
reading until I reread it in the Appendix.
• There are entirely too many acronyms in this section.
205
-------
Lyman L. McDonald, 14 November 1995
• Are secondary stressors "stressors" or "assessment endpoints"? It seems that they
could be both and maybe this point should be addressed in the definitions section.
• The concept of measurement error presented on line 3, second sentence, p. 63, is
incorrect. Measurement error is the difference between the measured value of a
variable on an experimental or sampling unit and the "true value".
o Repeated measured values of a variable on the same experimental unit will
not necessarily give the same "true value", but rather a distribution of values.
The characteristic of interest is not changing on the unit during measurement
(it is not random).
o To clarify the situation, I suggest introduction of the terms "experimental unit"
and "sampling unit". For example, the sampling unit may be a liter of
sediment from a point in a wetland. Variation from point to point in the
wetland is sampling variability. The liter has a true value for,-say lead; but,
no matter now ones tries, the liter can not be homogenized to the point that
repeated measurements of lead on the liter of sediment will be exactly the
same. This is measurement error on the sampling unit.
• Some general rewriting seems to be necessary. For example, what are "Mismatches
between hypotheses...." (line 17, p. 63), and "...Canonical or reference environment...."
(line 19, p. 68).
• In general, I liked the candor of this section. However, some editing and rewriting
seems to be necessary.
15. The discussion of multiple stressors would be improved with inclusion of a case study.
The Waquoit Bay Estuary case study may be appropriate.
Specific comments
• Lines 29-8, pp. 3-4. These seem to be objectives of risk assessment, not necessarily
results of risk assessment.
• Section 1.6, p. 11. Include examples from the case studies.
Line 10, p. 19. What are Values"?
• Be consistent with terminology. For example, in Box 2-4, p. 22, areal extent and
patch size of eel grass beds might be described as assessment endpoints.
• Lines 4-6, p. 30. Unclear to me.
• Section 3.4.2. Some assessment endpoints are inherently less variable than others.
206
-------
Lyman L. McDonald, 14 November 1995
Box 3-10, p. 42. Where is the turkey and deer example?
Line 8, p. 44. I have used the term "testable hypotheses" rather than "risk hypotheses"
in my past writing. To me, "testable hypotheses" is more intuitive.
Figure 4-5, p. 72. I missed the point of the figure.
Pp. 78-79. The discussion of the statistical methods would benefit with illustrations
from the case studies.
lines 17-19, p. 85. Rewrite. Specifically, the sentence starting in line 17 is misleading.
I suggest "Effects deemed significant at the 0.05 level have a 5% chance of a false
positive (Type II error) under the null hypothesis that there is no effect". As I
indicated above, I suggest rewriting the entire document with respect to hypothesis
testing; replacing results of hypothesis testing with confidence intervals (when
possible).
Line 9, p. 94. Mention identification of assessment endpoints and measurement
endpoints.
Box 4-19, p. 109. One major problem with indices not mentioned is that they do not
keep up with technology very well. An index is sensitive to the methods for collecting
data and when new data collection methods come along, it is extremely difficult to
compare an old index with new assessment endpoints or better measurement
endpoints.
line 25, p. 111. Emphasize the goal.
Lines 7-16, p. 117. This section seems to be out of place.
Lines 28-31, p. 117, and lines 17-30, p. 122. Again, confidence intervals on endpoints
is a better approach than testing of hypotheses.
Lines 25-27, p. 118. Unclear to me.
Line 7, p. 127. Transparency? Rearrange with lines 25-30.
207
-------
-------
Dwayne Moore
209
-------
Dwayne Moore
Comments on EPA's Ecological Risk
Assessment Guidelines
In my opinion, the guidelines are concise, well written and helpful to ecological risk
assessment (ERA) practitioners both within the US EPA and elsewhere. The careful and
logical process begun several years ago by the US EPA Risk Assessment Forum
(workshops, framework report, issue papers, case studies, etc) has obviously been
successful and is a model for others to follow.
Before addressing the questions raised in the Charge to Reviewers, I offer the
following comments that apply to the entire document,
(1) For a draft document, I was pleasantly surprised to find relatively few spelling or
grammatical errors. Further, all but two of the references cited in the text were
included in the references section.
(2) Most of the important contributions to the science of ERA by European scientists
have been ignored (or at least uncited). This is a serious oversight, particularly
given the many advances that have taken place in the Netherlands in
assessments of chemical stressors. For an overview of this work,! suggest the
recently published Risk Assessment of Chemicals; An IntrodusSon edited by Ce«s
van Leeuwen and Joop Hermens and published by Kluwer Academic Publishers.
In retrospect, it would probably have been a good idea to have had several
European scientists review the guidelines.
General Principles
(1) In general, I agree with the scope of the guidelines (i.e., not program-specific
guidance, not a text book). The guidelines, however, could be improved by.
specifying default positions on difficult topic areas. Many times in the guidelines,
issues were raised, briefly discussed and then left hanging. For example, the
guidelines correctly note that the use of hypothesis testing to determine NOELs
and LOELs in toxicity testing has been much criticized recently. The alternative
approach (regression analysis) is, however, inadequately described. What are the
210
-------
Dwayne Moore
recommended underlying models (e.g., probit, logistic, Weibull, etc)? How low
can the V in the EC* be before mode! dependence becomes a problem? When
is it appropriate to use hypothesis testing? When is it appropriate to use
regression analysis? More discussion.and recommendations could also have
been provided for other difficult issues (e.g., when and how to do quantitative
uncertainty analysis, recommended methods for characterizing sources and
releases, deriving uncertainty factors using empirical methods, recommendation*;
on the use of multivariate statistical techniques). If the guidelines are not
program-specific, not a textbook and rarely make recommendations on important
issues, then I am unclear as to how they differ from the framework report.
>
(2) In general, I liked the layout of the chapters, particularly the inclusion of text boxes
and figures to describe case studies. I would suggest, however, that figure titles
be more complete in order that readers can fully grasp what is being shown.
Guidelines Balance
(1) In general, I believe that the guidelines are balanced with regard to range of
stressors, ecosystem types and spatial/temporal scales. Assessing effects at the
community and ecosystem levels of organization was not discussed as much as
lower level effects, but this is probably more a reflection of the state of the
science, than author bias. I also found that the discussion of physical stressors in
chapter 4 to be somewhat superficial and so general that little was gained over
the preceding discussion of chemical stressors.
(2) The case studies in Appendix A are useful, particularly in giving readers an idea of
the breadth of problems currently being addressed by ERA. The case studies
would be even more useful if they had more detail to help readers understand why
certain decisions were taken, to illustrate complex issues, and to show how the
results were used in decision-making. Although I am somewhat biased on the
issue, I found it rather surprising that not one of the case studies included a
quantitative uncertainty analysis.
(3) The Superfund and carbofuran case studies are examples of terrestrial
211
-------
Dwayne Moore
assessments that incorporated field approaches in estimating risk. These case
studies should be expanded to illustrate how the field approaches were used in
estimating risk to the wildlife receptors.
(4) The discussion of uncertainty in various places throughout the document was
useful to a point and eertaintiy highlighted the importance of the issue to readers.
Where the discussion came up short, however, was the failure to describe and
illustrate specific methods for quantifying uncertainty (e.g., Monte Carlo
simulation, Bayesian methods, fuzzy arithmetic, etc). Even semi-quantitative
approaches for dealing with uncertainty (application factors, empirically-derived
uncertainty factors) were only briefly described with little guidance provided to
readers. Qualitatively describing uncertainty is a very useful exercise (as shown
in the case studies), but this is really only a preliminary step in many ERAS. The
state of the art in higher tier assessments is to quantify uncertainty.
Introduction and Scope
(5) I found that this chapter was well written and clear. Perhaps the discussion of
uncertainty under ERA "principles" is better placed elsewhere since it is not
principles that are being described here. Also, I would have finished the chapter
with a brief description of what follows, rather than finishing on a very small issue
(definition of stress regime). These are, however, minor comments.
(6) Key terms were well defined in chapter 1 and in Appendix B and I really have no
major difficulties with any of the definitions provided. Perhaps, my only
suggestion would be to change "measure of effect" back to "measurement
endpotnf since the latter is now well understood, and has found its way into
several key textbooks (e.g., Suter, Landis and Yu). Our guidance document for
assessments of priority substances in Canada also uses the term "measurement
endpoinf. There was also a tendency in places in the document to use several
terms where only one was required (e.g., ecological entity, assessment endpoirit,
ecological resource, endpoint response, receptors, etc in chapter 3).
(7) I like the ERA framework as it now stands. I would suggest only two minor
212
-------
Dwayne Moore
modifications. The discussion between the assessor and manager during
problem formulation should be: (i) expanded to include other interested parties or
stakeholders, and (ii) continued throughout the analysis and risk characterization
phases.
Editorial and Other Comments on Chapter 1
pt line 3 "draff and "proposed" are redundant
Text Box 1-1 title for case study a-2 is awkward
p3, line 29 sentence is wordy and awkward
p4, line 1 this point does not state how the evaluation of uncertainties is linked to
decision-making (see bottom of page 3)
p4, line 7 see preceding comment
p4, line 9 I don't see how the second sentence is implied from the first
p4, line 11 good point
Figure 1-2 label successive tiers; no need for 3D arrows
Figure 1-3 why is risk communication not included in the framework?
plO avoid writing in the second person
p10, line 22 sentence is awkward
'pi3, line4- delete "for* from "useful for in"
p13, line 22 define "delphic approaches"
p15, line 31 replace "agent" with "stressor" as per box 1-6
Risk Manager Interactions
(8) Chapter 2 provided a useful description of the risk manager's role in the initiation
of the ERA and setting appropriate management objectives. However, since
many assessments (particularly value-based assessments) involve community
groups, industries and other governments, it would be useful to describe how
these and other interested parties should be involved in the process. Text box 2-2
provides useful questions to be addressed by risk managers and assessors. A
somewhat related set of questions was listed in an editorial by Greg Biddinger and
myself that will be published in the December, 1995 issue of Environmental
Toxicology and Chemistry. These questions are listed below, not because they
213
-------
Dvfayne Moore
are any better, but for comparison purposes:
•> What management decision will the risk assessment support (e.g., permit
decision, decision to approve a new chemical, cleanup level for a
contaminated site)?
>• What are the time constraints on performing the risk assessment?
» What is the budget for the risk assessment, including the collection and
generation of additional data and/or modeling?
»• How many assessments will be made? Will more than one alternative be
examined?
> What is the maximum level of uncertainty that will still allow for a decision
to be made?
*• What are the reference conditions against which poosible adverse effects
will be compared?
These questions require numerous iterations before satisfactory answers are
obtained. Further dialogue involving stakeholders will lend credibility to the assessment
process in the eyes of industry, academia and the general public.
(9) The discussion of how ecological information may be related to risk management
decisions was useful as far as it went. However, more information could have
been provided on how to communicate risk to different audiences (many articles
have been published in Risk Analysis in recent years on this subject; also see
Morgan etal. (1992) ES&T26:2048-2056). Further, it would have been useful to
describe how ERA may be used to evaluate alternative risk mitigation strategies
once a problem has been identified, and how decision analytic tools may be used
for optimal decision making (see, for example, papers by Finkel [1994] in Risk
Analysis 14:751-761, and Frederick and Peterman [1995] in Canadian Journal of
Fisheries and Aquatic Sciences 52:291-306).
Editorial and Other Comments on Chapters 2 and 6
p19, line 25 first sentence should read "Management goals help to define the
ecological values to be protected" since assessors are also key in
identifying such values
Text box 2-2 "How likely will recovery occur?" is poorly worded (suggest "How likely is
214
-------
Dwayne Moore
recovery?") and also very difficult to answer prior to problem formulation
text box 2-2 suggest adding "lack of appropriate methods and/or data" as potential
constraints under questions for risk assessors
p21, line 26 water quality criteria are not risk assessments since they do not consider
exposure nor estimate risks
Text box 2-4 introduction to Waquoit Bay problem Is required so that reader can
understand the context of the management goal and subgoals
p22, line 25 re "Questions to ask..."; only one question follows, many more could have
been added
chapter 2 much of what is requested of the risk assessor during planning will change
during the course of the assessment; suggest re-emphasizing iterative
nature of planning
chapter2 this chapter was somewhat repetitive and could be shortened
chapter 6 what is the role of the risk assessor in risk management?
Problem Formulation
(10) The classification of assessments as stressor-initiated, effects-initiated or value-
initiated is useful and captures the major differences in how each of these
assessments is eventually formulated. For consistency sake, it would be nice to
have used the classification system used in the analysis chapter (chemical
stressors, physical streamers, biological stressors, etc). I am unsure which
classification system offers the most advantages and it may be worth discussing
at the workshop.
(11) For the most part, the discussion of assessment endpoints and their relationship
to management goals and measures was helpful and useful although I have
several criticisms. For example, I found that the term "assessment endpoinf was
often used in a vague manner. According to Suter (1993) the operational
definition of an assessment endpoint includes a subject (the ecological value to
be protected), a characteristic of the subject (e.ff., local extinction, reduction in
population size), and a numerical expression of the effect (e.g., probability of a
>10% reduction in harvestable yield). None of the examples in text boxes 3-2, 3-3
and 3-4 included a numerical expression and most were quite vague as to the
215
-------
Dwayne Moore
subject and characteristic of interest (e.g., "trophic status of freshwater ponds and
rivers" in text box 3-4).' I recommend that the text include more operational
examples of assessment endpoints. Similarly, several of the "management goals"
provided as examples are factual statements, not goals (e.g., text boxes 3-3 and
3-4). .Finally, I disagree with the sentiment that societal values have an important
role to play in the selection of assessment endpoints. I believe that the objective
of ERAs is to estimate risks to exposed receptors, whatever their perceived value
to society. It is during the risk management phase that societal values become
important, if for no other reason than the fact that "society" (community groups,
non government groups) is (or should be) present at the decision making table.
Perhaps the only exceptions are ERAs in which societal values were addressed in
the appropriate legislation or ERAs where community groups had a strong role in
the problem formulation stage (e.g., Waquoit Bay). Bottom line: I do not believe
that risk assessors and risk managers by themselves have the right to define
societal values.
(12) I do not feel that the discussion of conceptual models and risk hypotheses was
very useful. Most importantly, the use of risk hypotheses implies that the
objective of ERA is to derive conclusions (e.g., causes an effect or not).
However, the objective of ERA is to describe risk. That is, ERA is a descriptive
tool, not a conclusionary one. Further, much of section 3.5 was superficial (e.g.,
"Depending on what initiated the assessment, different elements are known and
unknown"), wordy and repetitive (e.g., section 3.5.1 just repeats what was stated
on the preceding pages). This section needs work.
Editorial and Other Comments on Chapter 3
p24, line 9 last sentence in paragraph is awkward
Figure 3-1 how are the bullets under "Source and Stressor Characteristics" any
different from the title; where are "measures of effect"?
Text box 3-3 title is awkward
pSO, line 7 paragraph repeats earlier material
p32,line24 'affects8 should be "effects"
p33, line 3 define "forcing functions*
216
-------
Owayne Moors
p35, line 24 sentence is a run on
Text box 3-7 good example
p38, Iine20 add "exposure" at the end of the sentence
Text box 3-8 good example
Text box 3-9 assessment endpoints require some sort of numerical expression (e.g.,
what is the baseline value of concern for species diversity, what is the
direction of concern for species abundance, since an increase in
abundance of one species can lead to a decrease in species diversity?)
;p42, line 28 Bamthouse et aL (1990) is not in the references section
p43, line 1 example is repeated from previous page
p44, line 31 "receptor8 is a better term than "ecological entity"
p45, line 11 point repeats what was stated on previous page
Text box 3-12 define PMN
p47, line 19 quotes should only be used to highlight a key point, not regurgitate a large
chunk of another paper
p47, line 24 Odum (1971) is not in the references section
Figure 3-3 figure is not very helpful unless it is explained; words are hard to read;
spell out "FORFLO"
Figure 3-4 what is the difference between items in squares and those in circles
p52, line 2 how does one show "degree of confidence" in a flow diagram?
p53, line 24 "important" should be "importance"
.Analysis
(13) The balance in the discussion of different stressor types was quite good, except
for the discussion of physical stessors, which I found ignored much of what has
been written in the literature on disturbance and, as a result, is somewhat
superficial.
(14) As I noted earlier, the discussion of analytical methods for the various types of
stressors was useful. However, principles or defaults were rarely stated (e.g.,
method A is preferred to method B, and method C is not to be used). Since ERA
is not prescriptive, the alternative route is to note the circumstances for which
certain methods are preferred. For example, in characterizing sources and
217
-------
Dwayne Moor*
releases, the guidelines could recommend measuring chemical levels in effluents
and multiplying by volumes released as the preferred approach for characterizing
point source releases. The second approach could be to use of models to predict
levels, and the third could be to use emission factors. It would also have been
useful to make a commitment to. full lifecycle characterization of sources and
releases. As it now stands, the discussion of sources and releases is to generic
to be helpful. Similarly, preferred approaches could be specified for characterizing
fate (e.g., preferred models for local and regional scales), characterizing toxicfty
test results (see earlier comment), population modeling, deriving empirically based
uncertainty factors, etc.
(15) The section on multiple stressors is in my opinion an accurate reflection of the
state of the science (i.e., not very well developed) and is well written. There was
an inconsistency with the preceding section because the multiple stressor section
did not finish with instructions on preparing a stessor-response profile.
Editorial and Other Comments on Analysis Chapter
pSS, line 22 why provide such a specific example?
Figure 4-2 what do the different box shapes indicate?
p62, line 5 what are "study group boundaries"? Also the strategy has little to do with
the example or source of uncertainty
p63 periods went missing on this page
p63, Iine24 how can "proportion of animals... exceed a toxicity threshold"?
p64. lineSO "ecological component of concern1' seems to be yet another term for
assessment endpoint
Figure 4-3 caption should be more explanatory; specify what different shapes of
boxes mean
Text box 4-2 this example is so generic that it is not very helpful
p68, lineQ delete "it" from "it the"
p68, line 19 define "canonical"
Sect. 4.2,2.2 although section is not intended to be educational, key fate and transport,
references could be cited to assist the reader
Rgure 4-4 suggest deleting figure since it is not helpful
218
-------
Dwayne Moore
p72, line 5 again, I suggest you provide key references
p73, line 11 I suggest you emphasize the importance of estimating exposure variability
here
, p73, line 24 how are important exposure pathways to be summarized?
p74, line 12 is the term "ecological response analysis" really necessary; seems to
overlap stressor response profile
p75, line 16 add "when" before'preparing"
Text box 4-6 Rand and Petrocelli reference should be by chapter since different authors
prepared different chapters
p77 l am a little surprised that a general discussion of QSAR could fail to cite
many of the important contributions from the Netherlands and others
outside USEPA
p78, line 17 what does "q.v." mean?
p78, line 24 my own experience with toxicity test modeling is that 5% is too low
because at this level the estimated effects concentration is usually model
dependent; much less dependence is observed at the 15 or 20% levels
p79, line 11 references to Suter 1993a throughout the text should be by chapter since
different authors prepared different chapters
p79, line 11 many others have recommended using regression analysis instead of
hypothesis testing (Environment Canada, the Netherlands, OECO, etc)
p80, line 29 John Cairns has argued in many papers that credible extrapolations
cannot be made at this time
p81, line 6 many others have contributed to the literature on extrapolating between
responses - is there some sort of rule, preventing citations from non US
residents?
Text box 4-7 without supporting text, this box is not helpful; also bioenergetic models
are ignored here and in the accompanying text
p81, line 17 the section on empirically deriving uncertainty factors is much too brief and
needs expansion
p32, line 29 many would argue that such extrapolations can be made with "fair
certainty"
p83, top a case study would help to illustrate the concept of allornetry •
p83, line 8 last sentence in this paragraph is very difficult to read
p83, line 21 I disagree with the notion that toxicity tests prevent mitigation of effects;
219
-------
Dwayne Moore
laboratory animals are well fed, free from predation and other stresses arid
therefore may be less sensitive to effect of chemicals than, organisms in
the natural environment
p83. line 26 this statement is pure nonsence; no such proof exists nor are any
references cited to support this point; also, delete "absent data to the
contrary*
p83, line 30 replace "as" with "an"
p84, line 28 the Suter and Bartell chapter on Ecosystem-level effects in Suter's ERA
book (1993) also reviews ecosystem models for the terrestrial
environment
p88, line 13 a paper by Glen Fox (1991) provid.es an excellent expansion to Hill's
criteria for evidence of causality [see J, Toxicol. Environ. Health 33: 359-
373]
pBB, line 7 also note that, where possible, variability and/or uncertainty should be
quantified (e.g., ECJO ± 95% confidence limits)
Figure 4-6 title should be more descriptive; difficult to read text inside shaded boxes;'
what do the box shapes indicate?
p92, line 27 add references for readers interested in exploring the use of GIS for
assessing disturbance effects
p93, line 8 guidelines correctly point out that identifying consequences of disturbance
is "challenging"; however, little help is provided below to assist with this
challenge
p94, line 24 note that many biological introductions have been accidental
Table 4-4 could add Lythrum salicaria (purple loosestrife) to the list as it has caused
serious effects to plant communities since its introduction to eastern North
American wetlands
Figure 4-7 title could be more descriptive; what do varying box shapes indicate?
p101,line11 very useful list
Section 4.4.2 section is well written
p105, line 11 "give" should be "given"
p105, line 13, provide citations for approaches to assessing effects of chemical mixtures
p106, line 20 Lynn McCarty and colleagues have written extensively on the subject of
assessing chemical mixtures using tissue residues and knowledge of
modes of action
220
-------
Dwayne Moore
p10S, line 21 "chemicals in" should be "chemicals is"
p107, line 9 statement is probably true, but not very helpful
Text box 4-18 box is not cited in text; drone and Pastorak (1993) is not the original
reference for the AET approach, please cite appropriate reference
p108,line24 define synoptic index
Text box 4-19 Qtt (1978) is missing from the references section
Risk Characterization
(16) As I noted above, I believe that quantitative uncertainty analytical methods should
have been discussed in much more detail (e.g., Monte Carlo simulation,
dependency bounds analysis, fuzzy math, etc) and recommendations provided on
when and how to use each of the various methods. Such methods are the current
state of the art for higher tier ERAs, Excellent general texts on the subject can be
found in Covello and Merkhofer (1993; Risk Assessment Methods: Approaches for
Assessing Health and Environmental Risks, Plenum Press, New York), Finkel
(1990; Confronting Uncertainty in Risk Management: A Guide to Decision-makers,
Resources for the Future, Washington) and elsewhere. Fora discussion of the
pros and cons of using deterministic versus probabilistic methods for estimating
risk, see the upcomihg debate in Human and Ecological Risk Assessment (e.g.,
Moore and Elliott (1996). Should uncertainty be quantified in human and
ecological risk assessments used for decision-making? HERA 2(1) in press;
Bartell (1996). Some thoughts concerning quotients, risks, and decision making.
HE/W 2(1) in press).
Decision analytic models should also have been discussed as these tools are now
being used to assist decision-making in the face of uncertainty (see, for example,
Frederick and Peterman (1995) Can. J. Fish Aquat. Soi. 52:291-306).
Communicating risk estimates is a tricky endeavor, particularly in complex
situations with high uncertainty. Guidance is required in this area-.'
(17) The list of criteria for assessing ecological significance is, in my opinion, complete
221
-------
Dwayns Moore
and useful. Discussion and guidance, however, could have been provided on the
use of population and ecosystem models to assess ecological significance.
Editorial and Other Comments on Chapter 5
Figure 5-1 suggest adding "and other stakeholders" to discussion box; can the tiered
nature of risk estimation be indicated in figure?; where is risk
communication?; what do the different box shapes indicate?
p113.Iine2 suggest "Risk Estimation" would be a better title
p115, line 3 quotient is semi-quantitative at best; therefore, the statement that "the
higher the quotient the higher the risk" is only true when quotients are
orders of magnitude apart
p116, line 10 much could be added to the list of quotient limitations; for example,
quotients provide little information to the risk manager (e.g., how many
resources should be allocated to a stressorwith a quotient of 1.5 is not at
all obvious)
p116, line 20 please cite original references
Text box 5-3 there are many better examples of "going beyond the quotient method"
than PDM3; an excellent example can be found in Macintosh et a/. (1994)
Risk Analysis 14:405-419
Table 5-1 title is not very accurate; strategy for unclear communication is too generic
to be helpful
p122, line 20 the phrase "change exceeds the variance" should more precisely stated
(e.g., inter-treatment variance exceeds intra-treatment variance)
p122, line 25 sentence just repeats preceding sentence
pl23, line 26 add appropriate references
p126, line 2 the first and third points in the list should be expressed in terms of the
assessment endpoint rather than in terms of the stressor
Text box 5-4 box not referred to in text
Comments on References and Appendices
p135, line 8 "Risk Anal/ should be "Risk Anal"
A-3, line 27 endpoint selected is vague '
222
-------
Dwayne Moore
A-4, line 19 define "net subsidence rate"
A-5, line 6 figures would be useful to illustrate model predictions
A-5, line 9 "general trend toward loss of wildlife values" is vague
A-6, line 21 "on" should be ttof
A-8, line 11 this example would seem to be amenable to a quantitative uncertainty
analysis
A-11, line 12 what does "quarantine importance" mean?
A-13, line 8 define endpoints more precisely
A-14, line 2 what exactly is a "fuzzy set9 approach?
A-18, line 21 why are soil biota, wildlife and plants not included as receptors potentially
at risk?
A-19, line 28 please provide examples of when lines of evidence converged or diverged;
figures would help
A-21, line 25 journal number, pages?
223
-------
-------
Gerald Niemi
225
-------
Gerald J. Niemi
COMMENTS ON DRAFT ECOLOGICAL RISK ASSESSMENT GUIDELINES
Prepared for Eastern Research Group, Inc, 110 Hartwell Avenue, Lexington, Massachusetts
02173- 3134, FAX - 617-674-2906.
Prepared by Gerald J. Niemi, Director, Center for Water and the Environment, Natural
Resources Research Institute, and Professor, Department of Biology, 5013 Miller Trunk
Highway, University of Minnesota, Duluth, MN 55811.
The format of this review follows the Charge to Reviewers document provided by the Eastern
Research Group; however, the questions are not restated. In addition, I have provided comments
relative to each section of the draft document.
Part I. General Principles ,
A. I appreciated the text box approach provided throughout the document. These provided
an easy means to clarify particular points and topics.
B. I found the document quite boring to read, primarily because there was much duplication
of the text material across the different sections. For example, how many times must it
be restated the various issues that a risk assessor must consider in the risk assessment
process. I would suggest that first a good technical writer and second a good technical
editor go through the document very carefully. I would especially encourage them to
look for duplicative material and a means to simply express the concepts presented. In
my opinion the document could be reduced in size by 40 % - at least this is what I would
suggest if I were doing a peer-review of the document.
C. I would have preferred to see a better literature review incorporated into the document.
The document is about "ecological" risk assessment, yet there is really very little
ecology presented. I am not suggesting that the document read like an ecological text. I
am suggesting that because many of the concepts, background, etc. have an ecological
basis, but there are few references to this ecological basis. For instance, most of the
226
-------
Gerald J. Niemi
references are reasonable; however, the same old material is cited such as a plethora of
EPA technical reports, and articles from Environmental Toxicology, Environmental
Management, and Chemistry and Aquatic Toxicology. There is a wealth of new
information in journals such as Ecology, Ecological Applications, Limnology and
Oceanography, and Conservation Biology. Moreover, a wealth of science in foreign
journals such as those from Canada (Canadian Journal of Fisheries and Aquatic
Sciences) or Europe (e.g., Nature, Oikos, and Oecologia). In my opinion, the document
should at least be a springboard to some of these other sources of information. This
information is badly needed to eventually do ecological risk assessment and especially to
improve it. Some examples include pg. 34, text box 3-6 on ecosystems at risk and
ecological effects, pg. 68,1. 9 on mercury, and pg. 73,1. 5 on toxicodynamic model (a
peer-reviewed citations here would be very useful). If too many references become a
problem go to a numbering system so as to not distract the reader.
Consistent with this relatively narrow focus of material cited, is the absence of some
material from the document. I was somewhat surprised to not see reference to
economics (e.g., Costanza's work), indices of biological integrity (indices are mentioned
briefly but again no good citations to relevant material), and meta-analysis to name a
few. I realize that economics would come into play during the risk management phase,
but the earlier we begin intergrating some of this information into the risk assessment
picture it might help in focusing on the most pressing issues. Meta-analysis most
assuredly will need to play a bigger role in doing ecological risk assessment because we
will be forced to combine, collate, and mix data from different studies and places.
D. I am concerned about what God-like person could ever do a reasonable ecological risk
assessment. A risk assessor should, at minimum, be a biologist, chemist, and ecologist
and certainly a data analyst, good communicator (written and oral), and manager. Show
me one. A few times within the document it is acknowledged that risk assessment
"probably" is a team effort - it must be a team effort. The document throughout should
emphasize and refer to the risk assessment management team.
E. Ecological risk assessment, as pointed out in the document, is likely easiest and maybe
227
-------
Gerald J. Niemi
possible for individual chemicals. The document makes reasonably candid remarks
about the lack of information about "real world" ecological system structure and
function and, hence, our inability to make make predictions of future events (risk) .
This', of course, is the major drawback in EPA's ability to actually succeed in doing
ecological risk assessment. The general feeling I had from reading the document is that
the author(s) were also pessimistic about this ability. Hence, the tone, if EPA is to
succeed in working through this process must be more positive. Of course, one must
identify the weaknesses/uncertainties, but a path to the necessary information or process
to do ecological risk assessment should be identified. The underlying assumptions and
default positions to do many of the more complex ecological risk assessments will be
based on professional judgment, educated guesses, delphic approaches, and concensus
methods, but most likely not on data or facts. Will this be the case until eternity?
Part II. Specific Questions
Guidelines Balance
1. The document hits on a wide variety of potential stressors, biological organization,
ecosystems and scales - it probably is a reasonable balance. However, the greatest
weakness is the lack of references to the primary literature on some of these topics. For
example, much has been written on disturbance in the ecological literature, yet there are
few references to this literature. Again, the document does not need to go into detailed
definitions, but a logical step would be appropriate references so that the reader could
pursue some of this material if necessary. Similarly, there are many good papers on
scale, yet these are not mentioned. One can only assume that the author(s) is not familiar
with this work. The document makes frequent reference to "appropriate consideration of
spatial and temporal scales must be included." Why not included more specific
reference to explicit acceptance of various spatial scales (e.g., patch, landscapes, biomes)
from the landscape ecology literature?
2. The case studies are blended reasonably well into the document. If they are emphasized
any more within the document, I think the generality of the document would become lost
in the specifics. I found Appendix A to be so abbreviated that it was less useful, perhaps
228
-------
Gerald J. Niemi
more details with an overall abstract for each would be better. If there is a particularly
"good" ecological risk assessment (e.g., reasonable amount of data and careful analysis),
then I would suggest that such a case study be included to represent a potentially good
example. This could also be contrasted with a data poor example.
3. Reiterating part of question 2, the document should be recognized as a general document
with selected examples. I do not see a great reason to include more "terrestrial"
assessments, but I would agree that a good example as well as reference to "good"
approaches' would be useful.
4. I did not find the tables very useful (e.g., Table 3.1) and again they contained much
duplicative material. I am also relatively confident that many of,the sections (e.g., pg. 9
and 10 - discussion of variability (pg. 9,1. 20) and measurement error (pg. 10,1 0-15))
were not written by a statistically-trained individual (e.g., "Systematic error cannot
usually be reduced."). There are a variety of techniques that can help to eliminate or
reduce systematic error. Note also pg. 85, on causality needs revision regarding
statistics. The author (s) have confused regression and correlation analysis. Regression
is used to examine the dependence between one or more variables., while correlation
analysis is concerned primarily with association, not necessarily dependence.
Introduction and Scope
5. In general, I have few comments for improving the introduction; however, it might be
useful to include some phraseology that was included in our "Charge to Reviewers"
which included a sidebar on what the Guidelines are and are not. Even though the
guidelines were primarily written for risk assessors at EPA, they most likely will be used
(abused) by many individuals who will find the guidelines useful.
6. Terminology is always troublesome. However, the definitions of key terms are included
in Appendix B. I am glad this wasincluded. My major suggestion would be to include
the source for each definition (even if it is Webster's dictionary - version ?). It was
adapted from U.S. EPA 1992a - but most do not have ready access to that document.
Minor comments: 1) (pg. 9,1. 17) - suggest change reducible to "uncertainty can be
229
-------
Gerald J. Niemi
controlled," and 2) (pg. 10,1.8) - "field mice" - either be more specific such as saying
Microtits spp. or be more general by just saying mice.
7. Fine.
Risk Manager Interactions
8 & 9. There are two concerns I have l)pg. 19,1. 18-19- there seems to be great latitude in
how the risk manager and risk assessor "interpret the goals of the ecological risk
assessment." For instance, they are using their information on the "ecology of the area"
and "publicly perceived environmental values." Hence, the iimitations or problems in
this interpretation will reflect the knowledge level of the manager and assessor regarding
knowing the ecology and their "self designed interpretation of the values of the public.
I suggest tightening this language so it is less subjective in interpretation. 2) Pg. 22 -1.
'15-22 (and other places) - ecological risk assessment and the outcome (decisions) is
clearly dependent on "financial resources available" and the "local economic impact."
Hence, a process that does not consider the economics along with ecological is flawed.
ft implies that economics is above the ecological considerations. In practice that may be
true, but it should not be so. In reality, there are a wide variety of scenarios such as risk
assessments with a) high ecological impact, but low economic impact, b) low ecological
impact, but high economic impact, or c) high ecological impact and high economic
impact. I like the analogy that the economy is a subsidiary of the ecosystem (including
its resources), especially in economies that are highly dependent on natural resources. In
this way, the strength of the economy is dependent on the healthy functioning of the
ecosystem.
Problem Formulation
10. I had no problem understanding either of these initiated risk assessments. However,
ecological-value is the most subjective which would ultimately lead to some controversy
in interpretation. The term "biodiversity" (pg. 28,1.31) is too vague to be used in this
230
-------
Gerald J. Niemi
context. A watershed, pond, vista, etc. are all distinct entitites. Biodiversity is not. It
would be better to use an endangered or threatened species as an example. The use of the
term "habitat destruction" (Pg. 28,1. 3) is also not value neutral. Many would interpret
logging as habitat "change" since the habitat is not necessarily destroyed. The old-
growth forest may have been destroyed or eliminated, but the habitat is free to again
become old-growth forest in the future (albeit a long time).
11. As pointed out in the document the most controversial aspect of defining assessment
endpoints is to define what has "environmental value." Everyone has their own value
system, so this is almost impossible to operationally define. I would suggest focusing on
the concensus-building process for defining assessment endpoints at the initial phase of
the risk assessment process. Concensus should be developed among a wide variety of
constitutents within the potentially affected area (e.g., business, scientists, and
environmentalists).
12. I continue to be concerned with the paperwork that is required of the risk assessor. For
instance, throughout the document the risk assessor needs to document many parts of the
process. On page 44,1. 18-19 - "justification for selecting and not selecting hypotheses
are documented." There are many other places within the document where the risk
assessor needs to justify this or that, or document reasons for this decision or for that
decision. All of these areas should be scrutinized and placed within the context of the
scale (size of area) of the risk assessment project. Some common sense could be
applied. Small projects may require less documentation, while large projects that affect
an area the size of a city, a state, or the nation would obviously require more
documentation.
Conceptual models or models in general are useful vehicles to convey the problem and
the connections between the various elements of the problem. However, models can
quickly become very complicated - Figure 3-3 (Pg. 49) is a good example. What
modelers and scientists usually know, however, is that certain connections are more
important than others. A focus on the most important connections can be a means of
simplifying the conceptual model.
13. The overall balance is reasonable.
231
-------
Gerald J. Niemi
14. In general, the document has done a reasonable job in recognizing the gaps in abilities to
do risk assessment on individual chemicals and the quantum leaps to attempt risk
assessment for many biological and multiple stressor problems. However, even with
chemical stressors the document focuses on, for example, using SAR's for risk
assessment (e.g., pg 77) and primarily cites some EPA documents as the primary sources
of information. Similarly, the description of physical stressors includes an analysis
using habitat suitability indices (HSI) (pg. 93). The problem here is that federal
documents continue to cite many of their own models and analyses, which then
perpetuate themselves more with the federal government and other agencies. As far as I
know HSI has never been peer-reviewed. Is there real science and science acceptability
behind these models? If these models are now to be used in ecological risk assessments,
then there should be some review or at least documentation that these methods and
models are accepted by the general scientific community.
Another general problem reflects some ecological naivete reflected in the document. In
several places (e.g., pg. 59,1.20-23; pg. 71,1. 11-18) there are suggestions that we
really know more "ecologically" than we do. In how many systems do we know how
abiotic factors will influence the degree of contact? How much do we really know about
competition (assuming here interspecific competition is implied?) and its effect on
resource utilization ("force some organisms to utilize contaminated areas"). These are
still interesting ecological problems - their usefulness in ecological risk assessment is
very, very limited.
15. Regarding multiple stressors (pg. 105,1.4), the document should distinguish between two
types of potential multiple stressors - 1) past stress plus present (or future) stressor which
I believe is "cumulative impact", and 2) multiple stressors - a combination of several
stressors applied coincidentally (e.g., mixtures).
The discussion on additivity or "synergy" among combinations of chemicals could be
clarified. A key point here is to identify the mode of action of a chemical which may or
may not allow for the prediction of toxicity, depending on the mode of action. Some
232
-------
Gerald J. Niemi
combinations can be predicted and for others with different modes of action the
uncertainty is very high. I am also surprised that there is no mention of the ASTER
system (a computer system for the analysis of risk) developed by the EPA's
Environmental Research Laboratory in Duluth. The system can make a,number of
predictions about chemicals including mode of action.
Some aspects of biocriteria or indices of biological integrity should be weaved into the
discussion on page 109. In addition, as previously mentioned, a meta-analysis approach
would logically fit into the discussion on weight of evidence approaches on pg 110.
Risk Characterization
16. Chronic toxicity should be weaved into the discussion on the Quotient Method. Also the
overall issue of recovery should be discussed throughout the document. For instance,
recover is finally mentioned on pg.. 123. If a stressor produces a high impact, but
recovers in a few days, then the stressor may not be a problem. However, if a stressor
has a relatively low impact, but the system may never recover, then this stressor may be
of more concern. Sorry to cite my own work, but Niemi et al. (1990) - Environmental
Management 14:571-587 and Detenbeck et al. (1992) - Environmental Management
16:33-53 would be highly relevant to this discussion.
The discussion on Scale (5.3.2.2) seems out of context. It primarily deals with habitat
and landscape issues which are appropriate, but there is not an appropriate lead into this
material. For example, landscape fragmentation to me appears.from nowhere. Page 125
-1.17-18- "continuous logging of old-growth forest will eventually eliminate the forest
ecosystem." No - logging reverts the forest.back into an early-successional stage, but
normally does not eliminate it as forest. Conversion of forest to a corn field or a
shopping mall eliminates the forest. Page 126 - 1. 20-29 - I suggest putting
anthropogenic disturbance into the context of the natural disturbance regime.
17. As discussed above - recovery should be emphasized more. Also ecological (biological)
233
-------
Gerald J. Niemi
significance must always use a common sense approach over "statistical" significance.
Statistical significance can simply be a result of a statistical experimental design with
high power for detecting differences (unfortunately this is seldom the case). In any
event, the risk assessor (scientist team) must be the statistical evidence in the proper
context.
234
-------
Richard Orr
235
-------
Date: November 13* 1995 Richard L. Orr
USDA APHIS PPD
Senior Entomologist
COMMENTS ON DRAFT ECOLOGICAL EPA RISK ASSESSMENT GUIDELINES
Part 1: General Principles
A "risk assessment' can be based on sound science but it is incorrect to claim that risk
assessment is a. "scientificprocess" (page 4, line 8; page 12 box 1-4). Indeed, many claim
that a good risk assessment is more art than science. The traditional methods/processes of
science are still sound and it would be inappropriate and dangerous to give the impression
that a risk assessment process could provide answers that are best answered by basic or
applied ecological research. For example, it will always be easier and cheaper to pool a
couple of "experts" and ask them what they think will happen and call it "professional
judgement" than it will be to apply the scientific method to address the same issue using an
experimental approach.
Consequently, the statement on page 17 (lines 3,4) "The purpose for an ecological risk
assessment is to[provide scientific information about ecological risks to managers
making environmental decisions" would be more correctly stated as "The purpose for an
ecological risk assessment is to organize scientific, and other pertinent information, into
a format that can be understood and used by managers in making environmental
decision^. The wording between the above two statements may be subtle but the
philosophical repercussions are major.
The scope of the definition of ecological risk assessment (on page 11 line 6 & 7, "The
process that evaluates the likelihood that adverse ecological effects may occur or are
occurring as a result of exposure to one or more stressors"} is too limited. Risk
Assessment is universally recognized as the likelihood and magnitude of an adverse event.
The EPA definition focuses only on the likelihood and not the degree of the severity which
will occur. Half of the risk formula is being played down in this definition.
236
-------
Part n. Specific Questions:
Page 40, line 6, states that "Assessment endpoints are not management goals". This is
too absolute and probably should read "Assessment endpoints are usually not
management goals" or "Assessment endpoints need not be management goals". For
example, the establishment of an exotic pest due to a specific pathway of introduction can
be both an assessment endpoint and a management goal.
Page 97, the Text Box states that "A team of six APHIS experts evaluated...". This is not
true, the experts were not from APHIS (Animal and Plant Health Inspection Service,
USD A) but were individuals from various disciplines brought together by the Forest
Service to conduct the assessment. It should read something like "A team of six experts
under the auspices of the Forest Service evaluated...". Also, page 103, in the Text Box
replace "APHIS" with "The assessors". Again this was not an APHIS document.
Page 99, last sentence, line 28 through 30. If "Agency" is referring to APHIS then the
statement in not correct. In fact, APHIS has completed nearly 100 environmental
assessments in support of management decisions to allow releases of nonindigenous
biological control agents. Each evaluation considered, in general, the potential ecological
impacts, and specifically considered the host ranges of the released organisms and how
host range would ultimately impact upon the ecology and agricultural of North America.
In addition, for the past several decades, APHIS has subjected all proposed releases of
phytophagous arthropods to very extensive review often consulting with a variety of
Federal agencies (including EPA), State agencies, as well as both academic and industry
representatives.
On page 100, line 13 and again on page 102, line 4, the word "bilge" should be replaced
by "ballast". Bilge water is a minimal pathway for aquatic organisms, ballast water is the
real problem.
237
-------
On page 100, line 26, the third bullet (line 24 & 25) needs to be changed to read "Are '
there current mitigation requirements or conditions that would kill or mitigate the pest
prior to entry, during transport, or at the port of entry".
Page A-10, line 2, the reference for the USDA process should be included (Orr et. Al.,
1993) after Inspection Service.
Page A-10, line 20 please replace "(the pest's capability to hitchhike on log shipments)
with "(the pest's capability to be,with the log shipment at origin)". "Hitchhiking" usually
refers to an organism that is not normally associated with the commodity, but gets on it
anyway. For example a wheat pest moving with logs.
Page A-ll, line 10, please replace "influences" with "repercussions from pest
establishment.
Page A-l 1, lines 27 & 28 reads "Although these species could be important, they are not
considered in the assessment." This is misleading, all organisms identified underwent
evaluation using the hazard identification process outlined on page 18 of the Chile log risk
assessment. Only those organisms meeting certain criteria were further evaluated using the
risk assessment process. Therefore, the sentence should read "Although some of these
species may become plant pests if established in North America they -were not assessed in
detail because the initial hazard identification process identified them as minimal risk."
Part ffl. Minor Corrections:
• Page 12, lines 10 & 20, should read "figure 1-1" not "figure 1-2".
• The end of the sentence is missing on page 39, line 20.
• Page 63 is missing periods.
On page A-9, line 11, please replace "7 CFR Parts 300 to 319" with "7 CFR
319.40".
238
-------
Kevin Reinert
239
-------
Kevin H. Reinert
11/15/95
Comments on the Draft Ecological Risk Assessment Guidelines
Specific questions listed in the Charge to Reviewers:
1. Certainly, more text is devoted to the discussion of chemical stressors.
Although more space should be devoted to physical and biological stressors, the state
of the science around these stressors is not as advanced nor are there as many risk
assessors experienced in these 2 areas. Additionally, more emphasis is placed on
single organism and population levels of organization. Again, more text should be
devoted to higher levels of organization (e.g., community and ecosystem), but these
have not been developed to the same extent as lower levels.
2. See comments specifically on Appendix A.
3. More examples of terrestrial assessments and field approaches should be
added; the aquatic matrix is better represented in the text. However, fewer studies
have been conducted in terrestrial and other field scenarios and therefore these areas
could probably not be enhanced without substantial effort.
4. Several comments on uncertainty are described in the following text. Most of
the uncertainty sections are written well.
5. These guidelines will certainly be used by more than just regulatory risk
assessors and managers. Use by others (e.g., industry) should be considered. See
comments specifically on section 1 in the following text.
6. The listing of Key Terms is essential. Possibly, more than one definition could
be listed for some terms as many programs may interpret these terms a bit differently.
7. I have attached revised versions of the framework figure as presented by the
American Industrial Health Council (AIHC) at the EPA colloquium held on May 3,1995.
Changes to the planning, risk management and data validation boxes should be made
in order to better define these aspects of the framework.
8. See comments in following sections concerning the role of the risk manager in
ecological risk assessment (ERA).
9. See number 8.
10. The categorization of assessments into 4 categories probably will not assist the
reader in understanding ERA concepts. New terminology is introduced that may in fact
confuse even the seasoned risk assessor. Additional comments are listed later in this
document.
11. Specific comments are listed later in this document.
240
-------
Kevin H. Reinert
11/15/95
12. Please see the specific comments for suggestions.
13. Please see specific comments.
14. Please see specific comments.
15. As discussed later, J. Foran at the ILSI Risk Sciences Institute should be
consulted on the issue of multiple stressors. He is currently working on a project
supported by EPA concerning multiple stressors.
16. Probabilistic ERA approaches have not been discussed in detail. Such methods
form a significant portion of many ERAs conducted in industry and are part of the
Aquatic Risk Assessment and Mitigation Dialog Group (ARAMDG) approach to aquatic
risk assessment (tier 2). Additionally, although an iterative approach is described in the
guidelines, the concept of tiers in ERA is not well developed in the document. Some
sort of screening assessment should also be described. Such a screen (as described
in the draft Superfund ERA guidance) is useful to set priorities and address those
issues that need specific ERA.
17. Ecological significance is a function of 3 factors, contaminant variables,
exposure pathways and site-specific factors (see. Brown and Reinert, 1992. Env. Tox.
Chem. 11:143-144) The four factors listed in the guidelines encompass these factors.
Page by page comments:
1. Introduction
P. 2, Text box 1-1: Although each of the case illustrations are useful, could examples
of case studies discussed earlier during the guidelines development been chosen that
more closely follow the ERA framework?
P. 2, line 31: Risk management (RM) always is addressed in another document or is
out of the scope. This is an extremely important aspect of the whole decisionmaking
process that must be addressed. Where and when will this issue get some text
devoted to it?
P. 3, line 1: Cite examples of other Agency-wide guidelines prepared by the EPA
Risk Assessment Forum (RAF).
—, line 16: delete issues.
241
-------
Kevin H. Reiriert
11/15/95
—, line 19-21: This statement is incorrect; all risk cannot be eliminated. Insert
'addressing and' before eliminating and delete all.
—.line 23: Insert risk before management.
P. 4, line 2: Uncertainty analysis helps focus both research and risk input refinement.
—, line 15: In some cases, a risk assessment (RA) may not be conducted because
the data cannot be gathered and RM alone takes care of the decision without the
benefit of RA (e.g., precautionary principle).
P. 5: The boxes on the right and bottom in this figure deal primarily with RM; during
the May 3 Colloquium, AIHC provided additional detail for each of these boxes which
begins to 'beef-up' the RM aspects of ecological decisionmaking. Aspects such as
review of existing legislation and regulation, international implications of the decisions,
corporate stewardship policies, and societal values and concerns should be discussed
during the planning stage. These aspects should be seriously considered for these
guidelines or at a minimum be included in an addendum to the ERA guidelines. The
figure is attached. The additions address the planning discussion prior to the conduct
of the ERA and the RM interface after the ERA has been completed. Discussions
between the risk assessors) and risk manager(s) should not only occur before and
after the conduct of the ERA, but continue throughout the process so that both parties
can chart the progress of the ERA and provide for fine-tuning during completion.
P. 6, line 27: Add 'often' before reduces.
—, line 28: Industry has also been conducting ERAs for many years, some
specifically under regulatory programs and other to better understand their products in
the environment Not only regulatory organizations should be listed here. Also,
reference to various groups in the European Union (EU) conducting ERAs should be
listed. .........
P. 7: I am not sure what this figure brings to the text. The iterative nature of ERA is
described directly in the text The tiered concept of ERA should also be discussed on
page 6 while discussing the iterative nature.
P. 9, line 6: Uncertainty should be thoughtfully addressed whether qualitatively or
quantitatively documented. Dr. G. Charley, Executive Director of the Congressional
Commission on RA and RM recently stated that the Commission did not want to see
uncertainty described quantitatively because most RMs are only able to understand a
qualitative or at best semi-quantitative discussion of ERA uncertainty.
P. 9: A subsection on Uncertainty would make this section read clearer.
P. 10, line 1: The use of a model not suitable for the scenario because other models
do not exist would also lead to uncertainty.
242
-------
Kevin H. Reiriert
11/15/95
-, line 7: Insert 'a' before quantity.
P. 11, line 31:1 am not convinced that measurement endpoint needs to be redefined as
measure of effect. I would prefer to keep the terminology consistent with Suter.
P. 12, line 6-18: This section further supports the need for continuing discussions
between the risk assessor and risk manager.
P. 13, line 22: Delphic is used several times throughout the document Perhaps a
better term such as obscure or ambiguous should be used to define these approaches.
.line 24: Add'al'to ecologic.
'—, line 27: This section should note that not all impacts cause a measurable effect.
P. 14, line 9: Hazard assessment only addresses the effects side of the ERA process
and I agree that the term hazard assessment should not be used in the guidelines.
However, as written, the text implies that it may be similar to ERA.
, line 25: What about enhanced responses? Enhancement of various properties
may not be detrimental or adverse directly, but could be negative with respect to
secondary effects (e.g., biological introductions, algal growth, etc.).
P. 15, line 28: A disturbance is still a stressor and I am not sure I agree that it should be
separately defined.
2. Planning
P. 17, line 10: The risk assessor and risk manager should discuss various aspects of
the ERA throughout the process, not just during problem formulation. This is essential
as a progress check, reality check and discussion concerning changes in outside
influences on the ERA needs and management goals.
P. 18, line 8: Add'the presence of before stressors.
—-, line 16: Is tolerate the correct term? Accept seems more appropriate.
P. 19, line 16: Both the manager and assessor also need to agree upon treatment of
uncertainty and additional data collection needs.
P. 20, text box: Goals and resource should be defined in the 3rd bullet for RM.
The existence of previous risk assessments should be specifically listed in bullet 5 for
RM.
243
-------
Kevin H. Reiriert
11/15/95
P. 21, line 18: From the goals, the risk assessor, in communication with the risk
manager, derives a series of pertinent, applicable assessment endpoints.
—, line 25: The risk assessor needs to know when the RA will be used to establish
national or other policy. This is part of the discussion concerning how the RA will be
used by the risk manager.
3. Problem Formulation
P. 24, line 9: Please add '4) poor or lack of uncertainty characterization'.
P. 25: The current diagram implies that one could construct the conceptual model
without addressing the assessment endpoint circle. Also, measurement endpoints
should be listed as a separate step prior to analysis. Additions to the box to the left of
the Problem definition phase are discussed in section 1 of these comments. Some of
the main points have been added.
P. 26, line 18: Replace product development with problem definition.
P. 28, line 4 and 26: Text boxes corresponding to physical and biological stressors
would be helpful. Physical and biological stressors are more difficult to understand and
less understood than chemical stressors.
—, line 16: When biomarkers are used, their ecological relevance and significance
should be understood.
—, line 28: Is resource-initiated more appropriate? I suppose there could be a
difference between ecological value and resource value.
P. 29, line 10: Replace value with 'resource"?
P. 31, bullet 5 in text box: Is land subsidence the correct term? It seems that
flooding and its periodicity are the primary factors.
P. 33, iast bullet in text box: Chemical fate may include transport; physical transport is
redundant.
P. 37, line 20: Reference?
—, line 27: If the assessment endpoints in an ERA are not ecologically relevant, the
assessment may predict risk, but not to ecologically significant components.
P. 38, line 20: Delete along with the and during and add 'occurring during a particular1.
—, line 21: Please refer the reader to text box 3-8.
244
-------
Kevin H. Reiriert
11/15/95
P. 39, line 8: Reference? ;
—-, line 12: Change is to'maybe'.
—.line 21: Delete through and measures and add 'ly1 to indirect.
P. 42, line 12: Change affect to 'effect1.
—, text box: The cited items should be referenced to the text in a clearer manner.
—, lines 23-28: Whose citation, Suter or Bamthouse?
P. 43, lines 1-6: Delete, repeated from p. 42, lines 19-23.
-—, line 25: Should the risk manager be involved with the assessor rather than have
the assessor develop assessment endpoints which may not be agreeable to the
manager?"
Should there be a separate section on measurement endpoints rather than have them
buried in the assessment endpoint section? See p. 53.
—, text box 3-11: Where is this box cited in the text?
P. 44, line 4: Change verbal to 'written.'
—, line 7: Doesn't the conceptual model also contain a description of the various
exposure scenarios and pathways?
P. 45, line 31: Add, 'but potentially important* before risk.
P. 47, line 10: Change 'to' to 'causing.'
P. 52, line 14: These flow diagrams should be presented/discussed with the risk
manager, addressing clarity, completeness and any changes before proceeding to the
analysis phase.
P. 53, line 7: The initial figure in this section needs to describe the selection of
measurement endpoints. I am not convinced that the term measures of effects is
necessary. Also, some thought concerning the measurement of the assessment
endpoints should occur while selecting assessment endpoints. A lot of time will be
wasted if assessment endpoints that are not clearly defined or cannot be measured are
chosen.
—, line 24: Change important to 'importance'.
245
-------
Kevin H. Reiriert
11/15/95
P. 54, line 26: This line is repeated below at line 29 and should be integrated or
deleted.
P. 55: Reference for the table is needed.
4. The Analysis Phase
P. 57: AIHC suggested some revision to the Data box on the right of this figure. These
are listed in the attached figure.
P. 58, line 14: Model calibration should be defined.
P. 59, line 3: This statement may be misleading; Is a stressor at natural background in
an ecosystem really a stressor? There may be some confusion regarding this concept.
P. 60: Is this example appropriate for demonstrating the relationship between primary
and secondary stressors? It seems that the primary stressor causes the smothering of
the benthic organisms (not just insects) from increased siltation due to logging and
concomitant erosion. I am not sure that a difference exists.
P. 62, line 7: Does high end mean upper 95% confidence limits (CL) or otherwise?
High end is not a useful description.
P. 63: Periods are missing from this page.
—, line 25: Change communication to communicate.
—, line 27: Add 'uncertainty' between quantitative analysis.
P. 64, line 5: Often risk managers and others are unable to understand complex
uncertainty presentations. The level of uncertainty analysis should be agreed upon
prior to the conduct of the ERA. See comments earlier on Congressional Commission.
—. line 6: Add to the end of the line -'in some manner appropriate to the ERA
necessary for decisionmaking and agreed to by the risk manager.'
P. 66, line 23: Should adventitious be changes to inadvertent?
—, line 28: Add 'of the primary stressor1 to the end of the sentence.
—: Where are text boxes 4-2,4-3 and 4-4 cited in the text?
P. 67, line 5: Why is background in italics?
—, lines 20-24: Are there any references that can be cited for this section?
246
-------
Kevin ri. Reiriert
11/15/95
—-, line 25: Should the paragraphs following this line be a separate subsection (e.g.,
chemical distribution)?
—, line 30: Add effluents to the complex mixture examples. Also, I would argue that
PCBs and dioxins do not have significant differing properties within their cluster; I do
agree that effluents would be different.
P. 69, figure 4-4: Does this figure add anything to the text discussion? It seems
obvious from the text.
—, line 15: No need for a colon or an ellipse on the next line.
P. 70, line 24: Some discussion or mention of the need to determine ecological
significance of biomarkers is needed.
-—, footnote: This should be restated from an ecological perspective and not human
as currently written.
P. 71, line 4: Again, need to discuss ecological significance of the assay.
—, line 9: No need for a new paragraph.
—, line 18: Such competition also creates additional stress for organisms.
—, line 19-24: Relates to a secondary exposure; could increase competition for
'clean areas.'
P. 72, text box 4-5: Where is this box cited?
P. 73, line 11: Can an example be cited?
—, line 24: Some additional verbiage should be inserted here concerning the
pathways.
P. 74: Inconsistent use of bullets in document.
—, line 13: Number crunching is not a technically precise term; substitute toxicity
inventory? Also, a separate section should be listed for Ecological Response Analysis.
P. 76, text box: The new Rand publication should be reviewed and cited if
possible.
—, line 23: What does the phrase 'test protocols may not always be followed'
mean? I find that this is not the norm.
247
-------
Kevin H. Reihert
11/15/95
—.line 30: LC values are more often used in ERA than LDs.
P. 78, line 16: F0 hatchability is also included in these studies as well as the growth of
theFiS.
—, line 25: Is lethality the correct term? The LC5, NOEC, etc. are used to assess
the risks of pesticides. The lethality is determined by the LC50.
P. 79, line 8-15: The NOEC and LOEC from hypothesis testing are artifacts of the
doses chosen and should be noted in this section.
—, line 19: Axenic culture does not necessarily imply single species, just bacteria
free. ,
P. 80, line 6: This line is not a sentence.
—, line 19: Confronted is a poor word in this situation; is 'presented' a better term?
—, line 28: Change paramecia to protozoa to be less specific.
P. 82, line 10: Change group to population.
P. 83, line 12: Reference?
—, line 22-23: Mitigating measures in the field are not only duration, but
bioavailability.
P. 84, line 16: Add 'to' before large.
—, line 20: Please cite the SETAC European Union (EU) document concerning the
design and use of microcosms and mesocosms
P 86, line 1: Cite reference for Koch.
—.line 11: Cite reference for Hill.
P. 89, Table 4-3 title: Change disturbances to stressors. I do not agree with the text
discussion on p. 88; it may confuse the reader.
P. 90, line 11: Are physical stressors different from other stressors in that the
characterization is described in 2 components here? Is this line necessary?
P. 95, line 7: Remove and and insert a comma.
P. 97, line 12: Use another term besides delphic.
248.
-------
Kevin H. Reinert
11/15/95
P, 101, text box 4-13: Some discussion or listing of whether the organism caused
problems in these countries should be added.
P. 105: The work being conducted by Jeff Foran (Risk Sciences Institute) for the EPA
concerning multiple stressors should be consulted.
P. 107, line' 9: Assuming additivity is certainly better than singularly assessing the
various stressors (without some sort of combination) that may be included in an ERA.
5. Risk Characterization
P. 112: More detail is needed in the risk estimation box. Also, see'the attached
additions to the box on the right of the figure.
P. 114, line 21-22: Either number or bullet these 2 items for clarity.
—, line 24: For ease in reading, the additional information and methods should be
listed here.
P. 115, line 13: Managers should not only be familiar with the method, but also
understand its derivation and limitations.
—, line 22: A 100 percent change of adverse effects occurring is not correct when
the quotient is exceeded; there is a high probability of an adverse effect occurring
based on the tier of ERA conducted. With further refinement (e.g., higher tier ERA,
additional input data) the risks may be lower.
P. 116, line 1-6: The risk assessor may be able to qualitatively approach this issue
with the ERA and the manager.
P. 117, line 26: Should 'effects' replace defects?
P. 119, line 29: Need to discuss why these were specifically excluded from the
ERA.
P. 123, line 10: The next items are discussed as a subset to ecological
significance. A lead in sentence is needed here listing the next four items.
P. 125, line 1: A reference is needed here.
P. 126, line 28-29: Perturbations in ecosystems may be extremely difficult to observe
(detect) among the various natural variations. This concept should be specifically
stated.
249
-------
Kevin H. Reiriert
11/15/95
6. Relating Ecological Information
P. 127: This section should be expanded to include more of the risk
management issues if a separate addendum is not prepared. There is a substantial
void in how the ERA information is used in decisionmaking and numerous
inconsistencies among the various programs within the Agency. There needs to be
feedback to the risk assessor. Is rework needed (i.e., another iteration; higher level tier
ERA)? Are additional data collection tiers needed? Does the ERA meet the needs of
the risk manager?
—, line 20: The items listed are not procedures but are check list items.
P. 128, line 4: Add 'and understandable' before manner
Appendix A
The specific EPA case study documents should be cited along with the specific authors
of the case studies.
Appendix B
Will a single definition suit all programs? Are the authors expecting each of the
program offices to rework and redefine these terms specific to their programs?
250
-------
W
» w t!
CO
o
O
°-
n&o o
^ **•• O
>» o b!
CQ CQ Q.
ca
•
O T3
on c 5?
••" -Z 1>
Q 3
"^ on 35
taO is c«
C ° •—'
•— OT >.
g S ^
S ^ =S
2
S Si
.|.2
' a -s .a
'| 1 ||_s
5<8:g-|
£ -§
g-e |S
1.2
o
ft-
oo
251
-------
^^^
f*J
o
^p? t _>^^
U) m C/5
3UJ K*£
O^*>
O^'frT
ffj ^M /Ik
v' C QJ
^^ C^j f"
•" • E~" l_
rlj £•••
^" ••&
flft ^X W^y
•"* **^~ f"
_^_ C/J ^^
^~ ^ •••• *^^^ ^^WB
•= Qi o &
o -u ^. ^
^^ ^^^ I^^^B ^1 •
•C C. C / \
co< § §
w to o >
fl) ^ ^ D-
iat Policy Issu
••^ • • M
Risk Assesso
ring And After
aracterization
i- >> = J=
S CQ O O
^^* mlmm •••• ^^
J
*»
X
-i.
&o c
o a
3 S
en S
H-. 0
>% c/
0 c/
~ <
scussion On Assessment And 1
irino Or Af- C*r\mn\f*1:tr\n Of Rio
Q C
c
v
V
a
.JiiirHMH ma il**Mtiuit J -^,,
• t f t
: v v 1
C | o O | f]
S S "" ^^ 18 s J c
e S ? ' ^*^ * ^ ^ — ii
O S ° " 2 ' : ** = f S ** J
: I Hi w 5 j|" 1
8 •§ """I = l"^'^ 1
£ 1 ' *
. ..,— ^ N
A , , , , , ' -v
s£i ss ^'' ^
"• c • 3 - •* N
§ i * . * E • \
0 £ I \
-» \
x " \
' \
\
\
\
o- o 2. o
a> . ^~ c
JT? t-< ' i}
N aj •-— ^
""••"• O
*i j^ co ^ C
g • S 5- o "^
S ^ c ^ .2
js 2 "-1 ••> o
O 5 c« J ^
>% ^ 3 — °
1 1 i-i 1 ^
| | §1 1 J'-B
-S -a O '5 • = ,5s °
O ^» ^^ *•< ^ ^^s^ •"*
e <; o js 2 ^22
i 1 f2^ i ^i
•a 1 -s^ ic~. io
a 1 ^|s Is, 11
a jl ijl ^ 1 ,11:
S H I-I -s ^s
s § sil s-g i^
.
252
-------
H-5-g
0
* *
C
O
< a: o <
2
CD
-o
C
re
o
* .5 w
"-1 c3 o
3J •*-> o,
HOW
^2=
0) C3
* Q
.
o
,u
W2
00
- ^ SP
"5 0 -E
§ « 3
'
5 £L
^J^ §>
< H < £:
I_
o
D)
-. «
« c
JZ (R
= s
O ^
O 0)
|E~
5--o-S
ffl =3
c « w
05? «
•— Q fy»
52 «j i=-
S2 w
= o>
° Crt
,2 S>
Q<
^c
en
E
anagement
Discussion
Between The
Risk Assessor
and
Risk Manager
(Planning)
-------
-------
William Smith
255
-------
Guidelines for Ecological Risk Assessment
Comments forEcoguide Workshop
William H. Smith
Yale University
November, 1995
I. Introduction and General
An appropriate definition of ecological risk assessment is "the qualitative or
quantitative estimation of the probability that some factor will cause an adverse effect .
on some ecological resource". In either an informal or formal sense human beings have
been making these estimations for a very long time. In some applications, for example,
agricultural and forest pest management, the literature and experience is quite extensive
(e.g..Baker etal. 1993, Byleretal. 1990, Gansner etal. 1994, and Maclean 1980).
Even in the absence of "formal" risk assessment guidelines, ecological risk assessments
will continue to be made. Ecological risk assessments are essential for effective natural
resource management. Ecological risk assessments are extremely useful for prioritizing
environmental regulatory, monitoring, research and education efforts; establishment of
environmental standards; and in conducting environmental cost-benefit assessments. To
quote Warner North (1990) "estimating risk is a process for summarizing science to
support decision making".
There are advantages to a systematic approach to ecological risk assessments.
These include consistency in assessments over time, comparability of assessments
conducted for various purposes, and completeness of assessments.
Ecological risk assessments should be explicit regarding four dimensions. These
dimensions include time, space, biological hierarchy, and societal values. Due to
variability in time, i.e. dynamic successional and natural disturbance processes
characteristic of all ecosystems, temporal bounding is essential for effective ecological
risk assessment. This is particularly true for long-lived ecosystems subject to multiple,
interactive stressors such as forests. Ecosystem management can pursue the
sustainability of dynamic ecosystems. Ecosystem management should not pursue the
256
-------
W.H. Smith
sustained maintenance of static ecosystems for extended time. Ecological risk
assessments should be place-based and the spatial boundaries of the ecological resource
must be clearly described. Surface waters, estuarine systems, dry lands, wet lands, and
forests exhibit enormous variability. Forests, for example, differ in soil type, climate,
aspect, elevation, species composition and age. Forests may be uneven aged, even aged,
all aged, or over mature. Forests may be reproduced by seed, by coppice, or by planting.
Some forests have their structure completely shaped by natural forces, some may be
influenced by human forces as well as natural forces, while other forests may be
completely artificial in design and establishment. Forest trees may be arrayed along a
continuum of human management efforts ranging from no management to intensive
management. All of this variability necessitates the need to spatially bound ecological
resources prior to risk assessments in order to focus the evaluation on specific type/range
of ecological resource variation. The endpoints of ecological risk assessments should be
clearly "nested" in the biological hierarchy from cell to biome. While species,
population, community, ecosystem and landscape scales are all conceptually convenient
"stopping points" within the hierarchy, the scientific understanding of stress response at
these various scales is very different. In general, scientific understanding declines rapidly
from species to landscapes. As a result, the most informed risk assessments will be
conducted at the species or population levels. All ecological resources are influenced,
directly or indirectly, by human beings. The human values associated with ecological
resources are extremely varied. In the case of forests, for example, values represent both
products and services provided by the forest (Table 1). Some ecological resource values
are traditional, long and widely appreciated, and quantifiable in standard economic
terms. Other values are of more recent appreciation, developing in acceptance and
sophistication, and are not quantifiable in standard economic terms. In any case,
endpoints selected for use in ecological risk assessments should be clearly linked to one
or more societal values.
The linkage between ecological risk assessments and stressor and ecological
resource monitoring is intimate. The full value of risk assessment will not be realized
unless and until we develop and maintain a scientifically sound and cost-effective
257
-------
W.H. Smith
environmental monitoring strategy.
It is less important to have an ecological risk assessment framework that fits all
stressor and ecological resources types, than to have a framework that works well for
some stressors and some ecological resources.
II. Major Elements of Ecological Risk Assessment
As proposed, the Ecological Risk Assessment Guidelines call for three major
elements for all risk assessments; problem formulation, analysis, and risk
characterization. Organizationally these represent reasonable elements for all
assessments.
A. problem formulation
The most important aspect of this element is the dear articulation of the answer
to the question, Why is this assessment necessary? The question is answered
interactively by the relevant parties which include scientists ("risk assessors" per
Guidelines) and environmental managers ("risk managers" per Guidelines).
The focus of the assessment is some ecological resource. It is imperative that the
ecological resource be clearly defined by bounding the spatial, temporal, biological
hierarchy and human value dimensions of the resource.
The stressor, or agent capable of imposing an adverse effect on the ecological
resource, must also be dearly defined. Sources should be identified if possible. Relevant
monitoring efforts that can reveal status and/or trends of the stressor must be identified.
Information regarding the interaction of the stressor with other agents capable of
causing adverse effects on the ecological resource must be reviewed.
The societal values assodated with the ecological resource ("endpoints" per
Guidelines) must be dearly defined. Quantitative measures ("indicators" per Guidelines)
of the status of these societal values must be established. Ambiguity of these measures
must be evaluated by reviewing other factors (stressors) that may influence the
indicators selected and by reviewing the uncertainty inherent in the information
available.
258
-------
W.H. Smith
Problem formulation must recognize that risk assessment generally contributes
to environmental and natural resource management decision-making and will be only
rarely the sole "driver" of environmental decisions. Ecological risk assessments have
great utility, as indicated in the Introduction for a very broad array of
environmental/natural resource managers and decision-makers. Only a fraction of these
users will be regulators.
It is very important to minimize the development and use of unnecessary jargon
in the development of risk assessment guidelines. Where possible, vocabulary should be
presented using common and conventional terms rather than inventing and defining
new ones. The Agency is to be complimented for keeping the "glossary of key terms"
(Appendix B) to 38 entries. Even in this case, however, selected terms such as
"community" and "ecosystem" could be deleted due to common appreciation; while the
"relative risk assessment" and "comparative risk assessment" terms could also be deleted
as they represent unnecessary linguistic "fine-tuning".
B. Analysis
This is the "science" section of the assessment. It is the section that provides all
relevant detail regarding the nature of the stressor ("exposure" per Guidelines) and
response of the ecological resource ("indicator(s)" per Guidelines) along with a
discussion of relevant uncertainty and incomplete information. While uncertainty can
be introduced in the problem formulation stage, it should be addressed in detail in the
analysis section.
With chemical stressors in particular, it is extremely important for the analysis
phase to emphasize the distinction between stressor presence in the environment and
stressor bioavailability in the environment. Equally important is the distinction between
the amount of stressors present, as revealed by environmental monitoring, and the
amount of stressors present at a potential site of biological action.
The development of exposure detail (e.g. intensity, time, and space dimensions)
and ecological effects detail (e.g. structure-activity relationships, single-species assays,
multispecies assays, and field experiments) should be presented as useful
259
-------
W.H. Smith
options/elements, but not as exclusive alternatives or as absolute requirements.
It is important to agree on the primary purpose of the risk assessment
framework. If the primary purpose is to inform environmental/natural resource
management decisions, rather than test scientific hypotheses, then the issue of causality
will be less central. Presumably, most risk assessments will be carried out to evaluate the
probability that an established cause-effect relationship will or will not occur in a specific
system. This will eliminate the need to discuss the application of "Koch's Postulates" to
pollutants rather than to microbes!
C. risk characterization
This is the most important part of the risk assessment framework as it provides
the product that will actually inform the decision-making process. As a result, ideal
charactristics of risk characterization include:
1. qualification
-regarding temporal, spatial, biological hierarchy, and social-value
boundaries
-regarding uncertainty and incomplete information
2. clarity
-minimal use of technical jargon
-danger of extrapolation to situations with other boundaries
-clear conclusions
3. conciseness
-usefulness and impact will be facilitated by brevity!
4. linkage
-risk assessment should be referenced in the context of similar
assessments done with the same or related stressors or ecological
resources
-risk assessment should be stated in the context that will most usefully
i
inform a management decision
260
-------
W.H. Smith
III. Specific Questions
1. Q.. Considering both the present state of the science and present and future
Agency needs, how well are the guidelines balanced regarding the range
of stressors, levels of biological organization, ecosystem types, and
spatial/temporal scales? Specifically, what would you emphasize or de-
emphasize?
A. The most effective risk assessments will be those conducted for ecological
resources that are explicitly defined regarding time, space, and biological
hierarchy and for which societal values are clearly articulated. Further, it
is more important tp develop risk assessment guidelines that will work
well for some stressors, than it is to develp a single framework that will
work for all stressors. A workable framework for chemical stressors, even
if it failed to work for biological/physical stressors would be a
contribution!
2. Q.. What would you suggest to improve the use of case illustrations in the
guidelines? How useful is Appendix A in illustrating a range of
applications of the risk assessment process?
A. Rather than retrofit previous case studies to the propsed framework, it
would be more useful to apply the framework to new (and preferably
actual) environmental decision-making needs. Ideally these case studies
would cover a range of ecological resources and a range of stressors.
3. Q. Some Agency reveiwers of the guidelines have suggested that more
examples of terrestrial assessments and field approaches (e.g.,
bioassessment techniques) should be used. Specifically, what, if anything,
should be added?
261
-------
W.H. Smith
A. Clearly the framework is intended to serve both "terrestrial" and
"aquatic" decision-making. As a result, both terrestrial and aquatic cases
are appropriate.
4. Q. Areas of uncertainty are summarized in the problem formulation (section
1.5), analysis (section 3.7), and risk characterization (section 5.2.4).
How useful is this approach in providing guidance on uncertainty issues?
A. The emphasis on "uncertainty" should be focused in the Analysis section.
This section presents the science and should detail the uncertainty. Also,
gaps in our knowledge should be highlighted in this section. The risk
characterization section must effectively convey the uncertainty and gaps
to the decision maker.
5. Q. How could the Introduction be modified to more clearly communicate
the scope and content of the guideline?
A. Some reduction in the length of the introductory section is desirable.
Sub-sections that could be reduced include; background (1.1), scope and
audience (1.2), and framework report (1.5).
6. Q. Terminology, especially related to endpoints and exposure, has always
been controversial. What changes, if any, would you recommend in
guidelines terminology?
A. Endpoint and exposure concepts are important and reasonabley defined.
In general, however, new terminology (jargon) and unnecessary
refinement (e.g. "comparative risk vs. "relative risk" should be minimized
wherever possible.
262
-------
W.H. Smith
7. Q. The overall framework figure for the ecological risk process has been
retained, although some changes have been made to the diagrams for
problem formulation, analysis, and risk characterization. What further
modifications, if any, are required?
A. The diagrams as presented are reasonable and useful.
8. Q. Please comment on the description of the risk manager's role at the
initiation of an ecological risk assessment and on the principles for
selecting management goals.
A. I prefer to susbstitute environmentaVnatural resource manager for "risk
manager" as I feel the .users of risk assessments will be a variety of
environmental and natural, resource professionals, with only a minority
being "risk managers" ie. environmental regulators. Because clarity
regarding "social values" at risk are central to problem formulation,
environmental mangers must be involved in the iniation of an ecological
risk assessment. Presumably, they know what their goals of professional
management are!
9. Q. What additional points, if any, should be covered about relating
ecological information to risk management decisions after the completion
of an ecological risk assessment?
A. No further perspective needed. I feel the "product" or end result of the
risk assessment process provides a qualitative/quantitative estimation of
the probability that some stressors will adversely impact some ecological
resource.
10. Q. How useful is the categorization of assessments as either stressor- or
263
-------
W.H. Smith
source-initiated, effects-initiated and ecological value-initiated?
A. Human values, associated with specific ecological resources, must be
clearly defined in order to initiate an ecological risk assessment, therefore
they are fundamentally important. The distinction between stressor-,
source-, or effects-initiated assessments is interesting but NOT of
fundamental importance.
11. Q. Please comment on the discussion of assessment endpoints and their
relationship to management goals and "measures".
A It is critically important that assessment endpoints be linked to
management goals which must be directly linked to societal values.
12. Q. What, if anything, would you add to the discussion of risk hypotheses
and conceptual models to give the reader a clearer understanding of their
nature and content?
A. In initially defining a risk assessment, the clarity of the question being
asked is critical. Most risk assessments will serve environmental/natural
resource decision-making NOT advancements of science. Hypotheses
testing is not a necessary ingredient of all assessments. Conceptual
models are very effective tools through which science can be summarized
for decision makers.
13. Q. How would you change this section, if at all, to improve the balance in
the discussion of the different stressor types?
A. I would not change it. I would emphasize, however, that it is not
necessary to develop a single framework that can be applied to all
264
-------
W.H. Smith
stressor/ecological resource situations. A "single-strategy" or "one-size-
fits-all" objective is NOT necessary. If EPA were to develop a framework
that would cover most chemical stressors this would be a useful
contribution!
14. O_. Are there additional points or principles that should be emphasized for
either chemical, physical, or biological stressors?
A. No, but reveiw answer to 13.
15. Q. What additional principles would you suggest, if any, for the analysis of
multiple stressors?
A As we move "up" the biological hierarchy (e.g. from species to
ecosystems) and as we lengthen the time boundary (e.g. from growing
season to decade) we must become more concerned with multiple
stressors and stressor interactions. The current "state-of-science" strongly
favors risk assessments focused on single species for relatively short time
periods.
16. Q, What additional principles should be highlighted in the discussion of risk
estimation techniques?
A. Any approach that links both stressor exposure and ecological resource
effects, whether qualitative or quantitative, can be considered "risk
estimation". Expert opinion, empirical approaches, quotient methods,
exposure/effects distributions, field surveys, and simulation appear to
cover the primary options.
17. Q. The guidelines propose four criteria for ecological significance. How
265
-------
W.H. Smith
should this list of tirteria be modified, if at all? What additional
guidance, if any, might be added to the discussion of these criteria?
A This is an important consideration! Numerous statutes include the term
or "concept" of ecological significance and very few include specific
reference to statistical significance. I feel ecological significance, absent
any reference to the value of specific ecological resources to human
beings, is of limited usefulness/importance. It is more reasonable to
identify specific human values (products/services) that flow from
ecological resources, and then evaluate adverse effects in the context of
these values. If this is done, then intensity, scale, recovery rate, and
sustainability become "useful" descriptors of the significance of specific
stressor interactions with specific values.
266
-------
W.H. Smith
Literature Cited
Baker, FA., D.L. Verbyla, C.S. Hodges, E.W. Ross. 1993. Classification and regression
tree analysis for assessing hazard of pine mortality caused by Heterobasidion
annosum. Plant Disease 77:136-139.
Byler, J.W., MA. Marsden, S.IC Hogle. 1990. The probability of root disease on the
Lolo National Forest, Montana. Canadian Journal of Forest Research 20:987-
994.
Gansner, DA, J.W. Quinby, S.L. King, S.L. Arner and DA. Drake. 1994. Tracking
changes in the susceptibility of forest land infested with gypsy moth. USDA
Forest Service, Res. Paper No. NE-690, Northeastern Forest Experiment Station,
Radnor, PA, 4 pp.
Maclean, DA. 1980. Vulnerability of fir-spruce stands during uncontrolled spruce
budworm outbreaks: A review and discussion. The Forestry Chronicle 56:213-
221.
Lorio, P.L. Jr. 1994. Southern pine beetle risk ratings. In J.B. Baker, ed., Proceedings of
the Symposium on Ecosystem Management Research in the Ouachita
Mountains: Pretreatment Conditions and Preliminary Findings. USDA, Forest
Service, Genl. Tech. Report SO-112, New Orleans, LA, pp. 162-171.
North, D. Warner. 1990. Personal communication. Decision Focus, Inc. Los Alto, CA.
Smith, W.H. 1993. Forests. Chapter 9. In, Preparing for an Uncertain Climate. Vol. II.
Publica. No. OTA-0-668, U.S. Congress, Office of Technology Assessment, U.S.
Govt. Printing Office, Washington, D.C., pp. 299-355.
267
-------
Table 1. VALUES ASSOCIATED WITH FOREST SYSTEMS:
Products
Services
Wood
Lumber
Partide Board/Plywood
Paper
Fuel wood
Mulch
Wildlife (game)
Wildlife (non-game)
Forage (wildlife, livestock)
Other
Seeds
Edible nuts
Syrup (sugar maple)
Drugs (e.g. Taxol)
Pesticides (e.g. Neem)
Chemicals (e.g. turpentine)
Christmas trees
Mistletoe
Edible mushrooms
Existence Value
Recreation
Tourism
Biological Diversity
Genes
Species
Communities
Ecosystems
Landscape Diversity
Amenity Function
Microclimate
Amelioration
Sound Attenuation
Aromatic
Hydrocarbons
Visual Attractiveness,
screening
Runoff/Erosion Management.
Soil/Nutrient Conservation
Pollutant Sequestration/
Detoxification
Smith (1995).
268
-------
Ralph Stahl, Jr.
-269-
-------
Ralph G. Stahl, Jr.
COMMENTS ON DRAFT ECOLOGICAL RISK ASSESSMENT
GUIDELINES
RALPH G. STAHL, JR
PART I. GENERAL PRINCIPLES
The ecological risk assessment guidelines can be improved to more
clearly indicate underlying assumptions and default positions in several
ways. The first is to reinforce the requirement for transparency
throughout the guidelines - that is, that the risk assessor be required t o
clearly document all underlying assumptions and default positions in
the assessment. Secondly, I believe that the guidelines could suggest that
the risk assessor have a separate appendix in the risk assessment that
includes example calculations, where default values and underlying
assumptions are detailed would also be a way to improve the clarity on
this point. Hand in hand with this documentation would be an
explanation or justification for using these defaults and assumptions so
that the reader is not left with the impression that the risk assessor uses
these without some basis or rationale. In the situation where these
defaults and assumptions are actually policy driven, not scientifically
driven, the text should so state.
I have made suggestions on how to change specific areas of the
document in my comments that follow.
PART IT. SPECIFIC QUESTIONS
A. Guidelines Balance
1. I found the emphasis on higher levels of biological organization to
be appropriate.
2. I found the case examples to be adequate for the purposes of this
document. They should not be overly detailed but illustrative of the
-270-
-------
Ralph G. Stahl, Jr.
various situations in which ecological risk assessments were conducted.
Improvements in the case examples would be to include two other
examples. One example would illustrate .how human health risk
assessment and ecological risk assessment are considered in parallel (or
tandem) and how the risk management options were developed. It
would be important for both ecological and human health risk assessors
to see their efforts interlinked rather than as separate. Another example
that should be included would be one where only a screening level
assessment was used. This would illustrate how the scope of an
assessment must be tailored to the particular issue.
3. It is true that there have been fewer ecological risk assessment on
terrestrial systems than aquatic., For that reason I would agree that a n
additional example of a terrestrially-driven assessment would be
beneficial. It may be appropriate to develop a case where soil is the
medium of concern since so little data have been developed in this area.
4. I found the approach of discussing uncertainty in the various
phases to be beneficial as I was able to understand uncertainty as it
pertained to the specific areas.
B. Introduction and Scope
5. I my opinion the scope and content of the guideline is adequately
addressed in the Introduction section.
6. There are no specific changes to terminology that I would suggest.
I believe that the guidelines provided the needed clarification of
assessment and measurement endpoints through additional discussion
and slight changes in terminology. This was very beneficial and
addressed two terms which I believe were greatly misunderstood.
7. At the EPA's May 3, 1995 Colloqium, modifications to the
framework figure were suggested by Rich Kimerle representing industrial
-271-
-------
Ralph G. Stahl, Jr.
viewpoints. I would encourage EPA to review Rich's written comments
and suggested changes to the framework figure.
C. Risk Manager Interactions
8. The added details and emphasis on the risk assessor / risk
manager interface were long overdue. I think that this has strengthened
the guidelines considerably and applaud the authors for doing this. I
believe however that there should be greater emphasis placed on what
the risk manager's role is when interacting with the public. For example,
the public should have input to the assessment endpoints or ecological
valuation process. The gathering of public input and the
communication about the need for and results of the risk assessment are
clearly the job of the risk manager. I believe this should be noted in the
guidelines.
9. Additional points that would be useful in the guidelines would
include those related to risk communication. Although these guidelines
are not for risk communication, they would benefit from some brief
discussion of potential communication options, problem areas, etc. that
should be known early in the risk assessment. Public discourse on the
risk assessment should be regular and open so that when the risk
management options are discussed with the public they have been part
of the process from the beginning. Ultimately the results of the risk
assessment must be communicated to the impacted stakeholders
directly, perfereablly in a face-to-face .format.
D. Problem Formulation
10. I'm not sure the categorization of the assessments into three
groups helped my understanding, but may prove useful to others who
view the assessements as fitting into one category or another. I tend to
view the assessments in three categories: screening, prospective, and
retrospective. , :,
-272-
-------
Ralph G. Stahl, Jr.
11. As I noted earlier, I believe the discussion of assessment endpoints
and their relationship to management goals and "measures" was long
overdue. I believe the discussion throughout the guidelines have
clarified some of the more confusing aspects of ecological risk assessment
/ risk management.
12. I think the discussion of risk hypothesis is adequate and may not
need additional work. However, the discussion of the conceptual model
would benefit from either an example or greater detailing perhaps in a n
appendix. Some still believe that the conceptual model is just that - a
"model" in the sense of a software program that one puts numbers into
and gets results out of. Using terms such as a summary of current
knowledge about the issue, or brief overview, or background, may dispel
some of the confusion that stems from using the word "model". Maybe
just saying "conceptual diagram" would help so that the risk assessor
knows that the purpose is simply to illustrate the situation at hand.
E. Analysis •
13. In later text, I have provided comments on specific paragraphs i n
this section. Briefly, I think the sections on non-chemical stressors could
be expanded and upgraded. I still have the impression that non-
chemical stressors are not considered very important by EPA, yet loss of
habitat (physical stressor) and exotic species' expansion into native
habitats (biological stressor) are probably responsible for more ecological
risk than the great majority of chemical stressors. Quite frankly, the
discussion of non-chemical stressors still "feels" like an also-ran, or add-
on, put there to placate critics rather than to provide a serious
discussion.
14. I believe that the section on biological stressors is still incomplete.
Perhaps it is my bias but I can't understand why there is no discussion or
mention of predation, competition, etc. as biological stressors. I
-273-
-------
Ralph G. Stahl, Jr.
recognize that we cannot regulate these to a large degree, but we should
at least acknowledge their importance as biological stressors that all
plants and animals face each day. I think that this needs to be stated in
the guidelines so that the risk "picture" is not missing some of its most
important components.
15. Multiple stressors is a difficult issue, and one that will not be
resolved in these guidelines. I have no additional principles to add.
F. Risk Characterization
16. There are several additional principles that could be useful. One
is that the risk assessor should clearly document (or reiterate)
underlying assumptions and default values that were used in the
estimation. Another is that the estimates be communicated in simple
terms, especially those which are understood by the public.
17. I believe there are two additional .points that should be included
in the discussion of ecological significance. The first is the time-scale.
Risks which may manifest themselves, over decades or longer are much
more significant than those which. are short-lived. The Concept of time
is not given much if any discussion in this section. Secondly, I believe
that risks to the sustainability of a population, community or ecosystem
are more significant that those which not threaten the sustainability.
PART III. GENERAL COMMENTS
• There needs to be some upfront discussion about the level of effort
involved in the risk assessment. For example, the guidelines should
note that risk assessments come in all shapes and sizes and will be
dependent on many factors. I find little discussion about this in the
current document and would prefer to see some discussion of when a
screening assessment would be appropriate compared to a more
extensive risk assessment.
-274-
-------
Ralph G. Stahl, Jr.
I am uncertain whether the concept of tiering the risk assessment is
clear to the reader. After reading the guidelines I certainly did not
feel as if the concept were clearly discussed, nor given an appropriate
level of visibility. I would encourage EPA to give the concept more
priority and visibility in the next draft.
I believe the guidelines should briefly note the linkage with h u m a n
health. This will be most important for retrospective risk assessments
where site clean ups are driven by human and ecological risks. For
example,- the guidelines could note that remedies to protect human
health can also provide some level of protection to ecological
receptors. In addition, the guidelines could indicate that there will
be situations where remedies to protect humans will not be protective
of ecological receptors.
I suggest that EPA consider adding a figure similar to the one shown
below to illustrate the various stressors present in the ecosystem. The
concept is that there is a .total stress in ihe system that can be broken
into its various components: biological, chemical (nutrients) physical
and contaminants. It is also important to help the readers
understand that the relative contributions of the various stressors is
dynamic and not evenly distributed. .
-275-
-------
Ralph G. Stahl, Jr.
CONTAMINANTS
SITUATION 1.
CONTAMINANTS
BIOLOGICAL SITUATION 2.
FIGURE < >. THE MAIN GROUPS OF
STRESSORS IN THE ENVIRONMENT. Note
that the relative proportions of the
stressors to the whole stress is dynamic
and not evenly divided.
-276-
-------
Ralph C. Stahl, Jr.
• I believe that the guidelines would benefit from the inclusion of a flow
diagram to illustrate the risk assessor / risk manager interactions. A
similar flow diagram was used by EPA OERRin their revised guidelines
for ecological risk assessment.
• I agree with the inclusion of simulation models in the guidelines and
would encourage EPA to more fully discuss this topic, either in the
text directly or in an appendix. Because these models offer such
promise (or opportunity for misuse in some cases) for improving risk
assessments I believe they deserve more discussion and visibility in
the document.
" I am most encouraged by EPA's use of several key terms in the
guidelines. The use of terms and concepts such as risk ranges instead
of point .estimates, transparency in the risk assessment, the separation
of science from policy, the focus on higher levels of biological
organization for the risk assessment (population, community,
ecosystem) all suggest, a greater appreciation by EPA of how to
improve risk assessment within the Agency.
• Finally, I am also highly encouraged by EPA's recognition that
destroying habitats to clean them up is not a viable approach in
most instances.
PART IV. SPECIFIC COMMENTS
• p. 3, L 2-4 Given that EPA acknowledges the lack of
understanding of ecological principles and processes of risk managers,
there should be a short preamble that would be specific to risk
managers. Some if this information may be provided in the Primer
for Risk Managers prepared by EPA's Risk Communication Group.
-277-
-------
Ralph G. Stahl, Jr.
p. 4, L 16 I agree with the statement. There should be a clear
understanding of the drivers behind why an ecological assessment is
being undertaken, and these should be acknowledged by the risk
assessor and risk manager before the risk assessment is undertaken.
pp. 11 - 12 The change in terminology from measurment
endpoint to measures of effect is an improvement due to its
clarification of what the endpoint really is (ie effect). However, this is
not clearly translated in Text Box 1-3 where the measure of exposure
is not as precise or clear to the reader. This section discussing
exposure and effects would benefit from greater parallelism between
the terminology for effects and exposure.
p. 13, L 2-4 The statements contained in this part suggest that it is
important for the risk assessor to continue to look for impacts until
they are found. I don't believe that this is what the authors
intended. The guidelines should not give the risk assessor the
impression that their job is to undertake an exhaustive and never
ending effort to find impacts regardless of their significance or
relevance.
p. 14, L 27 I believe that the definition of biological stressors is
too restrictive. What is the reasoning for not including other types of
biological stressors such as predation, competition, etc. in .the
assessment? I recognize that trying to do that would be difficult, and
may require far more knowledge about life history, etc., but at the
very least the guidelines should note that there are other biological
stressors at work and that they also play a significant role in the
overall stressor / receptor scheme.
p. 15, L 19 Does co-occurance really equal exposure? If the
material is biologically available, then perhaps it does. But without
knowing whether something is bioavailable, it seems inappropriate to
consider co occurance as equivalent to exposure.
-278-
-------
Ralph G. Stahl, Jr.
• p. 18, L 4 The timing for the risk assessment must also be
included as one of the key issues discussed by the risk assessor and
risk manager.
• p. 20, Text Box 2-2 Great! Glad to see this written as is.
• p. 21, Text Box 2-3 While sustainability is a very important
management goal, shouldn't other examples of management goals be
used as well? Could improvement or enhancement also be goals in
this process as well as sustainability?
• p. 24, L 17 This sentence would suggest that ecological risk
assessments could not be performed for natural phenomena such as
those not resulting from human activities. Is that really the case or
just an oversimplification? Why couldn't the guidelines and
framework be used to assess the risk of fire for example?
• p. 24, L 17-24 This section is a bit unclear to me. For example, what
about reaching agreement on Data Quality Objectives, expected
products, decision points, etc. ? In additipn, Text Box 3-1 would
benefit from expansion and greater detail reflecting the comments
made about lines 17 - 24.
• p. 27, L 28 What about the situation where there is no exposure?
Does that mean that there is no need to proceed with the risk
assessment? Shouldn't there be a statement that if there are no
potential exposure pathways, or complete exposure pathways that the
risk assessor may not need to go further in the assessment?
• p. 36, L 30 & p. 37, L 1 This discussion seems to imply that the
risk assessor should search or select assessment endpoints that will be
popular with the public, or at the very least should be changed t o
gain public acceptance. I believe this is the problem with the current
-279-
-------
Ralph G. Stahl, Jr.
practice today and should not be encouraged with language of this
sort. In my opinion it is up to the risk manager to make the case
with the public as to what is or is not valuable to them, as well a s
what should or should not be protected. Once this is agreed to, then
it is up to the risk assessor in concert with the risk manager t o
provide the public with a clear understanding of what that means.
Providing the public with a lay-level explanation of the risk
assessment and the ecological resources that will be part of the study
(amphipods for example) is a key responsibility of the risk assessor
and risk manager. Attempting to change the assessment endpoint to
gain wide popularity seems to be counter to the need for openess,
clarity and transparency in the entire process.
p. 38, L 9 The last sentence in this line seems to beg the question
of what types of information would be needed. I believe it would be
helpful to give some suggestions rather than to leave this open ended.
p. 38, L 21 I don't understand how "absence of contact" can still
be considered exposure as noted in this sentence. This should be
clarified.
p. 44, L 31 What is an ecological entity? Does this mean the
ecosystem at risk, the river or lake being impacted, all of the these or
something else?
p. 46, L 4 I think this section would benefit from a diagram
rather than text. It would be beneficial to- develop a diagram that
illustrates the iterative process and how the risk manager is kept in
the process loop.
p. 46, L 11 I think it would be useful to briefly discuss how
developing a testable hypothesis is or is not useful to the risk
assessment. While I agree with the intent behind doing this, I1 m not
sure it is very practical in all cases. The ability to develop a clearly
-280-
-------
Ralph G. StaW, Jr.
testable hypothesis requires a fair knowledge of the system under
study, confounding variables and other influences. This level of
understanding will seldom be present especially when the situation
involves multiple stressors.
p. 52, L 23 Both the feasibility of obtaining the data and the
relevance of that, data to overall risk must be taken into account.
This is particularly noteworthy in the case of using biomarkers where
the technology involved may only be managed by a few individuals
or laboratories and where it is nearly impossible, without substantial
effort, to link the results of biomarker measurements to risks to whole
organisms, populations or communities.
p. 53, L 30 How the results will be interpreted should also be one
of the important decisions made early in the process. It would be
useful to use Test Box 3-13 to illustrate this point. For example one
could say that " a 50% reduction in egg hatching rate will be
interpreted as...." to illustrate the point.
p. 58, L 6 While I agree with the concept that data used in the
assessement would be better having come from refined DQOs, I don't
want to restrict the risk assessor to using .only data obtained that
way. There may be scientifically valid data originating from non-
DQO type studies that the risk assessor should have the option of
using provided they give the scientific justification for their use and
can attest to their validity. When the risk assessor is planning new
studies to obtain new data then those studies should adhere to the
development of clear DQOs. '
p. 58, L 14 In addition to the points made in this sentence it
would be important to have the risk assessor provide example
calculations. Showing how the numbers are "crunched" allows the
reader to conduct simple "spot checks" on their own and thus gain a
better understanding of the results.
-281-
-------
Ralph G. Stahl, Jr.
p. 59, L 3 This particular section needs to be expanded. For
example there is no mention of the possible biological or physical
stressors that may be present in the ecosystem under study, or the
stressors associated with the particular land use of the area. The
focus is tied too heavily to other chemical stressors.
p. 63, L 9 - 17 This discussion would benefit were the risk
assessor to be required to establish criteria by which data would be
judged as to their scientific validity. At the very least the risk
assessor -should be required to demonstrate adequate knowledge about
the types of test data being utilized in evaluating the risk to a
particular receptor. While the risk assessor cannot be versed in all
facets of laboratory and field studies, they can establish criteria,
perhaps in concert with co workers or peer review groups, to judge
studies and data that will be used in the risk assessment.
p. 64, L 28 - 301 believe a table would more clearly illustrate the three
main objectives compared to the text that is currently, provided.
p. 67, L 6 The background is very important in retrospective risk
assessment such as those conducted for waste sites. For that reason
the discussion on background would benefit from some examples on
how one could apply background information to the risk assessment.
p. 71, L 8 It should be emphasized that these data are at best
qualitative in nature and will not be refined sufficiently to provide a
quantitative estimate of exposure. Even tissue residue studies, due to
the mobility of organisms and the materials in question, are not
highly quantitative unless they can be coupled with a substantial
amount of information on the life history of a particular organism.
p. 73, L 1-2 This point would benefit from an. equation. It is
difficult to grasp the point in the currently presented form.
-282-
-------
Ralph Q. Stahi, ft.
• p. 74, L 19 This point seems to beg the question of when it would
not be appropriate to analyze the cause and effect relationship.
Either give some examples of when / where it would not make sense or
delete the sentence.
• p. 76, L 25 I agree with the general concept put forth in this
section. However, some high quality tests are not conducted
according to Good Laboratory Practices and vice versa. Adherence to
GLPs tends to insure that one can reconstruct how the study was
done, not whether the study was scientifically valid. I believe it is
important for the risk assessor to understand the strengths and
weaknesses of testing protocols so that they can better appreciate the
results from protocol driven studies. Yet adherence to GLPs may or
may not be indicative of a good study or help an inexperienced risk
assessor understand the strengths and weaknesses of a particular
study.
• p. 81, L 18 I'd prefer to see some additional discussion about the
range of factors given in this sentence less the reader is left to believe
that only multiples of 10 can be used in the emphirical approach. It
would be useful to note that ranges from 10 to 100 for these values is
based primarily on historical precedent and not necessarily o n
scientific facts.
• p. 88, L 4 I still believe that causality needs to be discussed even
when detailing information from laboratory studies that maximize
the cause / effect relationship.
• p. 90, L 12 - 20 Why not reiterate the three dimensions of exposure as
noted earlier on page 71, lines 27 - 28? Was it intended for this
section to take a different view of exposure?
-283-
-------
Ralph G. Stahl, Jr.
• p. 91, Figure 4-6 Unfortunately I cannot read the inside text of
this figure and thus cannot comment on its usefullness.
• p. 103, L 23 Shouldn't all risk assessments involve a team
approach? Until this point in the guidelines the suggestion of using a
team .approach has not been made. Are biological stressors so unique
that they require a team approach while chemical and physical
stressors do not? If the risk assessment would benefit from a team
approach, then say so up front.
• p. 114, L 5 I whole heartedly agree that the risk should be
reported as a range and not as a point estimate whenever possible.
• p. 117, L 15 I believe that the risk assessor should provide a
range of quotients rather than just one. Even though the results /
data may not be continuous, using a range of quotients will
nonetheless give a better indication of the range of. hazard (risk)
involved.
« p. 118, L 9 - 31 This entire discussion would benefit from more detail
of simulation models and examples of applications. Perhaps adding
such a section to the appendices could be accomplished in the next
draft of these guidelines.
• p. 121, L 5 There needs to be a stronger wording of what the
risk assessor should do. Using the wording of "indication of" does not
convey the appropriate level of importance that this requires.
• p. 125, L 2 This end of this sentence is misleading as it
suggests the risk assessor should consider the time before
"introduction of a stressor". The more appropriate "before the
introduction of the stressor under study" should be used.
-284-
-------
Ralph G. Stahl, Jt.
p. 125, L 16 - 18 The example of old growth forests should be re-
written as it now conveys something that does not seem to be what
the authors intended. For example, the continuous logging of an old
growth forest may eliminate the old growth trees, but it will not
eliminate the forest ecosystem except if the trees and other plants are
not allowed to regrow. The forest ecosystem will not be eliminated,
but changed.
-285-
-------
-------
Randall Wentsel
-287-
-------
Randall S. Wentsel"
COMMENTS ON EPA DRAFT ECOLOGICAL RISK ASSESSMENT GUIDELINES
Randall S. Wentsel
U.S. Army
In 19,89, the EPA Risk Assessment Forum started on a process
to develop ecological risk assessment (ERA) guidelines. In
1992, they produced the "Framework" which I consider to be
the ecological equivalent to the "Red Book" for human health
risk assessment. Following their six year plan, they have
now produced draft ERA guidelines. I would like to commend
the Forum, and Bill van der Schalie in particular, for their
leadership, knowledge, and management of this complex
subject area. ,
Part 1
I don't think we need a "Framework II" at this point as much
as we need guidance for ERA. The refining of definitions,
the identification of additional types of uncertainty, and
changing rectangles to hexagons in the paradigm; may be
necessary but I don't think that is what we need to move the
process forward. On a spectrum of very applied to Very
theoretical, the discussion of ERA in this guidance is more
theoretical in scope. While I have read what the guidelines
are and are not intended to be, I still think more can be
done in the applied area to support increased quality and
consistency of ERA. For example, in the problem formulation
phase (or possibly in the risk manager/risk assessor
discussions) it is never discussed what criteria would have
to be met to determine that no ERA was necessary. In the
analysis phase, the use of benchmarks are never discussed.
In the risk characterization phase, the use of probabalistic
distribution is only briefly mentioned. These applied areas
are difficult to present and controversial, but I think they
need to be included.
-288-
-------
Randall S. Wentsel
I have several comments in the policy area of ERA within
EPA. First, the EPA goes in four or five directions in the
use of ERA. The EPA needs to present a more uniform
approach to ERA than they are currently doing. The Offices
and Regions need to brought in so that EPA speaks with one
voice on this subject. I do not believe most EPA
Headquarters Offices and Regional Offices have enough
personnel with the breadth of expertise required to generate
their own ERA guidelines. They need to be brought into the
HQ discussions to support the uniform approach.
Second, I think these guidelines should focus on the
statutes and areas where EPA is the primary Federal agency
involved, i.e. TOSCA, FIFRA, and Superfund. It seems like
these guidelines go out of their way to present examples in
watershed and land management areas where the Departments of
Interior and Agriculture have leadership. While these may
be more ecologically interesting in scope, critical issues
relevant to specific statutes, where EPA leads, should be
discussed.
Third, ERA is a tool to assist risk mangers in decision
making. Applied guidance is needed. While scientific
debate and theoretical issues have their place, eventually
factual information must be supplied from the risk
assessment to the managers. If ERA is to be an effective
tool in the risk management process, it must supply
information in a more standardized format in a timely
fashion to the risk manager. Endless debate and data
collection will reduce the value of ERA in environmental
policy decisions. The guidelines should ask the question -
Where is the science now in environmental chemistry,
environmental toxicology, and risk characterization?
Currently, too much expert judgment is used by the risk
assessor in conducting an ERA and too often scientific or
-289-
-------
Randall S. Wentsel
personnel values are used by the regulator to judge whether
an ERA is appropriate.
PART I GENERAL PRINCIPLES
o Specifically, how can the ecological risk assessment
guidelines be improved to more clearly indicate underlying
assumptions and default positions?
The ERA guidelines are currently written too generally to
establish default positions. Worst case assumptions and
default positions are controversial subjects. They are open
to criticism when they deviate from sound scientific
principles or are so conservative that they don't pass the
common sense test. I believe default positions imply a cook
book approach which ERA doesn't always fit.
o What changes would you recommend in each section of the
document?
ERA is referred to as a scientific process in the document.
I do not believe that is the most prevalent opinion.
PART II SPECIFIC QUESTIONS
Guidelines Balance
1. Considering both the present state of the science and
present and future Agency needs/ how well are the guidelines
balanced regarding the range of stressors, levels of
biological organization, ecosystem types, and
spatial/temporal scales? Specifically, what would you
emphasize or de-emphasize?
I would emphasize simple over complex ERAs. Discuss what we
can do. Phrases like "can be difficult" or "is complex" do
not provide anything new to the audience.
2. What would you suggest to improve the use of case
illustrations in the guidelines? How useful is Appendix A
in illustrating a range of applications of the risk
-290-
-------
Randall S. Wentsel
assessment process?
Case studies are too brief. The reader needs to know how
equations, models, and scientific judgment was used.
Consider using specifics from issue papers with page
numbers.
3. Some Agency reviewers of the guidelines have suggested
that more examples of terrestrial assessments and field
approaches (e.g./ bioassessment techniques) should be used.
Specifically, what, if anything, should be ad'ded?
A discussion of important issues for terrestrial assessments
is needed. What tools are available and what tools are
needed in the analysis phase for soils.
4. Areas of uncertainty are summarized in the problem
formulation (section 1.5), analysis (section 3.7), and risk
characterization (section 5.2.4). How useful is this
approach in providing guidance on uncertainty issues?
Communication and human error should be discussed but not
given equal treatment to the other areas. The three tables
should be combined into one. Specifics on uncertainty
reduction should be expanded.
Introduction and Scope
5. How could the Introduction be modified to more clearly
communicate the scope and content of the guideline?
6. Terminology, especially related to endpoints and
exposure, has always been controversial. What changes, if
any, would you recommend in guidelines terminology?
The Framework established common terminology. Refinement of
terminology has its place, but that must be balanced against
the confusion changes generate. I would place less priority
on terminology changes than in providing guidance to
assessors.
-291-
-------
Randall S. Wentsel
7. The overall framework figure for the ecological risk
process has been retained, although some changes have been
made to the diagrams for problem formulation, analysis, and
risk characterization. What further modifications, if any,
are required?
I think the workshop participants should review the
necessity of changes to the paradigm. Additional discussion
of the thinking behind the changes would be beneficial.
Risk Manager- Interaction
8. Please comment on the description of the risk manager's
role at the initiation of an ecological risk assessment and
on the principles for selecting management goals.
In initial risk manager and risk assessor discussions or
possibly in the problem formulation phase, I think guidance
is needed to address - Do we have a problem? Is an ERA
necessary?
9. What additional points, if any, should be covered about
relating ecological information to risk management decisions
after the completion of an ecological risk assessment?
Problem Formulation
10. How useful is the categorization of assessments as
either stressor- or source-initiated, effects-initiated and
ecological value-initiated?
It is an interesting discussion. Policy impacts from
specific legislation could be stressed as examples.
11. Please comment on the discussion of assessment endpoints
and their relationship to management goals and "measures.."
12. What, if anything, would you add to the discussion of
risk hypotheses and conceptual models to give the reader a
clearer understanding of their nature and content?
-292-
-------
Randall S. Wentsel
Tiers are discussed in the document. How would tiers be
addressed in the three phases of ERA?
Analysis
13. How would you change this section/ if at all, to improve
the balance in the discussion of the different stressor
types?
I would expand the exposure section to include more
specifics on pathways. I would slant the discussion toward
statutes where the majority of ERAs are prepared (probably
by number and dollar amount) and where more detailed
information can be presented. Areas where the science and
risk management decisions are complex and undeveloped should
receive less space.
14. Are there additional points or principles that should be
emphasized for either chemical, physical, or biological
stressors?
Environmental chemistry should be stressed. The impacts of
water and soil chemistry on the transport, fate, and
bioavailability of chemicals is important. Environmental
chemistry principles should be presented or referenced to
issues papers. Background concentrations are discussed on
the top of page 67. I think the guidance should consider a
definition or at least a more quantitative discussion of
addressing background. Without some specifics much of the
discussion in section 4.2.2. is of little value.
In section 4.2.3.2. in the discussion on uncertainty factors
for extrapolation between responses, factors between 10 to
100 are put forward. I believe these have been adopted from
human health risk assessments where protection of the
individual is the endpoint. I think uncertainty factors for
protecting endangered species may be similar to human health
but factors for non-threatened populations or ecological
systems would be less. This section should differentiate
-293-
-------
Randall S. Wentsel"
these issues.
15. What additional principles would you suggest/ if any,
for the analysis of multiple stressors?
Risk Characterization
16. What additional principles should be highlights in the
discussion of risk estimation techniques?
A further discussion of the impact of professional judgment,
scientific values, "and the incorporation of conservatism
into risk estimation should be presented. For example in
section 5.2.2.1. limitations of the quotient method do not
mention conservatism that is often built into the method.
Use of distributions in risk estimation needs to be
expanded.
17. The guidelines propose four criteria for ecological
significance. How should this list of criteria be
modified, if at all? What additional guidance, if any,
might be added to the discussion of these criteria?
-294-
-------
APPENDIX E
REVIEWER WORK GROUP ASSIGNMENTS
-------
-------
WORKSHOP ON THE DRAFT ECOLOGICAL RISK ASSESSMENT GUIDELINES
Washington, DC
Decembers-?, 1995
WORK GROUP ASSIGNMENTS
William Smith (Workshop Chair)
Each work group will address the following topics:
• Problem Formulation
• Analysis
• Risk Characterization
Work Group 1
Dwayne Moore (Chair)
Lawrence Barnthouse
James Clark
William Cooper
Peter deFur
Jim Donald
Anne Fairbrother
Alyce Fritz
Robert Goldstein
Timothy Kubiak
Richard Orr
Kevin Reinert
Work Group 2
Richard Kimerle (Chair)
Steven Bartell
John Bascietto
Nancy Bettinger
Joanna Burger
Peter Chapman
Charles Eirkson
Lawrence Harris
Lyman McDonald
Gerald Niemi
Ralph Stahl
Randy Wentsel
E-l
-------
-------
APPENDIX F
FINAL OBSERVER LIST
-------
-------
United States
Environmental Protection Agency
Risk Assessment Forum
Workshop on the Draft Ecological
Risk Assessment Guidelines
Holiday Inn - Georgetown
Washington, DC
December 6-7, 1995
Final Observer List
Todd Allen
Chemical Engineer
Radian Corporation
2455 Horsepen Road - Suite 250
Herndon.VA 22071
703-713-1500
Fax:703-713-1512
H. Kay Austin
Ecologist
National Center for
Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
401 M Street, SW
Washington, DC 20460
202-260-5789
E-mail: austin.kay@epamail.epa.gov
Tom Baugh
Environmental Scientist
Office of Research & Development
U.S. Environmental Protection Agency
401 M Street, SW (8205)
Washington, DC 20460
202-260-8936
Heino Beckert
Physical Scientist
Air, Water & Radiation Division
U.S. Department of Energy
1000 Independence Avenue, SW
(EH-4I2)
Washington, DC 20585
202-586-9643
Fax:202-586-3915
Frederick Betz
Senior Scientist
Jellinek, Schwartz & Connolly, Inc.
1525 Wilson Boulevard - Suite 600
Arlington, VA 22209
703-312-8522
Fax: 703-527-5477
E-mail: 75402.300@compuservexom
Melissa Bockelmann
Principal
Bockelmann Goellner, Inc.
9805 Peppermill Place
Vienna, VA 22182
703-255-0317
Fax: 703-281-2714
Christine Bunck
Manager, BEST Program
National Biological Service
U.S. Department of the Interior
1849 C Street, NW (3660)
Washington, DC 20240
202-482-3972
Fax: 202-273-0825
E-mail: chris_bunck@nbs.gov
Timothy Carlstedt
Attorney
N.W. Bernstein and Associates
2000 M Street, NW - Suite 745
Washington, DC 20036
202-466-8100
Fax: 202-466-8877
Patricia Casano
Consulting Attorney
General Electric Co.
1299 Pennsylvania Avenue, NW
Suite 1100
Washington, DC 20004
202-637-4228
Fax:202-637-4017
Patricia Cirone
Branch Chief
Evaluation Branch
U.S. Environmental Protection Agency
1200 Sixth Avenue (ES-098)
Seattle, WA 98101
206-533-1597
Fax:206-553-0119
Steve Cordle
Environmental Scientist
National Center for
Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
401 M Street, SW (8602)
Washington, DC 20460
202-260-5989
Fax: 202-260-3400
) Printed on Recycled Paper
F-l
-------
Richard Cunningham
Environmental Specialist
U.S. Department of Energy
1000 Independence Avenue, SW
(4B087)
Washington, DC 20585
202-586-7032
Fax: 202-586-6575
E-mail: rlchard.cunningham@dp.doe.gov
Sidney Draggan
Special Assistant for Science
Office of the Administrator
U.S. Environmental Protection Agency
401 M Street SW (1103)
Washington, DC 20460
202-260-4724
Fax: 202-260-4852
E-mail: draggan4idnex@epamall.epa.gov
Joseph Dulka
Global Regulatory Ecotoxicologist
DuPont Agricultural Products
Barley Mil) Plaza
Walker's Mill Building
P.O. Box 80038
Wilmington, DE 19880-0038
302-992-6271
Fax: 302-992-6470
Stephen Ells
Senior Environmental Scientist
Office of Superfund
U.S. Environmental Protection Agency
401 M Street, SW (5204G)
Washington, DC 20460
703-603-8822
Fax:703-603-9103
E-mail: eils.steve@epamall.epa.gov
Ronald Filadelfo
Research Staff
Center for Naval Analyses
4401 Ford Avenue
P.O.Box 16268
Alexandria, VA 22302-0268
703-824-2207
Kathryn Fogarty
Senior Scientist
Menzie-Cura and Associates, Inc.
I Courthouse Lane - Suite 2
Chelmsford, MA 01824
508-453-4300
Fax: 508-453-7260
Donald Goellner
Principal
Bockelmann Goellner, Inc.
9805 Peppermill Place
Vienna, VA 22182
703-883-1312
Fax: 703-883-0195
Allison Gore
Director, Environmental Affairs
Interstate Natural Gas
Association of America
555 13th Street, NW
Suite 300 West
Washington, DC 20004
202-626-3235
Fax: 202-626-3249
Michael Harrass
Environmental Toxicologist
Amoco Corporation
130 East Randolph Drive (PO69M4)
Chicago, IL 60601
312-856-5116
Fax:312-856-7584
Luis Hernandez
Senior Research Associate
Barrera Associates
733 15th Street, NW - Suite 1120
Washington, DC 20005
202-638-6631
Fax:202-638-4063
E-mail: lalberto@aoj.com
Stanley Hewins
Toxicologist
Air Force Center for
Environmental Excellence
8001 Arnold Drive
Brooks AFB, TX 78235
210-536-4755
Fax: 210-536-5989
Dexter Hinckley
5604 Bloomfield Drive - #102
Alexandria, VA 22312-2524
703-354-1342
E-mail: dhincklex@igc.apc.org
Robert Huggett
Assistant Administrator
Office of Research and Development
U.S. Environmental Protection Agency
401 M Street, SW (8101)
Washington, DC 20460
202-260-7676
Fax: 202-260-9761
Elizabeth Kelly
Environmental Statistician
Los Alamos National Laboratory
Mail Stop: K557
Los Alamos, NM 87544
505-667-2356
Fax:505-667-5531
Vince Leclair
Reporter
Pesticide and Toxic Chemical News
1803 19th Street, NW - #6
Washington, DC 20005
202-387-8504
Fax: 202-387-8504
Rose Lew
Environmental Scientist
U.S. Environmental Protection Agency
401 M Street, SW,
Washington, DC 20460
202-564-4147
Suzanne Marcy
Senior Scientist for Ecology
National Center for
Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
401 M Street, SW (8601)
Washington* DC 20460
202-260-0689
Fax: 202-260-1036
E-mail: marcx.suzanne@epamail.epa.gov
David Mauriello
Senior Ecologist
Health and Environmental
Review Division
Office of Pollution
Prevention and Toxics
U.S. Environmental Protection Agency
401 M Street, SW (7403)
Washington, DC 20460
202-260-2260 /
Fax:202-260-1236
E-mail: mauriello.dave@epamail.epa.gov
Margaret McVey
Program Officer.
Board on Environmental
Studies and Toxicology
National Research Council
2101 Constitution Avenue (HA 354)
Washington, DC 20418
202-334-2545
Fax: 202-334-2752
E-mail: mmcvex@nas.edu
F-2
-------
Paul Mehrte
Vice President & Director
Business Development
Analytical Bio-Chemistry
Laboratories, Inc.
7200 East ABC Lane
Columbia, MO 65202
314-443-9018
Fax: 314-443-9033
Reo Manning
American Industrial Health Council
2001 Pennsylvania Avenue, NW
Suite 760
Washington, DC 20006
202-833-2131 Ext: 3024
Fax:202-833-2201
Daniel Michael
Environmental Decision
Support Specialist
Neptune and Company
1505 15th Street-Suite B
Los Alamos, NM 87544
505-662-2121
Fax:505-662-0500
Susan Norton
Ecologist
National Center For
Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
401 M Street, SW (8602)
Washington, DC 20460
202-260-6955
Fax: 202-260-3955
Evelina Norwinski
Attorney
Hunton & Williams
2000 Pennsylvania Avenue, NW
Suite 9000
Washington, DC 20006
202-955-1603
Fax:202-778-2201
Roger Platt
Deputy Counsel
National Realty Committee
1420 New York Avenue, NW
Suite 1100
Washington, DC 20005
202-639-8400
Fax: 202-639-8442
Jane Mood/ Rachal
Risk Policy Report
Inside Washington Publishers
1225 Jefferson Davis Highway
Suite 1400
Arlington, VA 22202
703-416-8500
Fax: 703-4 J 6-8543
Donald Rodier
Biologist
Chemical Screening & Risk
Assessment Division
Office of Pollution
Prevention and Toxics
U.S. Environmental Protection Agency
401 M Street, SW (7402)
Washington, DC 20460
202-260-1276
Fax: 202-260-1216
Sara Thurin Rollin
Reporter
The Bureau of National Affairs, Inc.
1231 25th Street, NW
Washington, DC 20037
202-452-4584
Fax: 202-452-4150
Alan Rubin
Scientist
Water Environmental Foundation
601 Wythe Street
Alexandria, VA 22314
703-684-2438
Fax: 703-684-2492
Lee Salamone
Manager, Risk and Communications
Chemical Manufacturers Association
2501 M Street, NW
Washington, DC 20037
202-887-6944
Fax:202-778-4177
E-mail: lee_saJamone@mail.cmahq.com
Stephanie Sanzpne
Designated Federal Official
Science Advisory Board Ecological
Processes & Effects Committee
401 M Street, SW (1400F)
Washington, DC 20460
202-260-6557
Fax:202-260-7118
E-mail: sanzone.stephanie@epamail.epa.gov
John Schaeffer
Biologist
Health and Environmental
Review Division
Office of Pollution
Prevention and Toxics
U.S. Environmental Protection Agency
401 M Street, SW (7403)
Washington, DC 20460
202-260-1266
Fax: 202-260-1236
Anne Sergeant
Environmental Scientist
U.S. Environmental Protection Agency
401 M Street, SW (8602)
Washington, DC 20460
202-260-9376
Fax: 202-260-1722
E-mail: sergeant.anne@epamail.epa.gov
Victor Serveiss
Environmental Scientist
National Center for
Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
401 M Street, SW (8205)
Washington, DC 20460
202-260-5794
Fax:202-260-1722
E-mail: serveiss.victor@epamail.epa.gov
Michael Slimak
Associate Director for Ecology
National Center for
Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
401 M Street, SW (8601)
Washington, DC 20460
202-260-5950
Fax: 202-260-3427
Bill van der Schalie
Ecologist
National Center for
Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
401 M Street, SW (8602)
Washington, DC 20460
202-260-4191
Fax: 202-260-3955
F-3
-------
Margaret Spring
Sidle/ & Austin
1722 Eye Street, NW
Washington, DC 20006
202-736-8689
Fax: 202-736-8711
John Staunton
Reporter
Inside Washington Publishers
1225 Jefferson Davis Highway
Arlington, VA 22202
703-416-8564
Fax: 703-416-8543
Frank Vertucci
Certified Senior Ecologist
Environmental Toxicology
and Risk Assessment
ENSR
4413 West La Porte Avenue
Fort Collins, CO 80521
970-416-0916
Fax: 970-493-8935
Darin Waylett
Associate Scientist
TAS, Inc.
The Flour Mill
1000 Potomac Street, NW
Washington, DC 20007
202-337-2625
Fax:202-337-1744
William Wood
Director, Risk Assessment Forum
Office of Research and Development
U.S. Environmental Protection Agency
401 M Street, SW(8IOI)
Washington, DC 20460
202-260-1095
Fax: 202-260-3955
Maurice Zeeman
Chief
Environmental Effects Branch
Office of Pollution
Prevention and Toxics
U.S. Environmental Protection Agency
401 M Street, SW (7403)
Washington, DC 20460
202-260-1237
Fax:202-260-1236
F-4
-------
APPENDIX G
WORKSHOP CHAIRPERSON'S OPINION SURVEY
-------
-------
WORKSHOP CHAIRPERSON'S
"MODIFIED DELPHI" REVIEWER OPINION SURVEY
At the conclusion of the workshop, Workshop Chairperson William Smith conducted a
"modified Delphi" survey of reviewers' opinions on several issues discussed during the workshop.
He invited reviewers to record their reaction to 14 statements as follows; Strongly Agree (SA),
Agree (A), Disagree (D), Strongly Disagree (SD), or Unclear/Uncertain (?). Of the 25
reviewers, about 10 were out of the room or otherwise did not participate in the survey.
1. The Draft Guidelines represent a useful next step from the Framework Report.
SA(3) A (10) D(2) SD(0) ? (0)
2. The Draft Guidelines are not intended to be a technical manual.
SA(5) A (8) D(2) SD(0) ? (1)
3. The Draft Guidelines certainly can express preferences and they should, in fact, identify
criteria for selecting alternative techniques.
SA(6) A (10) D(0) SD(0) ? (0)
4. The Draft Guidelines include the key elements of what we perceive to be a reasonable
risk assessment.
SA(3) A (10) D(l) SD(0) ? (1)
::. The Draft Guidelines contribute to the hope for some degree of consistency of
application, at least within EPA.
SA (2) A (8) D (5) SD (0) ? (1)
6. Societal values translated via management or regulatory objectives bound/define risk
assessment questions.
SA(3) A (8) D(l) SD(0) ? (2)
'/, Revision of the Draft Guidelines should not be delayed by major supporting documents
(e.g., an Issue Paper on multiple stressors) judged to be needed.
SA(5) A (10) D(l) SD(0) ? (0)
G-l
-------
8. Teams, rather than individuals, represent "people" in the risk assessor and risk manager-
boxes in the framework diagram.
SA(6) A (9) D(0) SD(0) ? (0)
9. The importance of biological and physical stressors is recognized to be increasing.
SA(3) A (8) D(l) SD(0) ? (3)
10. Resources directed to technology transfer of the Guidelines would represent a wise
investment.
SA(4) A (6) D(l) SD(0) ? (4)
11. Unking risk assessments to ecological resource monitoring is important.
SA(2) A (7) D(l) SD(0) ? (5)
12. Minimal use of technical jargon and careful explanation of phrases with multiple
meanings (e.g., multiple stressors) will maximize the clarity of the Draft Guidelines.
SA(6) A (10) D(l) SD(0) ? (0)
13. As drafted, the Guidelines are suitable for "micro" site-specific assessments and "macro"
policy assessments.
SA (3) A (10) D (1) SD (0) ? (0)
14. The Draft Guidelines have no major "fatal flaws," and EPA should move ahead.
SA(6) A (8) D(0) SD(0) ? (1)
* o
»C.S. GOVERNMENT PRINTIHG OFFICE:1996-750-001/41041
------- |