&EPA
EPA/100/R-09/OQ6 | October 2009
www.epa.gov/osa
United States
Environmental Protection
Agency
Summary Report:
Risk Assessment Forum Technical
Workshop on Population-level
Ecological Risk Assessment
Office of the Science Advisor
Risk Assessment Forum
-------
-------
Summary Report:
Risk Assessment Forum Technical
Workshop on Population-level
Ecological Risk Assessment
Risk Assessment Forum
U.S. Environmental Protection Agency
Washington, DC 20460
-------
NOTICE
The statements in this report reflect the individual expert views and opinions of the workshop attendees,
together with summary observations and recommendations of an Agency technical panel. They do not represent
analyses or positions of the Risk Assessment Forum or of the U.S. Environmental Protection Agency.
This document has been reviewed in accordance with U.S. Environmental Protection Agency policy. Mention
of trade names or commercial products does not constitute endorsement or recommendation for use.
-------
Contents
Preface v
Authors, Contributors and Reviewers vi
Acronyms viii
Executive Summary x
1 Introduction 1
1.1 Background and Context 1
1.2 Workshop Objectives 4
1.3 Workshop Format 4
2 Summary of Opening Remarks and Presentations 7
2.1 Lee Hofmann, EPA Risk Assessment Forum, Executive Director 7
2.2 Wayne Munns, EPA Office of Research and Development, Workshop Chair 7
2.3 Charles Delos, EPA Office of Water 8
2.4 Edward Odenkirchen, EPA Office of Pesticide Programs 8
2.5 David Charters, EPA Office of Solid Waste and Emergency Response 9
2.6 Bruce Duncan, EPARegion 10 9
2.7 Steve Newbold, EPA Office of Policy, Economics and Innovation 10
2.8 Todd Bridges, U.S. Army Corps of Engineers 10
2.9 Jill Awkerman, EPA Office of Research and Development 11
2.10 Richard Sibley, University of Reading 11
3 Breakout Group Reports 13
3.1 Observational Approaches 13
3.2 Experimental Approaches 16
3.3 Modeling Approaches 21
4 Summary of Expert Opinions 33
4.1 Observational Approaches 33
4.2 Experimental Approaches 33
4.3 Modeling Approaches 34
4.4 Commonalities Across Approaches 35
5 Technical Working Group Recommendations for Future Progress 37
References 39
Appendix A. Workshop Attendees A: 1
Appendix B. Workshop Agenda B: 1
Appendix C. Breakout Group Charge C: 1
-------
List of Tables
Table 1. Developmental projects recommended for population-level
ecological risk assessment from an earlier RAF colloquium 2
Table 2. Observational methods and decision contexts 17
Table 3. Attributes of populations in assessment endpoints
(from Barnthouse et al. 2008) 18
Table 4. Applications of population models in environmental
management contexts 23
Table 5. Attributes of populations that can be evaluated using population models,
and data requirements of population models 25
Table 6. Types of models, their advantages and disadvantages, and
specific software implementations 26
List of Text Boxes
Examples of previous requests for training and educational exchange 3
An alternate opinion regarding toxicity data 27
-------
Preface
In June 2008, the Environmental Protection
Agency's (EPA) Risk Assessment Forum convened
a technical workshop on population-level
ecological risk assessment to consider whether the
current state of knowledge about this subject was
sufficiently mature to develop guidance, and if so,
to help to identify key actions needed to produce
such guidance. The purpose of this document is to
communicate the findings of that workshop.
In 1998, EPA's Risk Assessment Forum developed
its Guidelines for Ecological Risk Assessment
(U.S. EPA 1998) to help guide Agency programs
and practitioners in the performance of ecological
risk assessments. Public comment associated
with publication of the Guidelines indicated a
need for additional guidance for assessing effects
at the population, community and ecosystem
levels of ecological organization to serve as more
substantive guidance on protecting populations of
animals and plants. A survey of EPA ecological
risk assessors at that time ranked effects at higher
levels of biological organization, along with
assessment endpoints and measures of effect, as
having the highest priority for development of
additional guidance. The call for guidance has been
repeated in recent international efforts addressing
population-level ecological risk assessment (e.g.,
Barnthouse et al. 2008; Forbes et al. 2009). In
particular, Barnthouse et al. (2008) recommended
development of guidance to assist risk assessors,
risk managers and stakeholders in selecting,
applying, interpreting and communicating
population-level ecological risk assessment
procedures and analysis tools to cover a range of
environmental management contexts. Guidance
of this nature does not exist at this time, although
Barnthouse et al. (2008) felt that the state-of-the-
science was sufficiently mature to produce it.
The primary goal of this document is to inform
EPA in its decisions regarding development of
additional guidelines or best practice descriptions
for planning, implementing and interpreting
ecological risk assessments that involve population-
level assessment endpoints. It communicates the
individual opinions and insights of scientific experts
in the fields of population ecology and ecological
risk assessment as offered during the workshop.
It also communicates the recommendations of an
Agency Technical Panel concerning development
of guidelines or best practice descriptions for
population-level ecological risk assessment, and the
actions that can be taken to facilitate development
of such guidance. This document does not provide
technical guidance for population-level ecological
risk assessment, nor does it address the policy
issues attendant with performing or interpreting
such assessments. This report was prepared by a
Technical Panel and Workshop Steering Committee
under the auspices of EPA's Risk Assessment
Forum, and the Risk Assessment Forum is its
primary intended audience.
The Risk Assessment Forum was established to
promote scientific consensus on risk assessment
issues and to incorporate this consensus into
appropriate risk assessment guidance. To
accomplish this, the Forum assembles experts
from throughout EPA in a formal process to study
and report on these issues from an Agency-wide
perspective. Technical experts from outside the
Agency often contribute to this process as part of
workshops and other issue-oriented mechanisms.
This document, and the workshop it describes,
reflects the Forum's long-standing commitment to
advancing the concepts and practice of ecological
risk assessment.
-------
-------
Authors, Contributors
and Reviewers
This report was prepared by a Technical Panel and Workshop Steering Committee of EPA staff under the
auspices of EPA's Risk Assessment Forum.
Technical Panel
Wayne R. Munns, Jr. (Chair)
National Health and Environmental Effects
Research Laboratory
Office of Research and Development
U.S. EPA
Narragansett, RI 02882
Charles Delos
Office of Science and Technology
Office of Water
U.S. EPA
Washington, DC 20460
Bruce Duncan
Office of Environmental Assessment
Region 10
U.S. EPA
Seattle, WA 98101
Anne Fairbrother1
National Health and Environmental Effects
Research Laboratory
Office of Research and Development
U.S. EPA
Corvallis, OR 97333
Thomas Forbes
Immediate Office
Office of Environmental Information
U.S. EPA
Washington, DC 20460
1 The current affiliation of Dr. Fairbrother is
Exponent, Bellevue, WA 98007.
Steve Newbold
National Center for Environmental
Economics
Office of Policy, Economics and Innovation
U.S. EPA
Washington, DC 20460
Edward Odenkirchen
Office of Pesticide Programs
Office of Prevention, Pesticides, and Toxic
Substances
U.S. EPA
Washington, DC 20460
Anne W. Rea
Office of Air Quality Planning and
Standards
Office of Air and Radiation
U.S. EPA
Research Triangle Park, NC 27711
Donald J. Rodier
Office of Pollution Prevention and Toxics
Office of Prevention, Pesticides, and Toxic
Substances
U.S. EPA
Washington, DC 20460
Glenn W. Suter II
National Center for Environmental
Assessment
Office of Research and Development
U.S. EPA
Cincinnati, OH 45268
Randy Wentsel
Office of Research and Development
U.S. EPA
Washington, DC 20460
-------
External Contributors
Jerry Cura
The Science Collaborative
Winchester, MA 01890
Diane E. Nacci
National Health and Environmental Effects
Research Laboratory
Office of Research and Development
U.S. EPA
Narragansett, RI 02882
Robert Pastorok
Integral Consulting
Mercer Island, WA 98040
Mary T. Sorensen
ENVIRON International Corporation
Atlanta, GA 30339
Risk Assessment Forum Staff
Gary Bangs
Office of the Science Advisor
U.S. EPA
Washington, DC 20460
Colleen Flaherty (on detail)
Office of Pesticide Programs
Office of Prevention, Pesticides, and Toxic
Substances
U.S. EPA
Washington, DC 20460
Seema Schappelle
Office of the Science Advisor
U.S. EPA
Washington, DC 20460
EPA Reviewers
Mace Baron
National Health and Environmental Effects
Research Laboratory
Office of Research and Development
U.S. EPA
Gulf Breeze, FL 325 61
Marian Olsen
Region 2
U.S. EPA
New York, NY 10007
Cynthia Stahl
Region 3
U.S. EPA
Philadelphia, PA 19103
Charles Maurice
Region 5
U.S. EPA
Chicago, IL 60604
-------
Acronyms
ABM
ALMaSS
AQUATOX
ATLSS
BASS
CADDIS
CERCLA
CREM
D4EM
EC,
EMAP
EPA
ERA
ERAF
EU
I&E
IUCN
LCx
LTRE
LOAEL
NBII
NOAEL
OPP
ORD
agent-based model
Animal, Landscape and Man Simulation System
a general fate and effects model for aquatic ecosystems
Across Trophic Level System Simulation
Bioaccumulation and Aquatic System Simulator
Causal Analysis/Diagnosis Decision Information System
Comprehensive Environmental Response, Compensation, and Liability Act
U.S. EPA Council for Regulatory Environmental Modeling
Data for Environmental Modeling
concentration causing an effect level of x
Environmental Monitoring and Assessment Program
Environmental Protection Agency
ecological risk assessment
U.S. EPA Ecological Risk Assessment Forum
European Union
impingement and entrainment
International Union for Conservation of Nature
concentration causing a lethal response ofx
life table response experiment
lowest observed adverse effect level
National Biological Information Infrastructure
no observable adverse effect level
U.S. EPA Office of Pesticide Programs
U.S. EPA Office of Research and Development
-------
ORNL
OST
OSWER
PATCH
PATH
QA
RAF
RAMAS
RCRA
REMAP
RI/FS
SAB
SETAC
SSD
TMDL
ULM
USAGE
USFWS
Oak Ridge National Laboratory
U.S. EPA Office of Science and Technology
U.S. EPA Office of Solid Waste and Emergency Response
Program to Assist in Tracking Critical Habitat, also called HEXSIM
Plan for Analyzing and Testing Hypotheses
quality assurance
U.S. EPA Risk Assessment Forum
a family of commercially available software for population risk and other analyses
Resource Conservation and Recovery Act
Regional Environmental Monitoring and Assessment Program
Remedial Investigation/Feasibility Study
U.S. EPA Science Advisory Board
Society of Environmental Toxicology and Chemistry
species sensitivity distribution
total daily maximum load
Unified Life Models
U.S. Army Corps of Engineers
U.S. Fish and Wildlife Service
-------
Executive Summary
E.1 Background
The U.S. Environmental Protection Agency (EPA,
or the Agency) has adopted risk assessment as a
primary tool supporting environmental decision
making. To help maximize the value of ecological
risk assessment (ERA) to Agency programs,
EPA's Risk Assessment Forum (RAF) produced
the landmark Guidelines for Ecological Risk
Assessment (\].$. EPA 1998) which describe a
general strategy and framework for planning,
executing and interpreting ERAs. The 1998
Guidelines recommend a planning dialogue among
risk managers, risk assessors and other interested
parties as a critical first step toward initiating
an ERA. This dialogue is intended to produce
agreement on and understanding of management
goals and the types of decisions that the assessment
will support. It also establishes the scope,
complexity and focus of the risk assessment to be
conducted. Depending upon their context, planning
agreements might be established as a matter of
policy, or might be made on an ad hoc basis to
inform situation-specific management decisions.
Regardless, these agreements lead to selection,
during the problem formulation phase of the
assessment, of the endpoints to be evaluated during
the risk assessment. Assessment endpoints are
selected to describe valued ecological entities and
their attributes at levels of ecological organization
that are relevant and applicable to the decisions
being made, and in combination can encompass
single or multiple levels of ecological organization
in a single assessment.
The choice of ecological organization levels to be
evaluated in regulatory ERAs can be a challenging
one. Although the enabling legislation of many
of EPA's programs either explicitly or implicitly
identify protection of ecological populations
as management goals, most ERAs conducted
for chemicals by EPA, and indeed by most
organizations worldwide, focus on organism-level
entities and attributes (e.g., rainbow trout survival,
growth or reproduction) as assessment endpoints.
These endpoints are practical because they often
can be estimated through toxicological testing
and other means, and are expedient because they
are commonly presumed to provide protection of
population-level attributes (e.g., abundance and
persistence). Further, methods and practice are
well established for assessing risk to organism-
level assessment endpoints, but documentation of
consensus methods for population-level ecological
risk assessment is lacking. Consequently, risk to
populations has only occasionally been evaluated
directly by EPA (e.g., trout populations of
Adirondack lakes in the National Acid Precipitation
Assessment). This situation results from several
factors affecting assessment planning, including
the perceived relationships between assessment
endpoints and environmental management goals,
historical precedence, and importantly, the lack of
recognized consensus and guidance about how such
assessments should be performed.
The RAF conducted a colloquium in 1999 to help
identify the nature and scope of projects that would
advance development of ecological risk assessment
guidance in three broad areas: 1) effects at higher
levels of biological organization, including
landscape-level effects; 2) assessment endpoints
and measures of effect2; and 3) risk characterization
techniques. A broad theme emerging from
discussion of the first area was a focus on methods
for assessing risks to populations and interpreting
the results obtained by those methods. During the
colloquium, the needs of "on-the-ground" risk
assessors and risk managers led to identification
of several developmental projects (organized by
assessment phase in Table 1) related to population-
level ecological risk assessment and approaches
to implement them. These approaches reflected a
number of considerations, including the perceived
state-of-the-science, the types of intermediate
products deemed useful and the needs of the
Agency.
2 The RAF published guidance for Generic Ecological
Assessment Endpoints (GEAEs) for Ecological Risk
Assessment (EPA/630/P-02/004f) in 2003 (www.epa.gov/
raf/publications/geae.htm).
-------
In 2005, the RAF formed a Technical Panel
to explore a number of issues associated with
population-level ecological risk assessment.
Consisting of representatives of EPA Program
Offices, Regions and the Office of Research and
Development (ORD), this working group initially
identified three broad actions intended to enhance
the Agency's understanding of approaches for
assessing risks to populations. These actions are:
1. Expand training in population-level
ecological risk assessment - Since the
publication of the Guidelines, ORD has
received a number of requests for formal
training and educational exchanges
addressing topics related to population-
level ecological risk assessment. To help
meet this need, EPA's Risk Assessment
Forum sponsored the vendor-supplied
"Population Modeling Training Workshop,"
conducted at Region 5 's offices in Chicago,
Illinois in late 2004. This training was
coordinated through the Ecological
Risk Assessment Forum (ERAF) and
was attended primarily by regional risk
assessors who support hazardous waste
assessments under the Comprehensive
Environmental Response, Compensation,
and Liability Act (CERCLA) and the
Resource Conservation and Recovery Act
(RCRA).
In response to the positive reception of
the Chicago training workshop, the RAF
Technical Panel identified additional
training opportunities, open to all
interested Agency personnel, as a near-
term mechanism to enhance familiarity
by Program Offices and Regions with
population-level ecological risk assessment
concepts and methods. Additionally, such
training could facilitate identification
of issues requiring enhanced guidance.
Progress in this first action was made in
October 2006 when the RAF sponsored a
vendor-supplied "Population Ecological
Risk Assessment Training Workshop" in
Crystal City, Virginia.
2. Convene a technical workshop on
approaches for population-level ecological
risk assessment - As an action to be
completed in the mid-term, the Technical
Panel identified a multiple-day technical
discussion of the states of the science and
practice of population-level ecological
risk assessment to help inform the Agency
in decisions concerning development of
additional guidance supplemental to the
1998 Guidelines. Such an event would
bring together Agency and external experts
in population ecology and ecological
risk assessment in part to build upon the
previous discussions of this nature (e.g.,
Barnthouse et al. 2008). This workshop
was held in June 2008 and is the primary
subject of this report.
3. Develop best practices guidance for
population-level ecological risk assessment
- Development of best practices guidance
was envisioned by the Technical Panel
as a long-term (2-4 year) activity. The
specific projects and actions needed to
produce such guidance would be informed
by the workshop, by input received from
various training events and by other
developmental activities as needed. The
guidance would be developed by a cross-
program Technical Panel of the RAF to
supplement the 1998 Guidelines, and would
be responsive to the needs of Agency
Programs and Regions in their performance
of population-level assessments that inform
regulatory decisions. Technical Panel
recommendations for developing this
guidance are offered in Section 5 of this
report.
E.2 Workshop Objectives
The RAF Technical Panel organized this technical
workshop to achieve three specific objectives:
1. Identify the approaches, methods and
tools currently available for performing
population-level ecological risk assessment
in support of EPA programmatic and
regional decision making.
2. Identify the strengths, current limitations,
tradeoffs and outstanding research needs
associated with specific methods and
-------
tools currently available for performing
population-level ecological risk assessment
in support of EPA programmatic and
regional decision making.
3. Identify areas of need with respect to
development of written guidance for
performing population-level ecological risk
assessment to supplement the Guidelines
for Ecological Risk Assessment, and
the additional steps that can facilitate
development of such guidance.
These objectives were derived from the
recommendations of the 1999 colloquium and
the desires of practitioners, risk managers and
stakeholders for guidance in performing and
interpreting population-level ecological risk
assessment. The focus of the workshop was on the
technical matters of conducting, and the state-of-
the-science supporting, population-level ecological
risk assessment, and not on the policy issue of
levels of ecological organization appropriate for
environmental decision making. The individual
insights and opinions exspressed during the
workshop were intended to inform future RAF
projects and potential development of guidance.
The workshop itself did not produce guidance
or policy for any agency, nor did it develop
consensus opinions or group recommendations for
consideration by EPA.
The Workshop on Population-Level Ecological
Risk Assessment was convened on June 16-18,
2008 in Crystal City, Virginia. Thirty-two experts in
population ecology, ecological risk assessment and
risk management were invited from EPA Programs
and Regions, the U.S. Army Corps of Engineers,
academia and the private sector to reflect a range
of perspectives (Appendix A). A small number of
non-participating observers, exclusively from EPA,
were present on the first day of the workshop.
The workshop's format included both plenary
interactions and breakout group conversations
intended to facilitate information exchange. The
final workshop agenda is provided as Appendix
B. Plenary presentations and discussions during
the first day of the workshop were structured
to establish context and a common basis of
understanding by summarizing past efforts and
providing broad overviews from the perspectives of
EPA and other users of population risk information.
These presentations covered a wide range of
topics, and included descriptions of the needs and
approaches of individual Program Offices and
Regions, the perspectives of another federal agency
(the U.S. Army Corps of Engineers, a long-time
partner in population risk research), case study
illustrations of how population-level assessments
were used to inform decisions, and insights from
experiences in other countries. An evening poster
session on the first day of the workshop provided
the opportunity to explore case studies and
assessment approaches in greater detail. Summaries
of these presentations are offered in Section 2 of
this report.
Three primary approaches for obtaining
information about the population-level
consequences of human activity—observational,
empirical and modeling—provided the structure
of breakout groups charged primarily with
characterizing the states-of-science and practice of
techniques, methods and tools of each approach. In
this regard:
Observational approaches include those that
obtain data by monitoring the responses
of populations in the field to pollutants or
other anthropogenic stressors, and to natural
variables. The analysis of such data is
sometimes called "ecoepidemiology." These
approaches can be used to:
• Describe the condition of an
assessment population and determine
the causes of spatial and temporal
variation in population attributes
• Generate exposure-response
relationships directly from
observational data
• Provide data to parameterize process-
based models
• Provide data to test specific risk
hypotheses and the predictions of
process-based models
Experimental approaches involve controlled
experiments (e.g., toxicity tests) that expose
organisms or populations of organisms to
varying levels of chemical, physical and
biological agents to evaluate population
response. Experiments can be performed in a
laboratory, field or semi-field system. These
approaches can be used to:
• Derive understanding of population
responses directly from the data (e.g.,
-------
population growth rate, equilibrium
abundance)
• Provide data to parameterize process-
based models
• Provide data to test specific risk
hypotheses and the predictions of
process-based models
Modeling approaches involve application of
process-based population models to general and
specific risk problems to evaluate population
response to varying levels of chemical, physical
and biological agents, and to natural variables.
Process-based models are mathematical
constructs that estimate properties of biological
populations such as growth rate or time to
extinction, and are based on estimates of
underlying biological processes (such as
survival rates) and environmental change.
These approaches can be used to:
• Project or forecast population-level
consequences of changes in stressors
and other environmental conditions
associated with different management
scenarios
• Evaluate the population-level
consequences of changes in individual-
level attributes observed or measured
using observational and experimental
approaches
• Evaluate distributions of population
outcomes through time and across
space
• Inform the design of observational and
experimental approaches for assessing
population risk
The three breakout groups met throughout the
second day of the workshop to consider questions
relevant to workshop objectives (see Appendix C)
from the perspective of each individual's expertise.
Two breakout group leads, one from the Workshop
Steering Committee, the other invited from outside
of this committee, facilitated the discussions and
the expression of individual opinions. No attempts
were made to seek consensus among breakout
group members on any point or issue; rather, the
intention was to capture the diversity of expert
opinions and perspectives in each group relative
to their charge. Group membership consisted
primarily of experts with respect to the specific
approach to population-level ecological risk
assessment being considered. Two mechanisms
were used to help ensure a healthy level of cross-
fertilization in the discussions: 1) each group was
"seeded" with experts in the other two approaches,
and 2) the workshop chair, workshop facilitator and
RAF liaison each circulated among breakout groups
to communicate issues from the other groups. Each
group had a note taker to capture conversations.
The individual perspectives and opinions of
participants in the three breakout groups were
reported and summarized in plenary on the third
day of the workshop to address the workshop's
three primary objectives. Facilitated discussions
following the breakout group reports provided
yet another opportunity for the exchange of
perspectives and ideas.
E.3 Summary of Expert Opinions
The summaries provided in Section 3 communicate
the breadth of opinions and input expressed
by workshop attendees during breakout group
conversations. Section 4 communicates additional
observations, issues and suggestions that were
expressed during plenary discussions. Although
no attempts were made to seek consensus on any
particular issue or topic, certain commonalities
emerged over the course of breakout and plenary
interactions. Key opinions with respect to workshop
objectives are summarized here.
E.3.1 Experimental Approaches
Participants generally felt that the methods
employed to provide data for input to population-
level ecological risk assessments are sufficiently
well developed and informative to warrant
development of guidelines or best practices
descriptions. Several experimental methods are
available, and sometimes even standardized,
that can measure the responses of experimental
populations to stressors directly, or that provide
data that can be extrapolated to the population level
of biological organization. Even so, additional
design considerations might be required to help
ensure that key hypotheses regarding mechanisms
of effect and other important ecological processes
-------
can be evaluated as needed to inform environmental
decision making. In this regard, some experimental
designs likely have limited ability to incorporate
processes and interactions that can have important
population consequences, such as competition
and other forms of species interactions. Careful
planning during problem formulation of the
assessment will help to ensure use of experimental
designs and methods that provide the information
needed to quantify decision-relevant risk.
Because experiments inherently are abstractions
of nature and therefore cannot include all aspects
that might have ecological relevance, additional
research and development might be needed to
improve the value of experimental approaches to
population-level ecological risk assessment. For
example, the issue of cross-species extrapolation
was highlighted. Many species are not particularly
amenable to experimental manipulation, and
when assessment goals focus specifically on
populations of such species, their responses to
stressors will need to be extrapolated from those of
surrogates. Some progress could be made toward
resolving this issue by focusing upon mechanisms
of action and their ecological analogs, but there
likely will always continue to be meaningful
uncertainty whenever cross-species extrapolations
are required. Extrapolation of organism-level
measures to characterize risk to populations might
be less worrisome, because a variety of modeling
approaches are available that can accommodate
organism-level attributes to project population
dynamics. Even so, attention is needed to help
ensure that experimental data are collected in
the forms and temporal frames required by
extrapolation models.
Other areas of valuable research include
development of approaches and data that can link
certain types of measures—namely biomarkers
and organism dose concentrations—more directly
to the key demographic rates of reproduction
and survivorship that determine population
dynamics. In a similar vein, there might be
opportunities within the evolving technologies of
genomics and proteomics to develop approaches
that link data derived from these techniques to
population response. Advancements in this area
could produce efficiencies in the collection of
information for assessing population risk. Finally,
discussions emphasized the potential value to be
derived from combining experimental methods
(including more tightly coupled laboratory and
field experiments) with modeling and observations,
as these approaches provide complementary
and supplemental information about risks to
populations.
Several activities were identified that could support
communication of best practices. Included are case
study analyses, both comparing the informational
value of population-level measurement endpoints
versus organism-level measurement endpoints
when assessments include populations as
environmental values to be protected, and
evaluating the efficacy of population-level
assessments with respect to the outcomes of
decisions based upon them. Associated with this
was a sense that descriptions of experimental
designs that promote use of the resulting data in
modeling evaluations, and of how experimental
data can best be used in modeling applications,
should be developed to help focus experiments on
generating the most critical information needed.
Included were specific guidelines for performance
of life table response experiments (LTREs), bucket
tests and so on. Guidance about how experimental
results should be interpreted and communicated
with respect to population risk was also identified.
E.3.2 Observational Approaches
Workshop participants generally agreed that
observational methods are well established in
the fields of ecology and conservation biology,
and that approaches based upon them have
unique advantages in reflecting realism with
respect to population responses to stressors in
the environment. In this regard, information
obtained through direct observation reflects
the effects of multiple stressors and influences
of compensatory mechanisms (e.g., density
dependence), but the relative contributions of
various effects and processes usually are difficult
to tease apart. Additionally, the variability
inherent to natural systems could at times mask
detection of some important stressor effects.
Observational approaches were thought to be
-------
applicable to all tiers within a tiered assessment
protocol. Many noted, however, that the utility
of observational approaches might be limited
with respect to prospective assessments due to
imperfect transferability of study results beyond
the conditions and context within which they were
obtained. They also have limited value for helping
to evaluate decision alternatives, because the
information they produce reflects only the specific
circumstances in which they were conducted.
Data from observational studies can, however,
help to inform reassessments of past management
decisions. It was noted that new methods are
coming on line that can help to guide decisions
about the inferences that can be made using
observational data.
Developmental activities that were identified to
promote best practices included compilations of
case study examples of the use of observational
approaches to assess population risk, examples
of when such approaches failed to provide the
information needed to assess population risk, and
how observational studies influenced decisions.
Workshop participants highlighted the value
of catalogues and annotated descriptions of
available methods and observational data sets and
sources. Guidance in the form of decision trees
was suggested as being particularly helpful with
respect to assessment planning and interpretation
of observational study results. Participants also
noted that acceptance of the use of observational
approaches by decision makers could be facilitated
and enhanced through development of best
practices descriptions for effective communication.
E.3.3 Modeling Approaches
Contributors to the workshop expressed the
opinion that population models and the approaches
to deploy them within population-level risk
assessment are well established, and noted several
compilations of model descriptions and use
considerations in the recent literature. Opinion
was expressed that the stressors under evaluation,
and especially the decision context, influence
which models and approaches provide the most
valuable information. It generally was believed
that population models can be used to advantage
in any level within a tiered assessment protocol,
and that they are important integrators of data
and knowledge gained through observational and
experiment approaches. Important drawbacks
to modeling approaches, however, include the
skepticism often expressed by decision makers
about the degree of realism captured by models
and the accuracy of their outputs, and concerns
about assessment transparency when stakeholder
and decision maker understanding of modeling is
limited. A lively plenary discussion centered around
perceived inconsistencies in the level of acceptance
of population models relative to chemical fate and
transport models (of which acceptance is high).
Associated with this was continuation of the
ongoing debate centered around the meanings and
desirability of model verification, validation and
evaluation.
In spite of the generally high regard held by most
workshop participants for population models,
certain developmental issues were highlighted
as important. Among these was advancement
in coupling population models more directly
to exposure models, particularly with respect
to physiological-based dose-response models.
Additional exploration of modeling philosophy
and approaches addressing the effects of multiple
stressors would enhance model realism and
likely, accuracy. Issues associated with the form
and strength of density dependence as important
determinants of population response to stressor
exposure, although not directly ones of modeling
per se, might influence model realism and the
accuracy of assessment conclusions. Several
participants expressed the opinion that density
dependence might not be as important an issue as
some believe. Additional attention to developing
accessible implementation software and packages
also was highlighted as a need, although some
software is available commercially or as freeware.
A number of activities were identified that would
foster acceptance of good practices in the use of
modeling approaches in population-level ecological
risk assessment. Important among these were:
• Development of a decision framework for
model selection
• Development of best practices guidelines
for interpreting modeling results
• Identification of best practices for
-------
facilitating communication directly with
decision makers and stakeholders
• Documentation to guide design of
experimental and observational
studies performed in conjunction with
modeling approaches to ensure modeling
compatibility with data accessibility
• Development of guidelines for approaches
that extrapolate effects reflected in toxicity
data through time
E.3.4 Commonalities
Across Approaches
Several considerations expressed during the
workshop cut across assessment approaches,
and reflected the general sentiments of many of
the participants. Most importantly, in relation
to workshop objectives, was the sense that the
science underlying population-level ecological risk
assessment is sufficiently mature to support further
development of best practice guidelines. Although
the various approaches have perceived benefits and
limitations relative to different decision contexts,
and attention to certain developmental needs is
desirable, opportunities for applying existing
techniques to inform decisions were identified
within almost all of EPA's regulatory programs.
Participants often articulated the opinion that
the three assessment approaches should best be
treated as interdependent and complementary,
and that the power and value of population-level
ecological risk assessment as a decision-informing
tool are enhanced when approaches are used in
combination. Also expressed was the sentiment
that a primary advantage of focusing attention
more explicitly on measurement endpoints and
analysis techniques that address population
attributes directly is an assessment more relevant
to decisions involving protection of populations.
Most workshop participants promoted greater use
of population-level ecological risk assessment as a
tool to inform environmental decision making.
Documentation, communication and training
were felt to be components critical to credible
performance, advancement and acceptance of
population-level ecological risk assessment by
practitioners, decision makers and the public.
Important in this will be articulation of a
framework uniquely oriented toward planning,
implementing and interpreting results of
population-level ecological risk assessments.
This framework could include considerations
leading to selection of assessment approaches (i.e.,
combinations of experimental, observational and
modeling techniques) appropriate to the decision
and its context, potentially organized in the form
of a decision tree. Compilations and catalogues of
existing techniques, models, designs and data could
be linked to the decision tree to aid assessment
planning and performance. Programs could be
developed to help ensure that practitioners are
appropriately trained in relevant techniques and
models. Specific best practices guidelines would
help to direct interpretation of data and results,
focused on the decision they intend to inform.
These guidelines might summarize key aspects
of ecological theory and link to compilations of
case studies as illustrations of sound interpretation
approaches. Additional guidelines could support
communication of assessment results and their
meaning to the end-users of the assessment. And
throughout, materials and information should be
oriented toward or tailored to the unique decisions
and contexts of EPA's programs.
Several cross-cutting issues will require attention
if guidelines are to be developed. Key among
these are considerations associated with pragmatic
definitions of assessment population in various
decision contexts. Although reasonable approaches
to address this particular issue exist, definition of
the assessment population has been problematic
for Superfund and certain other programs.
Somewhat related to this are considerations
about spatial scale and context, and time horizons
appropriate to various management goals and
decisions. Attention also is needed for identifying
those measurement endpoints most relevant
to population assessment endpoints and the
nature of risks being evaluated. And finally, all
acknowledged that assessment populations do not
exist in isolation from other populations. Species
interactions can have important and substantial
influences on population performance and the
risks associated with anthropogenic stressors.
Some of the techniques explored during the
workshop (especially observational approaches)
reflect or accommodate species interactions more
-------
realistically than do others. Even so, the importance
of species interactions to assessment results, and
the uncertainties created when species interactions
are ignored, will require careful consideration as
the science of population-level ecological risk
assessment is employed.
E.4 Technical Panel
Recommendations
The resounding sentiment of the experts
assembled in this workshop was that EPA and
ecological risk assessment practitioners alike
would benefit from guidelines or best practices
documentation concerning population-level
ecological risk assessment. The science underlying
such assessments is sufficiently well developed
that guidelines could be created to promote
best practices with the understanding that such
guidelines would be updated on a regular basis as
the state-of-the-science and practice of population
level ecological risk assessment improves over
time. Based in large part on the opinions of these
experts, but also based on our individual and
collective professional perspectives, the RAF
Technical Panel recommends that the Forum
proceed with an effort to develop best practices
guidelines for population-level ecological risk
assessment. This section describes some of the
options and outputs that could be pursued in
such a project. Suggestions are offered only
generally about how best to accomplish individual
activities and the overall project. It is suggested,
however, that a phased implementation with
multiple intermediate products is likely to be most
successful.
The Technical Panel recommends a phased
approach to producing guidelines. Initial issue-
oriented white papers and summaries would help
to document the current states of science and
practice of technologies supporting population-
level ecological risk assessment, and could
suggest how EPA programs would benefit from
a more explicit focus on risk to populations.
Opinion statements would help to visualize how
regulatory programs could use information directly
communicating population risk to facilitate
understanding of the advantages and limitations
with respect to program mandates. Supporting
white papers could summarize EPA program policy
with respect to management goals, and how a
more explicit focus on population-level measures
could support the decisions to meet those goals.
Additional opinion papers, summarizing current
knowledge, could focus on inferences drawn about
risks to populations, and on projecting future
Agency practices that would be more inclusive of
population risk.
Development of best practice guidelines likely
will require directed conversations involving
ecologists, practitioners and users of assessment
results. Workshops that enable such interactions
likely will be important steps to developing best
practices guidelines. Topics for deliberation include
detailed evaluations of methods, best approaches
for combining methods in relationship to decision
contexts, and the decision criteria and processes
that could lead to a planning and implementation
framework specifically for population-level
ecological risk assessment. Equally important is
development of guidelines for interpreting results
and assessment outcomes. Such guidelines could
be organized by assessment endpoint attribute, and
could describe a nested hierarchy of considerations
and conclusions for interpreting lines of evidence
generated by multiple assessment approaches.
Retrospective analyses of cases in which risks
to populations were assessed would provide
both examples for future assessments, and
opportunities to evaluate the efficacy of various
approaches. Either as part of this or as a separate
effort, considerations of the informative value to
environmental decision making of population-level
ecological risk assessment and the approaches used
would provide additional insights supporting best
practices guidelines. Case study evaluations could
be commissioned from groups of experts, or could
be conducted in focused workshop settings.
In a related vein, assembly of information
describing the methods, models and data sources
would help to improve the accessibility of these
tools to risk assessment practitioners. Compilations
could include annotations describing acknowledged
advantages and limitations of methods and
models with respect to various risk problems,
environmental settings, stressors and decision
-------
contexts. Catalogues pointing to key sources of
toxicological data, demographic and life history
information and extrapolation relationships would
facilitate access to critical information and would
help to promote the quality of future assessments.
Attention to education, communication and
outreach will be critical to the success of an RAF
project that develops best practices guidelines
for population-level ecological risk assessment.
Although past Forum efforts to provide general
training in this area have been quite successful,
further development of training modules to focus
specifically on key topics and methods likely would
improve their value to practitioners. Modules
communicating best practices for using population
risk information would support understanding,
and perhaps further adoption, of population-level
ecological risk assessment by EPA programs.
Recognition of the roles and contributions of
stakeholder groups and the general public in
environmental decision making will be important as
education and outreach materials are developed.
The Technical Panel believes that each of the
activities described above will be important as the
project moves forward. It is also suggested that
a successful approach to supplementing existing
RAF guidelines will be to release products in a
phased manner as they are developed, rather than
to focus solely on a single major contribution at
the conclusion of the project. Such an approach
is likely to have several advantages, namely a
more rapid release of valuable information and
guidelines, an enhanced ability to incorporate
advancements in science and practice through time,
and a more timely and flexible responsiveness to
evolving Forum and Agency priorities.
-------
-------
1.0
Introduction
1.1 Background and Context
The U.S. Environmental Protection Agency (EPA,
or the Agency) has adopted risk assessment as a
primary tool supporting environmental decision
making. To help maximize the value of ecological
risk assessment (ERA) to Agency programs,
EPA's Risk Assessment Forum (RAF) produced
its landmark Guidelines for Ecological Risk
Assessment's. EPA 1998) which describe a
general strategy and framework for planning,
executing and interpreting ERAs. The 1998
Guidelines recommend a planning dialogue among
risk managers, risk assessors and other interested
parties as a critical first step toward initiating
an ERA. This dialogue is intended to produce
agreement on and understanding of management
goals and the types of decisions that the assessment
will support. It also establishes the scope,
complexity and focus of the risk assessment to be
conducted. Depending upon their context, planning
agreements might be established as a matter of
policy, or might be made on an ad hoc basis to
inform situation-specific management decisions.
Regardless, these agreements lead to selection,
during the problem formulation phase of the
assessment, of the endpoints to be evaluated during
the risk assessment. Assessment endpoints are
selected to describe valued ecological entities and
their attributes at levels of ecological organization
that are relevant and applicable to the decisions
being made, and in combination can encompass
single or multiple levels of ecological organization
in a single assessment.
The choice of ecological organization levels to be
evaluated in regulatory ERAs can be a challenging
one. Although the enabling legislation of many
of EPA's programs either explicitly or implicitly
identify protection of ecological populations
as management goals, most ERAs conducted
for chemicals by EPA, and indeed by most
organizations worldwide, focus on organism-level
entities and attributes (e.g., rainbow trout survival,
growth or reproduction) as assessment endpoints.
These endpoints are practical because they often
can be estimated through toxicological testing
and other means, and are expedient because they
are commonly presumed to provide protection of
population-level attributes (e.g., abundance and
persistence). Further, methods and practice are
well established for assessing risk to organism-
level assessment endpoints, but documentation of
consensus methods for population-level ecological
risk assessment is lacking. Consequently, risk to
populations has only occasionally been evaluated
directly by EPA (e.g., trout populations of
Adirondack lakes in the National Acid Precipitation
Assessment). This situation results from several
factors affecting assessment planning, including
the perceived relationships between assessment
endpoints and environmental management goals,
historical precedence, and importantly, the lack of
recognized consensus and guidance3 about how
such assessments should be performed.
Public comment prior to publication by EPA of
the Guidelines in 1998 indicated a desire by some
stakeholders for additional guidance on assessing
effects at the population, community and ecosystem
levels of ecological organization, and more
substantive guidance on protecting populations
of animals and plants. Also at that time, a survey
of EPA ecological risk assessors ranked effects
at higher levels of biological organization, along
with assessment endpoints and measures of effect,
as having the highest priority for development of
additional guidance. The call for guidance has been
repeated in recent international efforts addressing
population-level ecological risk assessment (e.g.,
Barnthouse et al. 2008, Forbes et al. 2009). In
particular, Barnthouse et al. (2008) recommend
development of guidance to assist risk assessors,
risk managers and stakeholders in selecting,
applying, interpreting and communicating
throughout this report, the terms "guidance" and
"best practice description" are used generically and
interchangeably to mean the documentation of techni-
cally credible and generally accepted approaches and
methods. No policy implications are intended through
their use.
-------
population-level ecological risk assessment
procedures and analysis tools to cover a range of
environmental management contexts.
The RAF conducted a colloquium in 1999 to
help identify the nature and scope of projects
that would advance development of ecological
risk assessment guidance in three broad areas: 1)
effects at higher levels of biological organization,
including landscape-level effects;
emerging from discussion of the first area
was a focus on methods for assessing risks to
populations and interpreting the results obtained
by those methods. During the colloquium,
the needs of "on-the-ground" risk assessors
and risk managers led to identification of
several developmental projects (organized
by assessment phase in Table 1) related to
population-level ecological risk assessment and
approaches to implement
Table 1. Developmental projects recommended for population-level ecological risk assessment from an earlier
RAF colloquium
Assessment Phase
Focus of Recommended Project
Approach*
Problem Formulation
Analysis
Risk Characterization
Selection of assessment endpoints - sensitivity of response
vs. level of organization
Conceptual model development
• consideration of scalar issues
• consideration of multiple stressors
• incorporation of multiple assessment endpoints
• delineation of scope, approach and boundaries
Ecological models
• basic principles underlying development and use
• rules for model selection and application
• procedures for model evaluation
• integration with other assessment tools
Extrapolation
• across levels of ecological organization
• through time and across space
Interpretation of assessment results
• relative to unstressed conditions
• in context of natural variation
• significance of changes in population attributes
1. Workshop
2. Guidelines
No recommendation
1. Annotated
bibliography
2. Application guidelines
1. Issue paper
No recommendation
* For some projects, multiple steps in approach were recommended. For others, no specific approach was
recommended.
-------
2) assessment endpoints and measures of effect4;
and 3) risk characterization techniques. A broad
theme them. These approaches reflected a number of
considerations, including the perceived state-of-the
science, the types of intermediate products deemed
useful and the needs of the Agency.
In 2005, the RAF formed a Technical Panel
to explore a number of issues associated with
population-level ecological risk assessment.
Consisting of representatives of EPA Program
Offices, Regions and the Office of Research and
Development (ORD), this working group initially
identified three broad actions intended to enhance
the Agency's understanding of approaches for
assessing risks to populations. These actions are:
1. Expand training in population-level
ecological risk assessment - Since the
publication of the Guidelines, ORD has
received a number of requests for formal
training and educational exchanges
addressing topics related to population-
level ecological risk assessment (see text
box). To help meet this need, ORD's Office
of Science Policy sponsored the vendor-
supplied "Population Modeling Training
Workshop," conducted at Region 5's
offices in Chicago, Illinois in late 2004.
This training was coordinated through the
Ecological Risk Assessment Forum (ERAF)
and was attended primarily by regional
risk assessors who support hazardous
waste assessments under Comprehensive
Environmental Response, Compensation,
and Liability Act (CERCLA) and the
Resource Conservation and Recovery Act
(RCRA).
In response to the positive reception of
the Chicago training workshop, the RAF
Technical Panel identified additional
training opportunities, open to all
interested Agency personnel, as a near-
term mechanism to enhance familiarity
by Program Offices and Regions with
population-level ecological risk assessment
4 The RAF published guidance for Generic Ecological As-
sessment Endpoints (GEAEs) for Ecological Risk Assess-
ment (EPA/630/P-02/004f) in 2003 (http://www.epa.gov/
raf/publications/geae.htm).
Examples of Previous Requests for Training
and Educational Exchange
2001 - W. Munns and M. Mitro presented an
overview entitled "Population Modeling to Estimate
Risks of Chemical and Other Stressors to Wildlife
and Aquatic Populations" as part of the ORD/OPPTS
Seminar Series in Washington, DC.
2002 - EPA's Regional Risk Assessors requested
"Assessing Risks to Populations at Superfund
Sites - Characterizing Effects on Populations" (W.
Munns and M. Mitro) at their 2002 annual meeting in
Philadelphia, PA.
2003 - The Ecological Risk Assessment Forum
(ERAF) requested an introductory training course
entitled "Assessing Risks to Populations - Population
Modeling" (W. Munns and A. Fairbrother) in
Washington, DC.
2004 - Region 1 's Biological Technical Advisory
Committee requested a presentation entitled
"Population-Level Risk Assessment" (W. Munns) in
Boston, MA.
2004 - A request from Region 9 through the ERAF
to ORD's Ecological Risk Assessment Support
Center (ERASC) lead to development of "Assessing
Risks to Populations at Superfund and RCRA Sites -
Characterizing Effects on Populations" (Munns and
Mitro 2004), which describes population concepts
and approaches for evaluating risk to populations at
hazardous waste sites.
2004 - The Office of Pesticide Programs requested a
presentation entitled "Population Modeling to Support
Ecological Risk Assessment: An Example Using
Mysid Toxicity Test Data" (J. Grear) in Crystal City,
VA.
2004 - The Office of Water requested a presentation
entitled "Population Modeling to Support Ecological
Risk Assessment: An Example Using Mysid Toxicity
Test Data" (J. Grear) in Washington, DC.
2004 - The ERAF requested a presentation entitled
"Assessing Population-Level Risk at Hazardous
Waste Sites" (W. Munns) at the annual Regional Risk
Assessors meeting in Boston, MA.
2005 - The Office of Pesticides Programs requested a
presentation entitled, "A Primer on Matrix Population
Modeling" (G. Thursby) in Arlington, VA.
-------
concepts and methods. Additionally, such
training could facilitate identification
of issues requiring enhanced guidance.
Progress in this first action was made in
October 2006 when the RAF sponsored a
vendor-supplied "Population Ecological
Risk Assessment Training Workshop" in
Crystal City, Virginia.
2. Convene a technical workshop on
approaches for population-level ecological
risk assessment - As an action to be
completed in the mid-term, the Technical
Panel identified a multiple-day technical
discussion of the states of the science and
practice of population-level ecological
risk assessment to help inform the Agency
in decisions concerning development of
additional guidance supplemental to the
1998 Guidelines. Such an event would
bring together Agency and external experts
in population ecology and ecological
risk assessment in part to build upon the
previous discussions of this nature (e.g.,
Barnthouse et al. 2008). This workshop
was held in June 2008 and is the primary
subject of this report.
3. Develop best practices guidance for
population-level ecological risk assessment
- Development of best practices guidance
was envisioned by the Technical Panel
as a long-term (2-4 year) activity. The
specific projects and actions needed to
produce such guidance would be informed
by the workshop, by input received from
various training events and by other
developmental activities as needed. The
guidance would be developed by a cross-
program Technical Panel of the RAF to
supplement the 1998 Guidelines, and would
be responsive to the needs of Agency
Programs and Regions in their performance
of population-level assessments that inform
regulatory decisions. Technical Panel
recommendations for developing this
guidance are offered in Section 5 of this
report.
1.2 Workshop Objectives
The RAF Technical Panel organized this technical
workshop to achieve three specific objectives:
1. Identify the approaches, methods and
tools currently available for performing
population-level ecological risk assessment
in support of EPA programmatic and
regional decision making.
2. Identify the strengths, current limitations,
tradeoffs and outstanding research needs
associated with specific methods and
tools currently available for performing
population-level ecological risk assessment
in support of EPA programmatic and
regional decision making.
3. Identify areas of need with respect to
development of written guidance for
performing population-level ecological risk
assessment to supplement the Guidelines
for Ecological Risk Assessment, and
the additional steps that can facilitate
development of such guidance.
These objectives derive from the recommendations
of the 1999 colloquium and the desires of
practitioners, risk managers and stakeholders for
guidance in performing and interpreting population-
level ecological risk assessment. The focus of
the workshop was on the technical matters of
conducting, and state-of-the-science supporting,
population-level ecological risk assessment, and
not on the policy issue of levels of ecological
organization appropriate for environmental decision
making. The individual insights and opinions
expressed during the workshop were intended
to inform future RAF projects and potential
development of guidance. The workshop itself did
not produce guidance or policy for any agency,
nor did it develop consensus opinions or group
recommendations for consideration by EPA.
1.3 Workshop Format
The Workshop on Population-Level Ecological
Risk Assessment was convened on June 16-18,
2008 in Crystal City, VA. Thirty-two experts in
population ecology, ecological risk assessment and
risk management were invited from EPA Programs
and Regions, the U.S. Army Corps of Engineers,
-------
academia and the private sector to reflect a range
of perspectives (Appendix A). A small number of
non-participating observers, exclusively from EPA,
were present on the first day of the workshop.
The workshop's format included both plenary
interactions and breakout group conversations
intended to facilitate information exchange. The
final workshop agenda is provided as Appendix
B. Plenary presentations and discussions during
the first day of the workshop were structured
to establish context and a common basis of
understanding by summarizing past efforts and
providing broad overviews from the perspectives of
EPA and other users of population risk information.
These presentations covered a wide range of
topics, and included descriptions of the needs and
approaches of individual Program Offices and
Regions, the perspectives of another federal agency
(the U.S. Army Corps of Engineers, a long-time
partner in population risk research), case study
illustrations of how population-level assessments
were used to inform decisions, and insights from
experiences in other countries. An evening poster
session on the first day of the workshop provided
the opportunity to explore case studies and
assessment approaches in greater detail. Summaries
of these presentations are offered in Section 2 of
this report.
Three primary approaches for obtaining
information about the population-level
consequences of human activity—observational,
experimental and modeling—provided the
structure of breakout groups charged primarily with
characterizing the states-of-science and practice of
techniques, methods and tools of each approach. In
this regard:
Observational approaches include those that
obtain data by monitoring the responses
of populations in the field to pollutants or
other anthropogenic stressors, and to natural
variables. The analysis of such data is
sometimes called "ecoepidemiology." These
approaches can be used to:
• Describe the condition of an
assessment population and determine
the causes of spatial and temporal
variation in population attributes
• Generate exposure-response
relationships directly from
observational data
• Provide data to parameterize process-
based models
• Provide data to test specific risk
hypotheses and the predictions of
process-based models
Experimental approaches involve controlled
experiments (like toxicity tests) that expose
organisms or populations of organisms to
varying levels of chemical, physical and
biological agents to evaluate population
response. Experiments can be performed in a
laboratory, field or semi-field system. These
approaches can be used to:
• Derive understanding of population
responses directly from the data (e.g.,
population growth rate, equilibrium
abundance)
• Provide data to parameterize process-
based models
• Provide data to test specific risk
hypotheses and the predictions of
process-based models
Modeling approaches involve application of
process-based population models to general and
specific risk problems to evaluate population
response to varying levels of chemical, physical
and biological agents, and to natural variables.
Process-based models are mathematical
constructs that estimate properties of biological
populations such as growth rate or time to
extinction, and are based on estimates of
underlying biological processes (such as
survival rates) and environmental change.
These approaches can be used to:
• Project or forecast population-level
consequences of changes in stressors
and other environmental conditions
modeled for different management
scenarios
• Evaluate the population-level
consequences of changes in individual-
level attributes observed or measured
using observational and experimental
approaches
• Evaluate distributions of population
outcomes through time and across
space
-------
• Inform the design of observational and
experimental approaches for assessing
population risk
The three breakout groups met throughout the
second day of the workshop to consider questions
relevant to workshop objectives (see Appendix
C) from the perspective of the group's focus. Two
breakout group leads, one from the Workshop
Steering Committee, the other invited from outside
of this committee, facilitated the discussions and
the expression of individual opinions. No attempts
were made to seek consensus among breakout group
members on any point or issue; rather, the intention
was to capture the diversity of expert opinions and
perspectives in each group relative to their charge.
Group membership consisted primarily of experts
with respect to the specific approach to population-
level ecological risk assessment being considered.
Two mechanisms were used to help ensure a healthy
level of cross-fertilization in the discussions: 1) each
group was "seeded" with experts in the other two
approaches, and 2) the workshop chair, workshop
facilitator and RAF liaison each circulated among
breakout groups to communicate issues from the
other groups. Each group had a note taker to capture
conversations.
The individual perspectives and opinions of
participants in the three breakout groups were
reported and summarized in plenary on the third day
of the workshop to address the workshop's three
primary objectives. Facilitated discussions following
the breakout group reports provided yet another
opportunity for the exchange of perspectives and
ideas.
-------
2.0
Summary of Opening
Remarks and Presentations
The following descriptions summarize briefly
the plenary presentations of the first day of the
workshop. Copies of all presentation material are
available at http://www.epa.gov/raf/population_
era_workshop .htm.
2.1 Lee Hofmann,
EPA Risk Assessment Forum,
Executive Director
Dr. Lee Hofmann opened the workshop by
welcoming attendees on behalf of EPA and the
RAF. She provided background information
regarding the RAF, including its organizational
structure, recent successes and current projects.
She briefly discussed the current state of ecological
risk assessment and the future directions of ERA,
including a growing emphasis on population-
level endpoints. Dr. Hofmann also explained the
workshop goals, processes and desired outcomes.
The goal of the workshop was to assess the current
state-of-the-science with respect to population-level
ecological risk assessment and to solicit individual
opinions from workshop attendees regarding the
maturity of population assessment methods and
tools.
2.2 Wayne Munns,
EPA Office of Research
and Development, Workshop Chair
Dr. Wayne Munns welcomed and thanked the
workshop attendees and observers, and provided
additional detail concerning workshop objectives,
approach and structure. He observed the current
lack of consensus guidance regarding approaches
for assessing risk to populations, and noted
that although such assessments are becoming
more commonplace, they are ad hoc and often
contentious in the absence of such guidance. Dr.
Munns described the workshop objectives: 1) to
identify and discuss approaches, methods and
tools available to population-level ecological
risk assessment, and in doing so to identify their
strengths, limitations and tradeoffs in use; 2) to
identify technical needs with respect to developing
guidance; and 3) to identify additional steps needed
to facilitate the development of guidance for
planning, conducting and interpreting the results
of population-level ecological risk assessment. He
described the opening plenary interaction as an
opportunity to review background information, to
describe perspectives and needs for information
regarding risk to populations, to illustrate case
studies and to identify issues that could be
addressed in the breakout group discussions of
the second day. He emphasized that the intention
of the workshop was to define the maturity of
science underpinning population-level ecological
risk assessment from a technical standpoint by
seeking the individual input of attendees, and that
the discussions would avoid recommendations of
specific approaches and issues of policy. Dr. Munns
concluded his overview by describing the products
expected from the workshop and their intended
uses.
In a presentation immediately following, Dr.
Munns described three key precursor activities
that helped to establish the context for the current
workshop. In the first, the RAF had sponsored
a colloquium shortly after publication of the
1998 Guidelines to inform future RAF projects
regarding selection of assessment endpoints, effects
at higher levels of biological organization and
risk characterization. The second was a Society
of Environmental Toxicology and Chemistry
(SETAC) "Pellston Workshop on Population-Level
Ecological Risk Assessment," held in Roskilde,
Denmark in 2003 (Barnthouse et al. 2008). That
workshop focused on advancing the acceptance
and practice of population-level ecological risk
assessment informing environmental management.
The objectives of that workshop were to evaluate
policy contexts for assessments, explore technical
-------
issues and opportunities, identify appropriate
empirical and modeling methods within varying
decision contexts, and to develop a framework for
conducting population-level ERA to inform risk
management decisions. The conclusions drawn
from the workshop included: 1) the science is
sufficiently mature to develop guidance; 2) specific
guidance should be developed for use of models
and data within a tiered assessment format; 3)
training programs should be developed; and 4)
acceptable levels of population risk in different
management contexts should be articulated. The
third activity, the SETAC "LEMTOX Workshop on
Ecological Models in Support of Regulatory Risk
Assessments of Pesticides," focused on the role of
population models to support pesticide registration
(primarily) in the European Union (Forbes et al.
2009). The conclusions identified in this 2007
workshop in Germany included the need to develop
guidance for good modeling practice. This group
recommended that aspects of such guidance should
focus on model development and evaluation,
documentation and communication, and analysis
and interpretation. Case studies were encouraged
to explore the value added to pesticide registration
decisions by using models in the assessment
process. Dr. Munns concluded that these precursor
activities positioned the current workshop to meet
its objectives.
2.3 Charles Delos,
EPA Office of Water
Dr. Charles Delos offered perspectives of the
Office of Science and Technology (OST) of EPA's
Office of Water in a presentation entitled "Is There
Potential for Using Population Modeling in Aquatic
Life Criteria Program?" He described the intention
of the aquatic life criteria and standards to define
biological goals in terms of community protection,
and to protect populations as opposed to individuals
within populations. Dr. Delos reflected on
considerations by OST and the Agency's Aquatic
Life Criteria Guidelines Committee over the past
15 years about how to incorporate population
modeling into criteria development, and indicated
that population modeling supported derivation of
the saltwater dissolved oxygen criterion in 2000.
Dr. Delos then described a case study addressing
time-variable exposures using population modeling.
This study involved a) a kinetic toxicity model
to translate between constant exposures used in
laboratory tests and the continuously varying
concentrations that occur in the field, and b) a
stage structured population model to extrapolate
measured effects on test organism survival and
reproduction to reductions in long-term population
abundance and growth rate. Population models took
both density-dependent and density-independent
forms, and results were compared. Dr. Delos
also described use of population models to
address assumptions used in criteria development
concerning the relative influences of various
demographic rates (namely, reproduction and
survivorship) on population abundance and growth.
He concluded his presentation with the observation
that overall the water quality criteria program
cannot be said to warmly welcome the additional
complexity introduced by population modeling.
2.4 Edward Odenkirchen,
EPA Office of Pesticide Programs
Dr. Edward Odenkirchen provided insights of
EPA's Office of Pesticide Programs (OPP) in
a presentation entitled "Population Modeling
in Ecological Risk Assessment - Regulatory
Perspective." He emphasized that the regulatory
context for incorporating population-level risk into
pesticide registration decisions is well established
in the United States by law and policy. Currently,
however, the program focuses on organism-
level attributes (e.g., survival, fecundity and
growth) assuming these to provide insight about
risks at higher levels of biological organization
(e.g., populations). He stated that for many risk
management decisions, these endpoints and their
inferences about population-level effects are
sufficient to inform the decision. Dr. Odenkirchen
went on to describe the potential benefits of
population modeling to his office's regulatory
process in the context of assessment tiering.
Such benefits include providing interpretation of
screening-level assessment results, supporting
refinement of problem formulation for future
assessments, allowing consideration of temporal
and spatial variability, supporting evaluation of
the consequences of pesticide exposure to species
of special concern (e.g., threatened or endangered
species) and supporting description of pesticide
-------
risks and benefits in common units. He then
outlined the requirements of population models for
use in OPP regulatory programs to include the use
of existing effects data sets (as provided through
the registration process), and their compatibility
with existing organism-level risk assessment tools.
Additionally, models should be adapted from other
programs or the open scientific literature when
practical, with proprietary models being avoided.
Communication of model assumptions, uncertainties
and limitations should be made explicit, and a
number of quality assurance requirements must be
met.
Dr. Odenkirchen also expressed the desire that
model architecture permit advancement of
model complexity and realism, in part to avoid
proliferation of tools across levels of assessment
refinement. Some challenges faced in use of
population models in pesticide regulation include
balancing model simplicity, realism and portability
across risk problems, and ensuring acceptable levels
of output uncertainty. Dr. Odenkirchen concluded
his presentation by describing past and ongoing
efforts by OPP and ORD to incorporate population-
level risk assessment methods into OPP's
refined risk assessment processes. These include
developmental activities to extract key demographic
information from avian reproduction tests to support
population model parameterization, construction
and evaluation of demographic models for aquatic
invertebrates and agricultural birds, and refinement
of a spatially-explicit population model for potential
use in assessing risks to bird populations in agro-
ecosystems.
2.5 David Charters,
EPA Office of Solid Waste and
Emergency Response
Dr. David Charters presented perspectives from
EPA's Superfund Program. He described the roles
of ecological risk assessment in this program as:
1) identifying and characterizing the current and
potential environmental threats of hazardous waste
spills at sites; 2) evaluating the ecological impacts
of alternate remedial strategies; and 3) informing
identification of cleanup goals for the remedy
selected that will protect natural resources. By
Office of Solid Waste and Emergency Response
(OSWER) directive, ERAs in the Superfund
Program utilize organism-level data to postulate
risks to populations and communities occurring
in specific habitats at hazardous waste sites. Dr.
Charters elaborated that the Superfund Program
extrapolates toxicity information (including
benchmarks) to potential impacts on site populations
based upon causal relationships, and that it is not
necessary to observe adverse effects onsite to make
determinations about ecological risk. Site cleanup
goals frequently are based on no or lowest observed
adverse effect levels as determined from toxicity
testing, or on toxicological benchmarks established
more generally. Dr. Charters went on to describe
some of the perceived challenges attendant to
assessing population risk at hazardous waste sites,
including issues associated with data collection and
the time frames allotted for Remedial Investigations/
Feasibility Studies (RI/FS), the need to associate
effects with hazardous substance releases and issues
surrounding reference comparisons. Jokingly,
Dr. Charters noted that Superfund ERAs "are
probably weakest in the terrestrial and aquatic
areas," implying substantial opportunity for
contributions by population-level risk science
to the program's assessment of ecological risk
at sites. He described useful contributions to
include short-term population studies that can be
completed in two years or less (consistent with the
RI/FS expectations), population metrics that are
useful for developing numerical cleanup goals,
methods to extrapolate from organism-level effects
to population-level response and insights into
problems associated with definition of and risk to
assessment populations when the organisms onsite
are part of more broadly distributed populations.
Dr. Charters concluded his presentation by offering
insights about how incorporation of population-level
ERA approaches and methods into the Superfund
Program could be facilitated.
2.6 Bruce Duncan, EPA Region 10
Dr. Bruce Duncan offered additional insights
relative to hazardous waste site assessments and
states' water quality standards approval processes
in a presentation entitled "Regional Perspective:
Population-Level ERA." As a regional risk assessor,
he expressed a goal common to both programs of
evaluating population-level effects associated with
chemical stress. He described the needs relative to
this goal to include approaches for evaluating risk
directly to populations, and those for extrapolating
-------
population-level effects from information obtained
at lower levels of biological organization (e.g.,
the organism level). Dr. Duncan elaborated on
several issues related to assessing population
risk, including those associated with definition
of assessment populations relative to the needs
of the decision being informed, identification of
population attributes best suited to the risk problem,
interpretation of assessment results in terms of
their ecological significance and the uncertainties
attendant to the assessment. Dr. Duncan concluded
his presentation by expressing his desire to leave
the workshop with specific ideas for facilitating
use of population-level ecological risk assessment
by regional programs, and with insights into
approaches to address some of the issues he had
described earlier in his talk.
2.7 Steve Newbold,
EPA Office of Policy,
Economics and Innovation
Dr. Steve Newbold provided a case study of the use
of population-level assessment to support Agency
rule making in a presentation entitled "Population
Modeling in Economic Analysis." The case study
he described supported benefit-cost analysis of the
Clean Water Act Section 316(b) rule developed
by the Office of Water. Dr. Newbold introduced
this study by linking some ecosystem services
to population phenomena, and suggested that
economic valuation of the ecological benefits of an
Agency action often will require population impacts
as inputs. He offered that economic analyses can
inform selection of assessment endpoints, and that
improvements in risk assessment practices should
also help to improve the benefits assessments
conducted by the Agency. Dr. Newbold then
described Section 316(b) as requiring application of
best technology to various aspects of power plant
cooling water intake design and construction to
minimize impingement and entrainment (I&E) of
aquatic organisms, and provided the context for this
case study. The benefits of the rule to be quantified
included expected increases in commercial and
recreational fish harvests. A fisheries yield model
was developed originally to evaluate fish biomass
foregone through I&E losses, using simplifying
assumptions of density-independence and
constancy of key model parameters. By allowing
feedbacks in the construction of a scalar population
dynamics model, Dr. Newbold demonstrated that
ignoring density-dependence does not always lead
to conservative estimates of risk.
2.8 Todd Bridges,
U.S. Army Corps of Engineers
Dr. Todd Bridges described efforts of the U.S.
Army Corps of Engineers (USAGE) Engineer
Research and Development Center, a long-
standing partner with ORD in development of
population-level ecological risk assessment
methods, in a presentation entitled "The Relevance
of Populations to USAGE." He introduced the
mission of USAGE to include management of
navigational dredging, hydropower and reservoir
management, ecosystem restoration and invasive
species management, and provided details about
each relevant to the workshop. In particular, Dr.
Bridges presented case studies in the development
of demographic models to extrapolate toxicity test
data to population-level effects to support dredged
material assessments responsive to the requirements
of the Marine Protection, Research and Sanctuaries
Act and Clean Water Act, in addition to research
to develop a spatially-explicit exposure model
for fish potentially utilizing a historic aquatic
disposal site to support selection of remediation
(capping) material. He next offered a case study of
population-level impacts on fish of hydroelectric
dam entrainment mortality which used a stochastic
demographic modeling approach with density
dependence. Dr. Bridges followed this with
descriptions of ecosystem restoration case studies
involving metapopulation models, population
viability analysis and habitat-based modeling. He
described the population-relevant issues pertaining
to invasive species management to include
quantifying the propensity for and conditions
of species invasions, predicting spread and
developing effective control strategies. Dr. Bridges
concluded his presentation with the articulation
of several issues to consider in population-level
ecological risk assessment, including: 1) ensuring
assessment relevance to the decision being made;
2) quantifying and using information about
uncertainty and establishing confidence in the
use of population models; 3) distinguishing the
influences of multiple factors on assessment results;
-------
4) defining the temporal limits of population
projections; 5) considering spatial aspects and
reflecting behavior and movement appropriately;
and 6) using descriptions of synthetic populations
as analogs of real populations to simplify risk
assumptions and characterizations.
2.9 Jill Awkerman,
EPA Office of Research
and Development
Dr. Jill Awkerman offered a case study from the
field of conservation biology in a presentation
entitled "Risks of Fishery Mortality to a Seabird
Population and Conservation Implications,"
focusing on the risks of fishery-induced mortality
of waved albatross (Phoebastria irrorata) in
South America. She explained that while many
albatross populations are declining in part because
of incidental longline bycatch, intentional waved
albatross capture in artisanal fisheries was reported
at an alarming rate. Dr. Awkerman described
parameterization of a three-stage stochastic matrix
model to evaluate albatross population growth
using current estimates of survival and fecundity
and discussed extinction risk under different
assumptions of additional mortality. She also
presented analyses of factors potentially influencing
differential mortality between sexes including
chick survival, foraging behavior and susceptibility
to fisheries capture. The results of these analyses
prompted creation of an Action Plan for Waved
Albatross conservation.
2.10 Richard Sibly,
University of Reading, U.K.
Dr. Richard Sibly, a population ecologist who
has supported numerous population-level
ERA initiatives in the European Union (EU)
and internationally, provided his perspectives
on selected issues regarding the state-of-the-
science. He noted that compensatory processes
within populations, and population dynamics
in heterogeneous environments, are two issues
requiring additional attention. Dr. Sibly went on
to describe novel research from his laboratory that
has addressed variation in carrying capacity as
influenced by environmental stressors, illustrating
with a contour plot the relationship of population
growth rate to experimental chemical concentration
and initial population density. He used a similar
approach to explore the natural distribution of
a daphnid by characterizing its ecological niche
in terms of pH and calcium concentration. Dr.
Sibly touched briefly on some potential roles that
microarray technology might play in helping to
quantify demographic rates of reproduction and
survivorship, and concluded his presentation with
highlights from a workshop held in 2004 in York,
England. That workshop focused on approaches to
assess risk to populations of birds and mammals
associated with pesticide use. Dr. Sibly described
an agent-based modeling approach that he felt holds
great promise for population-level assessments,
indicating that it illustrates that population
dynamics emerge as a result of local interactions
between organisms and their landscapes.
-------
-------
3.0
Breakout Group Reports
Breakout groups reflecting different approaches
to population-level ecological risk assessment—
observational, experimental and modeling—were
asked to address three broad questions from the
perspective of each group's analytical approach and
set of tools:
1. What specific approaches, methods
and tools are available currently for
performing population-level ecological
risk assessment? To what types of
environmental decisions, risk problems and
environmental situations do they apply?
2. Identify the strengths, current limitations
and tradeoffs associated with specific
methods and tools currently available for
performing population-level ecological
risk assessment in support of EPA
programmatic and regional decision
making. What technical issues currently
limit the usefulness to environmental
decision makers of information
developed using the methods of this
approach for population-level ecological
risk assessment? With what priority
should these issues be addressed to
improve population-level ecological risk
assessment?
3. Is the current state-of-the-science and
practice sufficient to support development
of guidance for performing population-
level ecological risk assessment? Up to
what point can that guidance be developed
(e.g., only broadly, detailed with respect to
certain (specified) tools, etc.)?
Issues considered when addressing these questions
are elaborated in Appendix C.
This section communicates the opinions of
breakout group participants as summarized initially
by the leads for each group. Individual group
reports are not intended to be comprehensive
literature reviews or syntheses, but rather provide a
sense of how the questions above were approached
and answered in group discussions. In some
instances, group summaries have been restructured
and edited for clarity and to address the charge
questions more directly. The original group notes
and group lead summaries have been retained to
ensure minimal loss of information, but are not
presented as part of this report. When appropriate,
strongly expressed opinions alternative to those
held by many in the breakout group are captured as
text boxes.
3.1 Observational Approaches
As a prelude to more specific deliberation of
observational approaches to population-level
ecological risk assessment, several breakout group
members reflected that problem formulation, and
the types of hypotheses developed in the problem
formulation, are critical to the success of any
approach taken to assess risks to populations. With
respect to observational approaches in particular,
studies need to be designed to generate sufficient
amounts of information over appropriate time
horizons, so that they provide the data needed
to address the risk problem adequately. A power
analysis is appropriate in this regard, and there
would be value to shifting designs away from
the standard null hypothesis approach to an
alternative inference approach to support a more
comprehensive evaluation of population-level
impacts using multiple hypotheses. A thorough
problem formulation is required to identify the
approaches and methods that are needed to inform
the environmental decisions to be made.
3.1.1 Methods, Tools and Applications
Observational approaches breakout group
discussions identified a number of tools that
are available for population-level assessments
(including mark/recapture methods, nest boxes
or nest monitoring approaches, telemetry/remote
sensing techniques, and a broad range of field study
designs and data interpretation methods), together
-------
with well developed supporting information to
guide planning and performance of observational
studies. In doing so, they acknowledged that
different types of assessment approaches have
different needs and uses with respect to these tools.
Two broad categories of approach were identified:
• General observational studies that identify
and document status and trends in the
environment (e.g., EPA's Environmental
Monitoring and Assessment Program
(EMAP) and similar monitoring efforts,
the US Geological Survey's North
American Breeding Bird Survey, harvest
databases and others). They can be used to
identify problems and associations that are
suggestive of causation.
• Targeted observational studies that provide
information to help resolve a specific
environmental problem. Three types of
targeted studies were identified: 1) those
that characterize an impairment but not its
cause (e.g., application of biocriteria); 2)
those that relate an impairment to a cause
of concern (e.g., relating impairments to a
waste site, a spill or a new pesticide); and
3) those that evaluate an impairment in
the context of multiple candidate causes
to identify the one most likely (e.g., using
EPA's Causal Analysis/Diagnosis Decision
Information System (CADDIS) to inform
total daily maximum load (TMDL)
calculations).
In breakout group discussions, consideration
focused primarily on three general types of
environmental decisions: 1) hazardous waste
site remediation; 2) pesticide and new product
registration; and 3) development of national/
regional/local criteria protective of the
environment. More broadly, however, Table 2
illustrates the views of at least one breakout group
member relating the types of observational studies
that can be conducted within a range of decision
contexts. A listing of population-level attributes
(Table 3.1 in Barnthouse et al. 2008; reproduced
here as Table 3) provided additional context for
group discussions.
In a general discussion of observational approaches
available to hazardous waste site decisions, the
breakout group noted that there are a number of
proven wildlife estimation methods that can support
determinations of population status within sites,
but that some require a fair amount of resources to
implement. The environmental decision to be made
will inform the level of expenditure warranted in
applying these methods. The group also noted the
existence of several sources of pre-existing data
that can support or augment targeted observational
studies, including:
• EPA's EMAP/REMAP
• State of the environment reports
• TMDL programs
• Fisheries and wildlife management harvest
data sets
• Integrated Natural Resource Management
Plans required by military bases
• Threatened and endangered species
databases maintained by USFWS and state
departments of natural resources
3.7.2 Strengths and Limitations
The breakout group identified a number of
advantages that observational approaches have
generally relative to experimental and modeling
approaches. Among these, observational
approaches:
• Are easy to understand by most managers
and the public because they are more
credible reflections of population dynamics
in real situations. For example, people can
identify with actual living beings more
directly than they can with model outputs.
• Rely on readily available methods that have
a long history of use in ecology.
• Can be performed in multiple ways to
identify population impairments, including:
o Comparing population status
to those at local reference sites
(e.g., contaminated sites versus
uncontaminated sites)
o Evaluating population status
against regional information
-------
(distribution of attributes in the
region)
o Comparing population status to
expectations
o Incorporating population models
to evaluate observations (do the
demographic parameters imply a
population decline?)
o Evaluating population status
across gradients (do population
attributes change along a gradient
of contamination?), which does not
require a reference location
• Inherently account for density dependence
and compensatory mechanisms in the
observations made, and as such require
no assumptions about the importance and
mechanisms of the processes at work.
• Are amenable to methods that assist in
the evaluation of multiple stressors (e.g.,
habitat suitability indices along with the
population measures).
Despite these relative strengths, the breakout
group identified a number of technical issues
associated with observational methods that either
present challenges to their use in population-level
ecological risk assessment, or limit their usefulness
in this context. Among these are:
• The magnitude of changes that should be
detected in observational studies to support
decision making is as yet unspecified.
Interpretive guidelines might be valuable in
this regard.
• Depending upon the circumstances of the
assessment, observational approaches can
be more costly to conduct than modeling
or experimental approaches. The financial
resources available to the assessment often
determine the design employed.
• Inherent variability of some parameter
measurements adversely affects analyses
of the significance of observed responses,
and often requires that other approaches be
included as part of the assessment.
• The potential effects of multiple stressors
are difficult to tease apart, rendering critical
the identification and use of an appropriate
study design.
• Risk conclusions drawn from field
observation tend to be situation-dependent,
and cannot be extrapolated to other
situations readily.
• Observational approaches are limited
in their ability to support prospective
assessments of risk (such as is needed for
pesticide registration, for example). In this
regard, prospective assessments must rely
on experimental and modeling techniques
to provide information needed to inform
decisions.
3.1.3 Sufficiency of Science for
Development of Guidelines
Breakout group discussions of the maturity of the
science supporting observational approaches to
population-level ecological risk assessment focused
largely on four methods and issues: 1) demographic
surveys; 2) biomarkers; 3) study design; and 4)
drawing inferences from observational studies. A
summary of opinions about these approaches and
issues includes the following and is consolidated in
Table 2:
• Demographic surveys - Methods for
counting and characterizing organisms
in populations are well developed and
accessible, and they have a long history
of acceptance for various uses, including
in ERA. They are transparent and easy to
communicate. There are numerous methods
being used in ecology and conservation
biology that may or may not have
current uses in ERA. Some issues remain
incompletely resolved, however, and could
form the basis of additional guideline
development, including how "population"
should be defined in different contexts
(although to some extent, this issue was
addressed in U.S. EPA 2002), and how
much impact a population can sustain
without adverse effect.
-------
Biomarkers and body burdens - These
organism-level measures are potentially
important for determining exposure
and whether observed differences can
be related to a toxic mechanism in
observational studies. There are many
techniques and tools readily available in the
literature, but standardization is lacking.
Additionally, little guidance is available
for utilizing such information in decision
making. Consideration is needed of how
biomarkers, body burdens and external
stressor concentrations relate to exposure of
individuals in populations.
Design of studies - Consideration of
spatial and temporal contexts are extremely
important when designing observational
studies; guidelines clarifying considerations
in this regard would be valuable.
Population-level ecological risk assessment
also would benefit from additional
consideration of: supporting information
that should be collected when applying
observational approaches (e.g., habitat
quality and distribution, spatial distribution
of stressors, etc.); designs for estimating
risk as opposed to testing hypotheses; and
designs that support model development
and parameterization. The group noted
that adequate guidelines for selecting
assessment and measurement endpoints
exist, but that some expansion specifically
focused on population-level endpoints
relevant to decision makers would be
valuable. A decision tree could provide
insights into data collection and analysis
activities to evaluate risks.
Drawing inferences - Some methods
for making inferences from the data are
accepted, but other methods for are far less
standard or straightforward, and clarity in
this regard would be beneficial. Elaboration
of basic decision considerations regarding
interpretation of changes in measurement
endpoints also would be helpful (e.g., for
changes in age structure).
The breakout group also discussed the desirability
of long-term educational efforts to aid stakeholders
in understanding that science is not going to be the
sole element informing decision making. Guidance
may be needed to enhance assessment transparency
and risk communication, and its development may
require involvement by social scientists.
3.2 Experimental Approaches
The experimental approaches breakout group
considered the contributions to population-level
ecological risk assessment of information gathered
from controlled manipulations in laboratory and
field settings, and in a combination of both. They
noted that numerical experiments—simulations
involving manipulation of models—potentially
could be another type of experimental approach,
and one likely to be considered perhaps more
appropriately as a modeling approach. Even
so, many of the issues concerning experimental
design, interpretation and usefulness addressed in
discussions of experimental approaches might be
re levant to numerical experiments.
3.2.7 Methods, Tools and Applications
Members of the breakout group noted that
controlled laboratory studies often are designed
to provide direct evidence of cause and effect or
to characterize stressor-response relationships.
Standardized toxicity tests have been applied in
many broad and site-specific regulatory decision
contexts, including regulation of chemicals and
assessments of risks of site-specific environmental
media associated with waste disposal and
hazardous waste sites. Historically, toxicity tests
have used a limited suite of species as biological
models to assess the toxicity of single chemicals,
tested in isolation from other stressors. There is a
large, accessible database of such information; full
stressor-response information would be more useful
for population modeling input.
-------
Table 2. Observational methods and decision contexts
Decision Context
Estimate risk
Develop cleanup
levels
Evaluate cleanup
alternatives
Prioritize sites in a
region
Validate (or not)
toxicity thresholds
and related decisions
Evaluate species
distribution, viability
and habitat
Evaluate resource
services
Diagnose causes of
observed impacts
Plan restoration
projects
Develop regional/
national criteria
Evaluate specific
disposal sites
Detect whether
perturbations are
meaningful
Feed into/inform
models
Detect unexpected
problems
Evaluate pesticides
for Registration or
Special Review
General Observational Approach
General
observational data
(existing databases;
secondary use when
applied to Superfund
site)
Possible, but most of
the time no
Possible, but most of
the time no
Possible, but most of
the time no
Possible, but most of
the time no
Possible, but most of
the time no
Possible, but most of
the time no
Possible, but most of
the time no
Possible, but most of
the time no
Possible, but most of
the time no
Possible, but most of
the time no
Possible, but most of
the time no
Possible, but most of
the time no
Possible, but most of
the time no
Possible
Not likely
Targeted
observational
design (ID a
problem without
ID of cause)
Yes*
Yes*
Yes*
Yes*
Yes*
Yes*
Yes*
No
Yes*
Yes*
Yes*
Yes*
Yes*
Yes*
Not really
Targeted observational
design (ID a problem
with one cause -
limited use of data if
no problem identified)
Yes*
Yes*
Yes*
Yes*
Yes*
Yes*
Yes*
Only limited to single
cause*
Yes*
Yes*
Yes*
Yes*
Yes*
Yes*
Not really
Targeted
observational
design (ID a
problem with
multiple alternative
causes)
Yes*
Yes*
Yes*
Yes*
Yes*
Yes*
Yes*
Yes*
Yes*
Yes*
Yes*
Yes*
Yes*
Yes*
Not really
"Ability of observational method to address the issue is critically related to design considerations in problem formation.
-------
Table 3. Attributes of populations in assessment endpoints (from Barnthouse et al. 2008)
Attributes of organisms
Attributes of populations
Demographics of individuals
• Mortality (e.g., living or dead)
• Reproductive state and output
(e.g., fecundity, births per female, potential seeds)
• Development rate (e.g., time for larval
development, time to maturity, weaning, ripening)
Abundance
• Population size (number or biomass)
• Population density
• Equilibrium (steady-state) abundance
• Carrying capacity
Extinction and recovery
• Probability of extinction
• Time to extinction
• Quasi-extinction
• Minimum viable population
Recovery time (from disturbance)
Physiologic characteristics
• Individual growth rate
• Respiration rate
• Ingestion rate
• Metabolism and excretion
Population growth rate
• Intrinsic rate of natural increase
• Finite rate of population increase
• Birth, death, immigration, and emigration rates
Demographics of individuals (continued)
•Age
•Size
•Sex
• Locations and "home range" or dispersal
of an individual
Population structure
• Age class distribution
• Size class distribution
• Sex ratios
• Spatial distribution of the population an individual
Genetic characteristics
• Individual genotypes
• Presence of particular alleles
• Heterozygosity
Genetic structure and variation
• Genotypic frequencies
• Heterozygosity
• Genetic diversity
Organism "health" or condition
• Condition factors (weight/length relationships)
• Morbidity
• Deformities
• Tumors and other histopathologic anomalies
Incidence (frequency, percent, or fraction) of the
population or distribution thereof with respect to
• Specified conditions
• Morbidity
• Effects (e.g., percent killed), and/) or exposures to
stressors
Ecology, behavior, and exposure
• Life history for an individual
• Habitat/food "preference" or location in space
• Locomotion, dispersal, migration (e.g., range), and
spatial extent of activity (e.g., home ranges) for
an individual
• Individual environmental exposure
Spatial distribution and habitat
• Spatial distribution across available habitat (may
involve distributions of age and/or size classes as
well as influences on genetic composition)
• Critical patch size
• Habitat requirements (quantity, quality,
fragmentation)
-------
Toxicity tests were designed to rank the toxic
potency of chemicals under idealized, standardized
conditions, and generally not to provide realistic
assessment of contaminant effects on populations
under varied or complex environments. Nor were
they designed to inform population models directly.
Several examples of studies that use toxicity test
data to parameterize models that project population
responses are available (see, for example, Akcakaya
et al. 2008). A variety of simple to more complex
methods have been used to translate toxicity test
responses into input for population models, often
required because of mismatches between test
data and model input needs (such as test versus
lifecycle stage duration). Some translations can
be accomplished using empirical relationships,
such as estimating chronic survival from acute
survival (e.g., Mayer et al. 1994). More specific
methods have also been described, such as when
activities that strongly affect vital rates are simply
missing from current testing. For example, seasonal
reproduction must be projected based on effects
when parental behavior is not considered explicitly
in the experimental design (Bennett and Etterson
2007).
Laboratory studies that have been designed to
collect information on the effects of stressors
(including chemicals) at the population level
include life table response experiments (LTREs),
in which information is summarized across
individuals, and so-called bucket tests which collect
information about groups of organisms. There is
a body of literature, particularly with respect to
pesticide effects on target/non-target species, that
provides examples of studies using these methods.
Studies designed to capture population endpoints
have been used more extensively in the EU than
in the United States, and it may be useful for EPA
to review examples where these have been used in
decision making. Some EU-based organizations
recognize the usefulness of these approaches, but
procedures have yet to be fully standardized.
Information from controlled laboratory studies has
also been used to provide demographic information
to population models. However, the performance
of laboratory populations reflects the "unnatural"
conditions of laboratory settings (e.g., excess food,
ideal environmental conditions, no predation,
etc.). Some examples demonstrate differences
in dynamics of populations under laboratory
conditions relative to dynamics in the field, and
highlight the artifactual nature of laboratory
responses. While laboratory-based studies provide
opportunity to gather data on stressor effects at
high resolution and fine-scale detail, more realistic
effects and more natural population dynamics
might be observed in field-based studies.
Field experiments have progressed historically
from massive and largely unreplicated studies to
smaller, replicated studies. They provide important
opportunities to gather "real-world" data but are
limited in their interpretability and portability
to alternate contexts, and often are prohibitively
costly. While some of these studies may not have
been designed to quantify stressor effects, they
have provided valuable information on life history,
abundance and habitat requirements of local fauna,
and might continue to serve this function. "Semi-
field" studies, such as mesocosm experiments,
can provide more feasible (and more controlled)
options to acquire realistic information about the
effects of stressors. For example, stream-side
studies (i.e., literally conducted beside streams
in mobile laboratories) provide realistic exposure
regimes in aquatic studies. Importantly, semi-field
experiments provide the opportunity to assess the
effectiveness of management alternatives at small
scales.
While few examples are currently available of
population-level ecological risk assessments,
studies designed with integrated laboratory
and field components can provide site-specific
information, or can facilitate tests of laboratory-
based predictions of field effects. Some examples
show that laboratory-based projections are
not coherent with field realities, but these
inconsistencies provide the opportunity to
reexamine conditional needs for specific population
model complexities. Because of their potential
to produce unique site-specific information, field
studies are potentially important in "hot spot"
(e.g., Superfund) assessments. It is particularly
advantageous to link laboratory and field
components early on in the assessment process (i.e.,
during problem formulation), because field studies
are more complex, customized and (typically) take
-------
longer to perform than do laboratory studies. When
results are clearly linked to decision outcomes, the
cost and time required for integrated laboratory
and field experiments (sometimes requiring two
or more years to complete) may be justified. For
some important chemicals of concern, critical body
residue studies may be useful to link laboratory and
field studies.
3.2.2 Strengths and Limitations
In considering the strength and current limitations
of experimental approaches, the breakout group
identified various strategies to make information
from experimental studies more useful to
population-level risk ecological assessment,
including:
• Simplified controlled laboratory systems
provide clear and easily understood
linkages between stressor exposure and
effects. They typically are inexpensive,
quick and easy. But, a population
perspective invites examination of
complexity, and the use of experimental
information to address issues associated
with multiple stressors, cumulative effects
and real-world population dynamics.
Some important issues to consider are
factors regulating populations, such as
disease and predation, and combinations
(and interactions among) chemical and
non-chemical stressors. There was also
recognition that background conditions
may be need to taken into account that
reflect real-world exposure to multiple
stressors for comparisons involving
background/reference populations.
Modeling approaches might provide
a framework directing the design of
experiments to do so.
• In the short term, currently standardized
tests with organism-level endpoints might
be appropriated for use in assessments
of population risk. For example, Bennett
and Etterson (2007) provide an example
of a modeling approach to translate
avian reproduction test data, collected to
support pesticides risk assessments, to
enhance projections of population-level
effects. The best toxicity test designs for
input to population-level ecological risk
assessment remain to be identified; whole/
partial lifecycle tests show promise,
but additional methods and validation
of their usefulness for this purpose are
required. Standardization of or best
practices documentation for existing tests
(e.g., LTREs) will add to their utility
immediately. Dialogue with modelers can
contribute to modifications to current test
designs so that the resulting data better
reflect the needs of population models. First
generation toxicity tests were designed
to capture toxicological variation among
species; next generation population-level
tests should try to capture demographic
variation among species as well.
Experimental methods can contribute
valuable information to other approaches to
population-level ecological risk assessment.
For example, experiments can provide
data needed to parameterize population
models. But, experimental studies also
can contribute to an understanding of
mechanisms and processes that affect
population dynamics and risk of stressors,
information valuable to decision making.
Experimental approaches almost always
utilize species amenable to experimental
manipulation. Unless the test species is
sufficiently representative of the assessment
population, or indeed is the same species
as the assessment population, some degree
of extrapolation of test and assessment
results will be required. The breakout group
did not address this potential limitation
in detail, other than to acknowledge that
extrapolation methods are available for
certain circumstances. It is likely that
population-level ecological risk assessment
would benefit from additional research
attention on extrapolation approaches.
-------
3.2.3 Sufficiency of Science for
Development of Guidelines
Despite the limitations previously noted, members
of the experimental approaches breakout group
felt that the science underlying toxicity testing,
LTREs and some field and semi-field experimental
designs are sufficiently well developed to move
forward with documentation of best practices
for their use in population-level ecological risk
assessment. Several of these methods have been
standardized with respect to their intended uses for
toxicity evaluations and so on, but not necessarily
for assessing population responses. The group
identified efforts that could advance development
of guidelines for population-level ecological risk
assessment (some of which are not limited to
experimental approaches):
• A synthesis of existing literature (including
conservation and gray literature) describing
cases in which population assessments
have been used in decision making could
illuminate best practices and limiting
considerations. Such a synthesis should
consider the types of decisions (e.g., setting
a cleanup goal, determining unacceptable
exposure concentration, etc.) informed
by the experiment approach employed,
describe how the test information was used,
and to the extent possible, evaluate whether
the decision was appropriate to meet
management goals.
• A second activity could evaluate the
efficacy of population-level ecological risk
assessment to describe risk through case
studies. For example, past assessments
conducted without consideration of
population assessment endpoints could
be performed anew with population-level
endpoints, with the results being compared
against post-decisional data sets (e.g.,
after-action monitoring data) that describe
the outcomes of decisions based on those
earlier assessments. Potential candidates
for this type of evaluation include certain
pesticide usage decisions (together with
their 5-year reviews) and water quality
criteria.
• A summary or catalogue of experimental
tools and data available to population-level
risk assessment could be very useful to
practitioners.
3.3 Modeling Approaches
Population models can be applied to a wide
variety of environmental management problems.
In predictive risk assessment, an advantage of
population modeling is that it provides the ability to
integrate multiple stressors and multiple endpoints
in a consistent fashion. Population modeling is not
only a practical approach, but it directly addresses
the (stated or implied) intent of regulatory statutes
to protect species populations as natural resources.
Population modeling might serve as a practical
compromise between simpler assessments mainly
based on organism-level endpoints and more
complex assessments based on fully integrated
ecological-economic models. The organism-level
approach is relatively easy to conduct but may be
overly simplistic and may leave many important
questions unanswered. Fully integrated ecological-
economic modeling can address many more
important questions in principle, but in practice
it often will require more time, data and other
resources than are typically available for most risk
assessments.
3.3.7 Methods, Tools and Applications
A wide range of population models is available for
use in EPA programs, including some that can be
accessed through EPA's Council for Regulatory
Environmental Modeling (CREM) Web site (e.g.,
BASS (Bioaccumulation and Aquatic System
Simulator) and PATCH (Program to Assist in
Tracking Critical Habitat, also called HEXSIM);
see http://cfpub.epa.gov/crem/knowledge_base/
knowbase.cfm). The modeling approaches breakout
group considered uses of population models
in a number of decision contexts, including:
1) developing risk estimates, cleanup levels or
evaluating alternative management actions (e.g.,
at a hazardous waste site or a dredged material
disposal site); 2) evaluating ecological risks of
pesticides or other chemicals; 3) evaluating species
distributions, population viability and habitat; 4)
determining resource services and land use actions;
-------
5) diagnosing causes of observed population or
habitat impacts; 6) developing criteria; and 7)
evaluating the spread of invasive species. Table
4 relates classes or types of population models
to applications in these and other environmental
management contexts. Formulations range from
simple models that aggregate across individuals and
space (i.e., unstructured models) to complex agent-
based models (ABMs) that are spatially explicit
and follow individual organisms through a set of
behavioral rules and in some cases physiological
processes. In ABMs and other individual-based
models, integrating across the pool of individuals
in a computer simulation creates the dynamics of
the population. Regardless of the basic approach,
spatial structure can be an important part of
population modeling, such as in metapopulation
models that include two or more subpopulations
located in different habitat patches at the same
time and that are linked by immigration and
emigration. Chapters 2, 3 and 9 of Barnthouse et al.
(2008) illustrate specific examples of management
decisions and related model applications. Pastorok
et al. (2002) and Akcakaya et al. (2008) provide
compilations of specific applications of population
models used in toxic chemical risk assessment and
other contexts (e.g., conservation biology, resource
harvest management).
Specific guidance is lacking for matching a
modeling approach to a particular risk problem.
Some ecologists apply unstructured models
to lower trophic levels (e.g., phytoplankton,
zooplankton), structured models to mid-trophic
levels (e.g., some fish, birds and mammals) and
individual-based models to higher trophic levels
(e.g., other fish, birds and mammals). This approach
was taken in the Across Trophic Level System
Simulation (ATLSS) modeling system (http://atlss.
org) applied to the Everglades Restoration Program.
With respect to model selection, a member of the
breakout group asked: "Can we characterize the
decision context in terms of endpoints of interest,
species of interest, degree of detail needed and the
relevant spatial and temporal extents and scales
(resolutions), and then let those characteristics
dictate the type of population model that would
be most useful?" The group identified the need to
develop a decision framework for selecting and
applying population models in EPA programs (see
Section 3.3.3 below).
Key points made by members of the modeling
approaches breakout group in response to Charge
Question 1 include the following:
• In principle, any type of model can be
used within any tier of an ecological risk
assessment, but in practice, development
of complex models typically will be
restricted to higher tiers (i.e., later phases
of an assessment requiring more detailed
evaluations).
• Models to evaluate products (e.g.,
pesticides) or develop national or
regional criteria need to be more flexible
and generic (and therefore likely more
simplistic) than site-specific models. Some
group members thought only scalar and
biologically structured models could be
used for pesticide registrations. Others
thought metapopulation and individual-
based models also could be useful.
• When evaluating specific sites with a
population modeling approach, two
analyses may be useful. First, treat the
individuals on the site as an independent
population, and then treat the site as one
unit of a metapopulation that includes other
sites that may be linked to the target site by
immigration and emigration.
• Population models can be used in a reverse-
mode risk assessment to estimate media
cleanup levels required to meet pre-defined
levels of acceptable risk. What defines
acceptable risk in terms of population
endpoints has not been clearly identified
for EPA programs, but general approaches
and specific metrics used in conservation
biology may be relevant.
• A useful reference for EPA programs
involving land use decisions is the
Akcakaya et al. (2004) compilation of
case studies of applications of population
modeling to issues in species conservation
and management.
• Models can be helpful in evaluating the
relative importance of alternative causes
-------
of observed population impacts (i.e., using
models in a diagnostic mode). However,
in conservation biology and similar
applications, models cannot be used by
themselves to determine the cause of past
declines. Historical data also are needed.
The multi-agency Plan for Analyzing
and Testing Hypotheses (PATH) project,
which used population models to evaluate
alternative hypotheses concerning causes
of declines in abundance of Snake River
Basin salmonid populations (Barnthouse
et al. 2000; Peters and Marmorek 2001)
illustrates the use of population models
to evaluate causes of species population
decline.
Table 4. Applications of population models in environmental management contexts
Decision Context
Estimate risk
Develop cleanup
level
Evaluate cleanup
alternatives
Prioritize sites in
region
Evaluate pesticides
for Registration
Evaluate pesticides
for Special Review
Verify toxicity
thresholds and
related decisions
Evaluate species
distribution,
viability and habitat
Evaluate resource
services
Diagnose causes of
observed impacts
Plan restoration
projects
Develop permit
criteria
Evaluate specific
disposal sites
Evaluate spread of
invasive species
Class of Model
Unstructured
(scalar)
X
X
X
Generic life
histories
Generic
X
X
X
X
Biologically
structured
X
X
X
Generic life
histories
X
Generic
X
X
X
X
X
Individual based
X
Estimating
exposures by
modeling behavior
X
X
X
X
X
X
X
X
Metapopulation
(occupancy)
X
X
X
X
X
Metapopulation
spatially explicit
X
Potential
(evaluate source/
sink dynamics)
X
X
X
X
X
X
X
X
X
-------
• Population models can be applied to
organisms and stressor effects in any kind
of environment, including terrestrial,
aquatic, aerial, wetland, etc.
Recognizing that a key strength of population
models is their ability to integrate multiple
endpoints and multiple stressors, the modeling
approaches breakout group offered the following
observations:
• Integration of multiple stressor effects
is straightforward if stressor exposure-
response functions are (or are assumed
to be) independent. Accounting for
dependencies between stressor effects
requires more data and more sophisticated
models.
• Physiologically-based population models
(Kooijman and Metz 1984; Kooijman
2000) can deal with bioaccumulation and
toxicity as well as illustrating population
dynamics. One such model already used by
EPA is the BASS model (Barber 2008).
• Some types of models are better than are
others for certain types of stressors. All
model types can evaluate risks of chemical
stressors. Habitat-related stressors are more
difficult to assess with some model types
(namely, spatially aggregated models).
The attributes of populations potentially evaluated
by modeling approaches are listed in Table 5.
By implication, the types of population-level
effects potentially evaluated using models are
reflected in changes in these attributes. Most of
these population-level effects can be expressed as
probabilistic outputs from software to implement
models.
3.3.2 Strengths and Limitations
Table 6, developed initially by R. Akcakaya, and
modified by the modeling approaches breakout
group, lists specific models types and gives
examples of their advantages and disadvantages for
use in population-level ecological risk assessment,
as well as specific software implementations
developed to date. Person (2002), Carroll
(2002), Regan (2002) and Akcakaya and Regan
(2002) provide detailed evaluations of strengths,
current limitations and tradeoffs of various
specific population models (also see Table 9.1 of
Barnthouseetal. 2008).
Technical issues that are likely to arise in applying
models to EPA regulatory programs include:
• The relative scarcity of models that link
exposure to stressors, stressor-response
relationships, and demography.
• Stressor-response data (e.g., toxicity data)
that are not measured or reported in a
format useful for population modeling.
• Uncertainty about the values of
demographic parameters for certain
species.
• Limited knowledge of the strength of
density-dependence and its interaction with
toxicity or other stressors.
Few population models include exposure, toxicity
stressor-response relationships and demographic
processes as part of their current structure (Person
2002, Carroll 2002, Regan 2002, Akcakaya and
Regan 2002). Software implementations for
population modeling could use improvement in this
aspect.
-------
Table 5. Attributes of populations that can be evaluated using population models, and data requirements of
population models
Modeled Attributes of Populations
Population-level Effects
Potentially Modeled
Key Data Inputs to Models*
Population abundance (size and
density)
Population growth rate
Population viability/extinction
probability/expected time to
extinction
Rates of birth, death, immigration
and emigration
Age/stage structure
Sex ratio
Spatial distribution
Genetic structure (e.g., genotypic
frequencies, heterozygosity, genetic
diversity)
Recovery time after disturbance
Change in abundance or population
growth rate
Change in age/stage structure
Change in population biomass or
production
Change in recruitment
Reproductive failure
Change in viability/extinction
probability/expected time to
extinction
Fragmentation of population
Spread of unwanted species
Increased variability of abundance
Change in sex ratio
Change in genetic structure of
population
Knowledge of the general life
history of the target species, such
as life span, length of life stages,
generation time, habitat distribution,
etc.
Intrinsic rate of population growth
Carrying capacity
Fecundity and survivorship rates at
key ages or stages
Body size or weight at each age/
stage category
Immigration and emigration rates
Individual behavior
Knowledge of stressor-effect
relationships for one or more of the
parameters above
* Not all of these data are needed for each population model and some of the listed parameters may be outputs from
some models and inputs for other models. The specific kinds of data needed depend on the type of model. Simple
population models, such as the Malthusian model or the Leslie matrix may require only readily available data or data
that can be easily obtained from a stressor-response test in the laboratory (e.g., a toxicity test), at least for some
species.
Some stressor-response data are not in the correct
units (e.g., a critical tissue residue value is reported
when a concentration in abiotic media is needed,
which also reflects a problem of mismatched
endpoints) or relevant to the time scales of attention
for population modeling. Full stressor-response data
often are not available, especially for toxic chemical
effects on vertebrates. Reported toxicity information
sometimes includes summary statistics (e.g., an LC5Q.
-------
o
e
o
'•i
^^
—
S.
'i
e«
O
;—
'S
8.
M
cS
r: o
1!
§1
5« .,,
^ CU
M 3
"
^ o
03
z
1
QJ
U
K
_O
a 3
£ c
£ s
o 3
50 "B.
g
Best for/when/if
£
S
£
Disadvai
ex
«
1
5
L
"8.
S
X
w
e.
^
"3
1
§
c
o
z
OH
O
OH
^3 O
t|-a
GO S 'C
•o
.c
'3
2
5
|1
• S o>
N
1 «
•S ~ •§ « i
c ° "^ K S -^
1 | ^"§ 1 ^
+^ •— i ^ g ^ ^ m
^ j c2 cS tS" S 2
T3*o^
C --H T3
.1 II
"3 H 8 S
t2 ""
»
O "^
1 8
S w
<^ <^
S S S
<; < j
• Demography or impact depends on
age/stage and
• Sufficient demographic data and
• No spatial heterogeneity
• Applied to vertebrates, plants and
some invertebrates
tj
•o
1 1
II
M
0
C
"^ ""
^
.0
c 5^
<3 Os
(D "^
3 c
1 &
*
o
S
a ^
1 |
S 0
OH
O
OH
S £2
S O
~J ~J
S S "
!?!
• Demography or impact depends on
age/stage and/or location (i.e., spatial
heterogeneity in demography or
toxicant distribution) and
• Sufficient demographic data
• Applied to vertebrates, plants and
some invertebrates
c
— «
.5 -o
i 1
II
.9 ! , a J
§ — M •« 'g .2 — a
. 2 3 o ^ „ l|1 §" I"
1 If 1 1 1 !l a 1"
f : ^f-^f -^
•0
1
_£
"« S
u g,
1 |
1 1
* *
c ^
o 13
S -g
3 OH -O
OH C3 2
i-, JJ 8
•Mt
S £ "^
.£ ^ «
||l
§* i
Q. rf i-3
S.S o
£ OJ 0
C/3
§1 " ° S
§1^
• Modeling behavior and movement to
estimate exposure, bioaccumulation,
genetic effects and/or dispersal rates
• Very small populations with abundant
demographic and behavioral data
• Usually applied to vertebrates
•o
oj C
'3 ji1 S "3 c
S =3 S o .0
-g g c s s
1 1 1 1 o,
"> "z, 0, 3 .3
(D
c
O
-------
NOAELs or LOAELs), but the underlying data
needed for population modeling (relationships
between demographic parameters and exposure)
might not be reported and can be difficult to obtain.
It was noted that uncertainty about the values
of demographic parameters (e.g., lack of data
for portions of a population, particularly for the
younger ages/stages) is typically lower than
uncertainty about exposure-response relationships
and the strength of toxicity effects (Barnthouse et
al. 1990). Reducing uncertainty associated with
toxicity effects is an important area for research
and development (one group member disagreed
with this position and felt that that the real issue is
whether population modelers effectively use the
available toxicity information; see text box). The
uncertainty associated with toxicity data comes
from several sources. Within-test variability is
often represented by confidence intervals about
a statistical endpoint (e.g., LC50); however, there
also is uncertainty associated with variability
from similar tests conducted at different times or
in different laboratories (among-test variability).
Often, only short-term toxicity data will be available
for estimating long-term exposure effects. There
also is a significant source of uncertainty associated
with toxicity data if extrapolation between species
is needed (for example, see Barnthouse et al.
1990). It may be difficult to reduce these sources of
uncertainty. However, use of stochastic population
models provides an opportunity to reflect this
uncertainty in information provided to decision
makers.
Clearly, density-dependence is important for most
natural populations, but some members of the
modeling approaches breakout group questioned
whether or under what conditions the inclusion of
density-dependence in population modeling would
change the qualitative conclusions of modeling.
Also, the level of density-dependence that matters
in a decision-making context for most species of
interest is unknown. Typical issues encountered
when incorporating density-dependence into a
population model for toxic chemical effects are:
What is the strength of density-dependence?
How does the relationship between vital rates and
population density change with different chemical
exposure levels? How sensitive are risk assessment
results to assumptions about the form and strength
of density-dependence?
An Alternate Opinion Regarding Toxicity Data
One breakout group member did not agree with all the
shortcomings of toxicity data suggested by other group
members. He felt that the overarching issue is whether
population modeling adds value to a good toxicity
assessment. If few exposure-toxicity test data exist,
then the toxicity assessment is inadequate and cannot
be used define a safe concentration, with or without
population modeling.
This breakout group member did not agree that toxicity
data are in the wrong format or the wrong units. He
did caveat this opinion by stating that specifying
survival sensitivity of older life stages offish or aquatic
invertebrates (and probably terrestrial animals) can
require some digging for appropriate data, and may
require some assumptions. Because young organisms
generally are assumed to be more sensitive, chronic
toxicity tests focus on the early life stages.
At least for the fish class of vertebrates, he felt that the
data reporting deficiencies were not critical. Although
acute testing customarily reports only the LC50, for
chronic tests it is customary to publish the raw data.
EPA likes to see the complete concentration-response
curve data. Where not published, EPA technical staff
can contact the original investigators to obtain it.
Several members of the modeling approaches
breakout group thought that one of the key issues
that limits application of population modeling to
regulatory decision making is the lack of clarity
of management objectives. If decision criteria are
vague, it is more difficult to develop or select an
appropriate type of model. Some members of this
breakout group thought that this issue was more
important than are any of the outstanding technical
issues, suggesting that the technical difficulties
associated with gathering data, and developing and
applying population models, are less severe than are
the communication difficulties between modelers
and decision makers.
One member of the group suggested that analysts
should select or propose decision criteria in the
-------
course of the analysis if the risk manager had not
provided clear criteria in planning discussions prior
to the assessment. For example, the EPA aquatic-
life criteria program's use of Species Sensitivity
Distributions (SSDs) did not develop in response to
a management objective to protect 95% of species.
Rather, it was EPA risk assessors who wanted to
use the SSD approach that suggested the level-of-
protection objective.
The modeling approaches breakout group identified
several research and development needs that are
critical for improving the value of population
modeling in ecological risk assessment, including:
• Development of toxicity databases with
population modeling and population-level
ecological risk assessment in mind:
o One example approach to
database development is Data for
Environmental Modeling (D4EM).
D4EM is a set of open source
software tools that obtains and
processes data for models. These
data include land use/land cover,
water bodies and networks, major
roads, and political boundaries
(http://www.epa.gov/ttn/chief/
conference/ei 16/poster/brandmeyer.
pelf).
o For existing toxicity data, it
is important to ensure that all
measurement units are consistent
and that any time scale differences
between the experiments and the
model simulation are reconciled.
Furthermore, impacts on survival
as represented by ECx or LCx
statistics may not be completely
independent (or perfectly
correlated) across time, so it will
be important to ensure that all
assumptions used to parameterize
the population model are explained
as transparently as possible and
that sensitivity analyses are
conducted as appropriate.
o Some older data are available from
experiments that were conducted
over longer periods of time than
are current testing regimes. The
Oak Ridge National Laboratory
(ORNL) work on linking acute
and full life lifecycle toxicity test
data to fish population models
(Barnthouse et al. 1987; 1990)
needs to be accessed.
o One member of this breakout group
suggested that research is needed
to evaluate whether (or when) to
use an individual effect dose versus
a hazard modeling (stochastic)
approach to incorporating
the effects of exposure to
contaminants. Under the former, a
modeled population might evolve
tolerance to a contaminant. Under
the latter it would not. Newman
and McCloskey (2000) concluded
that: "neither hypothesis alone was
the sole or dominant explanation
for the lognormal (probit) model."
Development of a database of population
parameters for species that have been
studied in the field or the laboratory. It also
would be valuable to consider developing
a set of "representative life histories," or
"demographic guilds," that can serve as
modeling surrogates for groups of similar
species when detailed species-specific
data are not available. It is important to
develop information on fecundity and other
reproductive parameters in real populations,
and on how toxicity and other stressors
affect these variables. The National
Biological Information Infrastructure
(NBII), a network of partners representing
the spectrum of governmental and
nongovernmental sectors, may have some
relevant information (http://www.nbii.gov/
portal/server.pt). The aquatic ecosystem
model AQUATOX (supported by EPA:
http://cfpub.epa.gov/crem/knowledge_base/
crem_report.cfm?deid=74876) may have
some information that would be useful to
population modelers.
-------
Compilation of information on variation
of demographic parameters overtime for
representative species.
Sensitivity analyses to identify data needs
for model parameters for specific models
as case studies. There is a need to use
modeling in designing experiments (and
by implication, observational studies)
for ecological risk assessment and for
parameterizing models. Simulations can
be conducted to evaluate various model
components and their relative importance
to population dynamics and spatial
distribution.
Development of model structures and
software implementations that link toxic
chemical (or other stressor) effects to
demographic parameters in exposure-
response functions based on observational
and experimental data. Some early research
on this issue suggests that the slopes of
exposure-response functions for toxic
chemicals may not vary much across
species, especially if the mode of action of
those chemicals is similar, but the locations
of the curves on the chemical concentration
axis do. Consequently, if estimates of LC50s
for different species are available, then
the analyst can characterize the rest of the
exposure-response curve with the common
slope for the group of species.
More individual-based exposure models
are needed, together with linkages between
these (or other exposure models) and
population-level effects models (e.g.,
Bennett and Etterson 2007).
Development of information on the
interactions of multiple stressors in eliciting
effects. For population modeling exercises,
a range of assumptions between pairs of
stressors may be plausible, including: 1)
independence (correlation = 0); 2) positive
dependence (sensitive individuals are the
same individuals for all stressor types;
correlation > 0); or 3) negative dependence
(different individuals are affected by
different stressors; correlation < 0). The
RAF has guidance on how to model effects
of mixtures of chemicals that could be
expanded to address issues relevant to
population modeling.
Guidance on model validation is clearly
needed as part of the overall process of
guidance development for population-
level ecological risk assessment5. There
is relatively little common understanding
among modelers, and especially between
modelers and decision makers, about what
model "validation" or "verification" means.
These terms have been used to refer to a
wide variety of activities, from confirming
the mathematical derivation of the model
structure or debugging the computer
model code, to sophisticated statistical
comparisons of model predictions to
observed outcomes. Standardized methods
and guidance regarding model validation
would be useful for practitioners, and
improved communication between
modelers and decision makers on this
issue is needed. Modelers should strive
to help decision makers to understand
what verification or validation of a
population model means. Validation of
most types of population models is not
needed. Unstructured, structured and
metapopulation models can be thought
of as "accounting frameworks" (e.g., net
growth rate equals birth rate minus death
rate). These modeling frameworks have
been used extensively and are credible.
However, specific applications of these
modeling frameworks might need to
be validated. Also, some verification or
validation of ABMs and other individual-
based models is needed.
Validation may be defined generally
as a comparison of model outputs to
observations of the system being modeled,
5Although not discussed during the workshop, this need
is partially met by EPA's Council for Regulatory Environ-
mental Modeling (CREM) Guidance on the Development,
Evaluation and Application of Environmental Models,
available at http://www.epa.gov/crem/model-evaluation.
html.
-------
for the purpose of evaluating the credibility
that should be accorded to results derived
from model applications. It has long
been understood that all models are
approximations to reality, so that no model
can be expected to provide accurate and
precise predictions under all circumstances
(Levins 1966; Mankin et al. 1975). The
purpose of validation is to identify the
specific circumstances under which the
model can be considered reliable. The
process may involve validating model
components (individual parameters and
functions) and software quality assurance
(QA), rather than validation of the
model structure itself. Recent literature
emphasizes the construction of alternative
models of the same population, applying
the alternatives to the same set of empirical
data, and then using statistical methods
such as Bayesian inference to identify the
model that is most consistent with the data
(Hilborn and Mangel 1997; Peters and
Marmorek2001).
• Development of a toxicological front-end
for RAMAS (family of commercially
available software for population risk
and other analyses) or a similar modeling
system to characterize the shapes of
stressor-response functions for a large
number of toxicants (or other stressors)
for various species and their life stages.
The initial version of such a front-end
for RAMAS is now under development
(funded by the US Army Corps of
Engineers) and will be available freely
online by spring 2009. The biggest research
and development need is to have data to
characterize the sensitivity of species,
entire stressor-response curves, and life
stages that are affected, while translating
the original time scale of the stressor
exposure-response test to other time
scales relevant for modeling. One member
expressed the opinion that there is a need
for the distribution of open-source, public-
domain software linking a toxicity model to
a structured population model.
3.3.3 Sufficiency of Science for Development of
Guidelines
The current state-of-the-science is sufficient to
develop guidelines to support use of population
modeling in EPA risk assessments. There is a
high degree of scientific acceptance of population
models and their application to environmental
management. Basic demographic models and their
structures are standardized. The biggest issues lie
in increasing understanding of the usefulness of
population-level ecological risk assessment and
facilitating acceptance by decision makers within
EPA. Modelers or model users need to learn how
to speak the decision maker's language to increase
transparency and understanding of model use.
Population modeling is a tool that is useful for
evaluating effects on many different endpoints.
There are accepted ways to perform assessments
of population-level endpoints. However, there
is no scientific consensus on a specific short list
of model endpoints because the nature of the
endpoints and related risk expressions depend on
the specific management question and the related
risk assessment objectives. Guidance is needed to
help managers work with population biologists to
develop risk expressions relevant to management
decisions and the criteria for interpreting results of
population model outputs.
Harmonization of best modeling practices across
EPA programs may be a desirable goal. It might
be possible to adapt existing standards and
guidance from the literature on constructing and
implementing other types of models (e.g., within
EPA programs, and the International Union for
Conservation of Nature (IUCN)). Also, available
guidance such as "Mistakes to avoid in population
modeling" (RAMAS.com/mistakes.htm) could
be incorporated into future EPA best practices
guidelines on population modeling.
Software for population models and analytical
techniques is generally available in the academic
community, but user-friendly software is limited.
It was recognized that if EPA adopts proprietary
software, strategies would be needed to make the
modeling process, tools and results transparent and
available to the public. Adobe Acrobat, a read-only
version of which is available free to the public, may
-------
serve as a business model for adopting proprietary
software without inconveniencing the public.
The modeling approaches breakout group identified
several actions as being critically needed to
advance development of guidelines or best practice
descriptions to support population-level ecological
risk assessment, including:
• Creation of an RAF subgroup or team to
develop population modeling guidance
for risk assessment. Population modeling
should be viewed as an integral part
of population-level ecological risk
assessment.
• Development of a decision framework
(e.g., decision tree) for using population
ecology assessment tools and developing
scenarios is needed.
• Convening of a workshop in the near future
to develop and review population-level
ecological risk assessment guidance.
• Development of case studies
documentation. Case studies can be
compiled readily from available books and
other literature, but the demonstration of
relevance to decision making needs to be
carefully documented.
• Communication to managers. A document
on applying population-level ecological
risk assessment to management decisions in
a "weight of evidence" framework should
be helpful. Application of population
models to inform management decisions
and the value added by using them needs to
be demonstrated to increase confidence in
ecological models.
• The need to convince managers that
population-level ecological risk assessment
is an important line of evidence to
enhance ecological risk assessment could
be addressed by commissioning an EPA
Science Advisory Board (SAB) review of
the issue.
• Development of guidelines for
extrapolation of toxicity data overtime.
• Development of guidelines for linking
exposure models and population ecology
models. One member also stated that EPA
could use existing software for population-
level risk assessment (e.g., PATCH,
RAMAS) or develop additional tools
throughout the process. Guidelines would
be most effective with facilitated access to
modeling tools and active training in their
use.
Summarizing the deliberations of this breakout
group, the state-of-the-science of population
modeling is sufficiently well developed to use
such models in ecological risk assessments
informing EPA programs. Several population
modeling approaches are applicable to a wide
range of environmental management issues.
Although several technical issues arise in
applying models, these issues can be overcome.
The most important issue likely is to be one
of communicating the general usefulness
and specific applicability of population
modeling to EPA decision makers and project
managers. The practice of population modeling
is sufficiently developed to support EPA's
preparation of guidelines for population-
level ecological risk assessment that includes
population modeling approaches.
-------
-------
4.0
Summary of Expert Opinions
The summaries provided in Section 3 communicate
the breadth of opinions and input expressed
by workshop attendees during breakout group
conversations. Additional observations, issues
and suggestions were expressed during plenary
discussions. Although no attempts were made
to seek consensus on any particular issue or
topic, certain commonalities emerged over the
course of breakout and plenary interactions. Key
opinions with respect to workshop objectives are
summarized in this section.
4.1 Observational Approaches
Workshop participants generally agreed that
observational methods are well established in
the fields of ecology and conservation biology,
and that approaches based upon them have
unique advantages in reflecting realism with
respect to population responses to stressors in
the environment. In this regard, information
obtained through direct observation reflects
the effects of multiple stressors and influences
of compensatory mechanisms (e.g., density
dependence), but the relative contributions of
various effects and processes usually are difficult
to tease apart. Additionally, the variability
inherent to natural systems could at times mask
detection of some important stressor effects.
Observational approaches were thought to be
applicable to all tiers within a tiered assessment
protocol. Many noted, however, that the utility
of observational approaches might be limited
with respect to prospective assessments due to
imperfect transferability of study results beyond
the conditions and context within which they were
obtained. They also have limited value for helping
to evaluate decision alternatives, because the
information they produce reflects only the specific
circumstances in which they were conducted.
Data from observational studies can, however,
help to inform reassessments of past management
decisions. It was noted that new methods are
coming on line that can help to guide decisions
about the inferences that can be made using
observational data.
Developmental activities that were identified to
promote best practices included compilations of
case study examples of the use of observational
approaches to assess population risk, examples
of when such approaches failed to provide the
information needed to assess population risk, and
how observational studies influenced decisions.
Workshop participants highlighted the value
of catalogues and annotated descriptions of
available methods and observational data sets and
sources. Guidance in the form of decision trees
was suggested as being particularly helpful with
respect to assessment planning and interpretation
of observational study results. Participants also
noted that acceptance of the use of observational
approaches by decision makers could be facilitated
and enhanced through development of best
practices descriptions for effective communication.
4.2 Experimental Approaches
Participants generally felt that the methods
employed to provide data for and input to
population-level ecological risk assessment are
sufficiently well developed and informative to
warrant development guidelines or best practices
descriptions. Several experimental methods are
available, and sometimes even standardized,
that can measure the responses of experimental
populations to stressors directly, or that provide
data that can be extrapolated to the population level
of biological organization. Even so, additional
design considerations might be required to help
ensure that key hypotheses regarding mechanisms
of effect and other important ecological processes
can be evaluated as needed to inform environmental
decision making. In this regard, some experimental
designs likely have limited ability to incorporate
processes and interactions that can have important
population consequences, such as interspecific
competition and other forms of species interactions.
Careful planning during problem formulation
of the assessment will help to ensure use of
experimental designs and methods that provide
the information needed to quantify decision-
relevant risk.
-------
Because experiments inherently are
abstractions of nature and therefore cannot
include all ecologically relevant processes,
additional research and development might be
needed to improve the value of experimental
approaches to population-level ecological
risk assessment. Highlighted was the issue
of cross-species extrapolation. Many species
are not particularly amenable to experimental
manipulation, and when assessment goals focus
specifically on populations of that species,
their responses to stressors will need to be
extrapolated from those of surrogates. Some
progress could be made toward resolving
this issue by focusing upon mechanisms
of action and their ecological analogs, but
there likely will always continue to be
meaningful uncertainty whenever cross-species
extrapolations are required. Extrapolation
of organism-level measures to characterize
risk to populations might be less worrisome,
because a variety of modeling approaches are
available that can accommodate organism-
level attributes to project population dynamics.
Even so, attention is needed to help ensure that
experimental data are collected in the forms
and temporal frames required by extrapolation
models.
Other areas of valuable research include
development of approaches and data
that can link certain types of measures—
namely biomarkers and organism dose
concentrations—more directly to the key
demographic rates determining population
dynamics. In a similar vein, there might be
opportunities within the evolving technologies
of genomics and proteomics to develop
approaches that link data derived from
these techniques to population response.
Advancements in this area could produce
efficiencies in the collection of information for
assessing population risk. Finally, discussions
emphasized the potential value to be derived
from combining experimental methods
(including more tightly coupled laboratory
and field experiments) with modeling and
observations, as these approaches provide
complimentary and supplemental information
about risks to populations.
Several activities were identified that could
support communication of best practices.
Included are case study analyses, both
comparing the informational value of
population-level measurement endpoints
versus organism-level measurement endpoints
when assessments include populations as
environmental values to be protected, and
evaluating the efficacy of population-level
assessments with respect to the outcomes
of decisions based upon them. Associated
with this was a sense that descriptions of
experimental designs that promote use of the
resulting data in modeling evaluations, and
of how experimental data can best be used in
modeling applications, should be developed
to help focus experiments on generating the
most critical information needed. Included
were specific guidelines for performance of
LTREs, bucket tests, and so on. Also identified
was guidance about how experimental results
should be interpreted and communicated with
respect to population risk.
4.3 Modeling Approaches
Contributors to the workshop expressed
the opinion that population models and
the approaches to deploy them within
population-level risk assessment are well
established, and noted several compilations
of model descriptions and considerations
for their application to risk assessment in
the recent literature. Opinion was expressed
that the stressors under evaluation—and
especially the decision context—influence
which models and approaches provide the
most valuable information. It generally was
believed that population models can be
used advantageously in any level within a
tiered assessment protocol, and that they are
-------
important integrators of data and knowledge
gained through observational and experiment
approaches. Important drawbacks to modeling
approaches, however, include the skepticism often
expressed by decision makers about the degree
of realism captured by models and the accuracy
of their outputs, and concerns about assessment
transparency when stakeholder and decision maker
understanding of modeling is limited. A lively
plenary discussion centered around perceived
inconsistencies in the level of acceptance of
population models relative to chemical fate and
transport models (of which acceptance is high).
Associated with this was continuation of the
ongoing debate centered on the meanings and
desirability of model verification, validation and
evaluation.
In spite of the generally high regard held by most
workshop participants for population models,
certain developmental issues were highlighted
as important. Among these was advancement
in coupling population models more directly
to exposure models, particularly with respect
to physiological-based dose-response models.
Additional exploration of modeling philosophy
and approaches addressing the effects of multiple
stressors would enhance model realism and
likely accuracy. Issues associated with the form
and strength of density dependence as important
determinants of population response to stressor
exposure, although not directly ones of modeling
per se, might influence model realism and the
accuracy of assessment conclusions. Several
participants expressed the opinion that density
dependence might not be as important an issue as
some believe. Additional attention to developing
accessible implementation software and packages
also was highlighted as a need, although some
software is available commercially or as freeware.
A number of activities were identified that would
foster acceptance of good practices in the use
of modeling approaches in population-level
ecological risk assessment. Important among these
were development of a decision framework for
model selection, and best practices guidelines for
applying models and interpreting modeling results.
Identification of best practices for facilitating
communication directly with decision makers and
stakeholders could help to establish confidence
in the use of population models to inform
decision making. Documentation developed to
guide design of experimental and observational
studies performed in conjunction with modeling
approaches would help to ensure compatibility with
modeling needs in the form and accessibility of the
data those studies generate. Some participants also
expressed a need for guidelines to approaches for
extrapolating the effects reflected in toxicity data
through time.
4.4 Commonalities
Across Approaches
Several considerations expressed during the
workshop cut across assessment approaches
and reflected the general sentiments of many of
the participants. Most importantly in relation
to workshop objectives was the sense that the
science underlying population-level ecological
risk assessment is sufficiently mature to support
furtherance of best practices guidelines. Although
the various approaches have perceived benefits and
limitations relative to different decision contexts,
and attention to certain developmental needs is
desirable, opportunities for applying existing
techniques to inform decisions were identified
within almost all of EPA's regulatory programs.
Participants often articulated the opinion that
the three assessment approaches should best be
treated as interdependent and complementary,
and that the power and value of population-level
ecological risk assessment as a decision-informing
tool are enhanced when approaches are used in
combination. Also expressed was the sentiment
that a primary advantage of focusing attention
more explicitly on measurement endpoints and
analysis techniques that address population
attributes directly is an assessment more relevant
to decisions involving protection of populations.
Most workshop participants promoted greater use
of population-level ecological risk assessment as a
tool to inform environmental decision making.
Documentation, communication and training
were felt to be components critical to credible
performance, advancement and acceptance of
population-level ecological risk assessment
by practitioners, decision makers and the
public. Articulation of a framework uniquely
oriented toward planning, implementing and
-------
interpreting results of population-level ecological
risk assessments is especially important. This
framework could include considerations leading to
selection of assessment approaches (combinations
of experimental, observational and modeling
techniques) appropriate to the decision and its
context, perhaps organized in the form of a decision
tree. Compilations and catalogues of existing
techniques, models, designs and data could be
linked to the decision tree to aid assessment
planning and performance. Programs could be
developed to help ensure that practitioners are
appropriately trained in relevant techniques
and models. Specific best practices guidelines
could help to direct interpretation of data and
results, again with an eye toward the nature of
the decision they intend to inform. These might
summarize key aspects of ecological theory and
link to compilations of case studies as illustrations
of sound interpretation approaches. Additional
guidelines could support communication of
assessment results and their meaning to the end-
users of the assessment. And throughout, materials
and information should consider, be oriented
toward or tailored to the unique decisions and
contexts of EPA's programs.
Several cross-cutting issues will require attention
if guidelines are to be developed. Key among
these are considerations associated with pragmatic
definitions of assessment population in various
decision contexts. Although reasonable approaches
to address this particular issue exist, definition of
the assessment population has been problematic
for Superfund and certain other programs.
Somewhat related to this are considerations
about spatial scale and context, and time horizons
appropriate to various management goals and
decisions. Attention also is needed for identifying
those measurement endpoints most relevant
to population assessment endpoints and the
nature of risks being evaluated. And finally, all
acknowledge that assessment populations do not
exist in isolation from other populations. Species
interactions can have important and substantial
influences on population performance and the
risks associated with anthropogenic stressors.
Some of the techniques explored during the
workshop (especially observational approaches)
reflect or accommodate species interactions more
realistically than do others. Even so, the importance
of species interactions to assessment results, and
the uncertainties created when species interactions
are ignored, will require careful consideration as
the science of population-level ecological risk
assessment is applied to EPA programs.
-------
5.0
Technical Panel
Recommendations for Future Progress
The resounding sentiment of the experts assembled
in this workshop was that EPA and ecological
risk assessment practitioners alike would benefit
from guidelines or best practices documentation
concerning population-level ecological risk
assessment. The science underlying such
assessments is sufficiently well developed that
guidelines could be developed soon to promote
best practices with the understanding that such
guidelines would be updated on a regular basis as
the state-of-the-science and practice of population
level ecological risk assessment improves over
time. Based in large part on the opinions of these
experts, but also based on our individual and
collective professional perspectives, the RAF
Technical Panel recommends that the Forum
proceed with an effort to develop best practices
guidelines for population-level ecological risk
assessment. This section describes some of the
options and outputs that could be pursued in such
a project. Suggestions are offered only generally
about how to accomplish individual efforts and
the overall project. We do suggest, however, that a
phased implementation with multiple intermediate
products is likely to be successful.
The Technical Panel recommends a phased
approach to producing guidelines. Initial issue-
oriented white papers and summaries would help to
document the current states of science and practice
of technologies supporting population-level
ecological risk assessment, and could suggest how
EPA programs would benefit from a more explicit
focus on risk to populations. Opinion statements
would help to visualize how regulatory programs
could utilize information directly communicating
population risk to facilitate understanding of
the advantages and limitations with respect to
program mandates. Supporting white papers could
summarize EPA program policy with respect
to management goals, and how a more explicit
focus on population-level measures could support
the decisions to meet those goals. Additional
opinion papers that summarize current knowledge
could focus on inferences drawn about risks to
populations, and projecting future practices that
would be more inclusive of population risk.
Development of best practice guidelines likely
will require directed conversations involving
ecologists, practitioners and users of assessment
results. Workshops that enable such interaction
likely will be important steps to developing best
practices guidelines. Topics for deliberation include
detailed evaluations of methods, best approaches
for combining methods in relationship to decision
contexts, and the decision criteria and processes
that could lead to a planning and implementation
framework specifically for population-level
ecological risk assessment. Equally important is
development of guidelines for interpreting results
and assessment outcomes. Such guidelines could
be organized by assessment endpoint attribute, and
could describe a nested hierarchy of considerations
and conclusions for interpreting lines of evidence
generated by multiple assessment approaches.
Retrospective analyses of cases in which risks
to populations were assessed would provide
both examples for future assessments, and
opportunities to evaluate the efficacy of various
approaches. Either as part of this or as a separate
effort, considerations of the informative value to
environmental decision making of population-level
ecological risk assessment and the approaches used
would provide additional insights supporting best
practices guidelines. Case study evaluations could
be commissioned from groups of experts, or could
be conducted in focused workshop settings.
In a related vein, assembly of information
describing the methods, models and data sources
would help to improve the accessibility of these
tools to risk assessment practitioners. Compilations
could include annotations describing acknowledged
advantages and limitations of methods and
-------
models with respect to various risk problems,
environmental settings, stressors and decision
contexts. Catalogues pointing to key sources of
toxicological data, demographic and life history
information and extrapolation relationships would
facilitate access to critical information and would
help to promote the quality of future assessments.
Attention to education, communication and
outreach will be critical to the success of a RAF
project to develop best practices guidelines for
population-level ecological risk assessment.
Although past Forum efforts to provide general
training in this area have been quite successful,
further development of training modules to focus
specifically on key topics and methods likely would
improve their value to practitioners. Modules
communicating best practices for using population
risk information would support understanding,
and perhaps further adoption, of population-level
ecological risk assessment by EPA programs.
Recognition of the roles and contributions of
stakeholder groups and the general public in
environmental decision making will be important as
education and outreach materials are developed.
The Technical Panel believes that each of the
activities described will be important as the project
moves forward. It also suggests that a successful
approach to supplementing existing RAF guidelines
will be to release products in a phased manner
as they are developed, rather than to focus solely
on a single major contribution at the conclusion
of the project. Such an approach is likely to have
several advantages, including perhaps a more rapid
release of valuable information and guidelines,
an enhanced ability to incorporate advancements
in science and practice through time, and a more
timely and flexible responsiveness to evolving
Forum and Agency priorities.
-------
References
Akcakaya, H.R. and H.M. Regan. 2002. Population
models - metapopulations. In: Ecological Modeling
in Risk Assessment: Chemical Effects on Populations,
Ecosystems, and Landscapes (R.A. Pastorok, S.M.
Bartell, S. Person and L.R. Ginzburg, eds.). CRC Press,
Boca Raton, FL, USA, pp. 83-95.
Akcakaya, H.R., M.A. Burgman, O. Kindvall, C.C.
Wood, P. Sjogren-Gulve, J. Hatfield and M. McCarthy
(eds.). 2004. Species Conservation and Management:
Case Studies. Oxford University Press, New York, NY,
USA.
Akcakaya, H.R., J.D. Stark and T.S. Bridges. 2008.
Demographic Toxicology. Methods in Risk Assessment.
Oxford University Press, New York, NY, USA.
Anderson, C.N.K., H.C. Hsieh, S.A. Sandin, R. Hewitt,
A. Hollowed, J. Beddington, R.M. May and G. Sugihara.
2008. Why fishing magnifies fluctuations in fish
abundance. Nature 452:835-839.
Barber, M.C. 2008. Bioaccumulation and Aquatic
System Simulator (BASS) User's Manual, Version
2.2. U.S. Environmental Protection Agency, National
Exposure Research Laboratory, Athens, GA. EPA/600/
R-01/035. (Available online: http://www.epa.gov/
ceampubl/fchain/bass/index. htm)
Barnthouse, L.W, D.R. Marmorek and C.N. Peters.
2000. Assessment of multiple stresses at regional scales.
In: Multiple Stressors in Ecological Risk and Impact
Assessment: Approaches to Risk Estimation (S.A.
Ferenc and J.A. Foran, eds.). SETAC Press, Pensacola,
FL, USA, pp. 1-26.
Barnthouse, L.W, WR. Munns, Jr. and M.T. Sorensen
(eds.). 2008. Population-Level Ecological Risk
Assessment. Taylor & Francis, Boca Raton, FL, USA.
Barnthouse, L.W, G.W Suterll and A.E. Rosen.
1990. Risks of toxic contaminants to exploited fish
populations: influence of life history, data uncertainty,
and exploitation intensity. Environmental Toxicology
and Chemistry 9:297-312.
Barnthouse, L.W, G.W. Suter II, A.E. Rosen and
J.J. Beauchamp. 1987. Estimating responses offish
populations to toxic contaminants. Environmental
Toxicology and Chemistry 6:811-824.
Bennett, R.S. and M.A. Etterson. 2007. Incorporating
results of avian toxicity tests into a model of annual
reproductive success. Integrated Environmental
Assessment and Management 3:498-507.
Carroll, S. 2002. Population models - life history. In:
Ecological Modeling in Risk Assessment: Chemical
Effects on Populations, Ecosystems, and Landscapes
(R.A. Pastorok, S.M. Bartell, S. Person and L.R.
Ginzburg, eds.). CRC Press, Boca Raton, FL, USA, pp.
55-64.
Person, S. 2002. Population models - scalar abundance.
In: Ecological Modeling in Risk Assessment: Chemical
Effects on Populations, Ecosystems, and Landscapes
(R.A. Pastorok, S.M. Bartell, S. Person and L.R.
Ginzburg, eds.). CRC Press, Boca Raton, FL, USA, pp.
37-53.
Forbes, V.E., U. Hommen, P. Thorbek, P. Heimbacj, P.J.
Van den Brink, J. Wogram, H.-H. Thulke and V Grimm.
2009. Ecological models in support of regulatory risk
assessments of pesticides: developing a strategy for
the future. Integrated Environmental Assessment and
Management 5:167-172.
Hilborn, R. and M. Mangel. 1997. The Ecological
Detective: Confronting Models with Data. Monographs
in Population Biology 28. Princeton University Press,
Princeton, NJ, USA.
Kooijman, S.A.L.M. 2000. Dynamic Energy and Mass
Budgets in Biological Systems. Cambridge University
Press, Cambridge, UK.
Kooijman, S.A.L.M. and J.A.J. Metz. 1984. On the
dynamics of chemically stressed populations: the
deduction of population consequences from effects on
individuals. Ecotoxicology and Environmental Safety
8:254-274.
Levins, R. 1966. The strategy of model building in
population biology. American Scientist 54:421-431.
Mankin, J.R., R.V O'Neill, H.H. Shugart and B.W.
Rust. 1975. The importance of validation in ecosystem
analysis. In: New Directions in the Analysis of
Ecological Systems, Part 1 (G.S. Innis, ed.), Simulation
Series Councils Proc. Ser. Vol. 5, No. 1, Simulation
Councils, Inc., LaJolla, CA, USA, pp. 63-71. (Reprinted
-------
in Shugart, H.H. andR.V. O'Niell (eds.). 1979.
Systems Ecology. Dowden, Hutchinson & Ross, Inc.,
Stroudsburg, PA, USA.)
Mayer, F.L., G.F. Krause, D.R. Buckler, M.R. Ellersieck
and G. Lee. 1994. Predicting chronic lethality of
chemicals to fishes from acute toxicity test data:
concepts and linear regression analysis. Environmental
Toxicology and Chemistry 13:671-678.
Munns, W.R., Jr. and M. Mitro. 2004. Assessing
risks to populations at Superfund and RCRA
sites - Characterizing effects on populations. U.S.
Environmental Protection Agency, Ecological Risk
Assessment Support Center, Washington, DC, EPA/600/
R-006/038. (Available online: http://cfpub.epa.gov/ncea/
cfm/recordisplay. cfm?deid= 154006)
Newman, M.C. and J.T. McCloskey. 2000. The
individual tolerance concept is not the sole explanation
for the probit dose-effect model. Environmental
Toxicology and Chemistry 19:520-526.
Pastorok, R., S. Bartell, S. Person and L.R. Ginzburg
(eds.). 2002. Ecological Modeling in Risk Assessment:
Chemical Effects on Populations, Ecosystems, and
Landscapes. CRC Press, Boca Raton, FL, USA.
Peters, C.M. and D.R. Marmorek. 2001. Application
of decision analysis to evaluate recovery actions for
threatened Snake River spring and summer chinook
salmon (Onchorhynchus tschcnvytscha). Canadian
Journal of Fisheries and Aquatic Sciences 58:2431-2446.
Regan, H.M. 2002. Population models: individual-based.
In: Ecological Modeling in Risk Assessment: Chemical
Effects on Populations, Ecosystems, and Landscapes
(R.A. Pastorok, S.M. Bartell, S. Person and L.R.
Ginzburg, eds.). CRC Press, Boca Raton, FL, USA, pp.
65-82.
U.S. EPA. 1998. Guidelines for Ecological Risk
Assessment. U.S. Environmental Protection Agency,
Risk Assessment Forum, Washington, DC, EPA/630/
R095/002F. (Available online: http://www.epa.gov/raf/
publications/guidelines-ecological-risk-assessment.htm)
U.S. EPA. 2002. Generic Ecological Assessment
Endpoints (GEAEs) for Ecological Risk Assessment.
U.S. Environmental Protection Agency, Risk Assessment
Forum, Washington, DC, EPA/630/P-02/004F. (Available
online: http://www.epa.gov/raf/publications/geae.htm)
-------
Observational Breakout Group
Appendix A.
Workshop Attendees
MEMBER AFFILIATION
Glenn Suter EPA
Working Group Lead Office of Research &
Development
Mary Sorensen
Invited Co-Lead
Anne Fairbrother
ENVIRON
International Corp.
Parametrix, Inc.
EMAIL
suter.glenn@epa.gov
msorensen@environcorp.com
afairbrother@parametrix.com
PHONE
513-569-7808
770-510-5010
425-458-6306
Colleen Flaherty
Jason Grear
Mark Johnson
Charlie Menzie
Brad Sample
EPA
Office of Pesticide
Programs
EPA
Office of Research &
Development
U.S. Army
Center for Health
Promotion and
Preventive Medicine
Exponent, Inc.
CH2M Hill
flaherty. colleen@epa. gov
grear.j ason@epa. gov
Mark.Johnson@amedd.army.mil
camenzie@exponent.com
bsample@ch2m.com
703-305-0389
401-782-9615
410-436-5081
571-214-3648
(cell)
916-920-0300
Randy Wentsel
Jill Awkerman
EPA
Office of Research &
Development
EPA
Office of Research &
Development
wentsel.randy@epa.gov
awkerman.jill@epa.gov
202-564-3214
850-934-9230
Observational approaches include those that obtain data by monitoring the responses of
populations in the field to pollutants or other anthropogenic stressors, and to natural variables. The
analysis of such data is sometimes called "ecoepidemiology." These approaches can be used to:
• Describe the condition of an assessment population and determine the causes of spatial
and temporal variation in population attributes
• Generate exposure-response relationships directly from observational data
• Provide data to parameterize process-based models
• Provide data to test specific risk hypotheses and the predictions of process-based
models
-------
Experimental Breakout Group
MEMBER
Tom Forbes
Working Group
Lead
Diane Nacci
Invited Co-Lead
Todd Bridges
Rick Bennett
Tom Chandler
Bruce Duncan
Sandy Raimondo
Richard Sibly
John Stark
AFFILIATION
EPA
Office of
Environmental
Information
EPA
Office of Research &
Development
U.S. Army Corps
of Engineers
Engineer Research and
Development Center
EPA
Office of Research &
Development
University of South
Carolina
EPA
Region 10
EPA
Office of Research &
Development
University of Reading
Washington State
University
EMAIL
forbes.thomas@epa.gov
nacci.diane@epa.gov
PHONE
202-566-0810
bennett.rick@epa.gov
chandlgt@gwm .sc.edu
tchandler@sc.edu
duncan.bruce@epa.gov
raimondo.sandy@epa.gov
r.m. sibly@reading .ac .uk
starkj@wsu.edu
401-782-3143
Todd.S.Bridges@erdc.usace.army.mil 601-634-3626
218-529-5212
803-777-5032
206-553-0218
850-9342424
01183788461
253-445-4519
Experimental approaches involve controlled experiments (like toxicity tests) that expose
organisms or populations of organisms to varying levels of chemical, physical and biological
agents to evaluate population response. Experiments can be performed in a laboratory, field or
semi-field system. These approaches can be used to:
• Derive understanding of population responses directly from the data (e.g., population
growth rate, equilibrium abundance)
• Provide data to parameterize process-based models
• Provide data to test specific risk hypotheses and the predictions of process-based
models
-------
Modeling Breakout Group
MEMBER
Steve Newbold
Working Group
Lead
Rob Pastorok
Invited Co-Lead
Resit Akcakaya
Larry Barnthouse
Charlie Delos
Lev Ginzburg
Ed Odenkirchen
Brenda Rashleigh
Glen Thursby
AFFILIATION
EPA
Office of Policy,
Economics and
Innovation
Integral Consulting Inc.
State University of NY
Stony Brook
LWB Environmental
Services, Inc.
EPA
Office of Water
Applied
Biomathematics
EPA
Office of Pesticide
Programs
EPA
Office of Research &
Development
EPA
Office of Research &
Development
EMAIL
ne wbold. steve @epa.gov
PHONE
202-566-2293
rpastorok@integral-corp .com
akcakaya@life .bio. sunysb.edu
Barnthouse@lwb-env.com
delos.charles@epa.gov
lev(3)ramas.com
odenkirchen. edward@epa.gov
rashleigh .brenda@epa.gov
thursby.glen@epa.gov
206-230-9600
631-632-8605
513-894-4600
202-556-1097
631-751-4350
703-305-6449
706-355-8148
401-782-3178
Modeling approaches involve application of process-based population models to general and
specific risk problems to evaluate population response to varying levels of chemical, physical and
biological agents, and to natural variables. Process-based models are mathematical constructs that
estimate properties of biological populations such as growth rate or time to extinction, and are
based on estimates of underlying biological processes (such as survival rates) and environmental
change. These approaches can be used to:
• Project or forecast population-level consequences of changes in stressors and other
environmental conditions modeled for different management scenarios
• Evaluate the population-level consequences of changes in individual-level attributes
observed or measured using observational and experimental approaches
• Evaluate distributions of population outcomes through time and across space
• Inform the design of observational and experimental approaches for assessing
population risk
-------
-------
Appendix B.
Workshop Agenda
Workshop on Population-Level Ecological Risk Assessment
Crystal City Marriott, Crystal City, VA
16-18 June 2008
Agenda
16 June - Opening Plenary
9:00 Welcome
9:15 Workshop Overview
Objectives
Workshop approach
Planned output of workshop
9:45 Review of Relevant Past Activities
EPA
SETAC Pellston Workshop
Other
10:15 Break
10:30 EPA Programmatic & Regional Needs and Case Studies
Office of Water
OPPTS
OSWER
12:00 Lunch (on your own)
1:00 EPA Programmatic & Regional Needs and Case Studies (cont.)
Lee Hofmann
Executive Director, Risk Assessment Forum
Wayne Munns
Workshop Chair
Wayne Munns
Charles Delos
Ed Odenkirchen
David Charters
Bruce Duncan
Region 10
2:00 Other Needs and Case Studies
EPAOPEI
USACE
A Conversation Case Study
A European Union Perspective
4:30 Charge & Instructions for Breakout Group Discussions
5:00 Dinner (on your own)
7:00 Poster Session (tentative) Bruce Duncan
Case studies and uses of population-level ERA in environmental decision making,
and examples of observational, experimental and modeling tools used in estimating
assessment population response to anthropogenic and natural stressors (posters will
remain up throughout the workshop).
Steve Newbold
Todd Bridges
Jill Awkerman
Richard Sibly
Jerry Cura & Wayne Munns
-------
17 June - Breakout Groups
9:00 Last Minute Issues Seema Schappelle &
Jerry Cura
9:15 Breakout Groups
Observational Approaches Glenn Suter & Mary Sorensen
Experimental Approaches Tom Forbes & Diane Nacci
Modeling Approaches Steve Newbold & Rob Pastorok
(timing of lunch and breaks optional for each group)
1:30 Questions & Issues in Plenary Seema Schappelle &
Jerry Cura
2:30 Breakout Groups continue (as needed)
18 June- Group Report Outs and Summary
8:30 Breakout Group Presentations
Observational Approaches
Experimental Approaches
Modeling Approaches
10:00 Break
10:15 Summary Discussion Jerry Cura
Commonalities, and relative strengths and limitations of the three approaches
Answers to the three breakout questions
12:00 Working Lunch
1:00 Summary continues
2:30 Final Observations & Next Steps Wayne Munns &
Seema Schappelle
3:00 Workshop Adj ourns
-------
Appendix C.
Breakout Group Charge
Workshop on Population-Level Ecological Risk Assessment
Breakout Group Charge
Workshop participants have been assigned to one of the three breakout groups prior to the
workshop. Each breakout group consists of 10-15 population-level ecological risk assessment
experts and stakeholders, and two co-leads. Support will be provided to each group to capture
salient issues, information and discussion points on flip charts. This material will be used to
support breakout group plenary presentations on the last day of the workshop. The workshop co-
chairs and potentially others will be moving among groups over the course of the day to facilitate
discussions and to address concerns.
We ask that the deliberations and plenary presentations of each group be structured to answer
the following three questions, each with specific issues to consider. These questions should be
answered from the perspectives of your group's analytical approach and set of tools (observational,
experimental, or modeling). Breakout group report outs (-20 minutes each) should focus on the
answers to each question.
Breakout Group Questions
From the perspective of your group's analytical approach and set of tools (observational,
experimental, or modeling):
1. What specific approaches, methods and tools are available currently for performing population-
level ecological risk assessment? To what types of environmental decisions, risk problems and
environmental situations6 do they apply? Specific issues to consider include:
• nature and types of decisions potentially informed by the methods and tools employed in
this approach
• types of stressors that can be addressed by the methods and tools employed in this
approach
• nature of population-level effects that can be evaluated by the methods and tools employed
in this approach
• attributes of populations that can be characterized by the methods and tools employed in
this approach
Specific decisions, problems and situations to consider include:
• hazardous waste site remediation decisions at sites ranging from small to large & complex
• registration of new products (e.g., pesticides)
• land use decisions at a variety of spatial scales
• development of national, regional and local environmental criteria
• resource protection decisions
• resource extraction decisions
• waste (e.g., dredged material, industrial wastes) disposal decisions
-------
• environmental contexts within which the methods and tools employed in this approach can
be used
2. Identify the strengths, current limitations and tradeoffs associated with specific methods and
tools currently available for performing population-level ecological risk assessment in support
of EPA programmatic and regional decision making. What technical issues currently limit the
usefulness to environmental decision makers of information developed using the methods of
this approach for population-level ecological risk assessment? With what priority should these
issues be addressed to improve population-level ecological risk assessment? Specific issues to
consider include:
• availability and accessibility of analysis techniques and methods
• data and information requirements, and their availability
• applicability to different tiers of assessment (ranging from screening to refined)
• ability to characterize cause and effect relationships, and to partition causes of effect
among multiple potential stressors
• availability of methods to account for temporal and spatial scalar issues
• availability of methods to account for stochasticity, compensation and other forms of
potential uncertainty
• transferability of analysis results among stressors, species, ecosystems and
environmental situations
• degree to which analysis techniques and methods have been evaluated
• critical research and development needs addressing key scientific uncertainties
3. Is the current state-of-the-science and practice sufficient to support development of guidance
for performing population-level ecological risk assessment? Up to what point can that guidance
be developed (e.g., only broadly, detailed with respect to certain (specified) tools, etc.)?
Specific issues to consider include:
• degree of scientific acceptance of current techniques, methods and tools
• degree of standardization of existing techniques, methods and tools
• degree of scientific acceptance in interpretation of resulting information
• extent of existing documentation
• degree of transparency and understandability of approach by decision makers and
stakeholders
• accessibility of analysis techniques and other necessary information
• critical actions needed to facilitate development of guidance
-------
xvEPA
United States
Environmental Protection
Agency
U.S. Environmental Protection Agency
Office of Research and Development
Washington, DC 20460
Official Business
Penalty for Private Use
$300
PRESORTED STANDARD
POSTAGE & FEES PAID
EPA
PERMIT NO. G-35
Recycled/Recyclable Printed on paper that contains a minimum of
50% postconsumer fiber content processed chlorine free
------- |