EPA/630/R-94/003
July 1994
A REVIEW OF ECOLOGICAL ASSESSMENT CASE STUDIES
FROM A RISK ASSESSMENT PERSPECTIVE
VOLUME H
Risk Assessment Forum
U.S. Environmental Protection Agency
Washington, DC 20460
Printed on Recycled Paper
-------
DISCLAIMER
This document has been reviewed in accordance with U.S. Environmental Protection
Agency policy and approved for publication. Mention of trade names or commercial products does
not constitute endorsement or recommendation for use. Case study data and interpretations were
current as of the peer review workshops held in the fall of 1992.
-------
CONTENTS
Foreword jv
Report. Contributors . . . . . . . . v
Summary ........ vi
PART I. CASE STUDIES OVERVIEW 1
1. Introduction 1
2. Guide to the Case Studies . 2
2.1. Background , 2
2.2. Case Study Highlights 2
2.2.1. Problem Formulation 5
2.2.2. Analysis . 7
2.2.2.1. Characterization of Exposure . 7
2.2.2.2. Characterization of Ecological Effects 8
2.2.3. Risk Characterization . 9
3. Key Terms •..•„.. 12
4. References 13
PART II. THE CASE STUDIES ; . . 14
1. Assessing the Ecological Risk of a New Chemical Under the Toxic Substances
Control Act (Short Title: New Chemical Case Study) 1-1
2. Risk Assessment for the Release of Recombinant Rhizobia at a Small-Scale
Agricultural Field Site (Recombinant Rhizobia Case Study) 2-1
3. Ecological Risk Assessment of Radionuclides in the Columbia River System—
A Historical Assessment (Radionuclides Case Study) 3-1
4. Effects of Physical Disturbance on Water Quality Status and Water Quality
Improvement Function of Urban Wetlands (Wetlands Case Study) . 4-1
5. The Role of Monitoring in Ecological Risk Assessment: An EMAP Example
(EMAP Case Study) . 5-1
in
-------
FOREWORD
Since 1990, the Risk Assessment Forum of the U.S. Environmental Protection Agency
(EPA) has sponsored activities to improve the quality and consistency of EPA's ecological risk
assessments. Projects have included development of Agencywide guidance on basic ecological risk
assessment principles (Framework Report, U.S. EPA, 1992) and evaluation of 12 ecological
assessment case studies from a risk perspective (U.S. EPA, 1993). To complement this original set
of case studies, several new case studies were recently evaluated to provide further insight into the
ecological risk assessment process.
As with the original case studies, each of the five new case studies was evaluated by
scientific experts at EPA-sponsored workshops. Two workshops were held in September 1992 (57
Federal Register 38504, August 25,1992); these workshops were chaired by Dr. Charles Menzie
and included reviewers from universities, private organizations, and industry.
The new case studies expand the range of the first case study set by including different
kinds of stressors (radionuclides, genetically engineered organisms, and physical alteration of
wetlands) and programmatic approaches (premanufacture notice assessments under the Toxic
Substances Control Act and the EPA's Environmental Monitoring and Assessment Program). In
addition, the authors and reviewers of the new case studies were able to use EPA's Framework
Report as background information. Both sets of case studies provide useful perspectives concerning
application of ecological risk assessment principles to "real world" problems.
Dorothy E. Patton, Ph.D.
Chair
Risk Assessment Forum
IV
-------
REPORT CONTRIBUTORS
Dr. William van der Schalie (EPA) and Dr. Charles Menzie (Menzie-Cura & Associates,
Inc.) prepared this report. Mr. Thomas Waddell and Mr. James Morash (The Cadmus Group)
provided review comments on the draft case studies, and Mr. Morash also edited the case studies
following their revision after the workshops. The workshops were organized by Dr. van der
Schalie and Mr. Waddell, with the assistance of Dr. Menzie and Ms. Deborah Kanter of Eastern
Research Group. Case study authors and peer reviewers are listed at the beginning of each case
study (part II). R.O.W. Sciences, Inc., under the direction of Ms. Kay Marshall, provided
editorial assistance in the preparation of this report. The Cadmus Group, Eastern Research Group,
Menzie-Cura & Associates, and R.O.W. Sciences, Inc., were EPA contractors or subcontractors
for this effort.
-------
SUMMARY
As with the previous case studies report (U.S. EPA, 1993), this document uses case studies
to explore the relationship between the ecological risk assessment process and approaches used by
EPA (and others) to evaluate adverse ecological effects. In contrast to the earlier report, the
authors and reviewers of these case studies were able to use EPA's Framework for Ecological Risk
Assessment (Framework Report, U.S. EPA, 1992) as background information. However, even
though the case studies have been structured as described in the Framework Report, most were not
originally planned and conducted as risk assessments. This should be kept in mind when
considering each case study's strengths and limitations.
Some of the contributions of the case studies in this report to a broader understanding of
the ecological risk assessment process are highlighted below.
• The application of the framework approach to nonchemical stressors is explored.
Examples include biological stressors (genetically engineered microorganisms),
physical stressors (alteration of wetland function by a variety of physical
disturbances), and radioactivity (radionuclides in water).
• The relationship of ecological risk assessment to a major EPA monitoring program
(Environmental Monitoring and Assessment Program—EMAP) is described.
• Regional scale assessments (EMAP, wetlands) are included.
• Conducting an ecological risk assessment in a tiered fashion starting with minimal
exposure and effects data is illustrated by the premanufacture notice (PMN) review
carried out under the Toxic Substances Control Act.
While these cases are representative of the state of the practice in ecological assessments,
they should not be regarded as models to be followed. Rather, they should be used to attain a
better understanding of ecological risk assessment practices and principles. These case studies and
others being prepared will be used along with the Framework Report to provide a foundation for
future Agencywide guidelines for ecological risk assessment.
VI
-------
PART I. CASE STUDIES OVERVIEW
1. INTRODUCTION
.In 1990, the Risk Assessment Forum initiated an effort to develop Agencywide guidance for
conducting ecological risk assessments. This effort consists of several parts, as described below.
• Basic principles and terminology for ecological risk assessment are described in the
report Framework for Ecological Risk Assessment (Framework Report) that was
published in 1992 (U.S. EPA, 1992). • -I
« Scientific/technical background information for development of future EPA
ecological risk assessment guidelines will be contained in a series of issue papers
based on the Framework Report that are now in preparation.
• Case studies are being developed to provide "real world" examples of how
ecological risk assessments can be conducted. The first set of 12 case studies has
been published (U.S. EPA, 1993).
This report includes five additional case studies that have been peer-reviewed and organized
according to the ecological risk assessment process as described in the Framework Report. As with
the first case studies report, this document should be useful to EPA regional, laboratory, and
headquarters personnel conducting ecological risk assessments, as well as to interested individuals
from other federal and state agencies and the general public. The Risk Assessment Forum plans to
continue development of other case studies as a means of illustrating the application of ecological
risk assessment principles.
-------
2. GUIDE TO THE CASE STUDIES
2.1. Background
The case studies presented in part n of this report illustrate several types of ecological
assessments. As summarized in table 1, these cases involve:
• studies done under several different federal environmental laws;
• spatial scales ranging from local impacts to national impacts;
• different types of stressors (chemical, physical, and biological);
• a variety of ecosystems, including aquatic (freshwater and marine), wetlands, and
terrestrial; and
* measurement endpoints reflecting different levels of biological organization, ranging
from effects on individual organisms up to and including effects on ecosystems.
(See part I, section 3 for definitions of measurement and assessment endpoints.)
These case studies expand the range of the first case study set (U.S. EPA, 1993) by
including different kinds of stressors (radionuclides, genetically-engineering organisms, and physical
alteration of wetlands) and programmatic approaches (Pre-Manufacture Notice assessments under
the Toxic Substances Control Act and the EPA's Environmental Monitoring and Assessment
Program).
2.2. Case Study Highlights
This section highlights some common themes and principles gleaned through development
and review of these case studies. This section is organized according to the framework for
ecological risk assessment provided in the Framework Report (U.S. EPA, 1992) (see figure 1):
• Problem formulation, which is a preliminary scoping process;
• Analysis, which includes characterization of both ecological effects and
exposure; and
» Risk characterization, which highlights qualitative and quantitative conclusions,
with special emphasis on data limitations and other uncertainties.
-------
Table 1. Case Study Characteristics
No."
1
2
3
4
5
Short Title
New Chemical
Recombinant
Rhizobia
Radionuclides
Wetlands
EMAP
Relevant
Federal
Legislation1"
TSCA
TSCA
CERCLA/SARA,
CWA
CWA, EWRA
-
Spatial
Scale of
Assessment
National
Local
Local
Regional
Regional
Stressor
Type'
SC
B
CM
P, CM
P,CM
Ecosystem
Typed
A/F
T
A/F
W.A/F
A/M
Level of
Biological
Organization0
Individual
Individual
Individual
Ecosystem
Community
" Numbers 1-5 refer to the sections of part n of this report
b Legislation
CERCLA/SARA: Comprehensive Environmental Response, Compensation, and Liability Act (1980)/
Superfund Amendments and Reauthorization Act (1987)
CWA: Clean Water Act (1977)
EWRA: Emergency Wetlands Resources Act (1986)
TSCA: Toxic Substances Control Act (1976)
c Stressor types
B: Biological ; ,
CM: Mixture of chemicals
P: Physical Stressor
SC: Single chemical
d Ecosystem types
A/F: Aquatic—freshwater
A/M: Aquatic—marine or estuarine
T: Terrestrial : •.
W: Wetlands
e Highest level of biological organization for the measurement endppints used.
-------
Discussion
Between the
Risk Assessor
and
Risk Manager
(Planning)
Ecological Risk Assessment
i
Characterization ' Characterization
of
Exposure
|
of
Ecological
Effects
'','
.~ ,,
RISK CHARACTERIZATION
' '
Data Acquisition; Ver
on and Monitoring
t
Discussion Between the
Risk Assessor and Risk Manager
(Results)
Rjsk Management'
Figure 1. The framework for ecological risk assessment (U.S. EPA, 1992). The ecological
risk assessment framework is the product of a series of workshops and reviews that
involved both EPA and outside scientists. While the Framework Report has been a
critical first step in developing ecological risk assessment concepts, evolution of the
framework concepts is expected and encouraged.
4
-------
2.2.1. Problem Formulation
Problem formulation is an initial planning and scoping process for defining the feasibility,
breadth, and objectives for the ecological risk assessment. The process includes preliminary
evaluation of exposure and effects as well as examination of scientific data and data needs,
regulatory issues, and site-specific factors. Problem formulation defines the ecosystems potentially
at risk, the stressors, and the measurement and assessment endpoints. This information then may be
summarized in a conceptual model, which hypothesizes how the stressor may affect the ecological
components (i.e., the individuals, populations, communities, or ecosystems of concern).
Two of the most important themes that emerged ;from a review of the 12 case studies (U.S.
EPA, 1993) and that were clearly evident in the review of the five case studies presented in this
document are as follows:
• Thorough formulation of the problem and development of the scope are essential
first steps for a successful risk assessment.
• It is important to clearly articulate management issues at the beginning of an
assessment.
The strengths and limitations of the case studies '• often were related to the care taken in
formulating the problem and articulating management issues at the beginning of the assessment.
Examples in this set of case studies that demonstrate careful implementation of these steps include
the New Chemical and Radionuclides case studies.
Monitoring Programs
Can Provide Data
Useful for Problem
Formulation
The ElMAP case study was unique in that it illustrated how monitoring
data can be used at the problem formulation stage of an assessment.
As indicated in figure 1, data acquisition and verification and
monitoring provide information that supports all phases of ecological
risk assessment. The EMAP Near Coastal program in the Virginia
Biogeographic Province is an example of a provincewide monitoring
program in which data are collected using a systematic, probability-
based design that facilitates detection of spatially distributed events but
does not estimate intraannual variability or short-term episodic events.
The monitoring program obtains data throughout the province on a
variety of exposure and effects indicators. The indicators were chosen
based on past monitoring experience with regard to environmental
conditions in coastal systems. Associations between exposure and
effects indicators imply neither causality nor direct effects from
anthropogenic stressors. As noted in the EMAP case study, "It is
important to recognize that monitoring data alone will not be sufficient
for establishing the causal relationships necessary for developing a
complete analysis of ecological risk." Taken along with other
evidence, however, associations between exposure and effects
-------
indicators can be used to direct further study and to aid in problem
formulation.
The Framework Can
Be Applied to Such
Diverse Stressors as
Radionuclides and
Genetically Engineered
Organisms
Iterative Approaches
Are Useful for Defining
Problems and
Allocating Resources
The EMAP case study also illustrates how information obtained from
provincewide monitoring can be used in the problem formulation
phase for more local or regional risk assessments. The monitoring
tools and the design employed within EMAP can be applied to these
smaller spatial scales.
The previous review of 12 case studies (U.S. EPA, 1993) indicated
that the framework can be applied to chemical and physical stressors.
This was demonstrated further with the present set of five case studies,
which includes assessments of the environmental release of a new
chemical substance and physical modifications of wetlands. The
Radionuclides case study showed that the framework is applicable to
radionuclides as well as to hazardous chemicals.
The authors and reviewers of the case study on the release of
recombinant rhizobia, a genetically engineered organism, concluded
that application of the framework to microbial stressors is possible. It
was generally agreed, however, that the unique properties and
complexities of a living, changing stressor should be acknowledged in
the framework and in subsequent case studies with a similar focus.
Stressors potentially associated with the rhizobia were characterized as
either biological (i.e., pathogenicity, altered legume growth, microbial
competition, and gene release) or chemical (i.e., toxins and detrimental
metabolites); the reviewers of the case study found this to be a useful
approach. The case study authors found it difficult to select endpoints
and to decide whether these represented assessment or measurement
endpoints.
As noted in the Framework Report (U.S. EPA, 1992), ecological risk
assessments are frequently iterative, with data collection and analysis
performed in tiers of increasing complexity and cost The New
Chemical case study illustrates this process. Ecological risk
assessments are conducted for new chemical substances under the
Toxic Substances Control Act in EPA's Office of Pollution Prevention
and Toxics (OPPT). In these assessments, there is a progression from
a simple screening approach to more resource-intensive evaluations
based on the results of the simpler analysis, consideration of associated
uncertainties, and identification of data gaps. The authors note that
because of the large number of PMNs received annually by OPPT, the
only practical approach is to use conservative screening estimates
initially and to proceed to more detailed assessments 'only when
necessary.
-------
All Important Exposure
Scenarios Should Be
Considered
Exposure routes should be carefully considered during problem
formulation to ensure that the risk assessment is properly focused.
For example, in two of the case studies, the reviewers suggested that
additional routes of exposure could have been included in the risk
assessments. In the New Chemical case study, exposure to suspended
sediments was suggested, while in the Radionuclides case study,
potential uptake from food could have been evaluated in addition to
direct uptake from water.
2.2.2. Analysis
Analysis includes the technical evaluation of data on both potential exposure to stressors
(characterization of exposure) and the effects of stressors (characterization of ecological effects).
Characterizing exposure involves predicting or measuring the spatial and temporal distribution of a
stressor and its co-occurrence, or contact, with the ecological components of concern; < 8
characterizing ecological effects involves identifying and quantifying the effects elicited by a
stressor and, to the extent possible, evaluating cause-and-effect relationships.
2.2.2.1. Characterization of Exposure
Models Provided Useful
Tools for
Characterizing
Exposure
"ReaUty Checks" Are
Important for Exposure
Estimates Based on
Models
As with the previous compendium of case studies, this set
demonstrates that simple as well as more complex models can help to
characterize the exposure field. Selection of models should be based
on the goals of the assessment as well as the availability of data and
resources. In the New Chemical case study, a simple dilution model
was initially used to estimate exposure concentrations in receiving
water. Based on the results from this model, which showed that
exposures could result in risk to aquatic organisms, a more complex
model was used to provide a more accurate but less conservative
estimate of exposure. ;
While exposure models can be useful, some degree of model
verification is important to reduce uncertainty. The Radionuclides
case study used a bioaccumulation model to estimate dose. When the
predicted doses were checked against a set of measurements, the
model was found to be conservative in some respects. The reviewers
observed that exposure may not be reliably predicted from
radionuclide activity in water, given the high variance found in
bioconcentration factors. In the ReCombinant Rhizobia case study,
field measurements conducted after the risk assessment was completed
verified the literature-based predictions concerning off-site migration
of the rhizobia microorganisms.
-------
Evaluating Exposure to
Genetically Engineered
Organisms Poses
Special Problems
Biological stressors were not addressed in the Framework Report, but
are the subject of the Recombinant Rhizobia case study, which
highlights some of the difficulties associated with predicting and
monitoring the spread of a stressor that is a living organism. Exposure
evaluation is most challenging because of the organism's capacity to
interact with its environment and to evolve. Moreover, because it is
capable of growth and reproduction, the stressor can increase in
amount over time as compared with amounts of chemical stressors,
which are either conservative or decrease with time and/or distance
from sources.
2.2.2.2. Characterization of Ecological Effects
Effects Information Is
Developed From
Predictive Methods,
Literature Values,
Laboratory Studies, and
Field Programs
Most Effects
Information Is
Developed for
Individual Organisms
in Single-Species Tests
Multiple Stressors
Complicate Evaluations
of Causality
The case studies demonstrate the range in sources of information used
for characterizing ecological risks. The Radionuclides case study and
the Wetlands case study relied primarily on existing guidelines or
literature values to characterize effects. The potential effects of
rhizobia were based on greenhouse studies, while the EMAP case
study used a suite of field studies. The New Chemical case study
utilized quantitative structure-activity relationships (QSARs) based on
molecular weight and log K^ as one source of information concerning
toxicity. QSAR methods were particularly useful in this application
given the large number of PMNs that need to be evaluated by EPA.
This case study also relied on laboratory bioassays. The author noted
that larger-scale studies (e.g., of mesocosms) have not been used •
routinely because of cost considerations. Nonetheless, OPPT is
initiating field mesocosm studies to evaluate the use of laboratory tests
for predicting effects in the field.
Most of the effects information presented in the case studies is based
on small groups of organisms tested as individual species. Because
effects data on mortality, growth, and reproduction are developed for
the individual, there is a general lack of information on effects at the
population level. Assessment endpoints, however, often are expressed
in terms of populations or communities of organisms. Similarly, data
from single species of organisms are used to derive stressor levels that
will be protective of communities or ecosystems, without consideration
of indirect effects or interspecies interactions. The use of such
extrapolations is a continuing area of controversy and discussion in
ecological risk assessment.
Individual stressors do not occur in a vacuum in the real world.
Rather, accompanying the stressor of interest may be a host of other
chemical, biological, or physical stressors that may alter or.confound
the effects and risks associated with the subject stressor. Thus the
EMAP case study noted that results of monitoring do not necessarily
-------
indicate causality. Reviewers of the New Chemical case study noted
that the effects of the chemical could depend on the presence of other
chemicals in a complex effluent. While the Radionuclide case study
concluded that radionuclides posed little risk to important fish species
in the Columbia River, the limited scope of the case precluded
consideration of other chemical and physical stressors that may pose a
much higher risk to fish populations. The Wetlands case study
examined the effects on wetland water quality status of a range of
stressors, including physical and hydrologic disturbances and loss or
conversion of wetland habitat. Several stressors were present at most
of the study sites. A multiple regression approach was used to relate
the effects of different stressors to water quality impacts.
2.2.3. Risk Characterization
Risk characterization uses the results of the exposure and ecological effects analyses to
evaluate the likelihood that adverse ecological effects are occurring or will occur in association
with exposure to a stressor. Essentially, a risk characterization highlights summaries of the
assumptions, scientific uncertainties, and strengths and weaknesses of the analyses. Additionally, a
risk characterization evaluates the ecological significance of the risks with consideration of the
types and magnitudes of the effects, their spatial and temporal patterns, and the likelihood of
recovery.
Most of the Case
Studies Used the
Quotient Method to
Integrate Exposure and
Effects Estimates
Risks to Populations
Were Qualitatively
Discussed
The Quotient Method was used in three of the five case studies: New
Chemical, Wetlands, and Radionuclides. While the Quotient Method
does not measure risk in terms of a likelihood of effects at the
individual or population level, it does provide a simple benchmark for
judging risk potential. As such, it has been widely used. The most
common application of the Quotient Method in aquatic ecological risk
assessments is to compare an estimate of a maximum exposure
concentration to a water quality criterion for a chemical. While
reliance on the Quotient Method in the present set of case studies is
consistent with the previous set of 12 case studies (U.S. EPA, 1993),
development and use of other ecological risk integration techniques
that can provide actual risk estimates should be encouraged. When
the Quotient Method is used, at least a qualitative description of key
study uncertainties and limitations should be provided.
Both the previous and present set of case studies made only limited
attempts at directly estimating population-level risks. Typically, risks
are assessed at the individual level, and population-level risks then are
inferred from the presence of risks to individuals. It is indeed
probable that when estimates indicate little or no risk to individuals,
there is little or no risk to the population. However, when there are
risks to individuals, there may or may not be risks to the population.
Thus the extrapolation from risks to individuals to risks to populations
-------
is frequently discussed as an area of uncertainty within the risk
assessments.
Stressor-Response
Models Are Useful in
Both Predictive and
Retrospective
Assessments
Major Sources of
Uncertainty Should Be
Identified
A Weight-of-Evidence
Approach Can Be
Useful in Risk
Assessments
The previous set of risk assessments (U.S. EPA, 1993) illustrated the
value of stressor-response models in quantitative risk assessment. In
the present set, the Wetlands case study used regression techniques to
develop stressor-response models for water quality impacts resulting
from a wide range of physical stressors. The reviewers of this case
study noted that this empirical statistical model was a key feature of
the case study and provided a predictive component. However,
because this model is based on a particular set of physical and
hydrological characteristics, predictions of the model may or may not
be applicable to other urban wetlands.
The EMAP case study was retrospective in nature because it examined
the relationship between indicators of the status of ecological resources
and an array of stressors. Although this case study was not a risk
assessment, it clearly showed that an understanding of stressor-
response relationships would be an important component of any future
risk assessment that evaluated the causal links between sources,
stressors, and observed effects.
Uncertainties associated with the use of available data for risk
assessments were mentioned in most of the case studies. The New
Chemical case study described the use of fixed "assessment factors" to
deal with extrapolations between different types of data. The EMAP
case study cautioned against assuming causality based on apparent
associations derived from monitoring exposure and effects indicators.
The authors and reviewers of the case studies frequently pointed out
potential problems in extrapolating between species and from the
laboratory to the field, in accounting for the combined effects of
multiple stressors, and in interpreting the results of field tests.
Although it is important to identify the major sources of uncertainty in
a risk assessment, the presence of uncertainty does not necessarily
preclude use of the risk assessment for risk management decisions.
The availability of multiples sources of information can help to
strengthen a risk estimate even when individual lines of evidence are
not conclusive. For example, in the Recombinant Rhizobia case study,
the reviewers felt that the data from the greenhouse studies and field
tests by themselves were not convincing. However, the availability of
information characterizing the rhizobia strains and documenting the
effects of previous releases of other rhizobia helped strengthen the
overall risk assessment conclusion that the small-scale field test of the
recombinant rhizobia should proceed. The EMAP case study also uses
a weight-of-evidence approach in problem formulation (not risk
10
-------
formulation (not risk characterization). Stressor and effects
information derived from the monitoring program are used to identify
areas of greatest concern that may be candidates for ecological risk
assessment.
11
-------
3. KEY TERMS (U.S. EPA, 1992)
assessment endpoint—An explicit expression of the environmental value that is to be protected.
characterization of ecological effects—A portion of the analysis phase of ecological risk assessment
that evaluates the ability of a stressor to cause adverse effects under a particular set of
circumstances.
characterization of exposure—A portion of the analysis phase of ecological risk assessment that
evaluates the interaction of the stressor with one or more ecological components. Exposure
can be expressed as co-occurrence or contact, depending on the stressor and ecological
component involved.
conceptual model—The conceptual model describes a series of working hypotheses of how the
stressor might affect ecological components. The conceptual model also describes the
ecosystem potentially at risk, the relationship between measurement and assessment
endpoints, and exposure scenarios.
ecological component—Any part of an ecological system, including individuals, populations,
communities, and the ecosystem itself.
ecological risk assessment—The process that evaluates the likelihood that adverse ecological effects
may occur or are occurring as a result of exposure to one or more stressors.
exposure—Co-occurrence of or contact between a stressor and an ecological component.
measurement endpoint—A measurable ecological characteristic that is related to the valued
characteristic chosen as the assessment endpoint. Measurement endpoints are often
expressed as the statistical or arithmetic summaries of the observations that comprise the
measurement.
risk characterization—A phase of ecological risk assessment that integrates the results of the
exposure and ecological effects analyses to evaluate the likelihood of adverse ecological
effects associated with exposure to a stressor. The ecological significance of the adverse
effects is discussed, including consideration of the types and magnitudes of the effects, their
spatial and temporal patterns, and the likelihood of recovery.
stressor—Any physical, chemical, or biological entity that can induce an adverse response.
12
-------
4. REFERENCES
U.S. Environmental Protection Agency. (1992) Framework for ecological risk assessment. Risk
Assessment Forum, Washington, DC. EPA 630/R-92/001.
U.S. Environmental Protection Agency. (1993) A review of ecological assessment case studies
from a risk assessment perspective. Risk Assessment Forum, Washington, DC.
EPA/630/R-92/005. .......
13
-------
PARTH. THE CASE STUDIES
Authors of the case studies
included in this section were asked to
follow the format shown in the box on
the right. As you read the case
studies, it is important to keep several
points in mind:
• The original case studies
were not necessarily
developed as risk assessments
as defined in the Framework
Report. EPA notes that the
case studies are often partial
risk assessments that focus on
available information without
discussing other relevant
considerations such as the
uncertainties defined by a
limited data base.
At the workshops, each case
study was evaluated as to
whether it (1) effectively
addressed the generally
accepted components of an
ecological risk assessment, or
(2) addressed some but not all
of these components or,
instead, (3) provided an
alternative approach to
assessing ecological effects.
Case Study Format
Abstract. The abstract summarizes the
major conclusions, strengths, and limitations
of the case study.
Risk Assessment Approach. This section
clarifies any differences between the
ecological risk assessment approach used in
the case study and the general process
described in the Framework Report.
Statutory and Regulatory Background. The
statutory requirements for the study are
described along with any pertinent
regulatory background information.
Case Study Description. This contains the
background information and objective for the
case study, followed by the technical
information organized according to the
ecological risk assessment framework:
problem formulation, analysis
(characterization of exposure and
characterization of ecological effects), and
risk characterization. A comment box is
included at the end of each major section.
References.
The strengths and limitations of each case study are highlighted in comment boxes at
the end of the problem formulation, analysis, and risk characterization sections.
Author's comments address issues raised in the preceding text or reviewer remarks from
the peer review of the case study. Reviewers' comments include strengths, limitations, and
general observations concerning the case studies.
The authors who compiled the case studies did not necessarily conduct the research
upon which the case studies are based. References to the original research are provided
in each case study.
14
-------
The general characteristics of the case studies are summarized in table 1 (in part I). Case
studies are referenced by the section of this report in which they appear. (The corresponding short
titles of the case studies are given in table 1).
15
-------
-------
SECTION ONE
ECOLOGICAL RISK ASSESSMENT CASE STUDY:
ASSESSING THE ECOLOGICAL RISKS OF A NEW CHEMICAL
UNDER THE TOXIC SUBSTANCES CONTROL ACT
-------
AUTHORS AND REVIEWERS
AUTHORS
David G. Lynch
Office of Pollution Prevention and Toxics
U.S. Environmental Protection Agency
Washington, DC
Gregory J. Macek
Office of Pollution Prevention and Toxics
U.S. Environmental Protection Agency
Washington, DC
J. Vincent Nabholz
Office of Pollution Prevention and Toxics
U.S. Environmental Protection Agency
Washington, DC
COMPILED BY
Donald Rodier
Office of Pollution Prevention and Toxics
U.S. Environmental Protection Agency
Washington, DC
REVIEWERS
Richard E. Purdy (Lead Reviewer)
Environmental Laboratory
3-M Company
St. Paul, MN
Gregory R. Biddinger
Exxon Biomedical Sciences, Inc.
East Millstone, NJ
Joel S. Brown
Department of Biological Science
University of Illinois at Chicago
Chicago, IL
Robert J. Huggett
Virginia Institute of Marine Science
The College of William and Mary
Gloucester Point, VA
Scott M. Sherlock
Office of Pollution Prevention and Toxics
U.S. Environmental Protection Agency
Washington, DC
Robert Wright
Office of Pollution Prevention and Toxics
U.S. Environmental Protection Agency
Washington, DC '.
Freida B. Taub
School of Fisheries
University of Washington
Seattle, WA
Richard Weigert
Department of Zoology
University of Georgia
Athens, GA
1-2
-------
CONTENTS
ABSTRACT 1-7
1.1. RISK ASSESSMENT APPROACH ................... . . . . . . . ...... 1-8
1.2. STATUTORY AND REGULATORY BACKGROUND . ...... . . ........... 1-8
1.3. CASE STUDY DESCRIPTION . ..-. ... . . . . . 1-11
1.3.1. Background Information and Objective 1-12
1.3.1.1. Chemistry Report 1-12
1.3.1.2. Engineering Report *.. j .....:.' 1-12
1.3.1.3. Environmental Exposure Assessment 1-12
1.3.1.4. Ecological Hazard Assessment 1-12
1.3.1.5. Ecological Risk Assessment . 1-13
1.3.2. Problem Formulation 1-13
1.3.2.1. Stressor Characteristics 1-13
1.3.2.2. Ecosystem Potentially at Risk ... :...;... . , ... . . , . 1-13
1.3.2.3. Ecological Effects ,.. .... ........ ... . 1-13
1.3.2.4. Assessment Endpoints ........:.... 1-15
1.3.2.5. Measurement Endpoints ... . . . . . . . . . , . ... . . ......... 1-15
1.3.2.6. Conceptual Model . i-15
1.3.3. Analysis, Risk Characterization, and Risk Management—1st Iteration 1-17
1.3.3.1. Analysis: Characterization of Exposure 1-17
1.3.3.2. Analysis: Characterization of Ecological Effects 1-18
1.3.3.3. Risk Characterization . . 1-19
1.3.3.4. Risk Management . 1-20
1.3.4. Analysis, Risk Characterization, and Risk Management—2?d Iteration 1-21
1.3.4.1. Characterization of Ecological Effects . 1-21
1.3.4.2. Characterization of Exposure 1-21
1.3.4.3. Risk Characterization 1-21
1.3.4.4. Risk Management . ... 4 ... r .... 1-22
1.3.5. Analysis, Risk Characterization, and Risk Management—3rd Iteration 1-23
1.3.5.1. Characterization of Ecological Effects .........:......,,. 1-23
1.3.5.2. Characterization of Exposure . . . . . . ......... . . 1-24
1-3
-------
CONTENTS (continued)
1.3.5.3. Risk Characterization: Risk Estimation and Uncertainty Analysis . . . 1-24
1.3.5.4. Risk Management 1-24
1.3.6. Analysis, Risk Characterization, and Risk Management—4th Iteration 1-24
1.3.6.1. Characterization of Exposure 1-24
1.3.6.2. Risk Characterization , 1-24
1.3.6.3. Risk Management 1-25
1.3.7. Analysis, Risk Characterization, and Risk Management—5th Iteration 1-25
1.3.7.1. Characterization of Exposure 1-25
1.3.7.2. Risk Characterization—Risk Estimation 1-25
1.3.8. Risk Management—Final Decision 1-27
1.4. REFERENCES 1-33
APPENDIX A—QSARS BETWEEN NEUTRAL ORGANIC CHEMICALS
AND FISH AND GREEN ALGAL TOXICITY 1-A1
APPENDIX B—INPUT AND OUTPUT PARAMETERS FOR EXAMS II 1-B1
1-4
-------
LIST OF FIGURES
Figure 1-1. Structure of assessment for effects of a PMN substance ..... . . ......... 1-9
Figure 1-2. Flow chart and decision criteria for the ecological risk assessment of a
PMN substance .,..'.. . . ;. . . . , . . . . . . . , ; ... ............ 1-10
i t. ,
: '"' V LIST OF TABLES •—•'"..:'; '-'"- ;
Table 1-1. Physical/Chemical Properties of PMN Substance 1-14
Table 1-2. PECs for PMN Substance G*g/L) . . . . . ....!... .".''../.......... 1-19
Table 1-3. PMN Substance Stressor-Response Profile ........... . . . . ../.!... 1-20
Table 1-4. Summary of Five Risk Characterization Iterations . . . . . . . . .... .;. .... . . i-21
Table 1-5. PDM3 Analysis . . . ./.'. . . . . /. . ... . f. . ... . .\. . . ; . . 1-23
Table 1-6. (Estimated) Stresspr-Ressponse Profile for Benthic Organisms . ..'. . . .;. ; . . . . 1-23
Table 1-7. EXAMS II Analysis ... . .......... 1-24
' "*" - " ....
Table 1-8. Stressor-Response Profile for Chironomus tentans 1-26
LIST OF COMMENT BOXES
Comments on Problem Formulation 1-16
Comments on Characterization of Exposure 1-27
Comments on Characterization of Ecological Effects 1-28
Comments on Risk Characterization . 1-30
1-5
-------
CBI
CC
ChV
CSRAD
EEB
EETD
EXAMS H
HERD
K™
MATC
OPPT
PDM3
PEC
PMN
QSAR
SAR
SNUR
POTW
LIST OF ACRONYMS
confidential business information
concern concentration
chronic value
Chemical Screening and Risk Assessment Division
median effect concentration
Environmental Effects Branch
Economics, Exposure and Technology Division
exposure analysis modeling system
Health and Environmental Review Division
soil/sediment organic carbon-water partition coefficient
octanol-water partition coefficient
median lethal concentration
maximum acceptable toxicant concentration
Office of Pollution Prevention and Toxics
probabilistic dilution model
predicted environmental concentration
premanufacture notice
quantitative structure-activity relationship
structure activity relationship
significant new use rule
publicly owned treatment works
1-6
-------
ABSTRACT
This case study is an example of how the Office of Pollution Prevention and Toxics
(OPPT) conducts ecological risk assessments for new chemical substances. The Toxic Substances
Control Act requires manufacturers and importers of new chemicals to submit a premanufacture
notice (PMN) to EPA 90 days before they intend to begin manufacturing or importing. Because
actual test data are not required as part of a PMN submission, EPA uses structure-activity
relationships to estimate both ecological effects and exposure.
The PMN substance is a neutral organic compound. This class of compounds elicits a
simple form of toxicity known as narcosis. The toxicity of neutral organic compounds can be
estimated through quantitative structure-activity relationships, which correlate toxicity with
molecular weight and the octanol-water partition coefficient (log Kow). The subject PMN substance
has a log K,,w of 6.7. Compounds with such a log K,,w are not expected to be acutely toxic (no
effects at saturation over short exposure durations) but are expected to elicit chronic effects.
Actual testing of the PMN substance confirmed these predictions.
The manufacturer identified processing, use, and disposal sites adjacent to rivers and
streams. Because it was expected that the PMN. substance would be discharged to such
environments, pelagic and benthic aquatic populations and communities were considered to be
potentially at risk. Therefore,, the assessment, endpoint used in this case study was the protection of
aquatic organisms (e.g., algae, aquatic invertebrates, and fish). Measurement endpoints used to
evaluate the risks to the assessment endpoint were mortality, growth and development, and
reproduction.
Initial exposure concentrations were estimated using a simple dilution model that divided
releases (kg/day) by stream flow (millions of liters/day). Subsequent exposure analyses used a
probabilistic dilution model (PDM3) and the exposure analysis modeling system (EXAMS II).
PDM3 was used to estimate the number of days a particular effect concentration would be
exceeded in 1 year, and EXAMS II was used to estimate concentrations hi the water column and
sediments using generic site data. *
In risk characterization, the quotient, method was used to compare exposure concentrations
with ecological effect concentrations. A ratio of I or greater indicates a risk. The case study
presents five iterations of analysis and risk characterization. The first four iterations identified an
ecological risk and resulted in the collection of additional ecological effects test data and more
information on potential exposure to the PMN substance. The, final outcome was that the PMN
substance could be used only at the identified sites because there was uncertainty as to whether the
concern level (1 /tg/L) might be exceeded at sites not identified by the manufacturer.
OPPT terminology differs from terminology in EPA-'s Framework for Ecological Risk ' •
Assessment (Framework Report; U.S. EPA, 1992). For example, OPPT uses "Hazard
Assessment" instead of "Characterization of Ecological Effects." Otherwise, the OPPT ecological
risk assessment procedure follows the approaches and concepts described in the first- and second-
order diagrams of the Framework Report.
1-7
-------
1.1. RISK ASSESSMENT APPROACH
This case study follows EPA's Framework Report (figure 1-1); that is, it is composed of
three phases: problem formulation, analysis, and risk characterization.
The Office of Pollution Prevention and Toxics' (OPPT's) overall approach to assessing the
risks of new chemicals is to compare exposure concentrations with ecological effect concentrations.
The process often begins with simple stream flow dilution models that typically result in a worst-
case scenario. If a risk is ascertained, more detailed analyses are performed (figure 1-2). Because
of the paucity of data associated with premanufacture notice (PMN) submissions (see discussion
under Statutory and Regulatory Background), there is a heavy reliance on the use of structure-
activity relationships (SARs) to estimate ecological effects and develop a stressor-response profile.
Figure 1-2 does not include risk management options. In addition to obtaining additional
exposure and ecological effects information, risk management options can include a variety of
regulatory enforcement actions such as banning discharges to water or requiring pretreatment. In
any event, risk assessors must ascertain that a risk exists before risk managers can exercise their
management options.
The case study has the following strengths: (1) it relates measurement endpoints to an
assessment endpoint; (2) it demonstrates that ecological risk assessments can be conducted with
minimal ecological effect and exposure data; and (3) it demonstrates the usefulness of SARs in
establishing a stressor-response profile.
One weakness of the case study is the lack of a true quantification of the effects to the
assessment endpoint (populations of aquatic organisms). However, this is a weakness only from
the scientific point of view; it was not needed from the regulatory point of view. Another
weakness is that the risk assessors expected the PMN substance to bioconcentrate, yet they did not
analyze the potential risks to predators that might ingest contaminated prey.
1.2. STATUTORY AND REGULATORY BACKGROUND
The Toxic Substances Control Act (TSCA) provides for the regulation of chemicals not
covered by other statutes (e.g., Food, Drug, and Cosmetic Act; Federal Insecticide, Fungicide, and
Rodenticide Act). Enacted in 1976, TSCA regulates industrial chemicals such as solvents,
lubricants, dyes, and surfactants. TSCA requires the assessment and, if necessary, regulation of all
phases of the life cycle of industrial chemicals: manufacturing, processing, use, and disposal.
TSCA regulates two categories of industrial chemicals: (1) chemicals on the TSCA
Chemical Substances Inventory List and (2) new chemicals. The TSCA Chemical Substances
Inventory includes chemicals in commercial production between 1975 and 1979, and chemicals
reviewed under the PMN program and commercially produced after 1979. New chemicals are
those substances that do not appear on the TSCA inventory. Section 5 of TSCA requires
manufacturers and importers of new chemicals to submit a PMN to EPA before they intend to
begin manufacturing or importing. EPA has up to 90 days to evaluate whether the substance will
1-8
-------
Risk Assessment
PROBLEM FORMULATION
Stressors: Neutral organic compound.
Ecosystem(s) at Risk: Freshwater rivers and streams.
Ecological Components: Pelagic and benthic aquatic organisms, including
fish, invertebrates, and algae.
Endpoints: Assessment endpoint is protection of aquatic life from unreasonable
adverse effects due to exposure to industrial chemicals. Measurement
endpoints are effects on mortality, growth, development, and reproduction using
surrogate species.
ANALYSIS
Characterization
of Exposure
Concentrations of the PMN
substance in the water column were
estimated with a simple dilution
model and PDM3. EXAMS II was
used to estimate concentrations in
the water column and sediments.
Characterization of
Ecological Effects
QSAR and test data for algae,
fish, daphnids, and chironomids
were used to establish a stressor
response profile.
RISK CHARACTERIZATION
The Quotient Method was used to integrate exposure and effects estimates.
Ecological effect concentrations of concern were established by applying an
uncertainty factor of 10 to the most sensitive measurement endpoint
concentration.
i
Risk Management
Five iterations before
final management
decision
Figure 1-1. Structure of assessment for effects of a PMN substance
1-9
-------
Stepl. FOCUS MEETING
• Determine the most sensitive species and endpoint using actual
test data or QSAR. Estimate a chronic value whenever possible.
• Apply an Uncertainty Factor to obtain a concern concentration (CC).
• Calculate a Predicted Environmental Concentration (PEC) using a
simple stream flow dilution model as a worst case scenario for
concentrations in the water column.
Drop from
Review
Yes
Step 2. STANDARD REVIEW
• Obtain more information about Production, Use, and Disposal of the
PMN substance.
• Obtain additional ecotoxicological data (testing, analogs, QSAR).
• Estimate a chronic value (ChV) for the most sensitive species.
• Adjust the ChV with a margin of exposure (typically 10) to obtain a
new CC.
* Use additional release data and the Probabilistic Dilution Model
(PDM3) to estimate the number of days in one year that the CC is
exceeded. Further analyses could employ EXAMS II.
Additional
ecotoxicity
or fate
tests
Is the CC exceeded
more than 20 times
in one year?
Drop from
Review
Step 3. RISK MANAGEMENT OPTIONS
• Control releases of the PMN substance pending additional testing.
• Ban manufacture or use under Section 5f of TSCA.
Figure 1-2. Flow chart and decision criteria for the ecological risk assessment of a PMN
substance
1-10
-------
present an unreasonable risk of injury to human health or the environment. With good cause, EPA
can allow an extension of up to 180 days for the evaluation of the chemical.
In addition to the short review time allowed, there are three major problems associated with
evaluating PMNs. The first is the confidential business information (CBI) protection afforded by
TSCA. Under this clause, manufacturers and importers can designate many characteristics of the
PMN substance, such as chemical name, structure, intended uses, and site of manufacture and use,
as CBI. This information is not available to the public, and only personnel with TSCA CBI
security clearance and members of Congress can access the information. There are strict
safeguards against disclosure of the CBI (see text box on page 1-12). The second problem is that
manufacturers and importers submit approximately 2,000 Section 5 notices to EPA annually. The
third and perhaps the most important problem is that only the following information must be
submitted: chemical identity; molecular structure; trade name; production volume, use, and
amount for each use; by-products and impurities; human exposure estimates; disposal methods; and
any test data that the submitter may have. The manufacturer does not have to initiate any
ecological or human health testing prior to submitting a PMN. Only 4.8 percent of the PMNs
reviewed to date contain ecological effects data, and most of those data consist of acute toxicity
tests performed on fish (Nabholz, 1991; Nabholz et al., 1993a; Zeeman et al., 1993).
1.3. CASE STUDY DESCRIPTION
This case study describes how OPPT evaluates the ecological risks of a PMN substance.
The risk assessment begins with a worst-case analysis using a stream flow dilution model to
estimate environmental concentrations. This is the typical approach taken by OPPT, and it results
in very conservative estimates. Investigators initially use SARs to assess ecological effects, and the
quotient method to integrate exposure and effects estimates.
Because the initial assessment identified a risk, additional analyses were performed using
actual test data and PDM3. The second risk characterization indicated risks to pelagic and benthic
aquatic life; therefore, investigators used the exposure analysis modeling system (EXAMS II) and
generic site data to estimate concentrations in both the water column and sediments. Investigators
estimated toxicity to benthic organisms using chronic test data for daphnids and assumed that the
sediments would decrease toxicity by a factor of 10. The results of these analyses identified a risk.
The manufacturer then supplied OPPT with more precise data on the use and disposal of
the PMN substance. Investigators input this new information into EXAMS II, and the results
indicated little risk to benthic organisms at the identified sites. OPPT was ready to issue a consent
order to restrict use of the PMN substance to the identified sites; however, the manufacturer chose
to perform an actual test on benthic organisms using chironomids as the surrogate species. The
results of the tests indicated moderate toxicity and little risk to benthic organisms at the identified
sites. The final outcome was that EPA restricted the use of the PMN substance to the identified
sites because there was uncertainty as to whether the concern level (1 /ig/L) might be exceeded at
sites not identified by the manufacturer.
1-11
-------
Confidential Business Information (CBI)
The CBI provisions ofTSCA are
intended to protect manufacturers and
processors. Disclosure of chemical
structures, uses, and even sites can
provide competitors with proprietary
information. However, CBI is available to
the personnel involved with processing and
evaluating Section 5 notices. This case
study cannot provide certain information
because of the CBI disclosure restrictions.
Thus, this report does not reflect all
available technical information, because
certain details cannot be revealed to
persons who are not cleared for CBI. For
example, the technical assessors know the
chemical name and structure of the PMN
as well as the uses, sites, and releases,
but such information cannot be revealed in
this case study. Therefore, CBI does not
hamper the ecological risk assessment
process by EPA scientists who must be
cleared initially for CBI before gaining
access to such information. In addition,
they must be certified on an annual basis
to maintain their access to CBI. Once
personnel move to positions that no longer
require access to CBI, their clearance for
access to such information is terminated.
1.3.1. Background Information and
Objective
OPPT performs the following analyses
in assessing the human and ecological risks of
PMN substances. For a more detailed
discussion of the process, see U.S. EPA
(1986), Nabholz (1991), and Nabholz et al.
(1993a).
1.3.1.1. Chemistry Report
The Industrial Chemical Branch of the
Economics, Exposure and Technology
Division (EETD) evaluates PMNs to ensure
that: (1) the chemical name matches
structure, (2) the chemical/physical properties
are accurate, (3) the information about
manufacturing and processing is accurate, and
(4) the uses are consistent with the chemical.
1.3.1.2. Engineering Report
The Chemical Engineering Branch of
EETD estimates worker exposure during the
life cycle of the chemical (manufacturing,
processing, use, and disposal) and estimates
releases of the chemical to the environment.
1.3.1.3. Environmental Exposure
Assessment
The Exposure Assessment Branch of
EETD evaluates available fate, transport, and
abiotic and biotic fate parameters. This is analogous to the exposure profile discussed in the
Framework Report. The exposure assessment estimates the environmental concentrations likely to
occur during the life cycle of the PMN substance. This includes an evaluation of potential
exposure from releases to surface waters, landfills, and land spray, as well as nonoccupational
exposures. Environmental concentrations can be site-specific or generic. PMN substances
frequently are discharged to water; therefore, most exposure assessments address aquatic
environments, chiefly rivers and streams.
1.3.1.4. Ecological Hazard Assessment
Also known as a toxicity assessment, the ecological hazard assessment is analogous to a
stressor-response profile and is performed by the Environmental Effects Branch (EEB) of the
1-12
-------
Health and Environmental Review Division (HERD). The initial ecological hazard assessment
evaluates the potential adverse ecological effects of a PMN substance and relies chiefly on SAR
For many classes of discrete organic chemicals (about 50 percent of which are neutral organic
chemicals), quantitative structure-activity relationships (QSARs) are available that permit an
estimation of acute and chronic effects to surrogate species such as fish, aquatic invertebrates and
algae (Auer et al., 1990; Clements, 1988; Nabholz et al., 1993a, b; Zeeman et al, 1993) HERD
will review the results of submitted test data and, if the results are valid, incorporate them into the
hazard assessment.
1.3.1.5. Ecological Risk Assessment
The Chemical Screening and Risk Assessment Division (CSRAD) conducts both human
health and ecological risk assessments. Ecological risk assessments are conducted in a tiered
fashion (figure 1-2). Initial hazard and exposure assessments are evaluated at a FOCUS meeting to
ascertain whether a potential risk exists. If the FOCUS meeting does not identify a risk the
chemical may be dropped from further review. If a risk is identified, the PMN substance undergoes
a more detailed assessment called a standard review. Alternatively, additional information may be
requested immediately following the FOCUS meeting. If a risk is still identified after all additional
information has been submitted, then risk management options are considered. Possible risk
management options are: (1) control options (such as no releases to water) pending further tests of
the PMN substance, (2) issuance of a significant new use rule (SNUR), and (3) direct control under
Section 5(f) (e.g., banning the manufacture or use of the PMN substance).
1.3.2. Problem Formulation
1.3.2.1. Stressor Characteristics
Table 1-1 lists the physical/chemical properties of the subject PMN substance. The
manufacturer declared the chemical identity, structure, intended uses, and sites of use as CBI This
particular example evaluated only the parent compound, because investigators did not expect the
PMN substance to degrade or be transformed into more toxic metabolites.
1.3.2.2. Ecosystem Potentially at Risk
The processing, use, and disposal sites are adjacent to rivers and streams. Investigators also
expected the PMN substance to be discharged to such rivers and streams. Thus, pelagic and benthic
aquatic populations and communities may be at risk.
1.3.2.3. Ecological Effects
The PMN substance belongs to a class of chemicals known as neutral organic compounds
These chemicals are nonelectrolyte and nonreactive and exert toxicity through a narcotic of
nonspecific mode of action (Auer et al., 1990; Lipnick, 1985; Veith and Broderius, 1990) Neutral
organic compounds can exert both acute and chronic effects. The toxicity of neutral organic
compounds has been correlated with molecular weight and the logarithm of the octanol-
1-13
-------
Table 1-1. Physical/Chemical Properties of PMN Substance
Property
Measured or Estimated Value
Chemical Class
Chemical Name
Chemical Structure
Physical State
Molecular Weight
LogKow
Water Solubility
Vapor Pressure
Neutral Organic
CBI
CBI
Liquid
232
6.7a
6.56b
0.051 mg/L (estimated)0
0.30 mg/L (measured)
< 0.001 Torr @ 20°Cd
"Estimated using CLOGP program (Leo and Weininger, 1985).
"Estimated by a regression equation developed by Karickhoff et al. (1979). The average method error
for the log K^ was 0.2 log K^ units over a log Koc range of 2 to 6.6.
°Estimated by a regression equation developed by Banerjee et al. (1980).
Estimated by a regression equation cited in Grain (1982).
water partition coefficient (Kow). Experimental data have shown that neutral organics with a log
K of 5 0 or more do not exert pronounced acute effects (toxic effects such as mortality or
immobilization within 4 days). This is mainly, due to the low water solubility of such compounds,
which results in decreased bioavailability to aquatic organisms. Because of the decreased
bioavailability, exposure durations of 4 days or less are insufficient to elicit marked acute effects
(e g as measured by a 96-hour LCso1 test). Because of the high Kow of this PMN substance
investigators expected only chronic effects to occur at or below the chemical's aqueous solubility
limit.
OPPT typically assesses ecological effects for three trophic levels: primary producers
(algae) primary consumers (aquatic invertebrates), and forage/predator fish. Investigators use the
most sensitive species and toxicological effect for the initial risk assessment. Unless only chronic
effects are expected, such as the PMN substance in this case study, OPPT usually assesses both
acute and chronic effects. The ecological effects characterization is based on effects on mortality,
growth and development, and reproduction. The rationale and approach used to assess these
effects are presented under Measurement Endpoints.
lrThe LCso is the median lethal concentration.
1-14
-------
1.3.2.4. Assessment EmHpoints
TSCA was intended to prevent unreasonable risks to health and the environment as a result
of the manufacture, processing, use, and disposal of industrial chemicals. The assessment endpoint
(Surer, 1990) used hi this case study is the protection of aquatic organisms (algae, aquatic
invertebrates, and fish). The investigators assumed that any effects from the PMN substance
would be exhibited at least up to the population level of organization.
1.3.2.5. Measurement Endpoints
Investigators used the following measurement endpoints (Suter, 1990) to assess the risks to
the assessment endpoint:
• mortality; v - .
• growth and development; and
• reproduction.
Clements (1983) and U.S. EPA (1984) present the rationale for selecting these endpoints.
To summarize, documented evidence indicates that xenobiotics can adversely affect these endpoints
both directly and indirectly. Since populations are governed by mortality, growth and
development, and reproduction, Investigators presumed that adverse effects to these measurement
endpoints would manifest themselves at least up to the population level of ecological organization.
Thus, there is a logical connection between the assessment endpoint (i.e., the protection of aquatic
life, at least up to the population level) and the measurement endpoints.
OPPT uses a tiered approach when testing the toxicity of a given industrial chemical (U.S.
EPA, 1983; Smrchek et al., 1993; Zeeman et al., 1993). The first tier consists of relatively
Inexpensive short-term tests that measure effects chiefly on mortality to fish and aquatic
invertebrates and population growth for green algae (the three trophic levels discussed under
Ecological Effects). The first tier or "base set" consists oif a 96-hour fish acute test, a 48-hour
daphnid test, and a 96-hour algal test. Because the algal test represents exposure across about
eight generations of algal cells, OPPT considers the algal test to be representative of chronic
toxicity to algal populations. Additional tiers consist of chronic tests, such as the fish early life
stage toxicity test that measures effects on mortality and growth and development, and the daphnid
chronic test that measures effects on survival and reproduction. Investigators must ascertain a risk
before proceeding to these additional tests.
1.3.2.6. Conceptual Model
Based on experience with neutral organic compounds and available QSARs, the high log
K,,w for the PMN substance indicated a risk of chronic toxicity to benthic and pelagic aquatic
organisms. Principal concerns were for effects on mortality, growth and development, and
reproduction. Investigators presumed that these effects would be manifested at least up to the
population level of organization (Clements, 1983).
1-15
-------
A preliminary exposure profile was developed through the use of simple stream flow
models. To characterize ecological effects, QSARs were used to develop an initial stressor-
response profile (Clements, 1988). The QSARs established which trophic level (i.e., algae, fish,
aquatic invertebrates) would be the most sensitive, and were developed from actual tests of neutral
organic compounds using surrogate species (U.S. EPA, 1982) that represented aquatic organisms in
rivers and streams.
Assessment factors (U.S. EPA, 1984; Nabholz, 1991; Nabholz et al., 1993a) were used to
address uncertainties in extrapolating from laboratory to field effects. Investigators used a quotient
method of ecological risk characterization to assess risk (Barnthouse et al., 1986; Nabholz, 1991;
Rodier and Mauriello, 1993). If the results of the risk characterization predicted an unreasonable
risk, investigators planned to perform a more in-depth analysis including fate and transport
modeling and ecological effects testing hi accordance with EEB ecological effect test guidelines
(U.S. EPA, 1985). The PDM3 and EXAMS II models would further characterize and refine
exposure, and additional ecological effects testing of the PMN substance would be based on the
criteria established by OPPT (U.S. EPA, 1983). Investigators would continue to use the quotient
method to characterize risks.
Comments on Problem Formulation
Strengths of the case study include:
• The process is scientific and judged to be adequate.
• The case study is a good example of the PMN process.
Limitations include:
• Much of the information is confidential and is unavailable to the reviewers.
• The problem formulation section should present more detail on potential
ecological effects.
• The PMN process appears to consider chemicals singly and not as part of a
complex mixture in the environment. Other chemicals might interact with the
chemical of interest, thereby changing exposure and/or toxicity.
• There should be some discussion as to the potential for transformation products
and what might be done if they were known to be produced.
General reviewer comments:
• This case study addresses all components of a risk assessment listed in the
EPA's Framework Report.
• Future PMN assessments should include fairfy realistic, yet simple,
bioaccumulation models.
1-16
-------
Comments on Problem Formulation (continued)
Author's comments:
« Using a general assessment endpoint, such as the protection of aquatic
organisms, helps to communicate the significance of risks determined with
measurement endpoints. Risk managers might not be familiar with the surrogate
species used in PMN testing or the significance of the test results (e.g., £C50,
MATC).
• Given the volume ofPMNs received annually, the approach of using
conservative methods initially and then proceeding to more detailed assessments,
as necessary, is the only practical approach.
« Generic assessments cannot identify specific biota at risk. This often is
considered a shortcoming; however, given the conservative exposure estimates
provided by the stream flow models, the lack of information about biota at
specific sites, and the use of assessment factors for projecting ecological effects,
it is not unreasonable to assume that the risk assessment will protect a wide
array of aquatic organisms.
« TSCA gives no legislative authority to regulate mixtures of chemicals. TSCAis
written to address each chemical individually.
9 OPPT always considers potential transformation products during assessments.
If a persistent and/or more toxic transformation product could be formed from a
PMN substance, OPPT would assess the product in the same way as the parent
compound was assessed. In this case, no transformation products of concern
were identified.
9 PMN assessments do include bioaccumulation models when they are needed.
Fish ingestion models by humans is a standard model run for all PMN
substances. Fish ingestion by predators is assessed if a potential concern has a
likely probability of occurring. In the early stages of this case, the assessor
knew that food chain transport could be a problem. Late in the assessment, the
company submitted fish bioconcentration data for a close analog, which showed
that the measured fish bioconcentration factor of the PMN substance would be
much lower than predicted. Therefore, exposure to human populations and
predators through fish ingestion was not evaluated further.
1.3.3. Analysis, Risk Characterization, and Risk Management—1st Iteration
1.3.3.1. Analysis: Characterization of Exposure
Because the use of the PMN substance is CBI, only the terms Manufacturing, Processing,
Use, and Disposal are used to describe the life cycle of the compound. The sites of manufacture,
1-17
-------
use, and disposal are CBI, and this draft considers the actual releases that were used to calculate
concentrations of the PMN substance hi receiving rivers and streams as CBI.
1.3.3.1.1. Stressor Characterization
The compound has low water solubility and is not expected to volatilize from water because
of the low vapor pressure. Photodegradation is negligible, and the compound is expected to sorb
strongly to sediments. The half-life for aerobic degradation could be weeks; anaerobic degradation
could require months or longer.
1.3.3.1.2. Exposure Analysis
In the first iteration, investigators used a simple stream flow dilution model to calculate
predicted environmental concentrations (PECs). The calculation was based on the following
algorithm:
Concentration = Releases (kg/day) / Stream flow (millions of liters/day)
The PEC calculations use both mean and low flow rates. In addition, the initial OPPT
exposure analysis typically ranks stream flow rates and uses the 10 percent and 50 percent flow
rates. The measured solubility limit of 0.3 mg/L was used.
Investigators determined that there would be no significant releases during the manufacture
of this PMN substance. The most significant routes of exposure would result from the use and
disposal of the chemical. Effluents containing the PMN substance would first be treated in publicly
owned treatment works (POTW), which are wastewater treatment plants that include primary and
biological treatment of the incoming waste stream. POTWs normally are located off-site or
between the processing plant and the receiving river. To assess the extent of removal of the PMN
substance by POTWs, investigators used data from laboratory-scale wastewater treatment
experiments and the output from mathematical wastewater treatment simulations. The results
indicated that removal would be due largely to adsorption to sludge; however, the analysis assumed
approximately 10 percent of the PMN substance released from treatment was in the effluent sorbed
to solids. This assumption was based on typical solids removal for secondary wastewater treatment
systems.
This study did not consider the fate and ecological effects of the PMN substance hi sludge.
1.3.3.1.3. Exposure Profile
Table 1-2 lists the PECs for the PMN substance during manufacture, use, and disposal.
1.3.3.2. Analysis: Characterization of Ecological Effects
OPPT initially used QSAR to estimate the ecological effects of the PMN substance. The
manufacturer contacted EPA prior to submitting the PMN and was briefed on concerns about
1-18
-------
Table 1-2. PECs for PMN Substance
Process
Manufacture
Use
Disposal
Mean Flow
10%a
0.0
9.0
52.3
50%
0.0
0.5
0.7
10%
0.0
68.0
90.2
Low Flow
50%
0.0
4.0
6.1
"Percentage of streams having flows equal to or less than the value used to calculate the PECs.
chronic effects. As a result, the manufacturer submitted a fish acute test and a fish early life stage
test.
1.3.3.2.1. Stressor-Response Profile
Table 1-3 summarizes the QSAR-derived effect concentrations and the results of the fish
acute and fish early life stage tests.
1.3.3.3. Risk Characterization
Five risk characterizations were performed in this case study. Table 1-4 provides a brief
summary of the assumptions, estimations, and types of uncertainty for each of the five iterations.
1.3.3.3.1. Risk Estimation (Integration and Uncertainty Analysis)
Investigators used the quotient method to estimate ecological risks. A quotient of 1 or
greater indicates a risk. The algorithm is given below:
Risk Quotient = PEC/CC
Normally, OPPT calculates the concern concentration (CC) by identifying the most
sensitive species and effect from the stressor-response profile and applying an assessment factor.
In this case, investigators used the measured chronic value (ChV) of 0.013 mg/L for the fathead
minnow rather than the estimated ChV of 0.004 mg/L for the daphnids (table 1-3). To account for
the uncertainty between chronic effects noted in the laboratory and those that might occur in the
field, an assessment factor of 10 was used (see text box on page 1-22). The ChV was divided by
the assessment factor to yield a CC of 0.0013 mg/L, which was rounded off to 0.001 mg/L or
1
In estimating risk, the CC of 1 /tg/L was compared to the PECs (table 1-2). As can be
seen, the CC was exceeded at both low and mean flow for 10 percent of the streams, and at low
flow for 50 percent of the streams. A risk was inferred based on mean flow.
1-19
-------
Table 1-3. PMN Substance Stressor-Response Profile
QSAR Estimated Toxicity"
Endpoint
Fish 96-hr LCJO
Daphnid 48-hr LC50
Green Algae 96-hr EC50b
Fish ChV°
Daphnid ChV
Algal ChV
Effect Concentration
No effect at saturation
No effect at saturation
No effect at saturation
0.002 mg/L
0.004 mg/L
No effect at saturation
Reference
Veith et al. (1983)
Hermens et al. (1984)
Appendix A
Appendix A
Hermens et al. (1984)
Appendix A
Actual Measured Toxicity
Fathead Minnow (Pimephales
r No effect at saturation
U.S. EPA (1993)
promelas) 96-hr Acute Test
P. promelas Early Life Stage 0.013 mg/L
Test, 31-day ChV (growth,
mean wet weight)
P. promelas Early Life Stage 0.061 mg/L
Test, 31-day ChV (survival,
growth [length])
U.S. EPA (1993)
U.S. EPA (1993)
•Based on molecular weight and log Kow.
bMedian effect concentration.
The ChV is the geometric mean of the highest concentration for which no effects were observed
and lowest concentration for which toxic effects were observed. The ChV is essentially the
geometric mean of the maximum acceptable toxicant concentration (MATC).
It should be noted that the initial risk assessment evaluates risks to aquatic species in the
water column only.
1.3.3.4. Risk Management
Because the results of the initial risk characterization identified a potential unreasonable
risk, investigators requested a chronic daphnid test to complete the chronic tier tests. EPA also
informed the submitter that a benthic test with contaminated sediments could be required if there
was a potential unreasonable risk to sediment-dwelling organisms. The concern for benthic
organisms was based on the high Kow, low vapor pressure, and low water solubility, which indicate
that the PMN substance was likely to partition to the sediments of rivers and streams, resulting in
1-20
-------
Table 1-4. Summary of Five Risk Characterization Iterations
Iteration
1
Estimates/Assumptions
Fish are the most sensitive suedes (
Uncertainty
Chronic \Vnrst-rras*» angltreio
4
5
effects at 1 /tg/L. PMN substance mixes
instantaneously in water. No losses.
Actual test data for daphnids still indicate a ChV
of 1 /tg/L. Determine how often this
concentration is exceeded using PDM3.
Estimate risk to benthic organisms using daphnid
ChV and mitigation by organic matter. EXAMS
II used to estimate concentrations.
Site-specific data obtained on use and disposal.
EXAMS II rerun with new data.
Actual test data for benthic organisms obtained.
Worst-case analysis.
Other species may be
more sensitive.
Generic production
sites. Actual data for
benthic organisms not
available.
Estimated toxicity for
benthic invertebrates.
Best estimates for
identified sites. May
not hold for other sites
or uses.
exposures of benthic organisms. EPA also requested a coupled units test (40 CFR 796.3300) to
simulate the effectiveness of a POTW in removing the PMN substance.
1.3.4. Analysis, Risk Characterization, and Risk Management—2nd Iteration
1.3.4.1. Characterization of Ecological Effects
A daphnid chronic toxicity test was conducted and found to be acceptable (i.e., it followed
OPPT guidelines and good laboratory practices). The ChV for survival, growth, and reproduction
was 0.007 mg/L.
1.3.4.2. Characterization of Exposure
The coupled units test is a measure of the ultimate biodegradation of the PMN substance
under conditions that simulate treatment in activated sludge. The POTW simulation conducted by
the manufacturer indicated that a POTW would remove from 95 percent to 99 percent of the PMN
substance.
1.3.4.3. Risk Characterization
Investigators used PDM3 (U.S. EPA, 1988) to estimate the number of days out of 1 year
that the CC will be exceeded. Like the simple stream flow model, PDM3 assumes that the
1-21
-------
Uncertainty Assessment Factors
OPPTuses assessment factors to
attempt to address three types of
uncertainty:
• Uncertainty regarding differences in
species sensitivity to toxicants.
• Uncertainty regarding the differences
between concentrations eliciting acute
effects and those causing chronic
effects.
• Uncertainty regarding comparisons of
laboratory studies to field conditions.
Assessment factors range from 1 to
1,000. The particular assessment factor
used for a chemical mil vary inversely with
the amount and type of data available.
Examples are shown below. A complete
discussion can be found in U.S. EPA
(1984).
Examples of Assessment Factors
Available Data
Acute toxicity QSAR
or test data for one
species
QSAR or test data for
fish, algae, and
aquatic invertebrates
QSAR or chronic
toxicity data for fish
or aquatic
invertebrates
Actual field study
Assessment Factor
1,000
100
10
1
chemical will mix instantaneously with water
and no losses will occur through any
physical, chemical, or biological
transformations. Flow rates were obtained
from the U.S. Geological Survey.
Investigators continued to use the CC
of 1 /ig/L, since the daphnid ChV of 0.007
mg/L divided by the assessment factor of 10
rounds off to 0.001 mg/L or 1 fig/L. Table
1-5 presents the results of PDM3.
1.3.4.3.1. Interpretation of
Ecological Significance
As a matter of policy, OPPT infers an
unreasonable risk to aquatic organisms if a
CC for chronic effects exceeds 20 days or
more. The 20-day criterion is derived from
partial life cycle tests (daphnid chronic and
fish early life stage tests) that typically range
from 21 to 28 days in duration. OPPT infers
a reasonable risk if the CC is exceeded less
than 20 days. It is important to remember
that the PDM3 model estimates only the total
number of days out of 1 year that the CC is
exceeded. The days are not necessarily
consecutive, and thus the 20-day criterion is a
conservative one. This iteration showed an
unreasonable risk to aquatic organisms from
the PMN substance because the CC was
exceeded 20 days for use and 39 days for
disposal (table 1-5).
1.3.4.4. Risk Management
EPA notified the company that a
potentially unreasonable risk to aquatic
organisms still existed. A meeting was held
to discuss possible benthic toxicity tests and
to clarify unanswered questions regarding
releases of the PMN substance through use
and disposal. It also was decided to evaluate
exposure further through the use of EXAMS
II (Burns, 1989).
1-22
-------
Table 1-5. PDM3 Analysis*
Process
Exceedance (days/year)
Manufacture
Use
Disposal
0
20
39
"Releases to water considered CBI. PMN substance was expected to be released 350 days/year,
and a 95 percent removal from POTW was assumed.
1.3.5. Analysis, Risk Characterization, and Risk Management—3rd Iteration
1.3.5.1. Characterization of Ecological Effects
Currently, there are no SARs for neutral organics and aquatic benthic organisms; however,
SARs do exist for neutral organics with earthworms in artificial soil. To estimate the ecological
effects of the PMN substance to aquatic benthic organisms, predictions from the fish 14-day LC50
QSAR (Konemann, 1981) were compared with the earthworm 14-day LC50 QSAR. The
earthworm 14-day LC50 was about 10 times higher than the fish 14-day LCj0. Investigators
assumed that the organic matter (i.e., ground peat) in the artificial soil mitigates the toxicity of
neutral organic chemicals by about 10 times.
Investigators further expected that the organic matter in natural sediments would mitigate
the toxicity of the PMN substance by at least a factor of 10, because natural organic matter in
natural sediments should be more efficient at binding neutral organic chemicals than freshly ground
peat in artificial soil. That is, sediment organic matter is likely to have a larger surface area-to-
volume ratio than ground peat and, therefore, have more sites to bind hydrophobic compounds.
Proceeding on the above assumption, the effective concentrations in the toxicity profile for water
column were multiplied by 10 to produce the stressor-respdnse profile for benthic organisms (table
1-6). This scenario used the best data available at the time for neutral organic compounds, and the
PMN submitter accepted the rationale for mitigation.
Table 1-6. (Estimated) Stressor-Response Profile for Benthic Organisms
Organism
Invertebrate
Invertebrate
Vertebrate
Endpoint
14-day LC50
21-day ChV
31-day ChV
Effect Level
(mg/kg dry weight)
0.3
0.10
0.3 to 1.0
1-23
-------
1.3.5.2. Characterization of Exposure
A preliminary EXAMS II analysis at the worst site indicated concentrations ranging from
11.2 to 21.8 mg/kg dry weight of sediment after 1 year of releases of the PMN substance.
Appendix B presents the critical input parameters for EXAMS II and an example of the output.
1.3.5.3. Risk Characterization: Risk Estimation and Uncertainty Analysis
The most sensitive endpoint was the invertebrate 21-day ChV of 0.1 mg/kg. An
assessment factor of 10 was applied to derive a CC of 0.01 mg/kg or 10 /tg/kg. The quotient
method was used. As can be seen from .the initial EXAMS II analysis, the exposure concentrations
exceeded the CC by factors of 1,000 to 2,000.
1.3.5.4. Risk Management
The manufacturer initiated an extensive site-specific evaluation of the releases of the PMN
substance during uses and disposal, and submitted new exposure information to OPPT for
evaluation. The report is CBI.
1.3.6. Analysis, Risk Characterization, and Risk Management—4th Iteration
1.3.6.1. Characterization of Exposure
OPPT used the additional information to conduct another EXAMS II analysis. Table 1-7
summarizes the results for three representative sites.
Table 1-7. EXAMS H Analysis
Water Column
Site Otg/L)
1 0.004
2 0.001
3 0.008
Sediments
(mg/kg)
0.019
0.014
0.038
1.3.6.2. Risk Characterization
There was not enough of a risk to benthic organisms to warrant a ban pending a testing
decision by OPPT.
1-24
-------
1.3.6.3. Risk Management
A decision was made to offer the company a consent order to allow manufacturing but
require a benthic/sediment toxicity test to confirm the toxicity profile and thus the risk assessment.
Prior to offering the consent order, the company volunteered to test with a benthic organism using
contaminated sediment. The submitter and OPPT agreed to a 28-day chironomid toxicity test.
1.3.7. Analysis, Risk Characterization, and Risk Management—5th Iteration
1.3.7.1. Characterization of Exposure
Table 1-8 presents the results! of the chironomid toxicity test.
1.3.7.2. Risk Characterization—Risk Estimation
A CC of 2.0 mg/L was set for the benthic community based on the most sensitive effect, a
ChV of 23 mg/kg for survival and emergence. The CC was 50 times higher than the highest PEC
for sediments, and the ChV was 600 times higher. Thus, there did not appear to be an
unreasonable risk to benthic organisms as a result of the use and disposal of the PMN substance
over 1 year.
As can be seen from table 1-7, concentrations of the PMN substance were three orders of
magnitude lower than the concern level of 1 jig/L for water column organisms at the specific sites
of use and disposal.
1.3.7.2.1. Uncertainty
In this case study, the three main types of uncertainty with regard to ecological effects are
variations in species-to-species sensitivity, uncertainty regarding acute versus chronic effects, and
uncertainty regarding extrapolating laboratory-observed effects to those that might occur in the
natural environment. U.S. EPA (1984) developed assessment factors specifically for establishing
concentrations of concern for PMN substances. Use of these factors is not intended to establish a
"safe" level for a particular substance, but rather to identify a concentration which, if equaled or
exceeded, could result in some adverse ecological effects. Such a finding provides the rationale for
requesting either actual testing of the PMN substance or more specific information about fate and
exposure. Naturally, there are other types of uncertainty, such as the effects of the PMN
substance on adult rather than juvenile fish. Such types of uncertainty are research issues.
In the case of the exposure profile, an important aspect of uncertainty has to do with the
actual duration of exposure. The PDM3 model predicts only the number of days out of 1 year that
the CC will be exceeded (table 1-4). These days are not necessarily consecutive days. Thus, only
flow rates could be used to account for seasonal variation. The presence or absence of critical life
stages of aquatic organisms cannot be accounted for with this type of analysis. In addition, the
generic nature of the assessment precludes identification of specific biota.
1-25
-------
Table 1-8. Stressor-Response Profile for Chironomus tentans
Endpoint
Effect Level
(mg/kg dry weight sediment)
14-day ChV
21-day EC50 emergence
25-day EC50 emergence
28-day EC50 emergence
28-day LC50 survival
ChV survival
ChV emergence
32
23
25
24
22
23
23
1.3.7.2.2. Risk Description—Ecological Risk Summary
This case study demonstrates the validity of QSAR in establishing toxicity profiles for
water quality organisms (fish, invertebrates, and algae). In this case, the chemical structure
indicated that the PMN substance was closely analogous to chemicals known to behave like neutral
organic compounds. The high Kow indicated that the compound would not be acutely toxic, and
this was confirmed by an actual test with a surrogate fish species. Actual chronic toxicity testing
confirmed the QSAR-predicted chronic toxicity (within an order of magnitude). EPA's experience
with other high-Kow compounds such as hexachlorobenzene and chloroparaffins further confirms
the chronically toxic nature of such compounds. The predictions for chironomid toxicity did not
agree with the actual test data. QSARs have not been developed for benthic organisms simply
because not enough test data are available to permit such analyses.
The use of QSAR is not limited to neutral organic compounds. Currently, there are
QSARs available for compounds that show more specific modes of toxicity or excess toxicity over
the neutral organics. Examples include acrylates, methacrylates, aldehydes, anilines,
benzotriazoles, esters, phenols, and epoxides (Auer et al., 1990; Clements, 1988).
Because the CCs were exceeded enough times out of 1 year, the PDM3 model indicated a
risk to aquatic organisms. When actual sites were analyzed using EXAMS II, no unreasonable
risks were identified.
1.3.7.2.3. Ecological Significance
There appears to be no unreasonable risks to pelagic and benthic organisms at the identified
use sites. The potential risk posed by the PMN substance bioaccumulating through the aquatic
food web was thought not to be significant.
1-26
-------
1.3.7.2.4. Spatial and Temporal Patterns of the Effects
CBI restrictions preclude revealing the uses and specific sites for the PMN substance
PiCK , teC^°al assessors ide°tified important river systems that could be affected by 'this
PMN substance. Thus, if there was a risk, the effects are not likely to be localized.
1.3.7.2.5. Recovery Potential
The PMN substance is a neutral hydrophobic chemical. This mode of toxicity is akin to a
simple narcosis type of action (Auer et al., 1990; Veith and Broderius, 1990) that Slversible if
exposure to the toxicant is terminated before lethality or death occurs.
? ""?* P°tential W3S n0t evaluated' Short-term pulsed exposure is not likely to cause
1S
1.3.8. Risk Management— Final Decision
The risk managers agreed that the PMN substance posed no unreasonable risks to pelagic
^ the^cmc^s of use and disposal. However, there could be risks at ofher
^ ?d dlSP°Sal °f *e PMN SUbstanCe" Therefore' the final di^ition was a
PMM I"/ f ^ SgamSt releashlg concentrati° suspended particle
-* feeding or bioconcentration.
1-27
-------
Comments on Characterization of Exposure (continued)
• The exposure analysis should have considered the fate of the sludge from the
POTW. Such sludge is often applied to agricultural or forest land.
• More detail about the discharges should be given, even if it is something like
"one large" or "several small"
• For chemical products that are mixtures, there may be a large number of
chemicals present, and this may contribute to variability in estimates as well as
measurements. The mixture can result in exposure conditions in the
environment different from those for the original material.
Author's comments:
• The simple stream flow model offers a conservative estimate of exposure by
assuming instantaneous mixing and dispersion of the chemical. The model does
not take into account any losses due to factors such as volatilization,
partitioning, or chemical or biological degradation after release. Because of
the paucity of data and information about exposure, the use of conservative
models is justified.
• The PDM3 model is an improvement over the simple stream flow models in that
the temporal nature of exposure can be evaluated. Thus, a risk manager can be
advised as to how often a particular concentration is likely to be exceeded.
• The above two models estimate chemical concentrations in the water column
onfy. As demonstrated in the study, more in-depth models such as EXAMS II
can be used to estimate chemical concentrations both in the water column and
sediments when sufficient data are available.
Comments on Characterization of Ecological Effects
Strengths of the case study include:
• The case study illustrates the iterative approach associated with the evaluation
of a PMN chemical.
Limitations include:
• Estimating a concern level for sediment organisms based on earthworm data is
not appropriate because earthworms exchange gases with air and sediment
organisms exchange gases with water. The statement that sediment organic
carbon will mitigate toxidty 10 times more than soil or peat carbon requires
additional support.
1-28
-------
Comments on Characterization of Ecological Effects (continued)
The authors should give a good rationale for their approach to estimating
sediment toxicity and tell why it is better than the equilibrium partitioning
method other offices in EPA are using, or they should use the equilibrium
partitioning method.
The case study should not state that it demonstrates the validity ofQSAR in
establishing toxicity profiles. The estimated 21-day chronic value of 0.100 was
230-fold lower than the test results (tables 1-6 and 1-8). A single case study
would not be sufficient to demonstrate the validity of using QSAR to establish
toxicity profiles, no matter what might have been shown.
Author's comments:
The use ofSAR is commonplace within OPPT because TSCA does not require
the completion of test data prior to the submission of a PMN. The track record
with SAR is extremely good (Nabholz et al., 1993b) and has resulted in dropping
low-risk chemicals from review and regulating high-risk chemicals without
measured data on the chemical. Provided sound expert judgment is employed,
SAR can identify whether acute or chronic tests (or both) are needed.
The use of surrogate species at different trophic levels (e.g., fish, daphnids,
algae, benthic organisms) permits one to evaluate which organisms are most
sensitive to a given xenobiotic. While many argue that the commonly used
surrogates may not be as sensitive as those in the wild, both industry and
government agree that it is the most practical way to evaluate the ecological
effects of chemicals. Because many industrial chemicals are used in a wide
array of industries as well as consumer products, identifying specific biota (at
the species level) is often impossible. \
The cost of larger scale studies such as laboratory microcosms and field
mesocosms has precluded their use to assess the ecological effects of new
chemicals. However, OPPT is initiating field mesocosm studies at EPA's Duluth
Environmental Research Laboratory to evaluate how well laboratory tests
predict effects in the field.
1-29
-------
Comments on Characterization of Ecological Effects (continued)
• Comparing a fish 14-day LC50 with an earthworm 14-day LC50 value in an
artificial soil was the only available way at" the time to estimate the effect that
organic matter would have on the hioavailabilitv of an organic chemical in
sediments. It was known that (1) earthworms interacted intimately with soil
pore water, (2) the toxicity of organic chemicals in soil toward earthworms
could be predicted by relating molar concentration in soil pore water to the
chemical's Kow (van Gestel and Ma, 1990), (3) K^ was highly correlated with
Koc, and (4) the amount of organic matter in sediments strongly influenced the
amount of organic chemical that could be absorbed by sediments. It was a
simple and valid extrapolation to use earthworms as a surrogate for benthic
organisms. In addition, when the OPPT assessment team conferred with the
submitter's assessors, the submitter accepted OPPT's best estimate given the
level of knowledge and available data that existed at the time.
• Sediment organic matter was expected to be more efficient than the ground peat
used in the artificial soil of the earthworm toxicity test because sediment organic
matter is generally more finely divided due to more processing by invertebrates
and partial degradation by microbes: Sediment organic matter was expected to
have a much greater surface area to volume ratio than the peat and, therefore,
a much greater absorptive area to reduce the bioavailability of organic
chemicals with high Kow values (i.e., >4.2).
Comments on Risk Characterization
Strengths of the case study include:
• The risk characterization appeared to be adequate for a management decision.
i
•' The case study illustrates the PMN risk assessment process.
Limitations include:
• The summary table of the major assumptions and estimates used at various
stages of the process (table 1-4) should have included some information on the
magnitude of uncertainty associated with each of these estimates or assumptions.
• The risk assessment methods employed do not distinguish risks to individuals
from risks to populations. In some cases "individuals" are the organizational
level of interest, while in other cases it is the "population. "
1-30
-------
Comments on Risk Characterization (continued)
« // was pointed out that assessment factors were developed in 1984 and continued
to evolve along with the PMN process. The case study should include the
method used to derive the assessment factors and a brief statement of the history
of these factors.
» The case study should clarify the difference between "uncertainty" in the
statistical sense and "uncertainty" as it is addressed by using "assessment
factors."
« Assigning an assessment factor of "1" to field toxicity data is not appropriate
because field data are site-specific, and the data may not be directly
transferable to other sites where the chemical might be used or released.
• The available information suggested that risks to benthic organisms were
probably more important than those to pelagic organisms. Yet, the process was
carried out in a specific manner that emphasized the studies on pelagic
organisms first. It was pointed out that this was policy.
• If disposal options were considered along with the risk assessment, then various
options could have been considered early in the process. Such mitigation could
be included as an iteration.
• If "acceptable" concentrations were first identified, then it would be possible to
estimate acceptable loadings.
General reviewer comments:
• The case study description of interactions between risk assessors and risk
managers led the reviewers to discuss the following:
Who is the risk manager? It appears to be a manager at EPA, but the
manager at the company also can manage risk by deciding not to test
further and abandoning the chemical, or he could deal with potential
exposure by treating the waste stream or making process changes.
— It might be useful to develop a framework for risk managers similar to
that for risk assessors. Both frameworks should contain sections on
interaction with the other and on mitigation.
• The example chemical does not indicate how well the process works for other
types of chemicals. Narcoleptic chemicals are the easiest chemicals to model
for toxicity. Reactive toxicants often cannot be modeled simply, if at all, and
they are usually more toxic or hazardous.
1-31
-------
Comments on Risk Characterization (continued)
• Reviewers discussed the use of "probabilistic" risk assessment. It was noted that
this is the direction in which EPA is going. A question was raised regarding
whether these quantitative methods would be understandable to the risk
manager. Some experience indicates that they would.
Author's comments:
• The Quotient Method is the most common ecological risk assessment method
used in OPPTfor new and existing chemicals. It also is used by the Office of
Pesticide Programs. The Quotient Method is easy to use, is mutually accepted
by industry and EPA, and is amenable to the ecological effects and exposure
data available to OPPT under TSCA.
• One disadvantage of this method is the uncertainty about the degree of risk
when quotients approach, but do not equal, 1. Also, it is difficult to quantify
risks to assessment endpoints when most ecological risk assessments under TSCA
are generic. While the risks to measurement endpoints can be quantified,
extrapolating such risks to the population or community level is impossible
unless simulation models are employed. OPPT is evaluating developmental
versions of population and ecosystem models for use with existing chemicals;
however, due to the volume ofPMNs, their use is not practical at this time.
This is particularly true for ecosystem models that require mainframe or high-
speed/high-memory computers. Thus, only qualitative inferences can be made
between measurement and assessment endpoints.
• Since 1979, OPPT has assessed the environmental toxicity of over 24,000
chemicals submitted under Section 5 of TSCA. Although only 4.8 percent of
those chemicals had any environmental toxicity information submitted with them,
OPPT has been able to use chemical structure and commonly measured
physical/chemical properties to model the aquatic toxicity of many classes of
reactive toxicants, including 64 classes of organic chemicals that have some type
of specific toxicity in addition to narcosis.
1-32
-------
1.4. REFERENCES
Auer, C.M.; Nabholz, J.V.; Baetcke, K.P. (1990) Mode of action and the assessment of chemical
hazards in the presence of limited data: use of structure-activity relationships (SAR) under
TSCA, Section 5. Environ. Health Perspect. 87:183-197.
Banerjee, S.; Yalkowsky, S.H.; Valvani, S.C. (1980) Water solubility and octanol/water partition
coefficients of organics. Limitations of the solubility-partition coefficient correlation.
Environ. Science Technol. 14:1227-1229.
Barnthouse, L.W.; Suter, G.W.; Bartell, S.M.; Beauchamp, J.J.; Gardner, R.H.; Under, E.;
O'Neill, R.V.; Rosen, A.E. (1986) User's manual for ecological risk assessment. Oak
Ridge National Laboratory. ORNL Publ. No. 2679.
Broderius, S.J.; Russom, C.L. (1989) Mode of action-specific QSAR models for predicting acute
and chronic toxicity of industrial chemicals to aquatic organisms. U.S. Environmental
Protection Agency. Duluth, MN: Environmental Research Laboratory. Deliverable No
81421.
Burns, L.A. (1989) Exposure analysis modeling system: user's guide for EXAMS II version 2.94.
Athens, GA: U.S. Environmental Protection Agency.
Clements, R. (1983) Environmental effects of regulatory concern under TSCA—a position paper.
Environmental Effects Branch, Health and Environmental Review Division (7403), Office
of Toxic Substances, U.S. Environmental Protection Agency, Washington, DC.
Clements, R. (1988) Estimating toxicity of industrial chemicals to aquatic organisms using
structure-activity relationships. U.S. Environmental Protection Agency, Washington, DC.
EPA 560-6-88-001. Available from: NTIS, Springfield, VA, PB89-117592.
Grain, C.F. (1982) Vapor pressure. In: Lyman, W.J.; Reehl, W.; Rosenblatt, D.H., eds.
Handbook of chemical property estimation methods, environmental behavior of organic
compounds. New York, NY: McGraw-Hill Co.
Hermens, J.; Canton, H.; Janssen, P.; DeJong, R. (1984) Quantitative structure-activity
relationships and toxicity studies of mixtures of chemicals with anesthetic potency: acute
lethal and sublethal toxicity to Daphnia magna. Aquat. Toxicol. 5:143-154. (also in
jClements, 1988).
Karickhoff, S.W.; Brown, D.S.; Scott, T.A. (1979) Sorption of hydrophobic pollutants on natural
sediments. Water Res. 13:241-248.
Konemann, H. (1981) Quantitative structure-activity relationships in fish toxicity studies. Part 1:
relationship for 50 industrial pollutants. Toxicology 19:209-221.
1-33
-------
Leo, A.; Weininger, D. (1985) CLOGP version 3.3. estimation of the n-octanol/water partition
coefficient for organics in the TSCA industrial inventory. Claremont, CA: Pomona College.
Lipnick, R.L. (1985) Validation and extension of fish toxicity QSARs and interspecies comparisons
for certain classes of organic chemicals. In: Tichy, M., ed. Toxicology and
xenobiochemistry. 8:39-52.
Nabholz, J. V. (1991) Environmental hazard and risk assessment under the United States Toxic
Substances Control Act. The science of the total environment. 109/110:649-665.
Nabholz, J. V. (in preparation) Estimating toxicity of industrial chemicals to aquatic organisms
using structure-activity relationships.
Nabholz, J.V.; Miller, P.; Zeeman, M. (1993a) Environmental and risk assessment of new
chemicals under the Toxic Substances Control Act (TSCA) Section Five. In: Landis, W.G.;
Hughes, J.S.; Lewis, M.A., eds. Environmental toxicology and risk assessment. American
Society for Testing and Materials, Philadelphia, PA. ASTM STP 1179.
Nabholz, J.V.; Clements, R.G.; Zeeman, M.G.; Osborn, K.C.; Wedge, R. (1993b) Validation of
structure-activity relationships used by EPA's Office of Pollution Prevention and Toxics for
the environmental hazard assessment of industrial chemicals. In: Gorsuch, J.W.; Dwyer,
F.J.; Ingersoll, C.G.; LaPoint, T.W., eds. Environmental toxicology and risk assessment.
American Society for Testing and Materials, Philadelphia, PA. ASTM STP 1216.
Rodier, D.J.; Mauriello, D. (1993) The quotient method of ecological risk assessment and
modeling under TSCA: a review. In: Landis, W.G.; Hughes, J.S.; Lewis, M.A, eds.
Environmental toxicology and risk assessment. American Society for Testing and Materials,
Philadelphia, PA. ASTM STP 1179.
Smrchek, J.; Clements, R.; Morcock, R.; Roberts, R. (1993) Assessing ecological hazard under
TSCA: methods and evaluation of data. In: Landis, W.G.; Hughes, J.S.; Lewis, M.A.,
eds. Environmental toxicology and risk assessment. American Society for Testing and
Materials, Philadelphia, PA. ASTM STP 1179.
Suter, G.W. (1990) Endpoints for regional ecological risk assessment. Environ. Manage. 14:9-23.
TSCA. (1976) Toxic Substances Control Act, PL 94-46, October 11, 1976, Stat. 90:2003.
U.S. Environmental Protection Agency. (1982) Surrogate species workshop. Environmental Effects
Branch, Health and Environmental Review Division (7403), Office of Toxic Substances,
U.S. Environmental Protection Agency, Washington, DC.
U.S. Environmental Protection Agency. (1983) Testing for environmental effects under the Toxic
Substances Control Act. Environmental Effects Branch, Health and Environmental Review
Division (7403), Office of Toxic Substances, U.S. Environmental Protection Agency,
Washington, DC.
1-34
-------
U.S. Environmental Protection Agency. (1984) Estimating concern levels for concentrations of
chemical substances in the environment. Environmental Effects Branch, Health and
Environmental Review Division (7403), Office of Toxic Substances, U.S. Environmental
Protection Agency, Washington, DC.
U.S. Environmental Protection Agency. (1985) Toxic Substances Control Act test guidelines final
rules. Federal Register 50(188):39252-39516.
U.S. Environmental Protection Agency. (1986) New chemical review process manual. Office of
Toxic Substances, U.S. Environmental Protection Agency, Washington, DC. EPA 560/3-
86-002.
U.S. Environmental Protection Agency. (1988) User's guide to PDM3, final report. EPA Contract
No. 68-02-4254, Task No. 117. Exposure Assessment Branch, Exposure Evaluation
Division (7403), Office of Toxic Substances, U.S. Environmental Protection Aeencv
Washington, DC.
U.S. Environmental Protection Agency. (1992) Framework for ecological risk assessment. Risk
Assessment Forum, Washington, DC. EPA 630/R-92/001.
i
U.S. Environmental Protection Agency. (1993) PMNEcotox Database: a data base of
environmental toxidty studies which are protected by confidential business information
(CBI). Environmental Effects Branch, Health and Environmental Review Division (7403),
Office of Toxic Substances, U.S. Environmental Protection Agency, Washington, DC.
van Gestel, C.A.M.; Ma, W.-C. (1990) An approach to quantitative structure-activity relationships
(QSARs) in earthworm toxici«y tests. Chemosphere 20:1023-1033.
Veith, G.D.; Call, D.J.; Brooke, L.T. (1983) Structure-activity relationships for the fathead
minnow, Pimephales promelas: narcotic industrial chemicals. Can. J. Fish. Aquat Sci
40:743-748. (also in Clements, 1988).
Veith, G.D.; Broderius, SJ. (1990) Rules for distinguishing toxicants that cause type I and type II
narcosis syndromes. Environ. Health Perspect. 87:207-211.
Zeeman, M.; Nabholz, J.V.; Clements, R.G. (1993) The development of SAR/QSAR for the use
under EPA's Toxic Substances Control Act (TSCA): an introduction. In: Gorsuch, F.J.;
Dwyer, F.J.; Ingersoll, C.G.; LaPoint, T.W., eds. Environmental toxicology and risk '
assessment: 2nd volume. American Society for Testing and Materials, Philadelphia PA
ASTM STP 1216. '
1-35
-------
-------
APPENDIX A
QSARS BETWEEN NEUTRAL ORGANIC CHEMICALS
AND FISH AND GREEN ALGAL TOXICITY
1-A1
-------
QSARS BETWEEN NEUTRAL ORGANIC CHEMICALS AND FISH CHRONIC VALUES
(Broderius and Russom, 1989)
• Log NOEC (mol/L) = -0.878 Log K,,,, - 2.40
n = 20 r2 = 0.911 s = 0.335
• Log LOEC (mol/L) = -0.862 Log K,,w - 2.16
n = 20 r2 = 0.913 s = 0.325
• Log ChV (mol/L) = -0.870 Log K,,w - 2.28
n = 20 r2 = 0.914 s = 0.327
• Log ChV (mg/L) = antilog ChV (mol/L) * mw
QSARS BETWEEN NEUTRAL ORGANIC CHEMICALS AND GREEN ALGAE
TOXICITY (GROWTH) (Nabholz, in preparation)
Log ChV (mmol/L) = 0.036 - 0.634 Log
n = 6 r2 = 0.99
• Log 96-h EC50 (mmol/L) = 1.48 - 0.869 Log
n = 22 r2 = 0.93
Please note: The QSARs referenced here and elsewhere in the report are now available as a
computer program called ECOSAR (EPA-748-F-93-002). Limited copies are available from the
National Center for Environmental Publications and Information, U.S. EPA, 26 West Martin
Luther King Drive, Cincinnati, OH 45268 (513-569-7985). In addition, copies may be obtained
from:
• U.S. Government Printing Office, Superintendent of Documents, ATTN:
Electronic Product Sales Coordinator, P.O. Box 37082, Washington, DC 20013-
7082 (202-512-1530);
1-A2
-------
Federal Bulletin Board, U.S. Government Printing Office of Electronic Information
Dissemination Services (202-512-1524); or
National Technical Information Service, U.S. Department of Commerce, 5285 Port
Royal Road, Springfield, VA 22161 (703-487-4650) (order as computer program
PB94-500485).
1-A3
-------
-------
APPENDIX B
INPUT AND OUTPUT PARAMETERS
FOR EXAMS H
1-B1
-------
INPUT PARAMETERS FOR EXAMS H
EXAMS II estimates exposure, fate, and persistence of an organic chemical after being
released into an aquatic ecosystem (Burns, 1989). EXAMS II requires input of data into three files
that describe the chemical, the environment, and the chemical loading to the environment.
Critical inputs to the chemistry file for this example were the water solubility, octanol-
water partition coefficient (K<,w), soil/sediment organic carbon-water partition coefficient (K,,,.), and
biodegradation rate constant. These parameters are important hi modeling the test chemical
partitions between the water column and sediments. This particular analysis used a log Kow of
6.56.
The environment file was culled from a set of predefined or canonical environments. Data
including stream geometry and surface water flow rates are included here. Two important
parameters that were user-defined in this example were the benthic and suspended sediment organic
carbon content and the concentration of microorganisms hi the sediments active in the
biodegradation of the compound. The mass of test chemical released per unit time is entered into
the loading file.
OUTPUT OF EXAMS H
The EXAMS II output includes tables summarizing test chemical properties; environmental
characteristics; chemical loadings; steady-state mean, minimum, and maximum concentrations hi
various environmental compartments; and an exposure analysis summary (see example below).
EXAMPLE OUTPUT OF EXAMS (NOT CASE STUDY PMN)
Exposure (maximum steady-state concentrations'):
Water column: 7.884E-03 mg/L dissolved; total = 8.255E-03 mg/L
Benthic sediments:
Biota ftig/g dry weight):
Pate:
Total steady-state accumulation:
Total chemical load:
7.377E-03 mg/L dissolved in pore water; maximum total
concentration = 33.1 mg/kg (dry weight)
Plankton: 7.68E+03
Benthos: 7.18E+03
494 kg, with 0.29 percent hi the water column and 99.71
percent hi the benthic sediments.
27 kg/month. Disposition: 0.00 percent chemically
transformed, 0.00 percent biotransformed, 0.00 percent
volatilized, and 100.00 percent exported via other pathways.
1-B2
-------
Persistence:
After 16.0 months of recovery time, the water column had lost 65.82 percent of its initial chemical
burden; the benthic zone had lost 60.08 percent; systemwide total loss of chemical = 60.1 percent.
Five half-lives (>95 percent cleanup) thus require about 60 months.
1-B3
-------
-------
SECTION TWO
ECOLOGICAL RISK ASSESSMENT CASE STUDY:
RISK ASSESSMENT FOR THE RELEASE OF RECOMBEVANT RfflZOBIA
AT A SMALL-SCALE AGRICULTURAL FIELD SITE
-------
AUTHORS AND REVIEWERS
AUTHORS
Gwendolyn McClung
Office of Pollution Prevention and Toxics
U.S. Environmental Protection Agency
Washington, DC
Philip G. Sayre
Office of Pollution Prevention and Toxics
U.S. Environmental Protection Agency
Washington, DC
REVIEWERS
Joseph E. Lepo (Lead Reviewer)
Center for Environmental Diagnostics
and Bioremediation
University of West Florida
Gulf Breeze, PL
Gregory R. Biddinger
Exxon Biomedical Sciences, Inc.
East Millstone, NJ
Joel S. Brown
University of Illinois at Chicago
Chicago, JJL
Herbert Grover
Benchmark Environmental Corporation
Albuquerque, NM
Thomas Sibley
Fisheries Research Institute
University Of Washington
Seattle, WA
Frieda B. Taub
School of Fisheries
University of Washington
Seattle, WA
2-2
-------
CONTENTS
ABSTRACT . ' 2-7
2.1. RISK ASSESSMENT APPROACH . . . . . 2-8
2.2. STATUTORY AND REGULATORY BACKGROUND 2-10
2.3. CASE STUDY DESCRIPTION 2-11
2.3.1. Background Information and Objective . . 2-11
2.3.2. Problem Formulation . . , 2-11
2.3.2.1. Planning ! 2-11
2.3.2.2. Stressor Characteristics 2-12
2.3.2.3. Ecosystem Potentially at Risk . 2-16
2.3.2.4. Endpoint Selection . 2-18
2.3.3. Analysis: Characterization of Exposure 2-21
2.3.3.1. Stressor Characterization 2-22
2.3.3.2. Ecosystem Characterization 2-23
2.3.3.3. Temporal Analysis 2-24
2.3.3.4. Exposure Analyses • • • •: 2-24
2.3.4. Analysis: Characterization of Ecological Effects 2-26
2.3.4.1. Evaluation of Effects Data 2-26
2.3.4.2. Evaluation of Causal Evidence . 2-26
2.3.4.3. Effects Needing Study in the Event of Significant Off-Site
Migration or Large-Scale Release 2-26
2.3.5. Risk Characterization 2-28
2.3.5.1. Risk Estimation 2-28
2.3.5.2. Uncertainty ; 2-29
2.3.5.3. Risk Description 2-29
2.4. DISCUSSION BETWEEN RISK ASSESSOR AND RISK MANAGER 2-32
2.5. RISK VERIFICATION . 2-32
2.5.1. Persistence 2-32
2.5.2. Competitiveness . 2-32
2.5.3. Dissemination From the Test Site 2-33
2.5.4. Effect on Alfalfa Yield During Field Test . . 2-33
2-3
-------
CONTENTS (continued)
2.6. KEY TERMS 2-33
2.7. REFERENCES 2-34
APPENDIX A—MONITORING AND ENUMERATION TECHNIQUES FOR RHIZOBIA 2-A1
APPENDIX B—PERSISTENCE IN THE RHIZOSPHERE AND NODULE OCCUPANCY 2-B1
APPENDIX C—RHIZOBIAL DISPERSION AND MIGRATION 2-C1
APPENDIX D—STRAIN COMPARISON AND COMPETITION TESTS 2-D1
APPENDIX E—RHIZOBIAL CULTURE VIABILITY 2-E1
APPENDIX F—ALFALFA YIELDS IN THE FIELD TESTS 2-F1
2-4
-------
LIST OF FIGURES
Figure 2-1. Structure of assessment for small-scale field tests with recombinant rhizobia ... 2-9
Figure 2-2. PMN data/information,, components, and decisions made . . . 2-13
Figure 2-3. Cassette diagram for RMB7103 with only primary sequences added 2-15
LIST OF TABLES
Table 2-1. Table of Recombinant Rhizobia for 1989-1990 Field Tests 2-14
Table 2-2. Linkages Among Assessment Endpoints and Data Needs Relevant to Endpoint
Evaluation . . . . 2-17
Table 2-3. Linkages Among Stressor, Monitoring, andjData Needs Relevant to Endpoint
Evaluation , 2-19
LIST OF COMMENT BOXES
Review Comments on Risk Assessment Approach 2-8
Comments on Problem Formulation 2-19
Comments on Characterization of Exposure 2-25
Comments on Analysis: Characterization of Ecological Effects 2-27
Comments on Risk Characterization 2-29
2-5
-------
LIST OF ACRONYMS
DNA deoxyribonucleic acid
EPA U.S. Environmental Protection Agency
FA fluorescent antibody
GEM genetically engineered microorganism
MDL minimum detection limit
MPN most probable number
OPPT Office of Pollution Prevention and Toxics
OTS Office of Toxic Substances ^
PMN premanufacture notice
t-
RDM rhizobia-defined medium
TSCA Toxic Substances Control Act
2-6
-------
ABSTRACT
This ecological risk assessment concerns a small-scale field test of genetically engineered
RMzobiwn meliloti strains. The strains were submitted hi 1988 as part of a prernanufacture notice
(PMN) to the Office of Toxic Substances (OTS, currently the Office of Pollution Prevention and
Toxics, OPPT) for tests to be conducted hi 1989-1990. The rhizobia were genetically modified by
the insertion of antibiotic resistance markers or by the addition of both antibiotic resistance and mf
genes to enhance nitrogen fixation. R. meliloti form nodules and fix nitrogen hi alfalfa (Medicago
sativa), sweet clover (Melilotus), and fenugreek (Trigonella). The surrounding agroecosystem near
Sun Prairie, Wisconsin, constituted the area of concern for ecological effects. Literature accounts
of rhizobial movement, field test site characteristics, and field test design indicated that the
microorganisms had only a minimal potential for migrating beyond the field test plot. The primary
assessment endpoint examined during the small-scale field test was the potential for these
recombinants to alter top growth of alfalfa. The ecological concerns for large-scale releases of
recombinant rhizobia—such as increased growth of nontarget legumes, decreased growth of target
legumes, spread of antibiotic resistance genes, nitrogen cycling disruption, and alteration of host
range—were of low concern for the agroecosystem around the test site.
Actual data obtained from the small-scale field study confirmed the predictions hi the OTS
PMN risk assessment conducted hi 1988 and those hi this ecological risk assessment. Little
horizontal, vertical, or aerial migration of R. meliloti occurred. The rhizobia primarily moved
with the alfalfa root system. When compared with unmodified strains, recombinant rhizobia did
not cause significant changes hi nitrogen fixation, as measured indirectly by alfalfa yield.
Recombinant strains did not out-compete parental strains, alleviating the concern for displacement
of the indigenous rhizobia.
2-7
-------
2.1. RISK ASSESSMENT APPROACH
This case study represents a typical risk assessment for a premanufacture notice (PMN)
received by the U.S. Environmental Protection Agency's (EPA's) Office of Pollution Prevention^
and Toxics (OPPT). OPPT effectively evaluates the potential risk using the paradigm of "Risk -
Hazard X Exposure" (Sayre, 1990). This paradigm is consistent with the Framework for
Ecological Risk Assessment (U.S. EPA, 1992). Figure 2-1 demonstrates how the assessment was
structured, using the framework report as guidance.
Since the framework report focuses on physical and chemical rather than biological
stressors, the report does not adequately address certain aspects of this risk assessment. These
include:
• the need for fate monitoring to build a data base of rhizobial behavior for larger-
scale releases;
• the potential that in some cases the introduced deoxyribonucleic acid (DNA) might
move from the genetically engineered microorganism (GEM) to other environmental
recipients;
• consideration of field site design that limited microbial dissemination beyond the
site, thereby alleviating the need for certain effects testing;
« the evaluation of exposure resulting from culturing and transporting OEMs to the
field site; and
• construct considerations.
Reviewer Comments on Risk Assessment Approach
• As currently formulated, EPA's Framework for Ecological Risk Assessment does
not address fundamental differences between biological stressors and chemical
and physical stressors. These differences include concerns unique to living
entities, such as replication, colonization, and genetic evolution. This case
study should provide a useful model for assessing risks of future limited releases
of genetically engineered microorganisms in agroecosystems.
• This case study does not address more general ecological risks, nor does it
consider risks of large-scale or commercial release of OEMs. However, the
study does serve as an important ground-breaking document because risk
assessments of OEMs released into agroecosy stems will become more common in
the near future.
2-8
-------
a* Rifk: A9roecosystem site for field test of recombinant
involving larger spatial scales were of lesser concern
aro?m ?nr phC°T-POnen?,: Na"Ve rhiz°bia'legumes within tne cross-inoculation
group for Rhizobium meliloti, legumes outside the cross-inoculation qrouo
nonlegumes affected by introduced sequences. lol-uiail°n group,
Endpoints: Assessment endpoints are listed in table 2-2. These also are
cons.dered to be measurement endpoints since they can be measured directly.
ANALYSIS
Characterization
of Exposure
Exposure potential was evaluated by
examination of literature on rhizobia,
PMN laboratory data, and
greenhouse data. Field test
procedures were developed to
minimize exposure in the field test.
Characterization of
Ecological Effects
Effects were evaluated based on
the literature, PMN greenhouse
studies, and PMN construct data.
RISK CHARACTERIZATION
Risk characterization focused on the exposure data that limited risk
S25?T ?Hri!?arily t0 smal|-scale issues- s"ch as effects on alfalfa yield,
which could be examined easily in the field test.
FIELD TESTING
mt H t Hf »!f .aSSeSSment> moniton'ng for microorganism movement was
mandated for the field test so that the test could be terminated if microbes
moved off site ,n great numbers or if adverse effects were noted. The field
test generally confirmed the predictions of the risk assessment. Dispersal
studies showed little off-site movement of rhizobia •*« .
Figure 2-1. Structure of assessment for small-scale field tests with recombinant rhizobia
2-9
-------
Reviewer Comments on Risk Assessment Approach (continued)
• The reviewers did not consider the application of the framework to microbial
stressors an insurmountable barrier. However, they agreed that both the
framework report and similar future case studies should acknowledge the unique
properties and complexities of a living, changing stressor. The reviewers
suggested that applying the framework to biotechnological risks—such as the
release of OEMs—might require providing the framework audience with the
biotechnological details needed for appraising the relative importance of the risk
factors.
2.2. STATUTORY AND REGULATORY BACKGROUND
The "Coordinated Framework for Regulation of Biotechnology" (Office of Science and
Technology Policy, 1986) explains that the Toxic Substances Control Act (TSCA) gives EPA the
authority to review certain classes of biotechnology products. Under the coordinated framework,
biotechnology products are regulated in accordance with the use of each product (Milewski, 1990).
Uses of microorganisms not covered by other existing authorities (U.S. Department of Agriculture,
Food and Drug Administration, EPA's Office of Pesticide Programs) are reviewed by EPA's
OPPT under TSCA; thus, TSCA serves as a "gap-filling" statute. TSCA's applicability follows
from the interpretation that microbes are chemical substances under TSCA. Candidates for review
are limited to those commercial microorganisms that have been altered to contain genetic
information from dissimilar source organisms. EPA describes as dissimilar those organisms
produced using DNA from different taxonomic genera. Such microorganisms are considered
"intergeneric." EPA does not regulate the use of naturally occurring rhizobial inoculants.
TSCA applies only to products developed for commercial purposes, whether for contained
systems or environmental releases. Under Section 5 of TSCA, manufacturers and importers of
intergeneric microorganisms must submit a PMN at least 90 days prior to beginning manufacture
or import. Under TSCA authority, OPPT can require information on microbial biotechnology
products in order to identify potential hazards and exposures. OPPT also can require testing a
microbial biotechnology product that may present an unreasonable risk of injury to human health or
the environment or that is produced in substantial quantities and may result in substantial
environmental release or substantial human exposure. Finally, OPPT can restrict the production,
processing, distribution, use, and disposal of a microbial biotechnology product if it presents an
unreasonable risk of injury to human health or the environment.
Because TSCA applies only to microorganisms developed for commercial purposes, EPA
currently requests that industry voluntarily comply with the PMN reporting requirements for any
commercial research and development field test that involves the release of intergeneric
microorganisms involving a TSCA use. As a result, the PMN submission for the small-scale field
test of genetically engineered strains of Rhizobium meliloti, the subject of this ecological risk
assessment, was submitted on a voluntary basis by Biotechnica Agriculture, Inc., in 1988.
2-10
-------
Approval of the field test resulted in the issuance of a 5(e) Consent Order, which bound the
company to the protocols, monitoring procedures, and data collection approved by EPA.
2.3. CASE STUDY DESCRIPTION
2.3.1. Background Information and Objective
This case study focuses primarily on the small-scale field testing of four recombinant
strains of R. meliloti. Rhizobia, a general term for various species of the genus Rfuzobium, are
Gram-negative, motile, rod-shaped, aerobic bacteria that infect legume roots. A symbiotic
relationship forms in which the bacteria fix atmospheric nitrogen, providing ammonium for protein
production hi the plant. In exchange, the bacteria obtain energy from the plant in the form of
photosynthate, specifically dicarboxylates.
The various species and biovars ofRhizobium have been designated according to the types
of legume plants they infect, such as alfalfa, clovers, beans, vetch, or lotus. The specificity of
infection by certain species or biovars ofRhizobium has led to the loose designation of "cross-
inoculation" groups (Alexander, 1977). For example, the alfalfa group consists of R. meliloti,
which is capable of infecting not only alfalfa (Medicago), but sweet clover (Melilotus) and
fenugreek (Trigonella).
The symbiotic relationship between rhizobia and legumes is of great importance in
agriculture, as legumes typically are not fertilized with nitrogen if rhizobia are present. In fact,
high nitrogen contents in soils actually suppress nitrogen fixation by the nodules. More important,
symbiotic nitrogen fixation contributes greatly to the nitrogen cycle. In association with alfalfa,
rhizobia fix nitrogen vigorously, perhaps fixing between 125 and 335 kg of nitrogen per hectare
each year (Alexander, 1977).
This case study has two purposes. First, the case study will examine the information
submitted and used during the OPPT risk assessment to determine whether the framework
assessment process can use the information as efficiently. Second, the case study will examine the
data generated from the field to determine how accurately the risk assessment process predicted
risks associated with the field test.
The area of concern for possible adverse ecological impacts is the surrounding
agroecosystem. Ecological concerns are contingent on the ability of the rhizobia to survive and
spread beyond the immediate area of the field site.
2.3.2. Problem Formulation
2.3.2.1. Planning
This risk assessment focused on determining the potential adverse effects of conducting a
small-scale field test with recombinant rhizobia. However, the data gathered from the field site
also may prove useful in projecting adverse effects that could result from a large- or commercial-
2-11
-------
scale release. Before such a release occurs, the potential ecological effects for a large-scale release
need to be addressed.
The PMN submission supplied laboratory data on the microorganism identity, construct
information, and microorganism characteristics and behavior. Greenhouse data addressed the
effects on alfalfa (yield data), survival, and competitiveness of the recombinant strains (nodule
occupancy). These data contributed to the decision-making process for approval of the small-scale
field test. Field test protocols and a site evaluation conducted prior to the field tests also helped
reach the decision to approve the small-scale field test of these recombinant rhizobia.
Figure 2-2 illustrates the various components of the risk assessment process for PMNs,
from the data submissions through the decision-making processes prior to approval for commercial
release The field test design included several approaches to evaluating adverse effects from small-
scale releases of recombinant rhizobia. Yields of alfalfa were an indirect measure of changes in
the nitrogen-fixing ability of rhizobial strains. Nodule occupancy indicated competitiveness of the
rhizobial strains with the indigenous rhizobial populations. The test design also included plans to
measure the persistence of the microorganism in the rhizosphere. Finally, to test the prediction
that only limited dissemination of the recombinant microorganisms beyond the site would occur,
the test design included monitoring both soil and air for the presence of these microorganisms.
2.3.2.2. Stressor Characteristics
The primary stressors in this case study are the recombinant rhizobia, as opposed to
chemical or physical stressors. In this study, the stressor has the potential to split into
subcomponents of a biological nature (pathogenicity, altered legume growth resulting from the
microbe) and subcomponents of a chemical nature (production of toxins, detrimental metabolites,
and overproduction of nitrate).
As with chemical stressors, characterizing the recombinant microbes to predict their
potential adverse effects constitutes a critical component for the risk assessment. For the
recombinant microorganism, characterization includes a description of the donor and recipient
microorganisms, including their taxonomic derivation. The phenotypic traits of most OEMs
reviewed in OPPT are encoded and analyzed with a PC-microcomputer version of the Micro-IS
software package; this data system was originally developed by the National Institutes of Health
(Segal, 1988). A description of the techniques used to construct the PMN microorganism also
contributes to characterization of the GEM.
The final step for GEM identification involves verifying that GEM DNA contains the DNA
of interest, along with additional vector DNA. This analysis is based on PMN submission
information that usually includes the following:
' • construction of the DNA cassette that codes for traits such as enhanced nitrogen
fixation;
• a complete description of the integration site hi the R. metiloti genome;
2-12
-------
§
•o
•a
42
I
1
2-13
-------
• construction of the vector containing the cassette;
• introduction of the vector carrying the cassette into the recipient microorganism;
and
• final construct and genetic stability of the PMN microorganism.
OPPT reviewed data from restriction digests, DNA probe verification, and phenotypic
analysis of recombinants for this step of GEM characterization. OPPT also used DNA sequence
data bases and software such as GENEMBL and DNA Star to examine introduced sequences.
These sequences are examined to determine functions of identified DNA and the potential for
unidentified DNA sequences (such as open reading frames) to encode known protein products.
Table 2-1 summarizes the recombinant rhizobia tested in the 1989-1990 field tests. The
genes inserted in the four recombinant strains were added to the same insertion site, the ino site.
Note that each wild-type R. meliloti recipient contains the usual complement of nif genes necessary
for nitrogen fixation.
Table 2-1. Table of Recombinant Rhizobia for 1989-1990 Field Tests
PMN No.
P88-1116
P88-1118
P88-1120
P89-280
Biotechnica
Agriculture,
Inc., Strain
RMB7101
RMB7201
RMB7401
RMB7103
Recipient
RCR2011"
PCb
UC445C
RCR2011
Modification
o/nega/strep/spec
omega/strep/spec
o/wega/strep/spec
omega/strep/spec/nif
Insertion Site
ino
ino
ino
ino
* Streptomycin-sensitive parent of R. meliloti strain Rml021. Strain Rml021 is a spontaneous
streptomycin-resistant mutant arising from strain RCR2011. Strain RCR2011 is derived from a
natural isolate, strain SU47 (Rothamsted Experimental Station collection).
b Natural isolate obtained in 1986 from a root nodule of inoculated alfalfa plant grown in soil from
the Chippewa Agricultural Station, Pepin County, Wisconsin.
c California soil isolate (UC445 or CA445) effective in the alfalfa cultivar Hairy Peruvian.
Each introduced sequence of the constructs was examined for the potential to cause adverse
impact. The insertion site, too, may disrupt recipient DNA. The altered or added DNA sequences
are noted below, along with their potential ecological impact (see figure 2-3 for additional
information about constructs):
• ino insertion site. The ino sequence encodes genes responsible for the metabolism
of myoinositol, a substrate usable as a carbon source during saprophytic growth. If
an introduced DNA sequence inactivates genes at this insertion site, rhizobia would
2-14
-------
promoterItePlnifH leader/nifA/omega/inp
ino = inositol region of R. meliloti located on megaplasmid pRmeSU47b
TiT2 = transcriptional terminator sequences from the rrriB gene that encodes the 5S rRNA of
Escherichia coli
nijD promoter = promoter derived from Bradyrhizobiumjaponicum
tet fragment = 200 bp of the tet* gene derived from pBR322
niJH = synthetic 21 bp oligomer linker with two added restriction sites plus the R. meliloti nifHDNA
that encodes untranslated leader RNA of niJH gene
nifA = R. meliloti nifA gene
omega = gene constructed by Prentki and Krisch (1984) derived from plasmid R100 originally
isolated from Shigellaflexineri; encodes a transcriptional terminator and resistance to streptomycin
and spectinomycin
Figure 2-3. Cassette diagram for RMB7103 with only primary sequences added (serves as
illustration of a gene cassette introduced into the ino site)
2-15
-------
have decreased survival in soil and senescing plant roots. No other adverse effect
would be expected.
• T^, nifH. Because the T^ termination sequence halts the transcription of the
introduced DNA into messenger RNA, this sequence limits the effects of the
cassette on surrounding DNA. Consisting of only a leader sequence for the nifA
and other genes, nifH has little likelihood of causing adverse ecological effects.
• nifA niJD. A regulatory gene, the nifA sequence controls the production of the
nitrogenase enzyme, which brings about nitrogen fixation in alfalfa. Altered
nitrogenase production could lead to decreased growth of alfalfa or increased
growth of weedy relatives. The same concerns apply to the cassette's promoter
sequence, nifD.
• tet, omega. The omega fragment encodes resistance to streptomycin and
spectinomycin. Transfer of these genes from OEMs to human or animal pathogens
would render them resistant to streptomycin and spectinomycin, but such resistance
transfer is not of concern in small-scale field trials. The tet gene does not encode
resistance to tetracycline because it is only a gene fragment.
Essentially, the construct analysis narrowed the concerns about Effects to the potential for
decreased yield in the target legume, alfalfa. The construct analysis did not eliminate concerns for
increased competitiveness or survival of the recombinant rhizobia relative to the wild-type strains.
The fate of the introduced DNA hi the GEM is an ecological concern. Natural gene
transfer of this DNA from recombinant strains to environmental receptors could produce secondary
stressors. However, in the case of these OEMs, careful analysis of the constructs, available
literature, and laboratory data indicated little need to monitor for the existence of secondary
stressors. For example, under optimal laboratory conditions, genetic transfer of the megaplasmid
containing the insertion point was not detected at a detection limit of 10* (Finan et al., 1986).
These constructs cannot transfer by means of transposition, because the omega fragment lacks
transposition functions. Finally, data on RMB7101 indicated a reversion frequency to a
streptomycin-sensitive phenotype of less than 6.3 x 1O8 (Sayre, 1988).
2.3.2.3. Ecosystem Potentially at Risk
The field test plots consisted of less than 1 acre in the northwest corner of a 14-acre parcel
leased from a 39-acre farm. This farm is located in Dane County, Wisconsin, directly north of the
city of Sun Prairie and 12 miles east of Madison. Sparsely populated agricultural land lies to the
north and east of the site. Residential areas lie within a mile to the south and 1.5 miles to the
west. Dane County is approximately 80 percent farmland, with approximately 80 percent of this
land in crops: corn for gram and silage, alfalfa, other hay, oats, sweet corn, and soybeans.
Potential biotic components of the agroecosystem include target and nontarget legumes
(including weedy legumes), rotational nonlegume crops, native rhizobia, and bacterial pathogens
that can acquire antibiotic resistance genes from the recqmbinant rhizobia (table 2-2).
2-16
-------
Table 2-2. Linkages Among Assessment Endpoints and Data Needs Relevant to Endpoint
Evaluation
Assessment Endpoint
1.
2.
3.
4.
5.
6.
7.
8.
Decreased alfalfa growth
Decreased growth of legumes outside
cross-inoculation group
Decreased growth of nonlegume crop
plants
Unanticipated effects of introduced
DNA sequences
Effects of introduced DNA on
recipient DNA at insertion site
Unanticipated effects of recipient
microbe
Effects of antibiotic resistance genes
Competitive displacement of native
Predictive Risk Endpoints Future
Assessment Monitored in Large-
Onformation Small -Scale Scale
used)' Test Issues
GH X
TX, CA
GH, CA, TX
CA
CA
TX
BSAC
X
X
X
X
"
Y
rhizobia if coupled with any hazards
listed in 1-3 or 9-10
9. Increased/decreased growth of sweet
clover
10. Increased/decreased growth of
fenugreek
11. Effects of coumarin on cattle
12. Effects on nitrogen cycle
X
X
X
X
a Legend:
BSAC = addressed by the EPA Subcommittee of the Biotechnology Science Advisory Board
CA = addressed by construct analysis
GH = addressed by PMN greenhouse data
TX = addressed by taxonomic analysis of recipient rhizobia
2-17
-------
Displacement of the indigenous rhizobia by recombinants also may alter ecosystem structure. Such
a change would adversely affect the ecosystem if the constructs have a lower nitrogen-fixing
capacity than native rhizobia. For example, Bradyrhizobium japonicum strain 123, which fixes
nitrogen poorly in the field, has out-competed and displaced native strains in the Midwest, resulting
in decreased nitrogen fixation in soybean plants (Tiedje et al., 1989). Displacement of native
rhizobia by recombinants having a greater nitrogen-fixing capacity also has the potential to affect
ecosystem function adversely. Increased soil nitrogen might disrupt the nitrogen cycle balance or
lead to localized pollution of ground water by nitrates.
2.3.2.4. Endpoint Selection
The assessment endpoints reflect the delineation of the ecosystem at risk: primary concern
focused on the area immediately surrounding the field plot, with some lessening concern for areas
farther removed from the field site. If the monitoring of the microorganisms during the field test
had shown significant off-site movement and spread (particularly if linked with decreased alfalfa
growth in the field or other adverse effects), then the field test would have been terminated and the
risk assessment expanded to include larger-scale issues.
In this ecological risk assessment, many of the 12 assessment endpoints listed in table 2-2
can be measured directly, eliminating the need to identify measurement endpoints for these
assessment endpoints. Table 2-2 links the assessment endpoints to the data needed to evaluate
them. Data relating to assessment endpoints originate from four sources: the literature, laboratory
studies, greenhouse studies, and the field test. Literature information and laboratory studies
conducted for the PMN can at least partially address assessment endpoints 4 through 7 of table 2-2.
Greenhouse studies for the PMN provide information on assessment endpoints 1 and 3, while the
small-scale field test concerns assessment endpoints 1 and 8. Assessment endpoints 1 through 3
and 8 through 12 concern information needed prior to large-scale release. Table 2-3 links the
OEMs to monitoring and data needs.
Comments on Problem Formulation
General reviewer comments:
The following factors in the current case study set it aside from risk assessments
of chemical and physical stressors:
the unique complexities of a microbial stressor;
- the real and imagined risks of genetically engineered microorganisms; and
— detecting off-site migration.
As risk assessment experience for microbial stressors accumulates, risk assessors
will gain facility in addressing these factors. The process should result in an
enhanced knowledge base that, can feed back into the risk assessment process
itself and can be implemented in the education of scientific and regulatory
communities as well as the general public.
2-18
-------
Table 2-3. Linkages Among Stressor, Monitoring, and Data Needs Relevant to Endpoint
Evaluation
Exposure
Element"
Detection of GEM in nodule
Survival of GEM in soil,
rhizosphere
Monitoring of GEM in soil, air,
water
Monitoring of gene transfer
Risk Assessment
(information
used)b
GH
GH,L
F
L, CA
Measurement
Endpoints for
Small-Scale Test
X
X
X
Future Large-
Scale Issues
X
Presence of the GEM or the introduced DNA in various media is necessary for linking the GEM
with the assessment endpoints in table 2-2.
b Legend: CA
F
GH
L
= addressed by construct analysis
= , addressed by examination of field prior to GEM release
= addressed by PMN greenhouse data
= addressed by PMN laboratory data
Comments on Problem Formulation (continued)
• Releases of genetically engineered rhizobia are probably the best available
model for initial release of OEMs. Thus, although supporting data for low risk
of small-scale field tests were weak (compromised or poorly designed
greenhouse studies), other factors contributed to the decision to issue consent for
the study, including:
the knowledge base on the generally innocuous nature of the rhizobia, e.g.,
the history of their application worldwide in the enhancement of symbiotic
nitrogen fixation;
site characteristics that would tend to inhibit spread of the introduced
strains; and
the nature of the genetic construct in the OEMs.
2-19
-------
Comments on Problem Formulation (continued)
• The case study set criteria for termination of the test and described monitoring
procedures. The authors may wish to point out the extent to which exotic
rhizobia have already been introduced in the United States and the
consequences. The authors do describe how the introduction of a highly
effective and competitive Bradyrhizobium japonicum (strain 123) became
problematic, out-competing indigenous rhizobia with a greater capacity for
nitrogen fixation. The effectiveness data, which addressed this particular kind
of risk, proved inconclusive. Although most introduced rhizobia have been
harmless, the legume kudzu (and its symbiotic rhizobia) has become an infamous
pest in the southern United States.
• In the characterization of the stressors, the authors split risks of the OEMs into
biological (i.e., pathogenicity, altered legume growth, microbial competition,
gene release) and chemical (i.e., toxins, detrimental metabolites such as
nitrate). This approach appears useful in addressing risks of plant-associated
microbes.
• One category of secondary stressors consists of microbial recipients that could
acquire introduced DNA by natural gene transfer, such as through
bacteriophages or conjugation. Several reviewers pointed out that the antibiotic
marker genes were likely to be of greater concern than the nif genes themselves.
• The review panel expressed interest in whether risk assessment for microbial
releases into an agroecosystem also should consider risk in the broader context
of general ecological effects; that is, should a more general range ofnontarget
animals, plants, microbes, or ecosystem function be incorporated as
measurement endpoints for the general health of the ecosystem? The case study
focuses on decreased production in a commercially important crop as opposed
to effects on surrounding ecosystems. However, a risk to surrounding
ecosystems may not be a risk to the agroecosystem. Although the study
discussed broader ecological risks outside the field site, the study considered the
risk of exposure beyond the site as minimal
2-20
-------
Comments on Problem Formulation (continued)
Authors'comments:
• At the time the Agency reviewed this proposed field test (1988), it was not
deemed necessary to assess risk in the broader context of general ecological
effects such as the general range ofnontarget animals, plants, microbes, or
ecosystem Junction for several reasons'. First, the field test was a small-scale
test that was expected to remain small scale given the data available in the
literature on rhizobial movement and specific site characteristics for this test.
Second, there was no reason to expect that the genetic modifications made to the
recipient rhizobial strains would result in any broad ecological consequences.
The genetic alterations of (1) enhancing the existing trait of nitrogen fixation
and (2) insertion of antibiotic resistance genes to serve as markers for detection
were not expected to confer on these microorganisms the trait of pathogenicity
to plants or animals, nor to alter the host range of plants these rhizobia can
infect (nodulate). Competitiveness of the recombinant rhizobia relative to the
parental strains and indigenous rhizobia was addressed in the field studies.
Similarly, in 1988, standardized validated protocols for assessing disturbances
in ecosystem junction were not available. Currently, the processes of (1)
identification of ecologically significant endpoints for assessing ecosystem
function, (2) the development of protocols/methodology for assessing those
endpoints, and (3) the interpretation of results from such tests are all still in
their infancy.
• Because EPA's framework report did not address several aspects of an
ecological risk assessment relevant to biological stressors, addressing these
aspects proves difficult. In this ecological risk assessment, two key facets in
particular were difficult to address: (1) the need to identify construct issues in
general and to use the construct information to lessen the concerns for fate and
effects and (2) the need to monitor the movement and survival of the OEMs in
different media. The exposure elements in table 2-3 were critical to identifying
the ecosystem at risk, but including the table proved problematic within the
context of the framework guidance.
• In addition, it was difficult to decide which table 2-2 endpoints to list and
whether these endpoints were assessment or measurement endpoints in this
particular case study.
2.3.3. Analysis: Characterization of Exposure
The exposure profile in this case study can contain specific information because both the
intended number of microorganisms to be applied and the area of application are known. The test
specified applying microorganisms by means of in-furrow spraying at the time of planting. The
2-21
-------
test
a total of approximately 6 X 1012 microorganisms to approximately 0.8 acre. The first
each Burred" on May 24, 1989, consisted of 5.52 X 10- cells for a strain comparison
test The second release, which took place on May 25, 1989, contained 5.52 X 10" cells for a
strain competition test. Monitoring these microorganisms continued for 2 years, collecting yield
data for alfalfa over two growing seasons. Post-termination monitoring of recombmant rhizobia in
soil extended months beyond the last alfalfa harvest.
Although the initial exposure is well characterized, uncertainty regarding exposure over
time arises as a result of microbial death, reproduction, and transport. Considerations Delude
survival in soil and root nodules and dissemination away from the planted rows within and beyond
the field plot as a result of vertical and horizontal movement through the soil or through wind-
vectoring of aerosolized microbes.
2.3.3.1. Stressor Characterization
Laboratory studies showed that recombinants in bulk field soil underwent a 1-log reduction
in survival over a 4-week period. Literature on rhizobial survival in soil available at toe time of
the PMN review indicated that only limited horizontal and lateral movement of rhizobia in soil
would occur Three studies indicated that lateral movement by wind, water, and bactenal motility
was on the order of only 2.5 to 5 cm (Kellerman and Fawcett, 1907; Robson and Loneragan
1970- and Brockwell et al., 1972, as cited in Madsen and Alexander, 1982). However some of
these studies have limited utility as a result of their qualitative nature, use of autoclaved soil, or
lack of proper controls.
Aside from survival as intact cells in soil, rhizobia exist in a morphologically altered form
(bacteroids) in root nodules. These intranodal rhizobia can survive saprophytically at the end of
the growing season, when the alfalfa senesces. These populations can then reinfect alfalfa in the
field the following year. Consequently, the rhizobial population in the soil shows seasonal
variations.
The present case study needed to determine how well the recombinants could be monitored
in the field. The 1988 PMN presented minimum detection limits and recovery efficiencies for tte
rhizobial strains, based on the technologies available at that time. At the time of this review ^ EPA
considered the use of selective antibiotic media as the appropriate method for monitoring rhizobial
numbers in this small-scale field test.
In the data submitted with the PMN, the company reported that use of selective antibiotic
media gave an actual minimum detection limit (MDL) between 2 x 104 and 2 x itf cells/g soil.
The use of fluorescent antibody (FA) technique lowered the MDL to 103 to 10* cells/g soil.
Appendix A provides additional details of the monitoring and enumeration techniques.
In circumstances with low rhizobial counts, such as horizontal dispersal studies field tests
may require a more sensitive detection limit. To meet this need, a most probable number (MPN)
enumeration procedure was developed. This MPN technique involved placing alfalfa plants in
growth pouches and infecting their roots with dilutions of soil suspensions. Any plants in which at
least a single nodule formed was scored as a positive. In some cases, laboratory personnel
2-22
-------
identified the rhizobial strain present in nodules on the plants exposed to the highest dilution that
resulted in nodule fonriation. Although the MPN technique inherently has a high statistical error, it
gives a minimum detection limit of approximately 10 cells/g soil using soil from the field test site.
The MPN technique was more sensitive but less quantitative than the other enumeration methods.
Prior to the field tests, the MPN technique was used to determine the number of indigenous
rhizobia in the field site soil. At the Sun Prairie site, the number of indigenous rhizobia was <10
rhizobia/g soil.
The PMN submission included both routine and emergency termination procedures, which
received EPA approval prior to the field studies. Routine termination procedures after completion
of the field tests included plowing under the test plots and, if necessary, applying glyphosphate
herbicide to kill any remaining alfalfa or weeds. Severe adverse effects such as die-off of the
alfalfa, tremendous increases in population density, or movement off-site would indicate a need for
emergency termination. Emergency procedures included treatment of the test area with methyl
bromide to rninimize the microbial populations. EPA did not specify the exact criteria that should
have triggered emergency termination procedures. Instead, EPA advised the company to report anv
"irregularities." v j v j
2.3.3.2. Ecosystem Characterization
The ecosystem under consideration was a 0.8-acre (275 ft. x 300 ft.) field site plus the
immediate surrounding agroecosystem in Dane County, Wisconsin (with lesser concern for areas
farther removed from the site itself). The test site lay 500 feet from a road and was separated from
it by a fence. The majority of the field was Piano silt loam that consisted of deep, well to
moderately drained soil on glaciated uplands. The soil contained high levels of phosphorus,
potassium, magnesium, manganese, iron, zinc, and copper. Organic matter was 3.3 percent, and pH
was 6.8. Organic nitrogen content was not supplied, but was roughly estimated at 0.19 percent.1
In a later PMN submission, the company stated that a nitrogen content of 0.20 percent was limiting
for alfalfa growth. &
The slope of the field was approximately 2 percent from east to west. Dane County
receives approximately 31 inches of rain each year. Although infrequent, some runoff from the test
plot was expected. The runoff would drain into a ditch south of the site and then enter a culvert
that empties into Koshkonong Creek and eventually into Koshkonong Lake. The study did not
monitor microorganisms in runoff water because their level was expected to be below detection
limits. The site area had no wells that could become contaminated by dispersing microorganisms.
As a very rough estimate of organic nitrogen levels, one may assume a conversion factor of 1 724
between organic matter and organic carbon (Broadbent, 1965). Therefore, the organic carbon content
should be approximately 3.3 percent/1.724 = 1.9 percent organic carbon. Most agricultural surface
soils have C:N ratios of approximately 10:1 (Breioner, 1965), suggesting that fee soil had a nitrogen
content of approximately 0.19 percent,,
2-23
-------
To address concerns that R. meliloti might infect nontarget legumes, the 14-acre test area
and the ditch separating the test area from the road were scouted before and during the field trials
for the presence ofMettlotus (sweet clover) and weedy Medicago species.
2.3.3.3. Temporal Analysis
The field trials ran for a maximum of 2 years, but the company reserved the option to
terminate the trials earlier if it so desired.
2.3.3.4. Exposure Analyses
To determine the spatial and temporal distributions of the OEMs at the field site, several
studies were performed before the field test. These included laboratory studies on survival of the
OEMs in pure culture, survival in soil, survival in rhizosphere soil in the greenhouse, and the
ability of the OEMs to infect alfalfa in greenhouse studies.
The PMN included laboratory survival data of several recombinant rhizobial strains hi
soils. Unfortunately, the studies employed soils obtained from areas other than the test site. In
addition, some of the studies failed to include parental strains as controls. R. meliloti strains
RMB7101 and RMB7201 showed a 1- to 2-log reduction in numbers over a period of 6 weeks hi
both Chippewa soil and hi soil obtained from another field. Later studies tested a streptomycin-
resistant spontaneous mutant of RCR2011 against the four recombinant strains used in the field
tests. Approximately a 1-log reduction hi numbers for all the strains occurred over 4 weeks, with
no significant difference between strains.
The PMN submission contained some data concerning the persistence of the recombinant R.
meliloti hi the rhizosphere. The rhizosphere samples were separated into two fractions, the inner
and outer rhizospheres. Soil aggregates that fell off the roots with vigorous shaking represented
the outer rhizosphere, while the inner rhizosphere consisted of the remaining root system and
associated soil. Recombinant rhizobia /persisted in both soil fractions and hi nonrhizosphere soil,
with only slight declines hi numbers over the 3-week study.
The PMN included two pilot tests of nodule occupancy to study competitiveness of the
rhizobial strains. Competitiveness, hi this context, means the ability to form nodules hi alfalfa
roots when competing with another RMzobium strain. The presence of a strain hi a nodule suggests
that it is the strain that caused the nodule to form. In one study, the two parental strains,
RCR2011 and PC, showed no significant differences in competitiveness when inoculated into
alfalfa hi a 1:1 ratio. The second test indicated no significant difference hi competitiveness
between a naturally occurring and a recombinant strain that was not one of the stressors hi the test
study.
EPA recommended including nodule occupancy as part of the field trials because of the
absence of greenhouse nodule occupancy data for the OEMs hi the field test. Also, nodule
occupancy data can link altered alfalfa yield with the recombinant rhizobia. The field data on
nodule occupancy showed no significant differences hi nodule occupancy between the recombinant
and the wild-type rhizobia (appendix D).
2-24
-------
Monitoring the microbe at the field site can indicate whether the GEM is associated with
changes in alfalfa yield and can track OEMs beyond the field site. The study monitored vertical
horizontal, and aerial dispersal of lie recombinant rhizobia by means of the strain comparison test
described in appendix C.
Analysis of exposure also included a strain comparison test to determine the efficacy of the
inoculahts. 3
Comments on Characterization of Exposure
General reviewer comments:
• The technology for monitoring the spread of introduced strains from the
inoculation site suffered from potential limitations of sensitivity and specificity.
Newer, more sensitive and highly specific technologies (e.g., pofymerase chain
reaction fPCRJ amplification of strain-specific sequences, strain-specific probes,
marker cassettes) could be brought to bear on these problems. The
manufacturer also could provide quality assurance/quality control of the methods
used to monitor these important endpoints (e.g., proper controls, background
levels of native rhizobia).
« Further work on developing the idea of meaningful estimates of exposure to a
microbiological stressor is needed. An examination of the uninoculated alfalfa
border plants for nodule occupancy by strains introduced within the field plots
might give another indication of their spread.
Authors' comments: \
* Newer, more sensitive methods such as gene probes, PCR, or marker cassettes
for detection of microorganisms in environmental samples have been developed
in recent years. However, at the time of this submission, in 1988, those
techniques were not routine laboratory analyses, and these laboratory research
techniques have just recently been refined for use in environmental matrices.
The use of antibiotic-selective media, supplemented with the fluorescent antibody
technique, and the use of the MPN growth pouch technique were deemed
appropriate by the Agency at the time of the review. The company was not
required to submit actual QA/QC documents, but its use of appropriate methods
and protocols, the use of proper controls as well as other aspects of its field
experimental designs, and determination of background levels of rhizobia was
reviewed by the Agency before the field test.
2-25
-------
2.3.4. Analysis: Characterization of Ecological Effects
2.3.4.1. Evaluation of Effects Data
The primary effects data reviewed prior to the field test consisted of greenhouse studies that
examined alfalfa yields resulting from infection with recombinant rhizobia. For these studies,
plants were grown in sterile vermiculite inoculated with RMB7103. Because vermiculite is
nitrogen limiting, nodule occupancy data were probably not needed to show a causal link between
rhizobia in the nodules and top growth of plants. One study demonstrated that no significant
difference occurred in the growth of alfalfa plants inoculated with the parental strain, RCR2011,
and a recombinant strain, RMB7101. Similarly, no significant difference in alfalfa yield occurred
for plants inoculated with parental strain RCR2011 or recombinant strain RMB7103. In one study,
recombinant strain RMB7103 gave a yield increase of 7.0 percent compared with RMB7101.
Another study using these same recombinant strains showed no significant difference in their effect
on alfalfa yield. Field yield studies also showed a lack of significant yield effects (appendix F).
However, the results of the greenhouse yield data were questionable for two reasons.
First, the studies reported data as fresh weight of alfalfa top growth rather than as dry weight.
Secondly, the studies were of short duration. Harvest of the alfalfa plants occurred 3 weeks after
planting, but it usually takes 11 days for nitrogen fixation to begin. Consequently, these data
demonstrated growth for only 10 days after the onset of nitrogen fixation hi the nodules, making it
difficult to interpret the effects of the inserted genes.
The study did not collect data on the growth of sweet clover or fenugreek, nor did the
1989-1990 field test generate data on effects on nonlegumes or on legumes outside the cross-
inoculation group. A 1987 PMN offered limited qualitative information that indicated a lack of
effects on such plants. The earlier PMN greenhouse studies exposed soybeans, peas, tender green
beans, and clover to inoculation levels of 109 rhizobial cells/g of soil. Results indicated no adverse
effects. Similarly, corn and ryegrass, crop plants commonly grown in rotation with alfalfa,
showed no adverse effects from such exposures.
2.3.4.2. Evaluation of Causal Evidence
This section evaluates the strength of the relationship between the stressor and the
measurement endpoint, yield of alfalfa. Problems associated with the greenhouse tests are noted in
this section. In addition to the problems already noted for the greenhouse studies, extrapolating
from the greenhouse to the field also presents difficulties. For example, such an extrapolation
must take into account that the greenhouse and the field differ in climate, soil, and pest species.
2.3.4.3. Effects Needing Study in the Event of Significant Off-Site Migration or
Large-Scale Release
This risk assessment assumed that only limited off-site migration of the rhizobia would
occur. If, however, this risk assessment had been conducted for large-scale releases or if large
numbers of rhizobia moved off-site, the risk assessment would need to address at least six main
ecological concerns.
2-26
-------
Increased competitiveness. If large numbers of rhizobia moved off-site, then the
risk assessment would need to examine whether the increased population resulted
from enhanced competitiveness relative to native rhizobia. Displacement of native
rhizobia by increased competitiveness would be a concern if the GEM decreased the
growth of alfalfa or increased the growth of weeds.
Increased nitrogen production. Increased nitrogen production by alfalfa and other
legumes may increase soil nitrogen enough to contribute to nitrate pollution of soil
or ground water.
Alteration of host range. Alteration of host range can result in effects on legumes
other than those that R. mettloti is known to infect. However, host range alteration
appears unlikely for the submitted OEMs because no manipulations occurred in the
loci important to host range specificity.
Effects on nonlegumes. Because naturally occurring rhizobia have no effect on
nonlegumes, including those grown in rotation with alfalfa, effects on nonlegumes
appear unlikely. In addition, information about the constructs gives no reason to
suspect such effects.
Effects on sweet clover and fenugreek. Increased growth of the sweet clover
(Metilotus) when it occurs as a weed in another crop could adversely affect the
agroecosystem by decreasing the quality of the planted crop or by increasing
production of coumarin, a secondary metabolite found in the sweet clover plant that
is hazardous to livestock. Decreased growth of fenugreek or of sweet clover (when
grown as a crop) also could adversely affect certain agroecosystems. The
greenhouse and field data on alfalfa yield would not be predictive of the effects of
rhizobia on these other legumes.
Spread of antibiotic resistance. Large-scale releases offer a greater opportunity for
transfer of these resistances to bacterial pathogens of humans and animals.
Comments on Analysis: Characterization of Ecological Effects
Strengths of the case study include:
• The body of knowledge on the effects of previous uses of rhizobia (rhizobial
inoculation has been practiced for almost a century) and the well-characterized
strains in the case study compensate, in part, for the weakness of monitoring
effects.
2-27
-------
Comments on Analysis: Characterization of Ecological Effects (continued)
Limitations include:
• Neither the yield data from poorly designed and implemented greenhouse studies
nor the highly variable data from the field tests themselves could reliably
comment on the efficacy of the introduced genetically engineered rhizobia.
• Table 2-2 in the study lists 12 assessment endpoints and the sources of
information used to evaluate them, but only alfalfa growth was addressed in the
study. Off-site migration was considered, and the field test itself contributed
data with regard to the movement of rhizobia off-site.
2.3.5. Risk Characterization
2.3.5.1. Risk Estimation
The risk of conducting the small-scale field test was considered low. The field test would
collect data on alfalfa yield and microorganism fate. Decreased alfalfa yields, increased
competitiveness, or movement off-site could have triggered termination of the field test.
The study did not evaluate several assessment endpoints because of the small likelihood of
off-site dispersal. Both the characteristics of the field site and the test protocol supported this
position (see section 2.3.3.) The field site's low slope minimizes surface water runoff, and the site
contains no wells. In addition, the test protocol also limited movement off-site through (1) the in-
furrow spraying technique for rhizobial application, (2) on-site decontamination of equipment and
disposal of plant material, and (3) growth of rye grass and uninoculated alfalfa borders around test
plots. Further, monitoring of soil and water evaluated off-site movement, while the test protocol
also established emergency termination procedures in the event that significant spread appeared
likely. Data collected during the small-scale field test confirmed the prediction that only limited
off-site movement of rhizobia would occur. Appendix C presents the results of the aerial, lateral,
and vertical dispersion studies.
Even if dispersal had occurred, the numbers of microorganisms required for legume
infection may have precluded effective nodulation of other legumes near the site. The PMN
submission suggests 103 rhizobial cells/seed for agricultural application. Others have noted
infection concentrations of 100 to 1,000 rhizobial cells/g soil for effective nodulation of legumes
(van Elsas et al., 1990). Consequently, the assessment did not address large-scale effects such as
effects on the nitrogen cycle and the spread of clinically important antibiotic resistances. To assess
enhanced growth of weedy legumes, the study examined the 14-acre site for sweet clover and
weedy Medicago species (as noted in section 2.3.3). The study did not assess exposure to the
legume fenugreek because this crop plant grows only in certain portions of the United States.
2-28
-------
Appendix B presents the data on the competitiveness and survival of the rhizobial strains,
as measured by nodule occupancy and persistence in the rhizosphere. The low viability of some of
the inoculant strains (appendix E) affects the data in appendices B and C. As predicted from the
greenhouse data, the rhizobial strains became established and survived well in the rhizosphere.
Nodule occupancy tests demonstrated that the inoculant strains were fairly competitive compared
with indigenous rhizobial populations. The recombinant and naturally occurring strains showed no
significant differences in survival or competitiveness.
2.3.5.2. Uncertainty
Both effects and fate data and information in the PMN had elements of uncertainty. For
the greenhouse yield data, uncertainty resulted from the protocol, the alfalfa cultivar relative to the
field trials, and the extrapolation to field results. The alfalfa yield in the field may not have
reflected the ability of the rhizobia to increase alfalfa growth because the test did not measure total
nitrogen in the field soil, and high levels of nitrogen can inhibit nodulation by rhizobia. Heavy
weed and leaf hopper infestations also may have confounded the alfalfa yield data.
Uncertainty also exists regarding the effects on weedy legumes and other crop legumes in
the cross-inoculation group for R. meliloti. For the OEMs undergoing field testing, no data existed
that would have indicated their competitive ability to nodulate alfalfa relative to native rhizobia.
Fate data and information in the PMN also had elements of uncertainty associated with
them. Extrapolation from pure laboratory culture and greenhouse studies to the field is
questionable. How well the monitoring techniques could distinguish the released rhizobia from
each other and from the indigenous rhizobia is also uncertain.
2.3.5.3. Risk Description
After completion of the field test, the risk assessment indicates that the likelihood of
adverse effects occurring either in the field or beyond the field border is considered low because of
limited dispersal from the site, site termination procedures, the number of rhizobia needed to infect
alfalfa plants, competition from native rhizobia, and the natural decline in cell populations expected
in the absence of further alfalfa planting. Other effects noted for large-scale release of rhizobia
will be addressed should large-scale releases become likely (see section 2.3.3).
Comments on Risk Characterization
Strengths of the case study include:
The case study characterized as low the risk associated with limited release of
genetically engineered rhizobia into a small-scale field site. This assessment
was based on the generally held view that rhizobia are fairly innocuous
bacteria, that the site would effect adequate containment of the released
bacteria, and that the genetic construct would preclude transfer of the
introduced nif genes as well as the antibiotic resistance markers to other strains.
The reviewers generally were satisfied with that assessment.
2-29
-------
Comments on Risk Characterization (continued)
Limitations include:
• The effects data on efficacy were lacking, and there was considerable
uncertainty in monitoring data because of the limitations of the chosen methods.
Some attempt should have been made to address these shortcomings.
General reviewer comments:
• The case study should have included a table that addressed the uncertainties
introduced by the assumptions made. For example, the study assumed that the
plate-counting technology used to estimate the spread of introduced strains can
distinguish between the introduced strains and indigenous R. meliloti. Similarly,
the study should either give a literature citation or acknowledge the following as
an assumption: an infective dose of l(f rhizobial cells/seed establishes a safe
level of escaped rhizobia at less than 10*/g soil.
• The case study might formulate action thresholds that would trigger the
termination of the small-scale field test. These thresholds should consider the
minimal infective dose and available data on persistence of the OEMs in the
soil. For example, a test would be terminated when plate-counting on medium X
detects more than 1,000 GEMs/g soil.
• As in all risk assessments, difficulty quantifying the hazard quotient leads
assessors to argue for reduced exposure. Reviewers generally agreed that the
risk assessors should attempt to bring quantification of the risk components of
stressors to state of the art.
• The reviewers also generally agreed that proper measurement endpoints should
make possible a meaningful characterization of risk in the restricted small-scale
test.
2-30
-------
Comments on Risk Characterization (continued)
Authors' comments:
• It is inappropriate to establish a level of safety for escaped rhizobia at itf
cells/g soil for several reasons. First, it is impossible to establish a safety level
of a certain number of microorganisms that is below the detection limit for that
microorganism. Second, it is not known exactly how many rhizobia are needed
for a nodule formation. As discussed in section 2.3.5.1, according to the PMN
submission, l(f cells/seed is the international standard inoculation rate for R.
meliloti. This rate is supposed to ensure that the inoculant strain will be able to
outcompete indigenous rhizobia. Another report in the literature suggested that
Itf to Iff cells/g of soil are needed for effective nodulation of legumes (van
Elsas et al., 1990); however, no data were supplied in this paper, and no
reference was given for where these particular data could be obtained. Third,
knowledge of the ecology of rhizobia indicate that rhizobial numbers are
greatest in the rhizosphere of leguminous plants and may drop off several orders
of magnitude in the bulk soil away from the plant. Rhizobia populations are
known to persist in soils at low numbers for long periods of time, but will
increase dramatically if the leguminous host plant is introduced into that soil
Consequently, it is inappropriate to establish any specific number as a safe level
of escaped rhizobia in soil, even if one defines the portion of the soil that one is
sampling, and even if one were to select a specific number that actually could
be measured in this study.
• Knowledge of rhizobial ecology precludes the formation of "action thresholds"
for rhizobia. It is inappropriate to put exact quantitative values on what level is
safe and what level would trigger emergency termination of the small-scale field
tests because of (1) a general lack of knowledge of exactly how many rhizobia
are needed for infection, (2) the variability in population densities in the
rhizosphere vs. soil at increasing distance away from the plant roots, and (3) the
ability to stimulate rhizobial growth even after several years by planting the
suitable leguminous host as discussed above.
• The reviewers again request that state-of-the-art methodology be used. As
discussed in the case study and above, at the time this review was conducted
(1988), antibiotic-selective media supplemented with the fluorescent antibody
technique and the MPN growth pouch methods were deemed appropriate for
these field tests. Great advances in methodology for detection of
microorganisms in the environment over the past few years may allow for
greater sensitivity in measurements for future studies.
2-31
-------
2.4. DISCUSSION BETWEEN RISK ASSESSOR AND RISK MANAGER
The 5(e) Consent Order (DCO 50-899004545) summarized how to conduct the field test
and which data to collect. The Consent Order specified the following items:
» The field test will use EPA-approved protocols that will describe test objectives,
field site, methods of transport of microbes to site, methods to limit dissemination,
methods for detection and identification, descriptions of sampling procedures, and
analysis of data.
• The test will provide data on the following (with proper controls): nodule
occupancy for all four recombinant microbes; alfalfa yield effects of all four PMN
microorganisms; persistence in the rhizosphere with RMB7101 and RMB7103;
vertical dissemination of RMB7101 and RMB7103; horizontal dissemination of
RMB7101 and RMB7103; and aerial dissemination of RMB7101 and RMB7103
beyond the test plot during inoculation and termination.
• The test will comply with applicable provisions of the Good Laboratory Practice
Standards (40 CFR 792).
• Microorganisms not used in the test will be disposed of in accordance with the NIH
Guidelines for Research Involving Recombinant DNA Molecules (51 FR 16958).
• Reports on progress of the field test will be provided every 3 months.
• The company will terminate the test if an event occurs indicating that the
microorganisms have caused an adverse effect that EPA believes presents an
unreasonable risk of injury to the environment.
2.5. RISK VERIFICATION
2.5.1. Persistence
The small-scale field tests verified the risk assessment conducted for this PMN submission.
As expected from knowledge of rhizobial behavior and from greenhouse data, the recombinant
rhizobia persist in the rhizosphere of alfalfa plants (see appendix B). The recombinant strains that
the field trials investigated for persistence—strains RmSF38, RMB7101, and RMB7103—survived
at rates of lOMO6 cells/g dry root into the second year of the field study.
2.5.2. Competitiveness
As an indication of competitiveness of the recombinant rhizobial strains relative to
unmodified strains, the study included nodule occupancy tests. Those conducted in the greenhouse
used recombinant strains similar to the subject GEMs, while those subsequently conducted in the
field used subject GEM strains. Neither set of occupancy tests indicated any significant difference
2-32
-------
in nodule occupancy for recombinant and parental strains (appendix C). However, in the strain
competition trials, the recombinant strains appeared somewhat less competitive than the wild types.
Interpretation of the data from this latter study proved difficult, however, because problems with
culture viability prevented the desired ratio of 1:1 for the application rate of recombinant:parental
strain (appendix E). i
2.5.3. Dissemination From the Test Site
Information in the literature suggested that little off-site movement of rhizobia would occur
during the test studies. The various dispersal studies conducted during the field trials confirmed
this prediction (appendix C).
2.5.4. Effect on Alfalfa Yield During Field Test
Appendix E presents the alfalfa yield from the field studies and compares these with the
greenhouse studies submitted as part of the PMN. This information is useful for validating both
the risk assessment done by OPPT and this case study performed under the framework guidance.
As predicted from the laboratory and the greenhouse studies, the construct analysis, and the
literature, no adverse effects on alfalfa growth occurred with any of the rhizobial strains tested.
Significant increases in yield also did not occur. Most importantly, no significant differences
occurred between the use of the wild-type parental strains and the recombinant rhizobial strains.
2.6. KEY TERMS
biovar—A group of bacterial strains that can be distinguished by special biochemical or
physiological properties that are consistent (but insufficient to justify a subspecies name for
the group).
cassette—Structural and regulatory DNA sequences introduced into a GEM that allow the GEM to
express a phenotypic trait of interest to the PMN submitter.
construct—1. (adj.) Information describing the DNA and genetic manipulations used to create the
GEM. Such information covers the cassette, site of cassette insertion, use of vector DNA,
intermediate recipients, and final recipients of cassette sequences. 2. (n.) The final genetic
makeup of a GEM, including information noted for use of this term as an adjective.
cultivar—A group of individual plants that differ from others within the species due to certain
consistent phenotypic traits (synonym: "variety").
vector—DNA sequences such as plasmids used to move the DNA of interest (usually cassette
DNA) from one organism to another.
2-33
-------
2.7. REFERENCES
Alexander, M. (1977) Introduction to soil microbiology. New York, NY: John Wiley & Sons.
Bremner, J.M. (1965) Organic nitrogen in soils. In: Bartholomew, W.V.; Clark, F.E., eds. Soil
nitrogen. Agronomy No. 10. American Society of Agronomy, Madison, WI, pp. 93-149.
Broadbent, F.E. (1965) Organic matter. In: Black, C.A.; Evans, D.D.; White, J.L.; Ensinger,
L.E.; Clark, S.E., eds. Methods of soil analysis. Agronomy No. 9. American Society of
Agronomy, Madison, WI, pp. 1397-1408.
Finan, T.M.; Kunkel, B.; DeVos, G.F.; Signer, E.R. (1986) Rhizobium meliloti genes required for
C4-dicarboxylate transport and symbiotic nitrogen fixation are located on a megaplasmid. /.
Bacterial. 170:927-934.
Kellerman, K.F.; Fawcett, E.H. (1907) Movements of certain bacteria in soils. Science 25:806.
Madsen, E.L.; Alexander, M. (1982) Transport of Rhizobium arid Pseudomonas through soil. Soil
Sci. Soc. Amer. J. 46:557-560.
Milewski, E. (1990) In: Nakas, J.P.; Hagedorn, C., eds. Biotechnology of plant-microbe
interactions. New York, NY: McGraw-Hill Publishing Company, pp. 319-340.
Office of Science and Technology Policy. (1986) Coordinated framework for regulation of
biotechnology; announcement of policy and notice for public comment. Federal Register
51:23302-23393.
Prentki, P.; Krisch, H.M. (1984) In vitro insertional mutagenesis with a selectable DNA fragment.
Gene 29:303-313.
Robson, A.; Loneragan, J. (1970) Nodulation and growth of Medicago truncatula D on acid soils.
Part 1. Effect of calcium carbonate and inoculation level on the nodulation of Medicago
truncatula D on a moderately acid soil. Aust. J. Agric. Res. 21:427-434.
Sayre, P. (1988) Ecological hazard assessment and construct analysis for PMN submission P88-
1115 through -1122. Office of Toxic Substances, U.S. Environmental Protection Agency,
Washington, DC.
Sayre, P. (1990) Assessment of genetically engineered microorganisms under the Toxic Substances
Control Act: considerations prior to small-scale release. In: Gresshoff, P.M.; Roth, L.E.;
Stacey, G.; Newton, W.E., eds. Nitrogen fixation: achievements and objectives. New
York, NY: Chapman & Hall, pp. 405-414.
Segal, M. (1988) In: Shoemaker, S.; Middlekauf, R.; Ottenbrite, R., eds. The impact of chemistry
on biotechnology. Washington, DC: American Chemical Society, pp. 386-396.
2-34
-------
Tiedje, J.M.; Colwell, R.K.; Grossman, Y.L.; Hodson, R.E.; Lenski, R.E.; Mack, R.N.; Regal,
P.J. (1989) The release of genetically engineered organisms: a perspective from the
Ecological Society of America. Ecology 70:298-315.
U.S. Environmental Protection Agency. (1992) Framework for ecological risk assessment. Risk
Assessment Forum, Washington, DC. EPA 630/R-92/001.
van Elsas, J.; Heijnen, C.; van Veen, J. (1990) The fate of introduced genetically engineered
microorganisms (OEMs) hi soil, hi microcosm, and the field: impact of soil textural
aspects. In: MacKenzie, D.; Henry, S., eds. The biosafety results of field tests of
genetically modified plants and microorganisms. Bethesda, MD: Agricultural Research
Institute, pp. 67-79.
2-35
-------
-------
APPENDIX A
MONITORING AND ENUMERATION TECHNIQUES FOR RHIZOBIA
2-A1
-------
APPENDIX A
MONITORING AND ENUMERATION TECHNIQUES FOR RHIZOBIA
The monitoring studies used only three strains: RmSF38, a spontaneously streptomycin-
resistant mutant of the parental strain RCR2011, and two recombinants, RMB7101 and RMB7103,
both of which are derivatives of RCR2011.
Selective antibiotic media differentiate the parental from the recombinant and from the
indigenous rhizobial strains. The basic medium proposed for enumeration of all rhizobial isolates,
RDM (rhizobia defined medium), consisted of the following (g/L): potassium gluconate 5.0,
K2HPO4 0.22, MgSO4-7H2O 0.1, sodium glutamate 1.1, 1,OOOX trace elements, 1,OOOX vitamin
stock, and agar. The parental strain RCR2011 was intrinsically resistant to kanamycin and
cinoxachi at 10 /ig/mL and 100 iiglroL, respectively. Medium A, proposed for enumeration of
RmSF38, consisted of RDM supplemented with kanamycin (10 /tg/mL), cinoxacin (100 j*g/mL),
and streptomycin (200 jtg/mL) as well as the antifungal agents cycloheximide and nystatin, both at
the rate of 75 jig/mL. Medium B, for enumeration of the recombinant strains RMB7101 and
RMB7103, was identical to Medium A except for addition of another antibiotic, spectinomycin
(100 /tg/mL). Spectinomycin was needed because both streptomycin and spectinomycin resistances
were carried on the 0 fragment that was inserted to make the recombinant strains.
Recovery studies revealed that 51 to 90 percent of added rhizobia were recovered from the
Sun Prairie soil 1 hour after addition to the soil. The PMN contained data from preliminary
laboratory studies indicating that indigenous rhizobia intrinsically resistant to the same antibiotics as
the OEMs occurred in low numbers and did not increase greatly in the presence of plant roots.
For the fluorescent antibody technique (and for future measurements during the field tests),
the study selected 20 colonies from each antibiotic plate to determine the percentage of colonies
formed on that plate by the inoculant strain versus the indigenous rhizobial populations.
Multiplying this conversion factor by the total number of colonies on the plates corrected for the
inoculants and eliminated the indigenous rhizobia.
Dr. E.L. Schmidt at the University of Minnesota prepared the immunofluorescent antibody
to the parental R. meliloti strain RCR2011 using antiserum collected from the first production bleed
of an immunized New Zealand white rabbit. The fluorescent antibody was a conjugate of the IgG
fraction of the antiserum to the fluorescent dye, fluorescein. Dr. Schmidt's laboratory titered the
fluorescent antibody to determine the highest antibody dilution that provided an acceptable
homologous cross-reaction against strain RCR2011. A 1:1 dilution of the antibody suspension in
glycerol was diluted 1:2, 1:4, 1:8, and 1:16 in filtered saline, and each dilution was applied to
microscope slides containing rhizobial smears. Cross-reactivity was rated as (-) = no reactivity, tr
— trace, and from (1+) to (4+) indicating very weak to strong cross-reactivity. The 1:16 dilution
exhibited cross-reactivity of 4+ with RCR2011 derivatives but no cross-reactivity with the other
rhizobial parental strains, their derivatives, or indigenous rhizobial populations. Consequently, this
dilution was used for all further work. Laboratory tests conducted prior to the field tests indicated
an MDL of 5 X 103 cells/g dry soil with this supplemental fluorescent antibody technique.
2-A2
-------
Colony morphology also distinguished between the RCR2011 derivatives and the indigenous
populations. The indigenous rhizobia produced mucoid colonies, whereas the RCR2011 derivatives
were always nonmucoid.
Although other aspects of population dynamics studies used all strains, dispersal monitoring
used only the RCR2011 derivatives. Neither the PC parent or derivatives nor the UC445 parent or
derivatives had good enough antigenic properties to produce a fluorescent antibody usable for
detection. In addition, the highly mucoid PC strains were indistinguishable from the indigenous
population. The RCR2011 strain and its derivatives served as an appropriate model for microbial
dispersal, making it unnecessary to investigate all strains.
2-A3
-------
-------
APPENDIX B
PERSISTENCE IN THE RfflZOSPHERE AND NODULE OCCUPANCY
2-B1
-------
Persistence in the Rhizosphere
The 2-year study followed the establishment and persistence of three strains—a wild-type
strain, RmSF38, and two recombinant strains, RMB7101 and RMB7103—by means of selective
media plating. All three strains were established hi the rhizosphere and remained stable through
the 1989 growing season at levels of approximately 10s hi the inner rhizosphere and 10s to 106
cells/g dry root hi the outer rhizosphere. The first sampling hi April 1990 revealed rhizobial
numbers hi the inner and outer rhizosphere similar to the levels for the last sampling of the 1989
season, indicating that the rhizobial steams either overwintered at these levels or recovered after
thawing hi the spring. Although all three steams persisted hi the rhizosphere through day 376, the
levels of the two recombinant steams were approximately tenfold lower than the level of indigenous
rhizobia. In summary, both the wild-type and the two recombinant steams became established hi
the rhizosphere and persisted into year 2, in general showing no population differences.
Nodule Occupancy
The strain comparison trial on October 3, 1989, entailed nodule occupancy studies. The
study measured the length of the root systems for 12 plants, with the root system being divided into
four sections: crown, top middle, bottom middle, and distal. A maximum of 24 nodules from
each section was screened for the presence of the inoculant. Unfortunately, the parental steam PC
and the indigenous rhizobia were indistinguishable. However, the other parental steams and the
recombinants could be identified. The data indicated that nodule occupancy ranged from 39 to 70
percent for the inoculated rhizobial steams, the remaining nodules being occupied by the indigenous
rhizobia. The percent nodule occupancy by the inoculant decreased with increased distance from
the crown hi all cases. No significant differences occurred between the wild-type and recombinant
strains. Plants collected hi the second year, 15 days prior to the second harvest, showed a decline
hi percent nodule occupancy for all inoculated treatments.
In the steam competition trial, parental and recombinant steams were inoculated together.
Nodule occupancy data showed that recombinant steams appeared somewhat less competitive than
the wild types. Because problems with culture viability prevented the desired inoculation ratio of
1:1, interpreting these data is difficult (appendix D).
2-B2
-------
APPENDIX C
RHIZOBIAL DISPERSION AND MIGRATION
2-C1
-------
Aerial Dispersal
Selective agar plates were mounted on posts located in all four compass directions at
various distances—4, 9, 50, 100, 200, and up to 500 feet—from the perimeter of the test plots on
days 0, 1, 2, 3, 4, and 6 after initiation of the strain comparison trial. Additional plates were
placed between the four compass points. No colonies appeared on the vast majority of plates
regardless of compass direction or distance. A total of 13 colonies appeared on Selective Medium
A over a cumulative exposure of 6 hours on day 0 for all compass directions and distances even
though a moderate wind blew on the day of application. Later samplings were for 2-hour
exposures only. On day 6, the number of colonies on Medium A from the west compass direction
(the direction with the highest counts) had dropped from 13 at the 4-foot distance to one colony at
both the 100- and 200-foot distances. Overall, little aerial dispersion of the PMN microorganisms
occurred. Likewise, aerial dispersion measurements taken at termination, when the fields were
being plowed, resulted in no detectable dispersal of inoculant from the test site.
Vertical Migration
Movement of the recombinant rhizobia downward through the soil profile past the
rhizosphere was measured by plating out soil obtained with a soil-coring device. Twelve-inch
cores were taken from control and treated plots hi an outside row, immediately adjacent to a plant
stalk. The top 2 and bottom 2 niches of the soil core were homogenized and subsampled for the
presence of added rhizobia.
Vertical monitoring used the plant MPN technique for enumeration at various time points
up to 312 days. Throughout the season, cell numbers ranged from 7 to > 138 cells/g dry soil in
the top 2 inches and from 3 to >524 cells/g dry soil in the 10- to 12-inch depth. Rhizobial
inoculants also occurred at a depth of 22 to 24 niches. Overall, only minimal movement occurred
beyond the root zone. No differences occurred in the vertical movement of the recombinant strains
versus the wild-type strain.
Horizontal Dispersion
The study monitored horizontal movement through the soil by sampling the top 2 inches of
the soil surface at a distance of 6 inches away from the edge of the plots in all four compass
directions on days 0, 11, and 34. Samples were examined for the presence of three strains:
RmSF38 and two recombinants, RMB7101 and RMB7103. Using selective media supplemented
with the fluorescent antibody method, samples contained no detectable inoculants. With the more
sensitive MPN enumeration technique, counts ranged from 0 to 57 cells/g dry soil. Consequently,
all subsequent analyses used the MPN technique. Up through day 123, cell counts never exceeded
250 cells/g dry soil, and nearly all counts dropped to 0 by day 159. These results indicate minimal
horizontal movement of the rhizobial inoculants throughout the study and no differences in the
behavior of the recombinant strains versus the wild type.
2-C2
-------
APPENDIX D
STRAIN COMPARISON AND COMPETITION TESTS
2-D1
-------
APPENDIX D
STRAIN COMPARISON AND COMPETITION TESTS
The strain comparison test used four recombinant strains and a single alfalfa variety. The
total area for the strain comparison trial was approximately 0.65 acre, with 0.07 acre treated with
recombinant rhizobia. The proposed design consisted of 13 treatments set up as a complete
randomized block design with six replicates. Each treatment occupied a plot measuring 5 X 25
feet. A 5-foot wide buffer strip of ryegrass separated plots from each other. A 5-foot wide border
of uninoculated alfalfa surrounded the experimental area. Alfalfa seeds were planted with a cone
planter in rows 6 inches apart and sown to a depth of approximately 0.25 to 0.5 inches. A carbon
dioxide-propelled bicycle sprayer, calibrated to deliver 10 mL/linear foot, sprayed 3.0 L of
suspensions of each rhizobial strain on the alfalfa seeds in the open furrows. The application rate
was approximately 10s bacteria per seed. This rate corresponded to 2.3 X 10" seeds per plot, for
a total of 5.52 X 1012 recombinant R. meliloti cells. Immediately following spraying of the
rhizobia, garden rakes were used to cover the furrows with soil.
The strain competition experiments took place on a 0.09-acre portion of the same field (48
X 78 feet). The proposed design consisted of 22 treatments set up as a randomized complete block
design with four replications. Each treatment consisted of one row, 6 feet long, with seeds spaced
every 0.5 to 1.0 inch. Rows were 3 feet apart. Because of the experiments' short duration (8
weeks), the ryegrass borders were omitted. As in the strain comparison trial, a 5-foot wide border
of uninoculated alfalfa surrounded the entire test area. A hand-held spray bottle sprayed 50 mL of
rhizobial suspension into each 6-foot furrow row. At an inoculum rate of approximately 105
bacteria per seed, each treatment had a total of 1.2 x 1010 rhizobial cells. This corresponded to a
total application of 5.52 x 1011 recombinant rhizobial cells. Then the furrows were covered with
soil.
Although the test design called for applying R. meliloti strains at a rate of 10s cells per seed
to obtain 100 times the international minimum standard for alfalfa of 103, the actual viable counts
applied in the field were significantly lower. For results of viability studies, see appendix E.
2-D2
-------
APPENDIX E
RHIZOBIAL CULTURE VIABILITY
2-E1
-------
APPENDIX E
RHIZOBIAL CULTURE VIABILITY
To determine the actual application rate of the R. mettloti strains sprayed on the seeds,
aliquots of the rhizobial suspensions were plated onto selective media to measure culture viability.
The following table summarizes the results.
Applied Strain
% of Anticipated Viable Cells
Strain Comparison Strain Competition
RCR2011 (parent)
RMB7101 (RCR2011 parent + fl)
RMB7103 (RCR2011 parent + 0 + mf)
97
113
89
60
10
20
PC (parent)
RMB7201 (PC parent + 0)
49
43
20
40
UC445 (parent)
RMB7401 (UC445 parent + 0)
77
50
5
5
Note that in some cases the numbers obtained are much lower than the number of viable
cells intended for application. This situation is particularly true for the strain competition trial.
Consequently, the strain competition trials often did not have the desired 1:1 ratios. The ratios of
parent: recombinant for recombinants RMB7101, RMB7201, RMB7401, and RMB7103 were
1:0.85, 1:1.1, 0.34:1, and 1:0.43, respectively.
2-E2
-------
APPENDIX F
ALFALFA YIELDS IN THE FIELD TESTS
2-F1
-------
Table Fl. Alfalfa Yields in Year One—First Cutting
Treatment
RCR2011
RMB7101 (RCR2011 parent + 0)
RMB7103 (RCR2011 parent + 0 + nij)
PC
RMB7201 (PC parent + 0)
UC445
RMB7401 (UC445 parent + 0)
Alfalfa Yield (kg/ha)
3,295
3,362
4,707
3,766
3,071
3,295
3,676
The naturally occurring and recombinant strains tested gave no significant differences in the
dry weight yield of alfalfa at the first cutting (coefficient of variance [C. V.] 42.45 percent). This
result may have occurred, in part, because of the variable stand of alfalfa often observed the first
year after planting. The test plots suffered heavy weed infestation (no preplant herbicide was
used), and the alfalfa plants also suffered stunting and chlorosis as the result of a heavy leaf hopper
infestation in early July. To allow for spraying for leaf hoppers, the first cutting occurred earlier
rather than the normal 10 percent bloom standard.
The table below presents the dry weight yields of alfalfa for the second cutting, which
occurred 44 days after the first cutting.
Table F2. Alfalfa Yields in Year One—Second Cutting
Treatment
Alfalfa Yield (ke/ha)
RCR2011
RMB7101
RMB7103
PC
RMB7201
UC445
RMB7401
3,295
3,049
3,362
3,004
2,892
3,049
3,049
2-F2
-------
Again, the naturally occurring and the recombinant strains resulted in no significant
differences in the dry weight yield of alfalfa (C.V. 11.74 percent). The test plots again showed
heavy weed infestation.
Alfalfa was harvested twice in the second year of the field tests, once on June 6 and 7 and
again on July 24. The table below presents data for dry weight yield.
Table F3. Alfalfa Yields in Year Two—First and Second Cuttings
Treatment
Alfalfa Yield fkg/hal
1st Cutting 2nd Cutting
RCR2011
RMB7101
RMB7103
PC
RMB7201
UC445
RMB7401
C.V. (%)
4,304
4,102
4,416
4,281
4,506
4,304
4,438
11.50
6,052
6,232
6,590
6,254
6,590
6,209
6,276
6.95
The second year of the strain comparison test showed no significant differences in alfalfa
dry weight yield with wild-type and recombinant R. meliloti strains. The second year's alfalfa
growth lacked much of the variation seen in the first year. Consistent trends, however, were not
evident.
The field data showed no conclusive trends toward either increased or decreased growth of
alfalfa as compared with the parent strains. Therefore, the field tests indicate that the recombinant
rhizobia posed little risk of decreasing alfalfa yields. The greenhouse data, although faulty, also
indicated little potential for decreased growth of alfalfa from the OEMs.
2-F3
-------
-------
SECTION THREE
ECOLOGICAL, RISK ASSESSMENT CASE STUDY:
EFFECTS OF RADIONUCLIDES IN THE COLUMBIA RIVER SYSTEM-
A HISTORICAL ASSESSMENT
-------
AUTHORS AND REVIEWERS
AUTHORS
Stephen Li. Friant
Environmental Science Department
Battelle Pacific Northwest Laboratories
Richland, WA
Charles A. Brandt
Environmental Science Department
Battelle Pacific Northwest Laboratories
Richland, WA
REVIEWERS
Thomas Sibley (Lead Reviewer)
Fisheries Research Institute
University of Washington
Seattle, WA
Gregory R. Biddinger
Exxon Biomedical Sciences, Inc.
East Millstone, NJ
Joel S. Brown
University of Illinois
at Chicago
Chicago, IL
Herbert Graver
Benchmark Environmental Corporation
Albuquerque, NM
Joseph E. Lepo
Center for Environmental Diagnostics
and Bioremediation
University of West Florida
Gulf Breeze, FL
Frieda B. Taub
School of Fisheries
University of Washington
Seattle, WA
3-2
-------
CONTENTS
ABSTRACT 3-7
3.1. RISK ASSESSMENT APPROACH , 3-8
3.2. STATUTORY AND REGULATORY BACKGROUND 3-8
3.3. CASE STUDY DESCRIPTION . . 3-10
3.3.1. Background Information and Objective 3-10
3.3.2. Problem Formulation 3-12
3.3.2.1. Stressors : 3-12
3.3.2.2. Biological Fate of Radionuclides 3-14
3.3.2.3. Ecosystem Potentially at Risk .....' 3-15
3.3.2.4. Endpoint Selection 3-15
3.3.2.5 Conceptual Model 3-15
3.3.3. Analysis: Characterization of Exposure 3-17
3.3.3.1. Sample Location . . . 3-17
3.3.3.2. Data Analysis 3-18
3.3.3.3. Exposure From Measured River Water Concentrations 3-18
3.3.3.4. Calculation of Organism Dose . . ] 3-18
3.3.3.5. Dose From Water Exposure . . ; . 3-23
3.3.3.6. Dose From Measured Tissue Concentrations 3-23
3.3.3.7. Dose From Measured Sediment Concentrations 3-27
3.3.4. Analysis: Characterization of Ecological Effects 3-27
3.3.5. Risk Characterization 3-31
3.3.5.1. Acute Exposure to Ionizing Radiation 3-31
3.3.5.2. Chronic Exposure to Ionizing Radiation 3-32
3.3.5.3. Uncertainty 3-32
3.3.5.4. Conclusions 3-33
3.4. REFERENCES 3-36
3.5. ADDITIONAL READING j 3-39
APPENDIX A—COLUMBIA RIVER FISH SPECIES AND FOOD WEB 3-A1
3-3
-------
CONTENTS (continued)
APPENDIX B-CRITR2 CODE CALCULATIONS AND
BIOACCUMULATION FACTORS
3-B1
3-4
-------
LIST OF FIGURES
Figure 3-1. Structure of assessment for effects of radionuclides 3-9
Figure 3-2. Location of the Hanford Reach of the Columbia River . . 3-11
Figure 3-3. Monthly concentrations for selected radionuclides in Columbia River grab
samples, 1963 3-20
Figure 3-4. Monthly concentrations for selected radionuclides in Columbia River grab
samples, 1964 3-21
Figure 3-5. Ranges of sensitivities of aquatic organisms to acute radiation exposure .... 3-29
Figure 3-6. Range of sensitivities of the early developmental stages of fish to acute
exposures 3-30
LIST OF TABLES
Table 3-1. Summary of Water Quality Data, 1957-1973 3-13
Table 3-2. Water Sampling Matrix (1963-1964) 3-19
Table 3-3. Maximum Grab Sample Water Exposure Concentrations for 1963-1964 Time
Period 3-22
Table 3-4. CRITR2 Code Calculation of Organism Dose From Water Exposure to
Various Radionuclides 3-24
Table 3-5. Calculated Dose Based, on Tissue Concentration for Selected Organisms of the
Columbia River ; 3-26
Table 3-6. Maximum Sediment Radionuclide Concentrations in the Hanford Reach and
Dose to an Organism Living in the Sediments 3-28
Table 3-7. Hazard Quotient for Early Development Stage of Fish and Adult Fish .... 3-32
LIST OF COMMENT BOXES
Comments on Problem Formulation 3-17
Comments on Characterization of Exposure 3-27
Commands on Characterization of Ecological Effects 3-31
Comments on Risk Characterization 3-34
3-5
-------
LIST OF ACRONYMS
BCF bioconcentration factor
CERCLA Comprehensive Environmental Response, Compensation, and Liability Act of 1980
DOE Department of Energy
HQ hazard quotient
NRDA Natural Resource Damage Assessment
PNL Pacific Northwest Laboratory
RM River Mile
USGS United States Geological Survey
3-6
-------
ABSTRACT
In 1943, nuclear production activities began at the U.S. Department of Energy's (DOE)
Hanfprd site in south-central Washington State. These activities continued for many years. During
this time, the site discharged radioactive effluents into the Columbia River, which runs through the
northern portion of the site and borders it on the east (the Hanford Reach). The DOE requested
the Pacific Northwest Laboratory (PNL) to conduct an ecological risk assessment to determine
whether the ecological risk assessment framework (EPA, 1992) used for hazardous chemicals is
applicable to radionuclides as stressors. PNL conducted this ecological risk assessment using
historical Hanford site monitoring data, which had been collected to characterize human dose. The
data characterized exposure by measuring radioactivity in water, sediments, and biota. The data
used in the current investigation were collected during 1963-1964, a period of peak production of
nuclear material. During this time, the maximum number of eight reactors were operational.
PNL employed two approaches in assessing ecological risk to Columbia River organisms.
The first approach used environmental exposure data (water concentrations for radionuclides) to
calculate dose to a variety of aquatic organisms, including the most sensitive receptors (fish). The
second approach made use of measured tissue concentrations of selected aquatic organisms to
calculate organism internal dose.
PNL used dose to assess potential toxic effects and assess regulatory compliance. Risk
characterization was developed by comparing dose levels in fish and other organisms found in the
Columbia River to known effect concentrations through a hazard quotient for acute dose and
possible developmental effects. The assessment endpoint was protection of fishes in the Columbia
River, and the measurement endpoint was increases in mortality and sublethal effects. One of the
most sensitive ecological receptors was the early developmental stage of chinook salmon.
The major conclusions of the study are:
• The ecological risk assessment paradigm is applicable to radionuclides as well as to
hazardous chemicals, as evidenced from the exposure, effect, and risk
characterization.
• The most sensitive life stage of fish (i.e., salmon embryo) did not appear to be at
risk from radionuclide exposure in sediments or water.
• During peak production at Hanford, releases of radionuclides did not result in any
measurable risk to the Columbia River ecosystem, as evidenced by indicator species
and regulatory benchmarks.
• Dose rates to Columbia River animals during the study period did not exceed the
DOE standard of 1 rad/d per DOE Order 5400.5 (DOE, 1989). Based on the
computer code CRITR2, only crayfish and a plant-eating duck received a dose rate
exceeding 1 rad/d. However, this risk assessment did not include ducks, and the
actual calculation of dose to crayfish from whole organism counts gave values
considerably less than both the modeled dose and 1 rad/d.
3-7
-------
3.1. RISK ASSESSMENT APPROACH
The ecological risk assessment follows the sequence of the U.S. Environmental Protection
Agency's Framework for Ecological Risk Assessment (EPA, 1992). This arrangement includes
problem formulation, analysis, and risk characterization, respectively (figure 3-1).
Exposure of aquatic organisms to radioactivity can elicit a toxic response depending on the
organism, level of dose, type of radionuclide, and habitat requirements of the exposed organism.
In this study, the assessment endpoint was defined as the maintenance of important recreational and
commercial fish populations in the Columbia River. The measurement endpoint from radioactive
dose was toxicological response. This assessment did not consider elemental chemical toxicity of
each radionuclide.
The major ecological components are benthic macroinvertebrates, zooplankton,
phytoplankton, and fish of the Columbia River. Fish species in the Columbia River are important
commercial, recreational, cultural, and regional assets.
Data analysis included exposure and effects characterization. Exposure characterization
consisted of an assessment of radioactivity at several river stations downstream from the Hanford
site. Measured river activity was used to calculate ionizing radiation dose from water to selected
organisms using bioaccumulation factors and computer modeling. A second and more direct means
of estimating dose to aquatic organisms used measured fish tissue concentrations. Available
sampling data included sediments, water, and biota.
Characterization of effects to aquatic organisms entailed using available toxicity data and
regulatory standards. The characterization was conducted at the individual level, qualitatively
interpreted, and applied to the population level of ecological organization. Risk characterization
was based on a hazard quotient (HQ), defined as the ratio of radionuclide organism dose (exposure
or tissue value) to benchmark dose values.
3.2. STATUTORY AND REGULATORY BACKGROUND
Although federal regulations do not require quantitative ecological risk assessments, they
can be used effectively to support regulatory requirements under nearly all of the major federal
environmental statutes (e.g., the Comprehensive Environmental Response, Compensation, and
Liability Act, CERCLA). Other potential applications include supporting compliance with federal
Executive Orders and with policy directives of various government agencies (e.g., DOE Orders).
A number of federal statutes have promulgated risk-based and technology-based standards
for the protection of ecological resources (e.g., water quality criteria under the Clean Water Act).
However, only one standard has been published for the protection of ecological resources from
exposure to radioactive materials. DOE Order 5400.5 (DOE, 1989) stipulates that the interim dose
limit for native aquatic animal organisms "shall not exceed 1 rad per day from exposure to the
radioactive material in liquid wastes discharged to natural waterways."
3-8
-------
PROBLEM FORMULATION
Stressprs: Ionizing radiation from radionuclides associated with the Hanford
site. Other chemical and physical stressors were not considered.
Ecosystem(s) at Risk: Columbia River downstream of the Hanford Site,
Richland, Washington
Ecological Components: Fish, zooplankton, phytoplankton, and benthic
macroinvertebrates in the Columbia River.
Endpojnts: Assessment endpoint was the maintenance of important
recreational and commercial fish populations in the Columbia River. The
measurement endpoints included dose-response information for radiation and
single species of aquatic organisms.
ANALYSIS
Characterization
of Exposure
Radioactivity of river water samples
was measured and used to calculate
ionizing radiation dose to selected
species using bioaccumulation
factors and models. Dose also was
determined by direct measurement
of fish tissues and sediments.
Characterization of
Ecological Effects
Radiation effects were evaluated
based on available laboratory
stresspr-response information on
mortality and developmental
effects and regulatory standards.
RISK CHARACTERIZATION
Hazard quotients were used to compare maximum exposure doses to the
lowest reported doses causing adverse effects to aquatic organisms. Major
uncertainties associated with this approach were described.
Figure 3-1. Structure of assessment for effects of radionuclides
3-9
-------
3.3. CASE STUDY DESCRIPTION
3.3.1. Background Information and Objective
It is generally assumed that human health risk standards for radionuclides protect wildlife
sufficiently. However, under some circumstances the risk to wildlife from radionuclides may need
to be considered, such as managing risks, developing cleanup strategies, and identifying injury
under the Natural Resource Damage Assessment (NRDA) process. The objective of this case study
is to evaluate the applicability of the ecological risk assessment paradigm for radionuclides as
stressors in the Columbia River.
The Hanford site, an area of slightly more than 1,400 km2 (560 mi2), straddles the
Columbia River just north of Richland, Washington. Three northwest-southeast-trending basalt
ridges cross this broad, relatively level gravel plain. The semiarid climate supports various
communities of shrubs—steppe and grassland.
The Columbia River extends 1,954 km (1,214 mi) from its origin in Columbia Lake in
British Columbia to its mouth at Astoria, Oregon, making it the fourth-longest river in North
America. Typical flow rates of the Columbia River at Priest Rapids Dam range from 2,800 to
3,400 cubic meters per second (cms), or 99,000 to 122,000 cubic feet per second (cfs) (Woodruff
et al., 1991).
The Columbia River has eight primary uses:
1. River navigation through navigation locks from the Pacific Ocean to the Port of Benton
in Richland.
2. Agricultural purposes, primarily irrigation. Approximately 6 percent of the Columbia
Basin's water is diverted for agricultural use.
3. Nonagricultural irrigation.
4. Electric power generation, provided by the system of 11 dams along the Columbia
River in the United States.
5. Flood control, also provided by the dams.
6. Fish and wildlife habitat, especially for anadromous salmon. The Hanford Reach
comprises the last major salmon and steelhead spawning area within the Columbia River
proper. The Columbia River also supports the vast majority of mesic terrestrial habitat
in the semiarid Hanford Reach.
7. Water supplies to numerous municipalities and industries.
8. Recreational use.
The Hanford Reach of the Columbia River runs from Priest Rapids Dam to just north of
the City of Richland and flows past the reactor areas of the Hanford site (figure 3-2). The average
annual flow of the Columbia River in the Hanford Reach, based on 65 years of record, is about
3,400 cms (120,100 cfs) (DOE, 1988). Flows in the Hanford Reach vary widely, not only because
of the annual flood flow but also because of daily regulation by the upstream power-producing
Priest Rapids Dam. Flow rates during the late summer, fall, and winter may vary from a low of
3-10
-------
Seattle f Spokane
Washington
Vancouver
-»X
Portland
Priest Rapids
Hanford
Town Site
\jTT*BPond
PUKEX
200 Areas
Hanford-
Site
Boundary
Advanced Nuclear Fuels
3QQQ Area
1100 Area
Rich]and Pumphouse
Arid Lands Ecology Reserve
Saddle Mountain
National Wildlife Refuge
Washington State
Department of Game Reserve
Figure 3-2. Location of the Hanford Reach of the Columbia River
3-11
-------
1,100 cms (36,000 cfs) to as much as 4,800 cms (160,000 cfs) each day. During the spring
runoff, peak flow rates from 4,800 to 20,000 cms (160,000 to 650,000 cfs) can occur.
The Washington State Department of Ecology classifies the Columbia River water quality
as Class A (excellent) between Grand Coulee Dam and the mouth of the Columbia River (DOE,
1988). Table 3-1 shows water quality data between Priest Rapids Dam and Pasco, Washington, for
the years 1957-1973. The dominant physical feature of the Columbia River through the Hanford
Reach is the high flow rate, which is subject to large, diurnal water-level fluctuations that change
the shoreline configuration and expose gravel substrate and periphyton to alternate periods of
wetting and drying. The Reach has a low level of suspended sediment, 1 to 7 mg/L.
The river-bottom sediments from Priest Rapids Dam to several kilometers below the
confluence of the Snake and Columbia Rivers are primarily mixed sands and gravels with some
cobbles (maximum diameter « 20 cm). Coarser sediments predominate from Priest Rapids Dam
through the reactor areas (DOE, 1988). The streambed near Richland consists of sand in deep
channels and a mixture of sand, silt, and some clay in shallow areas (DOE, 1988). Most of the
Hanford-produced cationic radionuclides are associated with suspended particulates and subsequent
fine sediments (Beasly and Jennings, 1984).
Because of the many dams on the Columbia River, the only free-flowing U.S. section
occurs between Priest Rapids Dam (River Mile [RM] 397) and McNary Reservoir (RM 351). The
Priest Rapids Dam immediately upstream from the Hanford site regulates flow. No significant
tributaries enter the stream in this section, which lies mostly within the Hanford site.
The main channel of the Hanford Reach is braided around the island reaches and
submerged rock ledges and gravel bars, causing repeated pooling and channeling. The riverbed
material is mobile and dependent on river velocities; it typically is composed of sand, gravel, and
rocks up to 20 cm (8 in) in diameter. Small fractions of silts and clays are associated with the
sands in areas of low-velocity deposition.
3.3.2. Problem Formulation
3.3.2.1. Stressors
The release of radionuclides from Hanford operations is one of several possible stressors to
the ecosystems of the Columbia River. Other possible stressors include thermal discharges from
Hanford reactors; varying river levels because of dams; the physical barrier to fish migration from
the dams; and heavy agricultural, commercial, and recreational activities along the river.
However, this assessment concerns only radionuclides as stressors of concern.
The cooling effluents of Hanford reactors contain over 60 radionuclides. Becker (1990)
has reported that during the period of maximum reactor production (mid-1960s), the Hanford site
discharged over 300,000 curies per year to the river. Radioactive decay influenced the relative
abundance of different radionuclides in the river (Becker, 1990). In fact, many of the
radionuclides discharged by the Hanford site have a short half-life and were not detected in the
effluent discharge. Others could not be detected in the river after dilution. Becker (1990)
3-12
-------
So*
OS
^
td
g
OS
eS
•2
Q
a
j
•s
CO
*
I
III
0X3
If
I~
js
II
6
o V §
o » "^
W N^ 3
K
a
e
II
o §
W &
1
I
tO
H ^
b|
0 3
M CO
fN O O
O »— i VJ
0 0 -<'
o § !2
0 0 0
o co r)
>n ON »— i
*o vo oo
^ co
o «n co
>o r~ ID
\Q f^ 00
g
CO ~
0 — CN
1
.2 -* «
S ^-1 *~" °^
^ ell
*T-1 O\ O\ ^
! 3 -' * -a
« ON »-< ^H .,-
(U S
•r" »N e .^
1 1 , 1 !
^o *g 3 x o
.2 S >2 M efl
« ,2: ^ 2 OH
«o ON r>-
O ••--< CO
o o o
S 2 8
0 0 0
o a:S
f^ CO ^P
^f t*» ON
OO
O 00 VO
oo •rt vo
vd oo 06
cs S
CN QQ
^H ^^ ^T
o ^ °-
• cs CN
CO ^ CN
„ 00 CO
vd 2 S
s 1
1 a 1
III
•«'l
d gl
Dissolved oxyj
Jackson turbidi
Most probable
Platinum-cobal
II II II II
O
3-13
-------
identified three radionuclides as being of concern because of their potential biological significance:
phosphorus-32, chromium-51, and zinc-65. Together they account for over 90 percent of potential
radiological dose to aquatic organisms. All are nuclear activation products that are activated as
Columbia River water cools the reactor core. The potential for some radionuclides to
bioaccumulate in aquatic food webs causes concern with respect to both the human exposure
pathways and potential ecosystem effects.
Among radionuclides, phosphorus-32 and zinc-65 are potential stressors because of their
biological importance and fate: they are essential elements for organism growth and are
incorporated into the aquatic food web. One study conducted in the Hanford Reach from 1961 to
1968 noted a seasonal pattern of uptake by algae, with higher radioactivity in winter and lower in
summer (Becker, 1990). This pattern reflects concentration and dilution phenomena from river
flows.
Unlike phosphorus-32 and zinc-65, chromium-51 is not considered a major biological
hazard. This radionuclide has a short half-life, low biological mobility (i.e., it has no known
essential role in the physiology of organisms), and weak radiations. It does not accumulate to any
extent in aquatic organisms and is transported with river-suspended paniculate material with little
dissolution (Becker, 1990). However, the risk assessment included it because it was a significant
activation product.
The half-lives of the three radionuclides considered in the risk assessment are:
• Phosphorus-32: 14.2 days
• Chromium-51: 27.8 days
• Zinc-65: 245.0 days
Phosphorus-32 is a beta emitter (negatrons); chromium-51 emits gamma radiation and
electrons; and zinc-65 is primarily a gamma emitter, but also emits positrons and electrons.
3.3.2.2. Biological Fate of Radionuclides
Phosphorus, including phosphorus-32, is a building block of various tissues and is a key
element in many biochemical transformations, especially energy transduction (ATP, ADP, GTP,
etc.). The element is comparatively scarce in the environment. Organisms can concentrate
phosphorus, including phosphorus-32, to levels that greatly exceed the concentration in the ambient
media. Phosphorus has a bioconcentration factor (BCF) of 24,000 for freshwater plants and 8,000
for freshwater animals (Becker, 1990).
Terrestrial plants take up little chromium-51 from soils, <0.5 percent (Becker, 1990). In
aquatic systems, this element sorbs to paniculate material and is transported along with it. Becker
(1990) reported that in biological systems chromium-51 has an affinity for the blood of fish.
Organisms accumulate a measurable fraction of zinc-65. In aquatic systems, this
radionuclide is transported through aquatic food webs. With chronic uptake, substantial tissue
accumulation can occur. In the Pacific Ocean, Becker (1990) noted BCFs of up to 103 for algae
3-14
-------
and 10s for certain molluscs. The BCF for plankton in the Columbia River ranges from 300 to
19,000 (Gushing and Watson, 1966; Gushing, 1967a, b), with adsorption as the primary means of
uptake. Because of its long half-life and biological mobility, zinc-65 can be transported through
food webs.
3.3.2.3. Ecosystem Potentfedly at Risk
The Columbia River supports a diversity of aquatic and terrestrial wildlife. The major
ecological components are benthic macroinvertebrates, zooplankton, phytoplankton, and fish.
Although a detailed description of the wildlife exceeds the scope of this effort, appendix A lists the
fish species and shows the generalized aquatic food web. This risk assessment focuses on the fish
of the Columbia River because they are aquatic organisms sensitive to ionizing radiation and
because the Columbia River supports a wide variety of fish, including several species that are
commercial, recreational, and cultural assets of the region.
3.3.2.4. Endpoint Selection
Exposure of aquatic organisms to radioactivity can elicit a toxic response depending on the
dose level, the length of exposure, the particular species, and the life stage at the time of exposure.
The magnitude of the response is proportional to radiological dose. In this study, the assessment
endpoint was the health and condition of local populations of selected fish species that were of
commercial, recreational, and cultural interest.
The risk assessment evaluated multiple measurement endpoints. They included literature
investigations of adverse effects on fish, such as acute mortality and sublethal and developmental
effects. Dose from ionizing radiation was evaluated in the maximally exposed individual fish and
fish in early developmental stages during the study period. Because no net increase occurred in the
concentration of elements, the assessment considered only toxicity resulting from ionizing
radiation, not toxicity resulting from chemical characteristics.
3.3.2.5. Conceptual Model
Radionuclides in the Columbia. River are partitioned between river water, sediment, and the
aquatic food web. Organisms become exposed through direct contact with river water, through
contact or ingestion of contaminated sediments, or through food web incorporation of
radionuclides.
Two organism exposure pathways exist for ionizing radiation. In the external exposure
pathway, an organism receives a dose from its external environment, such as ionizing radiation
from the water. If the energy of the radiation is high enough, it may penetrate the organism's
external tissue. In the internal exposure pathway, an organism receives a dose of ionizing radiation
as a result of uptake of a radionuclide. Consequently, exposure occurs to internal organs and
tissues. The significance of each exposure pathway depends on the aquatic fate of the
radionuclide, its concentration, the energy of its radiation, and also on the pathway of
bioaccumulation.
3-15
-------
The level of organism dose from either external or internal exposure depends on the length
of time an organism spends in the Hanford Reach feeding and breeding habitats, the degree of
interaction with the sediments (i.e., living on or in the sediments), the discharged levels of
radionuclides, and the river flows. Potential dose to aquatic organisms equals the sum of the total
ionizing radiation dose from multiple radionuclides.
Possible exposure scenarios include organisms living near or in reactor effluent discharges,
at various locations downriver of Hanford, and on or in contaminated sediments. A resident fish,
such as whitefish, can spend its entire life in the Hanford Reach. The adult chinook salmon, on
the other hand, is present only during selected periods of the year.
Generally, higher-level organisms such as fish have greater sensitivity to ionizing radiation
than lower-level organisms such as algae and invertebrates (Frank, 1973). Consequently, fish can
serve as indicators or benchmarks of the health of fish populations and the ecosystem. For fish,
sensitivity varies with developmental stage, (i.e., adult fish being less sensitive then juveniles),
amount of time required for various developmental stages, and number of fertilized eggs produced
(Whicker and Shultz, 1982). Species fecundity factors into extrapolating individual organism
effects to a population. For example, species with high fecundity rates most likely will not
experience adverse effects to the same degree as species with low fecundity rates. In addition, the
exposure of organisms to low-level ionizing radiation can promote injury repair mechanisms.
For Hanford, most of the available monitoring data for radionuclides were for river water
activity and tissue concentrations of selected species of fish, including mountain whitefish
(Prosopium williamsoni). One of the most fished species in the Columbia River, mountain
whitefish remains resident throughout the year, making it a useful biomonitor of radionuclide
incorporation into the human food chain. The food chain accumulation of radionuclides by
whitefish occurs in a three step process:
Water -* Algae -*• Insects -* Whitefish
Calculated dose to whitefish can be extrapolated to other fish species, such as adult chinook
salmon that occur seasonally in the Hanford Reach of the Columbia River. In the risk assessment,
whitefish served as an indicator or "generic" fish to develop a potential exposure/dose scenario.
Where available, the risk assessment incorporated data for other fish species along with supportive
or ecosystem descriptive data for phytoplankton, snails, and crayfish. Dose was estimated from
exposure to measured radionuclides in the river to salmon embryos, identified as one of the most
sensitive organisms to ionizing radiation.
3-16
-------
Comments on Problem Formulation
Strengths of the case study include:
This case study was well written and well organized. It is an ideal case for the
application of the EPA Risk Assessment Framework because discrete stressors
are easily identified and measured and substantial data are available on their
biological impacts. Assessment and measurement endpoints are identified and
fit nicely into the risk assessment paradigm.
Limitations include:
Because the Columbia River ecosystem has been affected by many other factors,
radiation may have a relatively small impact on salmon. Therefore, although it
may be valid to restrict the risk assessment to a single stressor that does not
reflect the "real world" situation, other stressors on salmon should be identified.
The authors should point out that data were developed for the specific case
study, rather than for a full-ranging risk assessment that could consider other
stressors. DOE and EPA need to know whether radionuclides are a major
problem or risk to the ecosystem is negligible.
The total biological community is not well characterized.
3.3.3. Analysis: Characterization of Exposure
Making use of the 1963-1964 data for sediments, water, and biota, the exposure
characterization employed two approaches to evaluate dose, which provided independent
assessments of dose. The first approach evaluated river radioactivity at several stations
downstream of the Hanford site. This approach then modeled organism dose using biological
accumulation factors for several "generic" aquatic organisms from measured radionuclide water
concentrations during the study period, 1963-1964. The second approach used measured
radionuclide tissue concentrations to calculate dose to whkefish. Directly measured tissue activity
has the advantage of considering all environmental pathways: water and food uptake, excretion,
sediments, etc. However, this approach has the disadvantage of measuring selected radionuclides
only in fish muscle tissue. As a result, the approach reflects the human pathway and places less
emphasis on effects to the fish. For example, although organs and bones also accumulate
radionuclides, they were not included in the dose calculation.
3.3.3.1. Sample Location
The initial exposure characterization was limited to the Richland Station (RM 344),
although ultimately all available data from the Hanford Reach were reviewed and considered. The
U.S. Geological Survey (USGS, 1966) indicated that the river is vertically and horizontally mixed
at this point. This approach was used because of the potential for large spatial and temporal
3-17
-------
variability of radionuclide concentrations upstream. This variability resulted from the discharge of
eight production reactors with individual production schedules. Once established, the relationship
between exposure and potential effects can be applied to upstream locations.
3.3.3.2. Data Analysis
The risk assessment reviewed three data sets to characterize exposure: measured
radionuclide river concentrations, measured sediment concentrations, and measured fish tissue
concentrations. The data were collected during routine monitoring of radionuclide concentrations
in the Columbia River system. River water was collected as composite, grab, or cumulative
samples. The sampling scheme varied over the 2-year period (table 3-2). Figures 3-3 and 3-4
show the monthly water grab sample concentrations for selected radionuclides over the 2-year
period. Water concentrations were generally highest during the winter and late fall and lowest in
the spring and summer.
3.3.3.3. Exposure From Measured River Water Concentrations
Exposure concentrations were established by reviewing measured river activity data to
determine the relationships among composite, grab, and continuous samples: that is, to see
whether one form of sampling yielded consistently higher water concentrations than another. The
results of this analysis showed that the highest river concentrations of radionuclides occurred in
whole-water grab samples.
An upper-boundary exposure concentration was derived by using the maximum observed
grab sample water concentration for the 2-year study period for each radionuclide shown in table
3-3. These concentrations were assumed to represent the maximum concentration for exposure of
river organisms. If the effect characterization indicated a potential risk, then more typical exposure
concentration scenarios could be developed.
The maximum sediment concentration measured for each radionuclide was used to calculate
organism dose.
3.3.3.4. Calculation of Organism Dose
The internal total-body dose rate to an organism from water exposure for a number (N) of
radionuclides is given as:
N
(3-D
where R<. is the dose rate to total body of organism c (rad d"1), bi(C is the specific body burden of
nuclide i in organism c (Bq kg'1), and Ei>0 is the effective absorbed energy rate for nuclide i per
unit activity in organism c (rad Ci"1 d"1):
3-18
-------
1
•
•a
*
a
«
fc
•a
*
•a
1
1
8
1*
§
i— 1
i— 1
£
.5
'a
s.
1
s
s
s
1
a
a
£
i
e
*
g
a
WJ
60
9
•"*
1
B
^
1
a
i
.Q
V
§
^
|
ll
•a S
<«,3
X X 0 *
x x o *
XX X 0 *
x x o *
XX XO*# OO
XX XO**OO
X X 0 * * 00
XX X 0 * 0 0
X X 0 * 0 0
x x o * o o
JS
XX X 0 * 0 0
x x o * o o
11 Jllll'l
"S "H B *o
, ^3 t9 s o
SI ^ c S| 3 tS
-1 22 a s| ;g 5
'
.
. .
S S 7 § |
O O 6 CJ <^ ^.g
• • •.; g a§ c
ON ON o> j? S
i_ ^ X 0 *
co co gj
13 "« «>
•4— » •*-* Ofl
3-19
-------
a
SB
(§)C3
i
1
.3
3
.3
1
§
1
5
3-20
-------
Th
1
CO
>
^ 4-
o t£
fr- 5
& _ cvi
<^i '^ PO
coOtog
^ o^^
o
s§
-Si"
cv
u
Q
3
O
•D
^
O
CO
s
I
3-21
-------
Table 3-3. Maximum Grab Sample Water Exposure Concentrations for 1963-1964 Time
Period (Dirkes, 1992; Nelson et al., 1964)
Radionuclide
As-76
Co-60
Cr-51
Cu-64
1-131
Na-24
Np-239
P-32
RE+Y
Sr-90
Zn-65
Concentration (pCi/L)
2,300
120
25,000
10,000
34
5,600
5,600
630
1,400
2.6
1,800
E:,. = eicMeV dis'1 X 3.70E10 dis s'1 Ci'1
1,C I,C
86,400 sd'1 x 1.602E-11 rad'1 MeV = 5.12E4 ei;C
(where e is the effective absorbed energy for nuclide i in organism c).
For a primary organism:
bi,c = Ci,cBi>C
(3-2)
where Q 0 is the concentration of nuclide i in the water to which organism c is exposed (Bq m'3)
and Bi>c is the bioaccumulation factor for nuclide i and organism c (m3 kg'1). Here the water
concentration already has been corrected for dilution and radioactive decay during transit from the
point of release into the receiving water body to the region of the organism's habitat.
Combining equations 3-1 and 3-2 yields the dose rate in rad/d to the primary organism, as
shown in equation 3-3 below. The calculation of internal dose from tissue concentration is the
same as equations 3-1 and 3-2, except the radionuclide-specific BCF is not used and correction for
decay and dilution is unnecessary.
3-22
-------
(3-3)
For a secondary organism, such as an herbivore or carnivore, an expression can be written
for a single radionuclide equating the change in body burden to the uptake and removal of the
radionuclide.
3.3.3.5. Dose From Water Exposure
Table 3-4 shows the CRITR2 code calculations of organism dose from water exposure to
various radionuclides. Appendix B provides a more detailed listing of CRITR2 code calculations
and bioaccumulation factors used. Water concentrations were maximum values for the 2-year
period. Table 3-4 indicates internal dose, immersion or surface dose (external water dose), and
sediment dose. Internal exposure gave the maximum dose. Since immersion and sediment doses
made only minor contributions, they were not considered in the risk characterization.
Table 3-4 summarizes dose for each organism. CRTTR2 default organisms are generic
plants, fish, crayfish, and ducks that eat plants and fish (DUCK-P and DUCK-F, respectively).
Plant-eating ducks had the maximum dose rate, followed by plants, crayfish, fish, and fish-eating
ducks. The dose rates to the plant-eating duck and crayfish exceeded the 1 rad/d level. The
maximally exposed fish had a dose rate of 0.42 rad/d.
The dose to salmon eggs was estimated from measured river water radionuclide activities
(table 3-3). Bioconcentration factors were estimated for salmon embryos from bioconcentration
data reported for developing plaice (Pleuronectes platessd) embryos with respect to various fission
product radionuclides (Woodhead, 1970). Concentration factors for day 4 of embryonic
development ranged from < 1 to 10 as a function of the radionuclide. This assessment used a
whole egg concentration factor of 10 for all radionuclides shown in table 3-3. Dose calculations
employed an overall egg diameter of 2 mm. Dose to whole eggs was 0.00442 rad/d.
3.3.3.6. Dose From Measured Tissue Concentrations
Table 3-5 lists calculated dose from measured tissue concentrations to selected organisms in
the Columbia River. Phytoplankton had the highest dose at 14 rad/d, followed by limpet hard
parts (shell) at 0.39 rad/d and caddisfly at 0.38 rad/d. The maximally exposed fish dose was
calculated to be 0.73 rad/d. For fish, table 3-5 concentrations used to calculate dose represent the
maximum values observed for whitefish during 1963-1964. Dose was evaluated for other species,
but whitefish had the highest body dose for the study period. Unfortunately, most of the fish data
were muscle tissue concentrations and therefore underestimated whole-body burdens.
Consequently, the assessment adjusted these values to whole-body values. Based on limited
Hanford data and published literature, the correction factors between whole body and muscle were
9:1 for phosphorus-32 and chromium-51 and 4:1 for zinc-65 (Poston and Strenge, 1989; U.S.
3-23
-------
Table 3-4. CRITR2 Code Calculation of Organism Dose From Water Exposure to Various
Radionuclides (Baker and Soldat, 1992)
OUT File Name: RMAX.OUT Created: 09:52 18-MAY-92
USR File Name and Header: RHAX.USR RMAX.USR Columbia River Max Concentrations
Version of Program used: V 1.0 of 26-Mar-92
18 May 92
*•*«•* CR1TR2 -• Aquatic Biota Screening Dose Rates
TITLE: Columbia River Max Concentrations -- Ecological Risk Assessment
Release
Plant
Fish
Organism Dose Rates
Crayfish Duck-P Duck-F
ci/y
AS-76
CO-60
CR-51
CU-64
1-131
NA-24
NP-239
P-32
SR-90
ZH-65
Ci/y
AS-76
CO-60
CR-51
CU-64
1-131
NA-24
NP-239
P-32
SR-90
ZH-65
Ci/y
AS-76
CO-60
CR-51
CU-64
1-131
NA-24
NP-239
P-32
SR-90
ZN-65
3.9E-02
2.7E-03
2.7E-02
1.6E-01
1.3E-04
3. OE-02
1.8E-02
1.1E+01
4.6E-04
1.6E-01
15P+fl1
• tC~ v 1
3.0E-05
9.3E-06
2.4E-05
5.6E-05
3.9E-07
7.9E-04
2.9E-05
O.OE+00
O.OE+00
3.2E-05
0 7P ftL
2.9E-06
1.8E-04
6.5E-05
2.8E-06
3.0E-07
3.4E-05
6.9E-06
O.OE+00
O.OE+00
4.3E-04
7.1E-04
3.9E-02
8.9E-04
•1.4E-04
2.0E-01
2.1E-05
3. OE-02
1.5E-01
3.8E-03
7.6E-06
5. OE-04
/ 2e.ni
** • £C U 1
3.0E-05
9.3E-06
2.4E-05
5.6E-05
3.9E-07
7.9E-04
2.9E-05
O.OE+00
O.OE+00
3.2E-05
97P-fli
• I C VH
2.9E-06
1.8E-04
6.5E-05
2.8E-06
3.0E-07
3.4E-05
6.9E-06
O.OE+00
O.OE+00
4.3E-04
7.1E-04
3.9E-02
2.9E-03
7.1E-03
2.8E-02
3.7E-05
2.2E-02
1.8E-03
2.2E+00
1.5E-05-
3.6E-02
?
Immersion or
1.5E-05
4.6E-06
' 1.2E-05
2.8E-05
1.9E-07
4. OE-04
1.4E-05
O.OE+00
O.OE+90
1.6E-05
4 OP* HA
• TC UH
. . . . C&/-J
5.9E-06
3.5E-04
1.3E-04
5.7E-06
6.0E-07
6.8E-05
1.4E-05
O.OE+00
O.OE+00
8.5E-04
1.4E-03
rna i i rac
3.1E-03
1.1E-03
1. OE-02
6.0E-03
1.4E-04
2.6E-03
6.2E-06
1.8E+01
4.8E-03
1.6E+00
1 OP+D1
1 . TC~W 1
Surface
1.7E-05
5.1E-06
1.3E-05
3.1E-05
2.1E-07
4.4E-04
1.6E-05
O.OE+00
O.OE+00
1.8E-05
_ , ...
'ment (r:
1.2E-06
7.1E-05
2.6E-05
1.1E-06
1.2E-07
1.4E-05
2.8E-06
O.OE+00
O.OE+00
1.7E-04
2.9E-04
VQJ —
6.1E-03
7.3E-04
.OE-04
.5E-02
.6E-05
.1E-03
.OE-04
.2E-02
.6E-04
.OE-02
/ nc n?
iraa/a; --- — ............................
1.7E-05
5.1E-06
1.3E-05
3.1E-05
2.1E-07
4.4E-04
1.6E-05
O.OE+00
O.OE+00
1.8E-05
,.,_-. .
ia/a; .....................................
1.2E-06
7.1E-05
2.6E-05
1.1E-06
1.2E-07
1.4E-05
2.8E-06
O.OE+00
O.OE+00
1.7E-04
2.9E-04
Grand Totals >»» 1.2E+01 4.3E-01 2.4E+00 1.9£+01 5.OE-02
3-24
-------
Table 3-4. CRITR2 Code Calculation of Organism Dose From Water Exposure to Various
Radionuclides (continued)
OUT File Name: RMAX.OUT Created: 10:21 18-HAY-92
USS File Name: RMAX.USR
Version of Program used: V 1.0 of 26-Har-92
Parameters and Water Concentrations
No dilution model used.
Bioaccumulation Factors for: Fresh No bioaccumulation factor corrections used.
Distance (m)
Mixing Ratio
Radius (cm)
Mass (kg)
Intake rate (g/d) -
Diet
Transit Time (h) --
AS-76
CO-60
CR-51
CU-64
1-131
NA-24
HP-239
P-32
SR-90
ZN-65
H. L.
26.32 H
5.271 Y
27.704 D
12.701 H
8.04 0
15.00 H
2.355 D
14.29 0
29.12 Y
243.9 D
Release Outfall
Concentration Plant Fish Crayfish Duck-P Duck-F
1
1
5.0
1
1
5.0
1
1
2.0
1
1
5.0
1.0
100
P
0
1
1
5.0
1.0
200
-f
0
Water Concentrations , ( Decay during transit included )
-i-i/y-
2
1
2
1
3
5
5
6
2
1
.3E-06
.2E-07
.5E-05
.OE-05
.4E-08
.6E-06
.6E-06
.3E-07
.6E-09
.8E-06
2.3E-06
1.2E-07
2.5E-05
1,, OE-05
3,,4E-08
5..6E-06
5..6E-06
6.3E-07
2..6E-09
1..8E-06
2.
1.
2.
1.
3.
5.
5.
6.
2.
1.
3E-06
2E-07
5E-05
OE-05
4E-08
6E-06
6E-06
3E-07
6E-09
8E-06
2
1
2
1
3
5
5
6
2
1
Ci/m3 or
.3E-06 2.3E-06
.2E-07 1.2E-07
.5E-05 2.5E-05
.OE-05 1. OE-05
.4E-08 3.4E-08
.6E-06 5.6E-06
.6E-06 5.6E-06
.3E-07 6.3E-07
.6E-09 2.6E-09
.8E-06 1.8E-06
uCi/mL .-•- —
2.3E-Q6
1.2E-07
2.5E-05
1. OE-05
3.4E-08
5.6E-06
5.6E-06
6.3E-07
2.6E-09
1.8E-06
3-25
-------
I
oo
o
ON
co
vo
S
CO
o
I
A
«
co.
CN o
CO
o
3
^
CN
o
to
CO
W ^
vo co
co • *™^
c- CN —
CO
oo
3
Soo W
ON^ '—i
^< O of co
c^ W
00 n o —• o >n
CO —I O CO CO -^
CN O >T) O CN O
•v • r» • «v •
«S O -^ O CN O
o cs
CN CN
CO O
CO
08^.
CO CN
CO
W
TJ-
^ *—j ^ ON or** co t-*
O\ CO ** « •* . *v .
CN ^—i SO O *~* O CN O
" ^ P 00 9
O VO O ON O
C? CO ^^^ ^^ Tt* ^" ""^J" IO IO ^O
.
T-H CN
o
I
m
oo in oo »o o
CN o\ >n ON i/"i
O *O O >T) O
• *r • *s •
O »-H O -HO
co
§cd
S
Chironimids
Dose, rad/d
«
S §
a Q
• ON . co
O co O ^
CQ
ft £*>
TJ
B
OH dT
.§ s
a Q
-------
Congress, Joint Committee on Atomic Energy, 1959). This correction introduces uncertainty into
the effects characterization, but uncorrected muscle values could underestimate individual dose.
3.3.3.7. Dose From Measured Sediment Concentrations
Table 3-6 shows the calculated dose from exposure to radioactivity reported in sediments of
the Columbia River. The calculated dose was quite small compared with other pathways.
Comments on Characterization of Exposure
Strengths of the case study include:
• The ability to evaluate the worst case (maximally exposed individual) at the most
sensitive life stage is an efficient method of screening for population-level
effects. This study also benefits from the availability of long-term data sets
collected on site.
Limitations include:
• Analysis also should consider potential uptake from food rather than only
exposure or direct uptake from the water. Large variance in BCF values
suggest that activity in water cannot reliably predict exposure.
• In the computer model scenario, algae, crayfish, and fish were not growing or
eating and did not accumulate a food chain dose. Although the computer code
included ducks, they were not included in the ecological risk assessment because
of limited data and limited ability to verify the model estimate.
3.3.4. Analysis: Characterization of Ecological Effects
Characterization of effects was based on dose-response information for fish from available
toxicity data and also on regulatory standards. Conducted at the individual level, the
characterization was interpreted qualitatively and applied to the population level of ecological
organization.
The general response of aquatic organisms to ionizing radiation occurs at both the cellular
and biochemical levels. Environmental factors also can affect the level of response. An NCRP
(1991) report, Effects of Ionizing Radiation on Aquatic Organisms, provided the basis for stressor-
response relationships developed in this report. Figures 3-5 and 3-6 were adapted from the NCRP
report and summarize the information on acute effects of ionizing radiation on aquatic organisms.
One would expect different fish species to accumulate different concentrations of
radionuclides based on their feedings habits, age, length of time spent at the site, and other factors.
Depending on the level of exposure, mortality can occur. The threshold level of radiation dose
3-27
-------
Table 3-6. Maximum Sediment Radionuclide Concentrations in the Hanford Reach and Dose
to an Organism Living in the Sediments (Dirkes, 1992; Haushild et al., 1966;
Nelson et al., 1964)
Nuclide
Cr-51
Co-60
Sc-46
Zn-65
Concentration
13,000
100
46
3,900
(pCi/Kg dry weight)"
*Total dose: Organism buried in sediment—0.16 rad/d.
Organism on surface of sediment—0.08 rad/d.
that can cause acute mortality occurs at approximately 100 rad (1 Gy) for amphibians and 1,000
rad (10 Gy) for crustaceans and fish (figure 3-5). Figure 3-5 summarizes the relationship between
organism dose and response and also shows the range for LD50s. Under no circumstances did
calculated dose to fish or other organisms exceed the boundary dose where acute effects would be
observed. Dose calculations based on tissue concentrations for selected Columbia River organisms
confirmed this finding. No aquatic animal organism used in the risk assessment exceeded the DOE
dose limit of 1 rad/d.
Few studies have evaluated the effects of chronic exposure to ionizing radiation. However,
it is known that the early developmental stages of chinook salmon are especially sensitive to
ionizing radiation. NCRP (1991) reported that exposure to 5.1 rad/d (51 mGy/d) for up to 69 days
produced no increase in mortality to chinook salmon embryos and alevins up to release as smolts.
Hershberger et al. (1978) reported lower return of spawning adult chinook salmon after exposure
of eggs and alevins at approximately 10 rad/d of gamma radiation. Gonadal development was
retarded in chinook salmon on exposure to 10 rad/d delivered to embryos (Bonham and Donaldson,
1972). Other laboratory research (Erickson, 1973) found that an exposure of 0.4 rad/d (4.0
mGy/d) reduced courting activity for male Poecilia reticulata exposed as embryos. Chronic
gamma radiation (190 days at an exposure of 18.5 rad/d) causes sterility in young adult Ameca
splendens (Rackham and Woodhead, 1984).
Based on available literature, the dose used in DOE Order 5400.5 appears sufficiently
conservative to protect most aquatic organisms. Consequently, unless future data indicate
otherwise, this dose can be considered protective of populations and the ecosystem in general. To
date, the sole qualifier is the work of Erickson (1973), who reported reduced male guppy courting
activity when exposed to 0.4 rad/d. Little other information exists with regard to behavioral
changes in fish exposed to ionizing radiation.
Figure 3-6 summarizes the effects of acute irradiation on development of fish. The
threshold for developmental effects on fish occurs at approximately 5 rad (0.05 Gy), as observed
for the one-cell-stage developing chinook salmon embryos. Radiosensitivity reportedly decreases
3-28
-------
10.000.0
1000.0
100.0
10.0
1.0
0.1
I
I
I
I
Freshwater Amphibians Crustaceans Molluscs Algae
fish
Organism
RadWien Mlwr hard x-ny or gamrnt.
Figure 3-5. Ranges of sensitivities of aquatic organisms to acute radiation exposure (adapted
from NCRP, 1991)
3-29
-------
100.0
10.0
5
-------
with increasing level of embryo development (Frank, 1973). Laboratory studies with the Chinook
salmon identify early life stages as the most sensitive for fish. Damage occurred when the dose
reached 9.64 rad/d (4 mGy/h) over an 81-day development period (Hyodo-Taguchi, 1980). Studies
have shown that 224 rad (2.24 Gy) reduced female germ cells in chinook salmon; a dose of 600
rad (6 Gy) produced the same effect in rainbow trout.
Comments on Characterization of Ecological Effects
Strengths of the case study include:
• Direct experimental observations (dose^response curves) were provided to
characterize effects. Figures 3-5 and 3-6 include ranges of acute toxicity data
for various taxonomc groups and different life stages of salmon.
General reviewer comments:
• It was suggested that more sensitive measures than mortality should be used to
assess effects. Dose-response curves could be provided to indicate the
conservative nature of the DOE regulatory limit.
• No data are presented to show that protecting salmon embryos protects the
ecosystem.
3.3.5. Risk Characterization
Ecological risk was characterized by assessing dose to fish and, as indicators of ecosystem
integrity, other aquatic organisms; by comparing doses to DOE Order 5400.5; and by comparing
doses to published toxicity data.
3.3.5.1. Acute Exposure to Ionizing Radiation
The level of potential risk from ionizing radiation was assessed for fish under both acute
and chronic exposure scenarios. The acute exposure considered mortality, while chronic exposure
considered developmental effects as measurement endpoints.
To determine the potential risk to fish, both water and organism concentrations of
radionuclides were converted to dose (tables 3-4 and 3-5, respectively). A comparison of these
values (0.43 and 0.73 rad/d) to the range of acute toxicity (LD50) reported for fish shows that no
acute mortality would be expected from these levels. To assess exposure effects on a developing
embryo, the whole egg dose was calculated to be 0.00442 rad/d.
3-31
-------
The characterization of the level of potential risk to fish during early developmental stages
and as adults was expressed as a hazard quotient (HQ), defined as the ratio of radionuclide
organism dose (exposure or tissue value) to a dose-response benchmark value:
HQ =
Exposure Dose
Dose Benchmark Value
(3-4)
If the HQ is equal to or greater than 1, the likelihood of an adverse effect or high risk
exists. The characterization was completed for the maximally exposed individual for the study
period. It was assumed that if risk to the individual was low, the population was not at risk.
The hazard quotients shown in table 3-7 for early developmental stages of fish and adults
were compared with toxicity values and DOE Order 5400.5. The maximum hazard quotient was
0.73 for adult fish. Assuming that this was the maximally exposed individual, the likelihood of an
adverse effect to an individual was low.
Table 3-7. Hazard Quotient for Early Development Stage of Fish and Adult Fish
Minimum Effect
Maximum Exposure Level
Hazard Quotient
Unfertilized ovum,
One-cell stage
Adult
0.00442
0.73
0.96," 0.4b
1
0.004,a0.11b
0.73a'c
"Based on recommendation of the NCRP (1991).
bBased on male courting activity in guppies (Erickson, 1973).
CDOE Order 5400.5.
3.3.5.2. Chronic Exposure to Ionizing Radiation
Mortality from chronic exposure presented minimal risk to fish. Chronic exposure to 5.1
rad/d for up to 69 days did not produce any mortality to chinook salmon embryos or alevins
(NCRP, 1991). Hershberger et al. (1978) reported lower return of spawning chinook salmon after
exposure of eggs and alevins to 10 rad/d and effects on gonadal development in chinook salmon
was reported to occur at 9.5 rad/d. Because the maximum dose rate to Columbia River adult fish
and developing embryos was 0.73 and 0.00442 rad/d respectively, no chronic effects or mortality
would be expected. Applying the behavior response noted for guppy embryo exposure (Erickson,
1973), the benchmark concentration would be 0.4 rad/d with an HQ of 0.1.
3.3.5.3. Uncertainty
Extrapolation of individual effects of radionuclides to populations and communities suffers
from the same constraints as similar extrapolations for hazardous chemicals. The quantitative
3-32
-------
relationship between potential effects to fish or fish embryos and population and community
response is not known. However, the effects data available for radionuclides showed that the
single-cell stage in salmon is one of the more sensitive indicators of irradiation effects in fish and
that protection of this stage of development should be protective of the population. Although
specific data were not available for salmon embryo, data for embryo development of plaice was
used to estimate dose.
The NCRP (1991) suggests that a "maximum dose rate 0.4 mGy/h (0.96 rad/d) would
provide protection for endemic populations of aquatic organisms in environments receiving
discharges of radioactive effluent." It further states, "adoption of a reference level of 0.4 mGy/h
appears to represent a reasonable compromise based on current literature, i.e., considering both the
nature of the effects observed at this dose rate and the limited amount of information on effects of
radiation in natural populations, including interactions between ionizing radiation and ecological
conditions." This value is also in agreement with DOE Order 5400.5.
Because whitefish are resident species in the Columbia River and can accumulate
radionuclides throughout their life cycle, the assessment assumed that the whitefish tissue dose
would be sufficiently conservative to extrapolate dose levels to other adult fish, including salmon.
Salmon, on the other hand, spend only a short period of time in the river and do not feed when
present. In addition, during the spring and early fall when salmon are present, river concentrations
of radionuclides were generally the lowest.
The risk characterization used the maximally exposed individual to calculate organism dose.
The risk characterization assumed that if an organism dose is below any known effect level with
some degree of certainty, then the likelihood of an adverse effect is minimal. (The assessment
endpoint was maintenance of important recreational fish populations in the Columbia River
measured by protection of fish populations and specifically salmon embryos.) Results indicate that
this is a reasonable assumption. Fish appear to be a suitable choice of receptor for screening risk
from ionizing radiation. In addition, a fish dose of less than 1 rad/d should be protective of the
ecosystem in general. However, since CRITR2 indicate that ducks could have received a dose
higher than 1 rad/d, further studies are warranted.
Another area of uncertainty in the risk assessment is the extrapolation of muscle tissue
concentration to whole fish concentrations for radionuclides. The assumption that protection of the
maximally exposed individual extrapolated to sensitive life stages constitutes an adequate measure
of the assessment endpoint also is a source of uncertainty. Alternatively, the hazard quotient is a
reasonable approach for radionuclides for baseline or screening assessments.
3.3.5.4. Conclusions
This study demonstrates that the ecological risk assessment paradigm is applicable to
radioactive substances. However, stressor-response data were limited to acute exposures; few data
addressed chronic sublethal exposures. Most endpoints used for hazardous chemicals are expected
to be equally appropriate for radionuclides. This study uncovered only one benchmark that
specifically addressed protecting aquatic organisms from exposure to radiation. DOE Order 5400.5
limits exposure to aquatic animals to 1 rad/d.
3-33
-------
Risk characterization did not indicate any measurable risk to the most sensitive aquatic-
organism (early life stage of chinook salmon) from exposure to radionuclides in sediments or water
in the Columbia River. 'During peak production at Hanford, releases of radionuclides to the river
did not result in a dose to fish that would exceed those specified in DOE Order 5400.5.
Dose calculations for radionuclide exposure from water and tissue concentrations provide
for two methods for assessing the potential risks. This study investigated both methods and found
that both provided reasonable results for fish, algae, and crayfish. Areas of uncertainty included
the relationship between muscle and whole fish concentrations, the lack of a strong data base for
organism exposure to chronic radiation, and a quantitative measure of ecosystem-level response to
radionuclides. During the study period, the major thrust of monitoring at Hanford was to protect
human health. Few studies examined ecosystem structure and function. Another significant area
of uncertainty was the use of adult whitefish tissue concentration as a surrogate for chinook
salmon. The study located no data suggesting that salmon accumulate a higher dose than whitefish,
which spend their whole lives in the Columbia River. Although using fish data tends to increase
uncertainty, fish are particularly sensitive to ionizing radiation and should provide a reasonable
level of protection for fish populations and communities (figures 3-5 and 3-6) and a screen or
benchmark indicator of ecosystem-level effects.
Comments on Risk Characterization
Strengths of the case study include:
The case study provides an opportunity to distinguish between screening
assessments and more rigorous (realistic) assessments. The CRITR2 computer
model is intended to provide a first pass that can be refined if there appear to
be significant concerns.
limitations include:
The hazard quotient should be described in more detail by addressing the
potential range of values, the establishment of confidence intervals, the degree
of confidence that the value of 1.00 is safe, etc. This study uses the most
sensitive individual to be conservative, but the selection of the most sensitive or
highest exposed individual biases the assessment. The establishment of
confidence bounds would result in a less biased measure of uncertainty.
Many assumptions are chained together in this case study to obtain highly
conservative assessments. A table should be developed that specifies these
assumptions and the types of uncertainties jhey introduce.
The focus on salmon limits an extrapolation to overall ecosystem effects.
3-34
-------
Comments on Risk Characterization (continued)
General reviewer comments:
• This section should emphasize that risk to the salmon populations is based on
an anafysis of risk to the most sensitive individuals and that risk from chemical
exposure or othe'r stressors was not evaluated. Nevertheless, risk from
radionuclides is addressed adequately.
• It would be helpful to have additional emphasis placed on estimating and using
variability and confidence intervals. This could be the primary content for the
section on uncertainty analysis.
3-35
-------
3.4. REFERENCES
- *
Baker, D.A.; Soldat, J.K. (1992) Methods far estimating doses to organisms from radioactive
materials released into the aquatic environment. PNL-8150. Richland, WA: Pacific
Northwest Laboratory.
Beasley, T.M.; Jennings, C.D. (1984) Inventories of 239, 240Pu, 241 Am, 137Cs, and 60Co in
Columbia River sediments from Hanford to the Columbia River Estuary. Environ. Sci.
Technol 18:201-212.
Becker, C.D. 0990) Aquatic bioenvironmental studies: the Hanford experience 1944-84. New
York, NY: Elsevier Science Publishers.
Bonham, K.; Donaldson, L.R. (1972) Sex ratios and retardation of gonadal development in
chronically gamma-irradiated chinook salmon smolts. Trans. Am. Fish. Soc. 101(3):428-
434.
Gushing, C.E. (1967a) Periphyton productivity and radionuclide accumulation in the Columbia
River, U.S.A. Hydrobiologia 24:121-139.
Gushing, C.E. (1967b) Concentration and transport of P-32 and Zn-65 by Columbia River
plankton. Umnol Oceanogr. 12:330-332.
Gushing, C.E.; Watson, D.G. (1966) Accumulation and transport of radipnuclides by Columbia
River biota. In: Guillon, A., ed. Disposal of radioactive wastes into seas, oceans and
surface waters. Vienna, Austria: International Atomic Energy Agency, pp. 551-570.
Dirkes, R.L. (1992) Columbia River monitoring data compilation. WHC-SD-EN-DP-024, Rev. O.
Prepared by Pacific Northwest Laboratory for Westinghouse Hanford Company, Richland,
WA.
Erickson, R.C. (1973) Effects of chronic irradiation by tritiated water on Poecilia reticulata, the
guppy. In: Radionuclides in ecosystems, vol. 2. Proceedings of the Third National
Symposium onRadioecology. May 10-12, 1971, Oak Ridge, TN. Nelson, D.J., ed.
Washington, DC: U.S. Atomic Energy Commission, pp. 1091-1099.
Frank, M.L. (1973) Sensitivity of carp (Cyprinus carpio) embryos to acute gamma radiation. In:
Radionuclides in ecosystems, vol. 2. Proceedings -o}the Third National Symposium on
Radioecology. May 10-12, 1971, Oak Ridge, TN. Nelson, D.J., ed. Washington, DC: U.S.
Atomic Energy Commission, pp. 1106-1112.
Haushild, W.L.; Perkins, R.W.; Stevens, H.H.; Dempster, G.R.; Glenn, J,L. (1966) RadionucUde
transport in the Pasco to Vancouver, Washington reach of the Columbia River, Jufy 1962 to
September 1963. U.S. Department of the Interior, Geological Survey. Portland, OR.
3-36
-------
Hershberger, W.K.; Bonkham, K.; Donaldson, L.R. (1978) Chronic exposure of Chinook salmon
eggs and aleyins to gamma irradiation: effects on their return to freshwater as adults.
Trans. Am. Fish. Soc. 107(4) :622-631,
Hyodo-Taguchi, Y. (1980) Effects of chronic g-irradiation on spermatogenesis in the fish (Oryzias
latipes), with special reference to regeneration of testicular stem cells. In: Egami, N., ed.
Radiation effects on aquatic organisms. Baltimore, MD: University Park Press, pp. 91-104.
National Council on Radiation Protection and Measurements. (1991) Effects of ionizing radiation
on aquatic organisms. NCRP Report No. 109. Washington, DC.
Nelson, J.L.; Perkins, R.W.; Nielsen, J.M. (1964) Progress in studies of radionuclides in
Columbia River sediments. HW-83614. Hanford Atomic Products Operation, General
Electric Company, Richland, WA.
Poston, T.M.; Strenge, D.L. (1989) Estimation of sport fish harvest for risk and hazard assessment
of environmental contaminants. Presented at Sixth National RCRA/Superfund Conference
and Exhibition, Hazardous Wastes and Hazardous Materials '89. New Orleans, LA.
Rackham, B.D.; Woodhead, D.S. (1984) Effects of chronic 7-irradiation on the gonads of adult
Ameca splendens (Oste»^hthyes: Teleostei). Int. J. Radlat. Biol. 45(6):645-656.
U.S. Congress, Joint Committee on Atomic Energy. (1959) Hearings, vol. 2. 86th Cong., first
session. Washington, DC: U.S. Government Printing Office.
U.S. Department of Energy. (1989) Order 5400.5: Radiation protection of the public and the
environment.
U.S. Department of Energy. (1988) Hazardous waste management plan. Defense Waste
Management, DOE.R1-88-01, U.S. Department of Energy, Richland Operations Office,
Richland, WA.
U.S. Environmental Protection Agency. (1992) Framework for ecological risk assessment*
Washington, DC. EPA/630/R-92/001.
U.S. Geological Survey. (1966) Radionuclide transport of the Columbia River, Pasco to
Vancouver, Washington Reach, July 1962 to September 1963. U.S. Geological Survey,
Progress Report. Portland, OR.
Whicker, F.W; Schultz, V. (1982) Radioecology: nuclear energy and the environment, vol. 1.
Boca Raton, FL: CRC Press.
Woodhead, D.S. (1970) 'The assessment of the radiation dose to developing fish embryos due to the
accumulation of radioactivity by the egg. Radiat. Res. 43:582-597.
3-37
-------
Woodruff, R.K.; Hanf, R.W.; Hefty, M.G.; Lundgren, R.E., eds. (1991) Hanford site
environmental report for calendar year 1990. PNL-7930. Richland, WA: Pacific Northwest
Laboratory.
3-38
-------
3.5. ADDITIONAL READING
Anderson, L.L.; Harrison, F.L. (1986) Effects of radiation on aquatic organisms and
radiobiological methodologies for effects assessment. U.S. EPA, Washington DC EPA
520/1-85-016. ' '
Auerbach, S.I. (1971) Ecological considerations in siting nuclear power plants: the long-term biotic
effects problem. Nuclear Safety 12(l):25-34.
Blaylock, E.G. (1969) The fecundity of a Gambusia afflnis population exposed to chronic
environmental radiation. Radial. Res. 37:108-117.
Blaylock, B.C.; Trabalka, J.R. (1978) Evaluating the effects of ionizing radiation on aquatic
organisms. In: Advances in radiation biology, vol. 7. Lett, J.T.; Aider, H., eds. New
York, NY: Academic Press, pp. 103-152.
Dauble, D.D.; Watson, D.G. (1990) Spawning and abundance of fall chinook salmon
(Oncorhynchus tashawytscha) in the Hanford Reach of the Columbia River, 1948-1988
PNL-7289. Richland, WA: Pacific Northwest Laboratory.
International Atomic Energy Agency. (1976) Effects of ionizing radiation on aquatic organisms and
ecosystems. Technical Reports Series No. 172. Vienna, Austria: International Atomic
Energy Agency.
National Research Council of Canada. (1983) Radioactivity in the Canadian aquatic environment
Pub. No. NRCC-19250. Ottawa, Canada.
Nelson, J.L.; Perkins, R.W.; Nielsen, J.M. (1964) Progress in studies of radionuclides in
Columbia River sediments. HW-83614. Richland, WA: General Electric.
Newcombe, H.B. (1972) "Benefit" and "harm" from exposure ofrvertebrate sperm to low doses of
ionizing radiation. Health Phys. 25(7): 105-107.
Ophel, L.L.; Hoppenheit, M.; Ichikawa, R.; Klimov, A.G.; Kobayashi, S.; Nishiwaki, Y.; Saiki,
M. (1976) Effects of ionizing radiations on aquatic organisms. In: Effects of ionizing
radiation on aquatic organisms and ecosystems. Technical Report Series No. 172. Vienna,
Austria: International Atomic Energy Agency, pp. 57-86.
Pearce, D.W.; Green, J.K., eds. (1965) Hanford radiological sciences research and development
annual report for 1964. BNWL-36. Richland, WA: Pacific Northwest Laboratory.
Rice, T.R.; Baptist, J.P. (1974) Ecologic effects of radioactive emissions from nuclear power
plants. In: Sagan, L.A., ed. Human and ecologic effects of nuclear power plants.
Springfield, IL: Charles C. Thomas, Publisher, pp. 373-439.
3-39
-------
Templeton, W.L.; Nakatani, R.E.; Held, E.E. (1971) Radiation effects. Radioactivity in the
marine environment. Washington, DC: National Academy of Sciences, pp. 223-239.
Templeton, W.L.; Bemhard, M; Blaylock, B.C.; Fisher, C.; Holden, M.J.; Klimov, A.G.;
Metalli, P.; Mukerjee, R.; Ravera, O.; Sztanyik, L.; Van Hoeck, F. (1976) Effects of
ionizing radiation on aquatic populations and ecosystems. Effects of ionizing radiation on
aquatic organisms and ecosystems. Technical Report Series No. 172. Vienna, Austria:
International Atomic Energy Agency, pp. 89-102.
Walden, S.J. (1973) Effects of tritiated water on the embryonic development of the three-spine
stickleback Gasterosteus aculeatus linnaeus. In: RadionucUdes in ecosystems, vol. 1.
Proceedings of the Third National Symposium onRadioecology, May 10-12, 1971, Oak
Ridge, TN. Nelson, D.J., ed. Washington, DC: U.S. Atomic Energy Commission, pp.
1087-1090.
Watson, D.G.; Gushing, C.E.; Coutant, C.C.; Templeton, W.L. (1970) Radioecological studies on
the Columbia River, Part II. BNWL-1377-PT2. Richland, WA: Pacific Northwest
Laboratory.
Wilson, R.H., ed. (1963) Evaluation of radiological conditions in the vicinity ofHanford April -
June, 1963. HW-78395. Richland, WA: General Electric.
Wilson, R.H., ed. (1963) Evaluation of radiological conditions in the vicinity ofHanford July -
September, 1963. HW-79652. Richland, WA: General Electric.
Wilson, R.H., ed. (1964) Evaluation of radiological conditions in the vicinity ofHanfordfor 1963.
HW-80991. Richland, WA: General Electric.
Wilson, R.H., ed. (1965) Evaluation of radiological conditions in the vicinity ofHanfordfor 1964.
BNWL-90. Richland, WA: Pacific Northwest Laboratory.
3-40
-------
APPENDIX A
COLUMBIA RIVER FISH SPECIES AND FOOD WEB
3-A1
-------
Table 3-A1. Fish Species in the Hanford Reach of the Columbia River
Common Name
Scientific Name
White sturgeon
Bridgelip sucker
Largescale sucker
Mountain sucker
Pumpkinseed
Bluegill
Smallmouth bass
Largemouth bass
White crappie
Black crappie
American shad
Prickly sculpin
Mottled sculpin
Piute sculpin
Reticulate sculpin
Torrent sculpin
Chiselmouth
Carp
Peamouth
Northern squawfish
Longnose dace
Leopard dace
Speckled dace
Redside shiner
Tench
Burbot
Acipenser transmontanus
Catostomus columbianus
Catostomus macrocheilus
Catostomus platyrhynchus
Lepomis gibbosus
Lepomis macrochirus
Micropterus dolomieui
Micropterus salmoides
Pomoxis annularls
Pomoxis nigromaculatus
Alosa sapidissima
Cottus asper
Cottus bairdi
Cottus beldingi
Cottus perplexus
Cottus rotheus
Acrocheilus alutaceus
Cyprinus carpio
Mylocheilus caurinus
Ptychocheilus oregonensis
Rhinichthys cataractae
Rhinichthys falcatus
Rhinichthys osculus
Richardsonius balteatus
Tinea tinea
Lota lota
3-A2
-------
Table 3-A1. Fish Species in the Hanford Reach of the Columbia River (continued)
Common Name
Scientific Name
Threespine stickleback
Black bullhead
Yellow bullhead
Brown bullhead
Channel catfish
Yellow perch
Walleye
Sand roller
Pacific lamprey
River lamprey
Lake whitefish
Coho salmon
Sockeye salmon
Chinook salmon
Mountain whitefish
Cutthroat trout
Rainbow trout (steelhead)
Dolly Varden trout
Gasterosteus aculeatus
Ictalurus melas
Ictalurus natatis
Ictalurus nebulosus
Ictalurus punctatus
Percaflavescens
Stizosieclion vitreum vitreum
Percopsis transmontana
Entosphenus tridentatus
Lampetra ayresi
Coregonus clupeafdrmis
Oncorhynchus Tdsutch
Oncorhynchus nerka
Oncorhynchus tshawytscha
Prosopium williamsoni
Oncorhynchus clarld
Oncorhynchus mykiss
Salvelinus malma
3-A3
-------
Death and Feces
(Bacterial Breakdown)
xv ^
/ Periphyton
( t
Sediments
(Inorganic and Organic)
Figure 3-A1. Columbia River aquatic ecosystem
3-A4
-------
APPENDIX B
CRITR2 CODE CALCULATIONS AND BIOACCUMULATION FACTORS
3-B1
-------
CRITR Code Calculation of Organism Dose from
Water Exposure to Various Radionuclides
CRITR OA Printout •-• User File: RMAX.USR
No Dilution Model used.
Run of: 09:52 18-MAY-92
DFSUIH DFSED
3.5E-12 3.2E-11
INUC K NUKSYHS NS
1 1 AS-76 P
t 2 AS-76 F
t 3 AS-76 C
1 4 AS-76 P
1 5 AS-76 F
DFSUIH DFSED
2.1E-11 1.7E-10
INUC K NUKSYHS NS
2 1 CO-60 P
2 2 CO-60 F
2 3 CO-60 C
2 4 CO-60 P
2 5 CO-60 f
DFSUIH DFSED
2.6E-13 2.SE-12
INUC K NUKSYHS NS
3 t CR-51 P
3 2 CR-51 F
3 3 CR-51 C '
3 4 CR-51 .P
3 5 CR-51 F
DFSUIH DFSED
1.5E-12 1.4E-V,
INUC K HUKSYHS NS
4 1 CU-64 P
4 2 CU-64 F
4 3 CU-64 C
4 4 CU-64 P
4 5 CU-64 F
DFSUIH DFSED
3.1E-12 3.0E-11
INUC K NUKSYHS NS
5 1 1-131 P
5 2 1-131 F
5 3 1-131 C
5 4 1-131 P
5 5 1-131 F
DFSUIH DFSED
3.8E-11 2.6E-10
INUC K NUKSYHS NS
6 1 NA-24 P'
6 2 NA-24 F-
6 3 NA-24 C •.
6 4 NA-24 P
6 5 NA-24 F
DFSUIH DFSED
1.4E-12 1.4E-11
INUC K NUKSYHS NS
7 1 NP-239 P
7 2 NP-239 F
7 3 NP-239 C
7 4 NP-239 P
7 5 NP-239 F
FSOLD FRUF
6.9E-02 0.2
CONOR IT
8.5E+04 3
8.5E+04 3
8.5E+04 3
8.5E+04 3
8.5E+04 3
FSOLD FRUF
6.9E-02 0.2
CONCRIT
4.4E+03 1
4.4E+03 3
4.4E+03 2
4.4E+03 1
.- 4.4E+03 3
FSOLD FRUF
6.9E-02 0.2
CONCRIT
9.3E+05 4
9.3E+05 2
9.3E+05 2
TB
3.7E+02
BIO
.OE-01
.OE-01
.OE-01
.OE-01
.OE-01
TB
KB
1.000
1.000
1.000
1.000
1.000
3.7E+02
BIO
.OE+00
.3E-01
.OE+00
.OE+00
.3E-01
TB
3.7E+
BIO
.OE+00
.OE-02
.OE+00
9.3E+05 4. OE+00
9.3E+05 2
FSOLD FRUF
6.9E-02 0.2
CONCRIT
3.7E+05 2
3.7E+05 2
3.7E+05 4
3.7E+05 2
3.7E+05 2
FSOLD FRUF
6.9E-02 0.2
CONCRIT
1.3E+03 3
1.3E+03 5
1 .3E+03 1
.OE-02
TB
KB
1.000
1.000
1.000
1.000
1.000
02
KB
1.000
1.000
1.000
1.000
1.000
3.7E+02
BIO
.OE+00
.5E+00
.OE-01
.OE+00
.5E+00
TB
3.7E+
BIO
.OE-01
.OE-02
.OE-01
1.3E+Q3 3.0E-01
1 .3E+03 5
FSOLD FRUF
6.9E-02 0.2
CONCRIT
.OE-02
TB
3.7E+
BIO
2.1E+05 1.0E-01
2.1E+05 1
2.1E+05 1
2.1E+05 1
2.1E+05 1
FSOLD FRUF
6.9E-02 0.2
CONCRIT
2.1E+05 3
2.1E+05 2
2.1E+05 3
2.1E*05 3
2.1E+05 2
.OE-01
.OE-01
.OE-01
.OE-01
TB
3.7E+
BIO
.OE-01
.5E+00
.OE-02
.OE-01
.5E+00
KB
1.000
1.000
1.000
1.000
1.000
02
KB'
1.000
1.000
1.000
1.000
1.000
02
KB
1.000
1.000
1.000
1.000
1.000
02
KB
1.000
1.000
1.000
1.000
1.000
BUILDUP
1.6E+00
RINTAKE
0.000
0.000
0.000
0.100
0.200
BUILDUP
3.4E+02
RINTAKE
0.000
0.000
0.000
0.100
0.200
BUILDUP
4.0E+01
RINTAKE
0.000
0.000
0.000
0.100
0.200
BUILDUP
7.6E-01
RINTAKE
0.000
0.000
0.000
0.100
0.200
BUILDUP
1.2E+01
RINTAKE
0.000
0.000
0.000
0.100
0.200
BUILDUP
9. OE-01
RINTAKE
0.000
0.000
0.000 .
0.100 '
0.200
BUILDUP
3.4E+00
RINTAKE
0.000
0.000
0.000
0.100
0.200
TTRANS EXP
0. OE+00 1. OE+00
ECRIT TBIO F1
1.5E-08 280.0 5. OE-01
1.5E-08 280.0 5. OE-01
1.5E-08 280.0 5. OE-01
1.5E-08 280.0 5. OE-01
1.5E-08 280.0 5. OE-01
TTRANS EXP
0. OE+00 1. OE+00
ECRIT TBIO F1
6.0E-09 9.5 3. OE-01
6.0E-09" 9.5 3. OE-01
3.3E-09 9.5 3. OE-01
6.0E-09 9.5 3. OE-01
6.0E-09 9.5 3>OE-01
TTRANS EXP
0. OE+00 1. OE+00
ECRIT TBIO F1
7.3E-11 616.0 1. OE-01
7.3E-11 616.0 1. OE-01
3.8E-11 616.0 1. OE-01
7.3E-11 616.0 1. OE-01
7.3E-11 616.0 1. OE-01
TTRANS EXP
0. OE+00 1. OE+00
ECRIT TBIO F1
2.1E-09 80.0 5. OE-01
2.1E-09 80.0 5. OE-01
1.9E-09 80.0 5. OE-01
2.1E-09 80.0 5. OE-01
2.1E-09 80.0 5. OE-01
TTRANS EXP
0. OE+00 1. OE+00
ECRIT TBIO F1
3.4E-09 • 100.0 1. OE+00
3.4E^09 100.0 1. OE+00
2.9E-09 100.0 1. OE+00
3.4E-09 100.0 1. OE+00
3.4E-09 100.0 1. OE+00
TTRANS EXP
0. OE+00 1. OE+00
ECRIT TBIO F1
1.5E-08 11.0 1. OE+00
1.5E-08 11.0 1. OE+00
1.1E-08 11.0 1. OE+00
1.5E-08 11.0 1. OE+00
1.5E-08 11.0 1. OE+00
TTRANS EXP
0. OE+00 1. OE+00
ECRIT TBIO F1
2. 9E-09 39000.0 1.0E-03
2. 9E-09 39000.0 1.0E-03
2.8E-09 39000.0 1.0E-03
2.9E-09 39000.0 1.0E-03
2. 9E-09 39000.0 1.0E-03
LAHC
0. OE+00
0. OE+00
0. OE+00
6.3E-01
6.3E-01
LAHC
0. OE+00
0. OE+00
0. OE+00
7.3E-02
7.3E-02
LAHC
0. OE+00
0. OE+00
0. OE+00
2.6E-02
2.6E-02
LAHC
0. OE+00
0. OE+00
0. OE+00
1.3E+00
1 .3E+00
LAHC
0. OE+00
O.OE+00
0. OE+00
9.3E-02
9.3E-02
LAHC
O.OE+00
O.OE+00
O.OE+00
1.2E+00
1.2E+00
LAHC
O.OE+00
O.OE+00
O.OE+00
2.9E-01
2.9E-01
MASS
0.0
0.0
0.0
1.0
1.0
HASS
0.0
b.o
0.0
1.0
1.0
HASS
0.0
0.0
0.0
1.0
1.0
HASS
0.0
0.0
0.0
1.0
1.0
HASS
0.0
0.0
0.0
1.0
1.0
MASS
0.0
0.0
0.0
1.0
1.0
MASS
0.0
0.0
0.0
1.0
1.0
DOSECRIT
3.9E-04
3.9E-04
3.9E-04
3. IE-OS
6.1E-05
DOSECRIT
2.7E-05
6.9E-06
2.9E-05
1.1E-05
7.3E-06
DOSECRIT
2.7E-04
1.4E-06
7.1E-05
1.0E-04
1.0E-06
DOSECRIT
1.6E-03
2.0E-03
2.8E-04
6.0E-05
1.5E-04
DOSECRIT
1.3E-06
2.1E-07
3.7E-07
1.4E-06
4.6E-07
DOSECRIT
3.0E-04
3.0E-04
2.2E-04
2.6E-05
5.1E-05
DOSECRIT
1.8E-04
1.5E-03
1.8E-05
6.2E-08
1.0E-06
PLANT
FISH
CRAUDAO
DUCK-P
DUCK-F
PLANT
FISH
CRAUDAD
DUCK-P
DUCK-F
PLANT
FISH
CRAUDAD
DUCK-P
DUCK-F
PLANT
FISH
CRAUDAO
DUCK-P
DUCK-F
PLANT
FISH
CRAUDAD
DUCK-P
DUCK-F
PLANT
FISH
CRAUDAD
DUCK-P
OUCK-F
PLANT
FISH
CRAUDAD
DUCK-P
DUCK-F
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
eats
.
»
»
P
F
.
.
.
P
F
.
.
•
P
F
.
.
.
P
F
.
.
.
P
F
.
.
.
P
F
.
.
.
P
F
3-B2
-------
OFSUIN OFSEO FSOLD FRUF TB
O.OE*00 O.OE+00 6.9E-02 0.2 3.7E+02
INUC 1C NUKSYMS NS CONCRIT BIO KB
1 1 £"lf ? H6*04 5-OE*02 i.ooo
I 1 £"ff £ 'HE*04 1-7E-01 1.000
f / £"« ' £ 2.3E+04 1.06*02 1.000
I £ £"31 c 2.3E+04 5.0E+02 1.000
8 5 P-32 F 2.3E+04 1.7E-01 1.000
DFSWIM DFSED FSOLD FRUF TB
O.OE*00 O.OE+00 6.9E-02 0.2 3.7E+02
INUC K »UKSY« NS CONCRIT BIO KB
9 1 SR-90 P 9.6E+01 3.0E+00 1.000
9 2 SR-90 F 9.6E-01 5.0E-02 1.000
B ? !f'«2 C 9-6E+01 1.0E-01 1.000
9 4 SR-90 P 9.6E+01 3.0E+00 1.000
9 5 SR-90 F 9.6E+01 5.0E-02 1.000
OFSUIM OFSED FSOLD FRUF TB
4.8E-12 4.1E-11 6.9E-02 0.2 3.7g
INUC
10
10
10
10
10
K
1
2
3
i>
S
NUKSYMS
2N-65
ZN-65
ZN-65
ZN-65
ZN-65
NS
P
F
C
P
F
CONCRIT
6.7E+04
6.7E+04
6.7E+04
6.7E+04
6.7E+04
BIO
2.0E+01
6.4E-02
1.0E*01
2.0E+01
6.4E-02
+02
KB
1.000
1.000
1.000
1.000
1.000
BUILDUP
2.1E+01
R INTAKE
0.000
0.000
0.000
0.100
0.200
BUILDUP
3.6E+02
R INTAKE
0.000
0.000
0.000
0.100
0.200
BUILDUP
2.3E+02
R INTAKE
0.000
0.000
0.000
0.100
0.200
TTRANS EXP
O.OE+00 1.0E+00
ECRIT TBIO F1
9.6E-09 257.0 8.0E-01
9.6E-09 257.0 8.0E-01
9.6E-09 257.0 8.0E-01
9.6E-09 257.0 8.0E-01
9.6E-09 257.0 8.0E-01
TTRANS EXP
O.OE+00 1.0E+00
ECRIT TBIO F1
1.6E-08 4000.0 3.0E-01
1.6E-08 4000.0 3.0E-01
1.6E:.08 4000.0 3.0E-01
1.6E-08 4000.0 3.0E-01
1.6E-08 4000.0 3.0E-01
LAHC
O.OE+00
O.OE+00
O.OE+00
5.1E-02
SilE-02
LAMC
O.OE+00
O.OE+00
O.OE+00
2.4E-04
2.4E-04
HASS
0.0
0.0
0.0
1.0
1.0
MASS
0.0
0.0
0.0
1.0
1.0
DOSECRIT
1.1E-01
3.8E-05
2.2E-02
1.8E-01
1.2E-04
DOSECRIT
4.6E-06
'7.6E-08
1.5E-07
4.8E-05
1.6E-06
PLANT
FISH
CRAWDAD
DUCK-P
DUCK-F
PLANT
FISH
CRAUDAO
DUCK-P
DUCK-F
TTRANS EXP
O.OE+00 1.0E+00
ECRIT
1.2E-09
1.2E-09
5.3E-10
1.2E-09
1.2E-09
TBIO
933.0
933.0
933.0
933.0
933.0
F1
5.0E-01
5.0E-01
5.0E-01
5.0E-01
5.0E-01
LAMC
O.OE-00
O.OE+00
O.OE+00
3.6E-03
3.6E-03
HASS
0.0
0.0
. 0.0
1.0
1.0
DOSECRIT
1.6E-03
5.0E-06
3.6E-04
1.6E-02
1.0E-04
PLANT
FISH
CRAUDAD
DUCK-P
DUCK-F
eats -
eats -
eats •
eats P
eats F
eats -
eats -
eats •
eats P
eats F
eats -
eats -
eats -
eats P
eats F
Notes C Units:
Hanford biofactors used.
Ho bfoaceutnulation factor corrections used.
LAMDA Bad. Decay constant
DFSUIM limterston DF
DFSEO .Sediment DF
FSOLD Nuctide sed. buildup rate
FRUF Roughness factor
TB Sed. Buildup time
BUILDUP Sed. Buildup
TTRANS Transport time
EXP ....
Vd
Sv/d per Bq
Sv/d per Bq
m3/m2-d
Fractional decay during trans.
d
Bq-d/Bq
d
CONCRIT Cone, in Water
BIO Bioaccum factor
RINTAKE Intake rate
ECRIT Energy absorbed
TBIO Biological half time
F1 Fraction to total body
LAMC Effective decay const.
HASS Organism mass
DOSECRIT Organism Dose
Bq/m3
m3/kg
kg/d
J/Bq-d
d
1/d
kg
Gy/d
3-B3
-------
tn i— i
m •— « «-«
«/» o
o o oo
m
«•-<•— l
CM to «-«
to o CM
ot
OOOOOO-iOVOOOO
oooooo o ooo
CO O »-. O O O O OOO
««r t-i o o ro o »-i o
in 1-4 o in
§00000000000
ooooooooooo
roou->ooo«—"inroooo
«• CSI t-i O OOO
»-t O <—I CM
tn
to
oooooooooooo
O «• ro o o o o to o O O
n t-i ooo ir>—< o o o
•-i tn o io tn o o
CO * O CO
10 0)
OOOOOOOOOOOO
OC3OOOOOOPOOC3O
PO o •—"«•••—io«-iin ooo
to a)
I >
000000000000
oooooooocoooo
ro o **•*•*+ o <-i in
CM
ooo
o ~*o
I C/>
(ft
«/» a
o o
o> e
u-sr
0) C «•?-
o e
3-B4
-------
SECTION FOUR
ECOLOGICAL RISK ASSESSMENT CASE STUDY:
EFFECTS OF PHYSICAL DISTURBANCE ON WATER QUALITY STATUS AND
WATER QUALITY IMPROVEMENT FUNCTION OF URBAN WETLANDS
-------
AUTHOR AND REVIEWERS
AUTHOR
Naomi Detenbeck
Environmental Research Laboratory-Duluth
U.S. Environmental Protection Agency
Duluth, MN
REVIEWERS
Richard Weigert (Lead Reviewer)
Department of Zoology
University of Georgia
Athens, GA
Gregory R. Biddinger
Exxon Biomedical Sciences, Inc.
East Millstone, NJ
Joel S. Brown
Department of Biological Science
University of Illinois at Chicago
Chicago, IL
Robert J. Huggett
Virginia Institute of Marine Science
The College of William and Mary
Gloucester Point, VA
Richard E. Purdy
Environmental Laboratory
3-M Company
St. Paul, MN
Freida B. Taub
School of Fisheries
University of Washington
Seattle, WA
4-2
-------
CONTENTS
ABSTRACT 4 ?
4.1. RISK ASSESSMENT APPROACH 4_g
4.2. STATUTORY AND REGULATORY BACKGROUND 4_g
4.3. CASE STUDY DESCRIPTION , 4_10
4.3.1. Background Information and Objective 4_10
4.3.1.1. Study Area 4_10
4.3.1.2. Site Selection 4-12
4.3.2. Problem Formulation . . ." 4_12
4.3.2.1. Stressor Characteristics 4-12
4.3.2.2. Ecosystem Potentially at Risk 4_17
4.3.2.3. Endpoint Selection 4_lg
4.3.2.4. Conceptual Model . . 4_2Q
4.3.3. Analysis: Characterization of Exposure 4_24
4.3.3.1. Stressor Characterization . 4_24
4.3.3.2. Ecosystem Characterization . 4_24
4.3.3.3. Exposure Analysis 4.25
4.3.3.4. Exposure Profile 4_2g
4.3.4. Analysis: Characterization of Ecological Effects 4_29
4.3.4.1. Evaluation of Relevant Effects Data 4-29
4.3.4.2. Ecosystem Response Analyses 4_32
4.3.4.3. Analyses Relating Measurement and Assessment Endpoints 4-35
4.3.4.4. Stressor-Response Profile 4.35
4.3.5. Risk Characterization 4_3g
4.3.5.1. Risk Estimation 4_3g
4.3.5.2. Uncertainty 4_4g
4.3.5.3. Risk Description: Summary and Interpretation of
Ecological Significance . . 4_4g
4.4. REFERENCES 4_54
4-3
-------
LIST OF FIGURES
Figure 4-1. Structure of assessment for physical or hydrological impacts on wetland water
quality status and function in the eight-county Minneapolis/St. Paul
metropolitan area
Figure 4-2. Map of eight-county Minneapolis/St. Paul, Minnesota, metropolitan area
Figure 4-3. Path analysis diagram of cause/effect relationships
Figure 4-4. Frequency of disturbance regime by wetland type for study sites in
Minneapolis/St. Paul metropolitan area
4-9
4-11
4-21
4-26
Figure 4-5a. Cumulative frequency distribution of average mid-wetland turbidity values
over the growing season. Predisturbance turbidity distribution (P) is
compared to peak-disturbance distributions for wetlands in watersheds with
(C) and without (NC) construction activity •
4-40
Figure 4-5b. Cumulative frequency distribution of average mid-wetland turbidity values
over the growing season. Predisturbance turbidity distribution (P) is
compared to first-year postdisturbance distributions for wetlands in
watersheds with (C) and without (NC) construction activity
4-41
Figure 4-6. Regression line and 95 percent confidence interval for relationship between
In (construction/wetland area) and springtime In (peak+postdisturbance
pb/predisturbance pb) for urban wetlands affected by storm-water
additions
Figure 4-7a. Isopleths for mid-wetland TP threshold values of 40 /ig P/L or 107 /*g P/L;
predictions for springtime mid-wetland TP, depth change cases
Figure 4-7b. Isopleths for mid-wetland TP threshold values of 40 /ig P/L or 107 /ig P/L;
predictions for growing season mid-wetland TP, depth change cases, target
level of 40 /ig P/L
Figure 4-7c Isopleths for mid-wetland TP threshold values of 40 /ig P/L or 107 /ig P/L;
predictions for growing season mid-wetland TP, depth change cases, target
level of 107 /xg P/L
Figure 4-7d. Isopleths for mid-wetland TP threshold values of 40 /ig P/L or 107 /ig P/L;
predictions for growing season mid-wetland TP, storm-water cases
Figure 4-8. Isopleths for spring mid-wetland color target levels of 113 PCU, 268 PCU,
and 583 PCU, based on predicted response of median predisturbance values
and combinations of two stressors, depth change and change in
watershed/wetland area ratios
4-43
4-44
4-45
4-46
4-47
4-49
4-4
-------
LIST OF TABLES
Table 4-1. Characteristics of Wetland Disturbance Study Sites in the Minneapolis/St Paul
Metropolitan Area '
Table 4-2. Summary of Wetland Disturbance Intensities for Study Sites in the
Minneapolis/St. Paul Metropolitan Area
Use
Table 4-5. Water Quality Values Associated with Mean Light Requirements of 21 4
Percent Incident Radiation for Submerged Aquatic Vegetation in Northern
Lakes ;
Table 4-6. Summaty of Results of MANOVAs Testing for Significant Difference in
Water Quality Change Among Disturbance Classes for Each of Four Time
Periods
Table 4-7. Equations Predicting Change in Mid-Wetland Water Quality as a Function of
Disturbance Intensity
Table 4-8. Mean and Range of Growing Season Mid-Wetland Water Quality Values Prior
to and Following Disturbance
Table 4-9. Uncertainties Affecting Measurement of Risk to Urban Wetland Water Quality
Matus and Water Quality Improvement Function Related to Physical
Hydrologic Disturbance
4-13
4-15
Table 4-3. Endangered, Threatened, and Special Concern Species in the Upper Midwest
That Are Associated With Wetland Habitats . . . . 4.19
Table 4-4. State of Minnesota Water Quality Criteria for Surface Waters by Designated
or
Table 4-10. Summary of Risk to Urban Wetland Water Quality Status and Water Quality
Improvement Function Assessed Against Loss or Conversion of Wetland
Habitat
4-31
4-31
4-33
4-36
4-39
4-50
4-51
LIST OF COMMENT BOXES
Comments on Problem Formulation
Comments on Characterization of Exposure
Comments on Characterization of Ecological Effects
Comments on Risk Characterization
4-22
4-28
4-37
4-53
4-5
-------
DP
DOC
EPA
FTU
FWS
MANOVA
MNDNR
MPCA
NTU
NWI
NWP
PCU
SRP
TCMA
TDS
TP
TSS
U.S. ACOE
LIST OF ACRONYMS
dissolved phosphorus
dissolved organic carbon
Environmental Protection Agency
formazin turbidity units
Fish and Wildlife Service
multivariate analysis of variance
Minnesota Department of Natural Resources
Minnesota Pollution Control Agency
nephelometric turbidity units
National Wetland Inventory
Nationwide Permit
platinum cobalt unit
soluble reactive phosphorus
Twin Cities metropolitan area
total dissolved solids
total phosphorus
total suspended solids
U.S. Army Corps of Engineers
4-6
-------
ABSTRACT
This case study demonstrates an empirical approach to quantifying the regional risk to the
water quality of wetlands and adjacent surface waters based on the frequency, type, and intensity
of physical disturbances. The case study describes an investigation, which began in the fall of
1988, to determine the effects of physical and hydrological modifications on wetland water quality
function in the eight-county Minneapolis/St. Paul metropolitan area. Investigators identified the
incidence of potential stressors to wetland water quality function through surveys of the U S Army
Corps of Engineers (U.S. ACOE) 404 permits under the Clean Water Act, state and county
agencies, and local watershed management organizations.
The study addressed 33 wetland sites potentially affected by deposition of fill, dredging,
imppundment, sedimentation, and storm-water or pumped ground-water inputs during the
succeeding year. Stressor intensities were quantified as wetland fill area, percentage wetland
filled, change in water depth due to dredging or impoundment, changes in the ratio of watershed to
wetland area, changes in the ratio of impervious surface area (urban or residential land use) to
wetland area, and the ratio of construction area (bare earth) in the watershed to wetland area.
Assessment endpoints were potential water quality effects relative to wetland biota (reduced
transparency, altered ionic strength, low dissolved oxygen/high ammonia stress, and lead toxicity)
and potential water quality impacts on downstream surface waters (eutrophication, reduced
transparency, nitrate/nitrite toxicity, and lead toxicity). Measurement endpoints were changes
between pre- and postdisturbance conditions in the following mid-wetland water quality parameters-
temperature, dissolved oxygen, conductivity, turbidity, orthophosphate, nitrate plus nitrite,
ammonia, dissolved and total nitrogen, phosphorus, organic carbon, total and volatile suspended
solids, and total extractable lead. Sampling was conducted for up to 1 year prior to disturbance,
during the peak-disturbance period, and over a 1- to 2-year postdisturbance or recovery period. '
Investigators used a multiple regression approach to quantify stressor-response
relationships. Change in a water quality variable between pre- and postdisturbance or recovery
periods was regressed against measurements of disturbance intensity. The y-intercept in these
regression equations represented annual changes in water quality in the absence of disturbance
(e.g., due to interannual climate variability), while the slope of the relationship represented the
response to increasing intensities of disturbance.
Risk characterization required integrating cause-effect relationships identified through site-
specific investigations with information on regional distributions of stressor type and intensity,
One of the greatest uncertainties associated with evaluating risks to wetland water quality in the
study area was estimating the true incidence or intensity of unregulated or incompletely regulated
physical or hydrologic disturbances, especially with respect to small, isolated headwater wetlands
Estimates of ecological risk to aquatic biota in wetlands also were hampered by problems in
extrapolating water quality standards derived primarily for different classes of surface waters to
wetlands.
4-7
-------
4.1. RISK ASSESSMENT APPROACH
This case study represents a regional risk assessment of the impacts of physical and
hydrological disturbance on the water quality status and function of freshwater emergent wetlands
in the eight-county Minneapolis/St. Paul metropolitan area. The study was not designed to fit the
complete U.S. Environmental Protection Agency (EPA) ecological risk assessment framework
(U.S. EPA, 1992). In particular, investigators could not fully identify or quantify stressor
characteristics during the problem formulation phase because of a lack of good background
information. Thus, problem formulation was refined in conjunction with the stressor
characterization portion of the analysis phase.
The study analyzed wetland water quality status and function, i.e., ecosystem-level effects.
Ecological impacts on specific wetland biota were not the focus of the initial research, but
investigators were able to analyze ecological risks to wetland biota and biota of downstream surface
waters by comparing study area data with state water quality criteria and critical effects levels
derived from the literature for relevant wetland biota (U.S. EPA, 1986). Figure 4-1 provides a
summary of the assessment approach used.
4.2. STATUTORY AND REGULATORY BACKGROUND
One of the goals of the Clean Water Act is to restore and maintain the chemical, physical,
and biological integrity of the waters of the United States. A panel of wetland experts broadly
defined wetland integrity as ". . . the persistence of physical, chemical, and biological conditions
that sustain the long-term processes and structure of the regional wetland resource ..." (Adamus,
1989). Similarly, the Emergency Wetlands Resources Act of 1986 promotes "the conservation of
the wetlands of the nation in order to maintain the public benefits they provide." Wetland-related
activities within EPA focus on assessing and protecting wetland processes associated with water
quality, flood control, and habitat functions of wetlands (Leibowitz et al., 1992).
In practice, the only federal regulatory framework consistently applied to protect wetlands
is the program established under Section 404 of the Clean Water Act, which controls the disposal
of dredge or fill material in wetlands. Much of the wetland fill activity in urbanizing areas was
covered under Nationwide Permit (NWP) 26, which authorizes wetland fill of up to 10 acres in
isolated or headwater wetlands, with no predischarge notification required for fill of less than 1
acre. Subsequently, as part of the 401 certification process, all NWP 26 applications filed in the
State of Minnesota must include a predischarge notification (U.S. ACOE, 1992).
In spite of the wide range of disturbances to which wetlands are subjected (Leslie and Clark
1990), the assessment of long-term impacts on inland wetlands has been restricted to the loss of
wetland area through fill or drainage (Tiner, 1984). Urban wetlands in particular are exposed to a
wide range of physical modifications and hydrologic disturbances—filling, draining, dredging,
impoundment, and storm-water or pumped ground-water inputs—yet little research or synthesis of
information has been done to assess risks to these systems.
4-8
-------
PROBLEM FORMULATION
dredging, impoundment, storm water, pumped
Ecosystem(s) at Risk: Wetlands in the Twin Cities (MN) metropolitan area
Endpoints: Assessment endpoint is maintenance of wetland water quality
/urement endR°'nts are mid-wetland physical-chemical
(e.g., suspended solids, specific conductivity, organic carbon)
ANALYSIS
Characterization
of Exposure
Natural resource managers were
surveyed and 404 permit notices
reviewed. Indices of disturbance
measured were: fill area, % wetland
filled, changes in water depth, ratio
of watershed, urban/residential, or
construction area to wetland area.
Characterization of
Ecological Effects
Mid-wetland water quality was
measured each season. Changes
among predisturbance, peak-
disturbance, 1st year post-
disturbance, and 2nd year
recovery phase were calculated.
RISK CHARACTERIZATION
Peak- or postdisturbance water quality conditions were compared to the
predisturbance phase. Mid-wetland water quality was compared to state
surface water quality criteria and tolerance levels cited in literature for dominant
wetland vegetation. Rate of recovery of mid-wetland water quality was
assessed.
Figure 4-1. Structure of assessment for physical or hydrological impacts on wetland water
quality status and function in the eight-county Minneapolis/St. Paul metropolitan
arpa - r
area
4-9
-------
4.3. CASE STUDY DESCRIPTION
4.3.1. Background Information and Objective
The antidegradation clause in the Clean Water Act requires the maintenance of wetland
ecological integrity, while the Emergency Wetlands Resources Act promotes the conservation of
public benefits (i.e., functions) of wetlands. A complete risk assessment of the impacts of physical
or hydrologic disturbance on urban wetland status and function would require an examination of
effects on wetland hydrologic functions (flood control, ground-water recharge), habitat functions,
and water quality improvement functions.
Traditionally, the loss of wetland function has been monitored as a net change in wetland
area (Dahl and Johnson, 1991). Minnesota's 1990 report to Congress under Section 305(b) of the
Federal Water Pollution Control Act estimated that mitigation activities under the 404 permit
program resulted in a statewide net loss of 61 acres of wetlands (of the 5.02 million acre total)
during 1988-1989, with an additional 4,000 acres of wetlands restored or "enhanced" (MPCA,
1990). Similarly, a comparison of previous rates of wetland loss from drainage with recent rates
shows a decrease. However, the loss of wetland function can occur through type conversions (with
no loss of wetland area) as well as through degradation of existing conditions. Therefore, this risk
assessment explicitly targets an information gap—the potential degradation of wetland water quality
status and function due to common physical disturbances. Where possible, effects on wetland
habitat (loss or conversion) are discussed, but quantification of these impacts was beyond the scope
of this study.
The case study summarizes the results of a 3-year, $280,000 research project on the
impacts on, and recovery of, mid-wetland water quality from physical or hydrologic disturbance in
the eight-county Minneapolis/St. Paul metropolitan area. Stress-response curves derived from this
study were supplemented with literature- and permit-based surveys of the incidence of physical or
hydrologic disturbance activities hi this region. Investigators also supplemented water quality
criteria values with a literature review of tolerances of relevant wetland-dependent biota to
measured water quality parameters. The Wetland Function Project (U.S. EPA Wetland Research
Program) provided funding for the original research. At the time, the Wetland Function Project
focused on wetland water quality and water quality functions; therefore, site investigations of
potential habitat effects were limited to qualitative descriptions of dominant plant species or cover
and to an assessment of changes in wetland type.
4.3.1.1. Study Area
The study area encompasses both the 7,330 km2 Minneapolis/St. Paul metropolitan area and
adjacent Wright County (figure 4-2). The population of the region is over 2,000,000, with the
heaviest densities in the central cities of Minneapolis and St. Paul. Land use is 27 percent urban,
43 percent agricultural, and 30 percent open space (Ayers et al., 1985). Urbanization is rapidly
spreading into agricultural and open areas, with greatest population increases now occurring in
Anoka and northern Dakota Counties.
4-10
-------
N
10 KM
Washington '•
36
31
Hennepin { «x R^p
39 X ST.
.MINNEAPOLIS. \ * r^^
AUL
Carver
' 40 3T
,cT r -
V
Scott
21
24
27
Dakota
i
Figure 4-2. Map of eight-county Minneapolis/St. Paul, Minnesota, metropolitan area
(adapted from Detenbeck, et al. 1992). Sites are listed by number and
characterized in table 4-1.
4-11
-------
Omernik (1986) defines the Twin Cities metropolitan area (TCMA) as part of the North
Central Hardwood Forest ecoregion, with portions extending into the Western Cornbelt Plains.
Topography consists of gently undulating, glaciated uplands dissected by the St. Croix, Minnesota,
Rum, and Mississippi River valleys. The region is characterized by terminal moraines and glacial
outwash with wetlands in areas of high water tables, in glacial kettle depressions, and along major
rivers and associated tributaries (Ayers et al., 1985). Agricultural and urbanization pressures
resulted in the filling or draining of many wetlands, and by 1969 only half of the presettlement
wetland area remained (Anderson and Craig, 1984). Wetlands now constitute about 7.6 percent of
the region (Owens and Meyer, 1978).
4.3.1.2. Site Selection
The study design limited the selection process to those wetlands that could be sampled
before, during, and after disturbance within the two growing seasons of the original study time
frame (September 1988 to October 1990). The lack of legal access eliminated only four of 53
wetlands identified as suitable for the project. Investigators also eliminated wetland disturbances
adjacent to the St. Croix, Minnesota, and Mississippi Rivers because of the slight chance of
observing a measurable impact to the riparian wetlands of these large, lotic systems. Impacts on
water quality status and function of large riverine wetlands are better handled through cumulative
impact assessments than site-specific or population studies (e.g., Gosselink and Lee, 1989; Osborne
and Wiley, 1988).
Investigators identified 31 wetlands for the study by surveying wetland fill 404 permit
notices and by requesting information on additional disturbance activities (dredging, impoundment,
draining, storm-water inputs) from the Minnesota Pollution Control Agency (MPCA), Minnesota
Department of Natural Resources (MN DNR), county (drainage) ditch commissioners, and
watershed management organizations.
4.3.2. Problem Formulation
4.3.2.1. Stressor Characteristics
Disturbance activities identified through surveys of area resource managers included
wetland fill (16), impoundment or dredging (9), and diversion of storm water or pumped ground
water into wetlands (14; table 4-1). Construction activity in the watershed was quantified after the
fact, when monitoring demonstrated that severe sedimentation problems existed at some sites.
Investigators calculated physical or hydrologic disturbance intensities for each site based on field
observations, 404 permit notices, and topographic maps combined with land-use maps derived by
classifying aerial photos (1:9600) taken before and after disturbance activities (table 4-2).
In cases of dredging or impoundment, the disturbance was defined as a step change in
water depth, based on field observations and design criteria contained in permit notices.
Investigators used Circular 39 (Shaw and Fredine, 1956) definitions to classify pre- and
postdisturbance wetland types. The difference in wetland types before and after disturbance was
used as a measure of the intensity of dredging or impoundment. For example, a change from a
type 3 shallow marsh to a type 5 wetland pond would have an intensity value of +2. Nonriparian
4-12
-------
I
I
I
•a *
4-13
-------
I
^w,
s
"8
° S
o" 0
g
E H w
S3
I
M
=§
1
O
-
NO
5 « eu
•< W •-»
§
g
CO
«?
cs
C: CJ
i \O \O
—i O
W
«
o co Z
U ^ C
>, O U
3
S I
H!H
8.iS S o
4-14
-------
I
i
•
•*•*
.1
s
£
.=
I
t
1
8
•a
I
.a.
}]]
ill
S S 8 8 3
o o o o o
2 2
o o
5 D S 3 S S $ $
oddddooodoooddo
o oooo- iJ ^10
S
o o o o — o o
v> >n «i _
«. 2 3 » o S
9 9 9 d d -
S S 3 3 3
oooooooo*-!<-3i-i
' S2
*
CQ P3
$ J? 5? ^ ^
ft ft ft 6; ft
en <»> <^ « m ^N en
4-15
-------
I
o
I
1
1
1
a
1-8
8 £3
*>H i-( O
8 S
I
8 S
S S
?)
o o o o o o
OOOOOOOO-~"'OOOOOOOO
CO
g g g g 9 2 2
g S 2
a a °
o o o o
ooooooooo
8
I
CO TO W
s s s s
CO CO CO
ooooooooo«-
-------
shrub-scrub (type 6) and woody (type 7) wetlands typically have water table levels equivalent to
type 2 wet meadows and were assigned a hydrology factor of 2 on the intensity scale. '
Investigators measured the intensity of fill disturbances as fill area, the percentage of
wetland area filled, and the distance from the sampling point to the nearest area of wetland filled.
Public notices published as part of the U.S. ACOE 404(c) permit program provided information on
area of fill. Wetland areas were obtained from 404(c) permit notices, watershed districts, or
National Wetland Inventory (NWI) maps. Distance from sampling point to nearest filled area was
calculated from permit notice site maps, topographic maps, or NWI maps.
Storm-water inputs to a wetland are related to the degree of urbanization in the watershed.
Increases in impervious surface area and point-source storm-sewer inputs increase the volume of
storm water entering a wetland. To quantify the increase in urbanization, watersheds were gridded
into 0.25- to 16-hectare cells on 1:25,000 U.S. Geological Survey topographic maps (depending on
watershed size). Using this map, investigators identified the number of cells classified as urban or
residential before and after disturbance, based on an examination of aerial photos (Detenbeck et
al., 1991a). The change in the ratio of urban and residential area in the watershed to the area of
each study-site wetland was used as one indicator of storm-water disturbance intensity. Because
the creation of storm-sewer systems can involve connecting previously isolated watersheds, the
change in watershed/wetland area was calculated as an additional index of hydrologic disturbance.
Investigators also used the gridded map to quantify erosion inputs. Construction zones with
surfaces of freshly disturbed bare earth have the largest erosion potential. Therefore, the ratio of
construction area in each watershed to postdisturbance wetland area was calculated for each
wetland site, based on an examination of aerial photos.
Stressor impacts are determined not only by the incidence and intensity of physical or
hydrologic disturbances but also by the frequency and duration of stressors, incidence of multiple
stressors including increased chemical loadings from watershed development, and time since initial
disturbance (recovery period). Ecosystem response also depends on tolerances of existing species,
which may be related to prior disturbance history, including both anthropogenic and natural
(climatic) disturbance regimes. Moderating factors include season, antecedent wetland type,
vegetation, watershed conditions, and the use of best management practices (e.g., preservation of
vegetated [upland] buffer strips).
4.3.2.2. Ecosystem Potentially at Risk
Wetlands in this area can be classified by either water depth or predominant vegetation type
(e.g., Shaw and Fredine, 1956; Cowardin et al., 1979). Most of the freshwater wetland types
identified by Cowardin occur in the study area (Owens and Myer, 1978; Werth et al., 1977),
although bogs are extremely rare. Some calcareous fens occur in Dakota and Scott Counties in the
southern part of the TCMA and contain plant species listed as endangered, threatened, or species
of concern in Minnesota (Eggers and Reed, 1987). While wetland vegetation communities have
been inventoried for the TCMA (Owens and Myer, 1978; Werth et al., 1977), few faunal
inventories are available. A number of amphibians and reptiles are found in the study region,
including eastern newts (Notophthalmus viridescens), tiger salamanders (Ambystoma tigrinum),
4-17
-------
leopard frogs (Rana pipiens), striped chorus frogs (Pseudacris triseriata), green frogs (Rana
clamitans), wood frogs (Rana sylvatica), spring peepers (Hyla crucifer), snapping turtles (Chelydra
serpentina), painted turtles (Chrysemys pictd), and the smooth green snake (Opheodrys vernalis)
(Niering, 1985). In all, 35 of the animal (27) or plant (8) species listed as endangered, threatened,
or of special concern within Minnesota are associated with wetland habitats (MN DNR, 1984); 18
of these species have ranges that overlap with the study region (Niering, 1985; see table 4^3).
Hydrologic classifications for wetlands in the study area include (a) isolated wetlands, with
no inlets or outlets; (b) intermittent-flow wetlands, with inlets and outlets that flow only during
snowmelt or major storm events; or (c) flow-through systems, with a fairly continuous movement
of surface water in and out of the wetland. Distinct differences in water chemistry exist among
these hydrologic wetland types in the TCMA, with higher nutrient, carbon, and conductivity levels
in isolated wetlands; thus response may differ by wetland type (Detenbeck et ah, 1991a).
Therefore, impact analyses should consider initial (predisturbance) wetland water quality as a
reference condition.
4.3.2.3. Endpoint Selection
Surface water inputs and outputs to wetlands often are intermittent and cannot be rigorously
quantified without intensive instrumentation and monitoring (e.g., Brown, 1985). Thus, mid-
wetland water quality variables were chosen as the best set of measurement endpoints to indicate
wetland condition and potential inputs to downgradient ground water or downstream surface
waters.
Measurement endpoints were chosen as indicators of four components of mid-wetland water
quality (transparency, trophic status, potential heavy metals toxicity, and redox status) and three
components of downstream or downgradient surface water or ground-water quality (transparency,
eutrophication, and potential toxicity to humans [nitrate] or aquatic biota [lead]). Within wetlands,
reduced transparency from high dissolved organic carbon (DOC) or suspended solids will limit the
growth of submerged macrophytes (Chambers and Kalff, 1985). Sedimentation can inhibit
germination from seedbank sources (Galinato, 1985), which may already be depleted by dredge and
fill activities. Qualitative records of dominant vegetation and plant cover at study-site wetlands
suggested that recovery of submerged aquatics was delayed by >2 years following initial impacts
(Detenbeck et al., 1992). Regional or local declines in submerged aquatic communities elsewhere
have been attributed to eutrophication and reduced water clarity (Dennison et al., 1993).
Phosphorus often is the limiting nutrient to primary producers in metropolitan area lakes, which are
already predominantly mesotrophic or eutrophic (Metropolitan Council, 1981); thus, any increased
loading to downstream lakes could be considered detrimental. Productivity of area wetlands can be
either nitrogen- or phosphorus-limited; if it is nitrogen-limited, then increased nitrate loadings
would also have an impact on wetlands.
Lead was chosen as a measurement endpoint because it is a common contaminant in urban
environments. Lead levels are already elevated in metropolitan area lakes to levels exceeding
water quality criteria (Metropolitan Council, 1981), and high lead levels are associated with urban
storm- water runoff in this region (time-weighted annual average = 39 /ig Pb/L; Johnston et al.,
4-18
-------
Table 4-3. Endangered, Threatened, and Special Concern Species in the Upper Midwest
That Are Associated With Wetland Habitats (derived from MN DNR, 1984, and
Niering, 1985)
Scientific Name
Common Name
Podiceps auritus
Pelecanus erythroorhynchos
Botaurus lentiginosis
Buteo lineatus
Pandion hattaeteus
Gray canadensis
Rallus elegans
Coturnicops noveboracensis
Gallinula chloropus
Phalaropis tricolor
Sterna forsteri
Asia flammeus
Ammospiza caudacutus
Homed grebe
American white pelicanb
American bittema>b
Red-shouldered hawk3
Osprey
Sandhill crane
King rail"
Yellow rail
Common moorhen3
Wilson's phalarope
Forster's tern
Short-eared owl"
Sharp-tailed sparrow
Clemmys insculpta
Chelydra serpentina
Crotalus horridus
Acris crepitans
RarlU catesbiana
Rana palustris
Wood turtle"
Snapping turtlea>b
Timber rattlesnake
Northern cricket frog
Bullfrog3
Pickerel frog3
Clossianafrigga saga (Staudinger)
Epidemia dorca dorcas (W. Kirby)
Eribia disa mancinus (Doubleday and Hewitson)
Oeneis jutta ascerta (Masters and Sorensen)
Proclossiana eunomia dawsonii (Barnes and McDunnough)
Frigga fritillary
Dorcas copper
Disa alpine
Jutta arctic
Bog fritillary
4-19
-------
Table 4-3. Endangered, Threatened, and Special Concern Species in the Upper Midwest
That Are Associated With Wetland Habitats (continued)
Scientific Name
Common Name
Notropis emilae (Hay)
Pofyodon spathula (Walbaum)
Scaphirhynchus platorynchus (Rafinesque)
Pugnose minnow
Paddlefish8
Shovelnose sturgeon"
Arethusa bulbosa L.
Cephalanthus occidentalis L.
Decodon verticillatus (L.) Ell.
Hydrocotyle americana L.
Pinguicula vulgaris L.
Orchidaceae8
Rubiaceae8
Lythraceae
Apiaceae8
Lenibulariaceae*
"Occurring within study region according to range maps in Niering (1985).
bObserved at least once in study site(s).
1990). Wetlands can efficiently retain paniculate lead, thereby protecting downstream lakes but
posing a potential risk to wetland biota (Detenbeck et al., 1991b).
Nitrate was chosen as a measurement endpoint because denitrification is an efficient and
significant water quality improvement function associated with wetland ecosystems. A buildup of
nitrate in urban wetlands would indicate a breakdown in normal wetland water quality function as
well as a risk to users of downstream surface waters or ground water.
Measurement endpoints included changes in specific conductivity, temperature, dissolved
oxygen, and ammonia because any of these might affect the suitability of wetland habitats. The
analysis did not include measurements of sodium concentrations or ratios of monovalent to divalent
cations, which would be expected to increase with an influx of road salt and could have deleterious
effects on wetland phytoplankton or macrophyte vegetation (Wetzel, 1975). In particular, inputs of
storm water to extremely dilute bogs or alkaline fens would be expected to effect ecologically
significant changes in mid-wetland water quality (Rushton, 1991).
4.3.2.4. Conceptual Model
Elements of the conceptual model for this assessment are listed in figure 4-1, and the path
diagram (figure 4-3) outlines the relationships between disturbance indices (stressors) and wetland
water quality response. Construction activity is a potential source for sediment, phosphorus, and
nitrate supplies in urban wetlands, while wetland fill area can serve as a source of sediment or
phosphorus prior to revegetation. Construction activity also can increase loadings of dissolved
4-20
-------
Jill
II §1
P
PH U .3 U
4-21
-------
organic carbon by disrupting soil structure and promoting degradation of soil organic matter. A
large fraction of lead in urban runoff is in paniculate form (paniculate toxicity), thus lead loadings
should be related to sediment inputs. Impacts of nutrient, organic carbon, sediment, and lead
inputs are inversely proportional to wetland area; i.e., the same loadings will have a larger impact
on a small wetland than a large wetland. Internal loading of sediment and phosphorus may
increase as wetlands become shallower and resuspension increases, although development of
emergent vegetation in the shallow marsh zone and of submerged vegetation may limit
resuspension.
Losses of sediment, paniculate-associated contaminants (lead), and nutrients (phosphorus)
are controlled by hydraulic retention time (flow-through) and the time required for particles to
settle out of suspension (settling), which is a function of particle size and wetland depth (Walker,
1987). Flow-through rates are dependent on runoff (a function of watershed area, percentage of
impervious area, and precipitation) relative to wetland volume (surface area x depth).
Within each wetland, transparency is a function of dissolved organic carbon (color),
turbidity (suspended solids), and to some extent, chlorophyll a. Chlorophyll a probably plays a
lesser role in reducing water clarity in wetlands than in lakes, because algal production in the water
column becomes inhibited by light limitation from suspended solids and water color. (Shading by
floating algal mats would be an exception.) High turbidity, color, and trophic status (high
chlorophyll a) within wetlands potentially limit the development or recovery of submerged
macrophytes by limiting the depths at which sufficient light is available for growth (Chambers and
Kalff, 1985; Dennison et al., 1993).
Redox status within the water column and surficial sediments will be reflected by levels of
dissolved oxygen and by the proportion of dissolved inorganic nitrogen as ammonia. Under low
dissolved oxygen conditions, relative levels of nitrate will decrease because of denitrification
occurring in anaerobic sediments and because of inhibition of nitrification (conversion of ammonia
to nitrate).
Comments on Problem Formulation
Strengths of the case study include:
General information on regional wetland types and species composition was
summarized thoroughly.
4-22
-------
Comments on Problem Formulation (continued)
Limitations include:
• The risk assessment was based on research that was primarily focused on
wetland water quality junction and attendant protection of downstream surface
waters. Insufficient information was available to fully assess the impacts on
wetland biota. An analysis of the biota found in specific study sites and an
assessment of the relative sensitivity of different classes of biota would have
strengthened the risk assessment. These data were not attainable given
available resources. Research is under way to characterize the wetland
macroinvertebrate communities that are affected by storm-water inputs in this
region.
• Stressors and endpoints. Water quality might not be an appropriate endpoint to
evaluate the impacts of disturbance to wetland ecosystems. For example,
productivity is determined by the throughput, or turnover, of nutrients and can
be affected without significant effects on standing stocks of free nutrients.
Temperature is a relative factor; its impact depends on the system and its
ground state. Also, the presence or absence of individual species is unlikely to
be a sensitive indicator because of prior impact; however, changes in abundance
might signal important changes.
• Wetland values. Wetlands are valuable for more reasons than serving as a
buffer for downstream water quality. For example, wetlands provide habitat for
migratory birds and ecotone species and help maintain ground-water levels.
The case study should indicate how the stressors affect variables such as these.
It also should be noted that partial fill is a loss of wetland habitat.
General reviewer comment:
• Although a path diagram is included (figure 4-3), a statistical path analysis
could not be completed. The ecological literature tends to emphasize causality
models that perform path analysis. If the data on wetlands are numerous
enough or amenable to such analyses, a path analysis would be useful to
decompose direct and indirect effects. The resulting diagram would be useful by
showing linkages between dependent and independent variables. Rather than
pattern hunting, statistics could be used to test specific hypotheses. The path
analysis diagram could be used as a basis for the conceptual model.
4-23
-------
4.3.3. Analysis: Characterization of Exposure
4.3.3.1. Stressor Characterization
Among the study sites, dredging impacts ranged from minimal (no change in water depth
class; e.g., Minnehaha site 39, type 5) to an increase in three depth units for site 36. In the latter
case, a type 2 wetland was dredged to form a type 5 wetland pond. Impoundment in the absence
of dredging was relatively rare, occurring at only two study sites, and resulted in a step change of
only one unit in the scale of relative water depth (1-5).
Wetland fill area varied from 0.01 ha (Comma, site 38) to 2.0 ha (Colonial Pond, site 1),
with percent wetland area filled ranging from less than 1 percent (Comma) to 71 percent (Colonial
Pond). The greatest potential erosion impacts occurred at Centerville (site 31), with a ratio of 120
for construction zone to wetland area, indicating a high potential loading of sediment per unit
surface area of wetland.
New storm-water inputs were common in urbanizing regions as the area of impervious
surface increased and point-source storm sewers were built to divert storm water into wetlands.
The potentially greatest storm-water impacts occurred at JP-26W (site 29), with an increase in the
ratio of urban plus residential area to wetland area of 56. In two cases, ground water also was
pumped into wetlands as a means to de-water adjacent construction sites. These inputs were
temporary and sporadic and could not be quantified easily.
4.3.3.2. Ecosystem Characterization
Wetlands ranged in size from 0.01 ha to over 112 ha, and watersheds varied from 3.3 ha to
8,864 ha. Predominant land use in each watershed was classified as agricultural, urban or
residential, or undeveloped or open space. Overall land use ranged from 0 to 92 percent
agricultural, from 0 to 45 percent forested, from 0 to 48 percent urban, from 0 to 49 percent
residential, from 1 to 58 percent water (lake plus marsh), from 0 to 70 percent construction area,
and from 0 to 9 percent orchard in any given watershed. Average watershed slope varied from 1.4
percent for the Coon Creek watershed within the Anoka Sand Plain (site 23) to 13.2 percent for
JP-25 (site 40) in the hilly terrain of Eagan. Soil erodability varied from an average K-factor of
0.16 (Coon Creek watershed, site 23) to an average K-factor of 0.33 in JP-68 (site 28) within
Eagan.
Only a small proportion of the sites modified by physical or hydrological disturbance had
buffers of undisturbed vegetation left surrounding the wetland. Typically, construction extended to
the edge of or directly into the wetland. Only six wetlands had vegetated buffer zones left between
the impact and the sampling point; these buffers ranged in width from 3 to 8 meters.
Effects of physical or hydrologic stressors will be moderated or exacerbated by climate,
particularly the amount and temporal distribution of precipitation. Climate in the TCMA is
continental, with mild, humid summers and relatively long, severe winters. Most rain comes in
frontal storms or warm-weather convective storms, with May and June typically the wettest months
4-24
-------
and February the driest (Brown, 1984). Normal annual precipitation is 68.6 cm, including the
water content of 111.8 cm average winter snowfall. Annual precipitation varied greatly during the
study. In the drought year of 1989, annual precipitation was only 59.2 cm, while heavy summer
rains brought the 1990 total to 83.9 cm (U.S. Weather Service, 1991).
Impacts on wetlands from physical and hydrologic disturbance depend in part on initial
hydrologic and vegetation conditions at each site. Investigators classified the range of wetland
types in the study area using the definitions in Circular 39 (Shaw and Fredine, 1956). Predominant
vegetation in the wet meadows included reed canary grass (Phalaris arwdinacea), smartweeds
(Polygonwn spp.), and stinging nettle (Urtica dioicd). Cattail (Typha spp.) dominated both shallow
and deep marshes, with sizable inclusions of softstem bulrush (Scirpus validus), giant reed grass
(Phrqgndtes australis), and arrowhead (Sagittaria latifolia). Various pondweeds (Potamogeton
spp.), coontail (Ceratophyllum demerusum), and water milfoil (Myriophyttum spp.) were common
among submersed vegetation, with lotus (Nelumbo luted), yellow water lily (Nuphar variegata),
white water lily (Nymphaea odorata), and duckweed (Lemna spp.) floating on the surface of
wetland ponds. Invasions of purple loosestrife (Lythrum salicarta) were found in scattered areas.
Typical shrubs in type 6 wetlands included red osier dogwood (Cornus stolonifera), speckled alder
(Alms rugosa), and numerous willows (Salix spp.). Of the species noted, seven were rated as
moderately tolerant to turbidity and pollution by Kadlec and Wentz (1974), while only one species
(Potamogeton natans) was rated as relatively intolerant. Typha spp. and Phragmites australis,
common at many of the sites, are considered to be invasive species that often appear in disturbed
areas.
4.3.3.3. Exposure Analysis
Frequency of impacts due to physical or hydrologic disturbance to wetlands in the TCMA
was quantified by two related survey approaches. First, information on all impending physical or
hydrologic disturbances to wetlands in the TCMA for the period of fall 1988 to fall 1990 was
requested from area resource managers (see above). Second, all individual and nationwide U.S.
ACOE 404 permits requiring predischarge notification received in 1988 and 1989 by the St. Paul
District U.S. ACOE Office were reviewed for information on factors related to permit success
(Taylor et al., 1992).
The frequency of disturbance regimes at sampling stations was tabulated by Circular 39
wetland type (figure 4-4). Most sites received multiple impacts. Almost all study sites (79
percent) were potentially affected by sedimentation from construction activity immediately
surrounding the wetland or from physical modifications to existing wetlands. Nearly two-thirds of
the study sites were partially filled or affected by storm-water or pumped ground-water inputs.
Added water inputs were most common for wetland pond or deep marsh systems, while water level
changes due to dredging or impoundment were most common for shallow marsh or wet meadow
systems.
A total of 114 fill permits were reviewed, of which 86 (75 percent) were approved.
Investigators identified 30 (35 percent) approved permits for which additional disturbances at the
wetland site were anticipated, either as part of a construction project, mitigation action, or water
level manipulation for waterfowl management. Eighteen (60 percent) of the additional disturbances
4-25
-------
sedimentation
fill water inputs depth change
DISTURBANCE REGIME
Wetland type
wet mead/seas'y fl
| | pond
shallow marsh
shrub/woody
deep marsh
Figure 4-4. Frequency of disturbance regime by wetland type for study sites in Minneapolis/
St. Paul metropolitan area (Detenbeck et al., 1991a). Physical or hydrologic
disturbance regimes are categorized as sedimentation (i.e., erosion from
construction activity or resuspension), partial fill activity, storm-water or pumped
ground-water inputs, and water depth changes due to dredging or impoundment.
Wetland sampling stations are categorized by Circular 39 classifications: wet
meadow and seasonally flooded wetlands (types 1, 2), shallow marsh (type 3),
deep marsh (type 4), wetland ponds (type 5), and scrub-shrub or woody wetlands
(types 6, 7).
4-26
-------
involved dredging open water areas, five (17 percent) involved new storm-water inputs, four (13
percent) involved dredging channels, three (10 percent) involved impoundment but no dredging,
and three (10 percent) involved wetland drainage (two cases of temporary drainage).
Most of the wetland sites identified in this study received more than one physical or
hydrologic disturbance, and many sites obviously had been affected by past alterations. Without a
longer term record, however, it was not possible to determine the frequency of disturbance to
wetlands in the TCMA over time. The season during which the physical activity creating physical
or hydrologic disturbances ends is probably the most critical aspect of timing that will affect
wetland recovery. At 34 (72 percent) of the 47 mid-wetland stations monitored, the physical
activity producing the wetland disturbance ended outside of the growing season, i.e., in the fall,
over the winter, or during snowmelt.
It is clear that physical or hydrologic disturbances affect some wetlands in the TCMA more
heavily than others. However, there is no evidence that fill permit success is significantly
associated with wetland type, adjacency to large wetland complexes, adjacency to calcareous fens
(which have special protection status in the state of Minnesota), or state-protected status (types 3,
4, and 5 wetlands; Taylor et al., 1992). Permits to fill wetlands immediately surrounded by
industrial or commercial land or by open land (on the suburban fringe) are significantly more likely
to be approved than those for wetlands immediately surrounded by residential, mixed residential, or
agricultural land use (Taylor et al., 1992). Storm water-related disturbances to wetlands are
prevalent in the Eagan area (northern Dakota County), which has a relatively steep topography and
a rapid rate of growth through residential development (figure 4-1).
An exact percentage of the area of wetland resources in the TCMA affected annually by
physical or hydrologic disturbance cannot be easily quantified until automated data are available
from NWI map digitization. Wetlands of 10 to 500 acres in size were partially catalogued in 1967
by MN DNR for fish and wildlife management (MN DNR, 1967). According to their records,
approximately 745 type 2, 3, 4, or 5 wetlands were found in the TCMA. No quantitative
inventory of smaller wetlands is available. If 86 wetlands are partially filled in a 2-year period, 30
of which experience physical or hydrologic disturbances, this represents an incidence of
approximately 11.5 percent of wetlands affected by partial fill and 4 percent of wetlands affected
by additional physical or hydrologic disturbance over a 2-year period. This obviously is an
overestimate, however, because many of the wetlands affected by fill and related disturbances are
much smaller than 10 acres in size, and this fraction of the wetland resource has not been well
quantified in the TCMA.
Total wetland losses in Minnesota resulting from 404 permits (individual, general, and
nationwide) equaled approximately 1,196 acres out of a total of 5.02 million acres, or 0.024
percent per year in 1988-1989. During the same period, Minnesota also saw a gain due to
mitigation activities of 1,135 acres, for a net loss rate of 0,0013 percent per year (MPCA, 1990).
In comparison, wetland losses due to drainage in a 10-county area during 1974-1980 were
estimated at 0.02 percent per year for wetland ponds, 0.6 percent per year for deep marshes, and
2.3 percent per year for shallow marshes (MPCA, 1990). Loss of specific wetland types as the
result of conversions to other wetland types has not been quantified for this region of the country.
Nationwide, 0.1 percent and 1.3 percent of forested jpalustrine wetland area (swamps) have been
4-27
-------
lost by conversion to nonvegetated wetlands (ponds) and marshes, respectively, while 0.2 percent
of marshes has been lost to conversion to ponds (Dahl et al., 1991).
4.3.3.4. Exposure Profile
The most commonly recorded physical or hydrologic disturbances to wetlands in the
TCMA are, in order of frequency, sedimentation from excessive erosion, wetland fill, deepening
by dredging or impoundment, and storm-water impacts. Up to 11.5 percent of wetlands in the
TCMA were permitted for partial or complete filling over a 2-year period, with up to 4 percent of
all wetlands affected by additional physical or hydrologic disturbances. Wetlands on the suburban
fringe or those surrounded by industrial or commercial land uses are most likely to be filled.
Storm-water inputs are probably most common in areas of rapid residential growth and relatively
steep topography, but the use of wetlands for storm-water management is not well documented on a
regional basis. Most disturbances to wetlands occur or terminate during a period outside of the
growing season, thus maximizing potential recovery time.
Comments on Characterization of Exposure
Strengths of the case study include:
• Spatial and temporal variability in exposures is described.
• The causes of uncertainty in exposure estimates are documented.
Limitations include:
• A complete exposure profile for the TCMA wetlands would require that the
potential resources affected be better quantified in terms of wetland number,
area, and type. In addition, a more complete sample of physical and
hydrologic disturbance frequency and intensity, particularly for partially
regulated or nonregulated disturbances (drainage, impoundment, dredging,
storm-water or pumped ground-water inputs) is needed.
• A more complete exposure profile also would include site-specific information
on particular wetland populations and communities exposed to physical and
hydrologic disturbance, as determined by the overlap of their temporal and
spatial distributions. Some of this information will be available from an
ongoing study of effects of storm-water and nonpoint-source pollution on
wetland macroinvertebrate communities in the TCMA.
4-28
-------
4.3.4. Analysis: Characterization of Ecological Effects
4.3.4.1. Evaluation of Relevant Effects Data
Investigators judged the relevance of the impacts of disturbance on wetland water quality on
the basis of (1) the statistical significance of effects and (2) the potential ecological significance of
effects. The statistical significance of changes in water quality was tested both as a verification of
cause-and-effect relationships and as a means of comparing water quality values against a reference
(predisturbance) condition. Comparisons against reference conditions are appropriate when water
quality varies regionally as a function of landscape or climatic conditions or when there is a high
level of uncertainty associated with the magnitude of critical effect levels. For example, reference
conditions by ecoregion have been used in deriving regional lake water quality standards for the
State of Minnesota (Heiskary and Wilson, 1990). Water quality criteria provide critical effect
levels, but these often are derived based on tests of nonwetland species and under testing conditions
(high dissolved oxygen, low dissolved organic carbon, circumneutral pH) that are atypical of
wetlands (Hagley and Taylor, 1991).
The realism inherent in field-scale manipulations or observations is accompanied by spatial
(geographic) variability among study sites as well as temporal (climatic) variability between pre-
and postdisturbance periods. Analyses of predisturbance wetland water quality identified wetland
type, hydrologic class, contact with sediment (pore water vs. surface water), season (snowmelt vs.
growing season), and surrounding land use as factors with significant contributions to variability in
wetland water quality variables (Detenbeck et al., 1991a). Thus, paired before-and-after
comparisons were used to factor out spatial variability among sites. Predisturbance conditions at
each site served as a reference against which peak- or postdisturbance conditions were compared
using a parametric multivariate analysis of variance (MANOVA). A nonparametric Kruskal-Wallis
test was used when data could not be normalized with log transformations* (Sokal and Rohlf, 1981).
MANOVAs were used in place of paired t-tests to reduce the probability of Type II errors, which
increases as the number of tests performed increases. Repeated analysis of variance measures by
variable type would be an ideal test to use here to determine time to recovery because these tests
would correct for possible carry-over effects (serial correlation), but the number of observations
available without missing data for all of the time periods of interest was very low (n=8-9).
The disadvantage of a paired before-and-after comparison approach is that interannual
climatic variability can bias changes between pre- and postdisturbance periods. Thus, only those
water quality variables showing both (a) a significant change from predisturbance periods and (b) a
significant difference in response among disturbance classes were considered to be affected by
physical disturbances. Alternatively, one could employ paired before-and-after comparisons with
sites distributed along a disturbance gradient, somewhat analogous to an analysis of covariance
approach. This approach regresses the change in mid-wetland water quality at each site against
indices of disturbance intensity. Thus the y-intercept represents the expected change in water
quality due to interannual climate variability alone, while the slope of the regression represents the
response to increasing disturbance intensity. A multiple regression approach can be used to factor
out the effect of multiple disturbances as long as colinearity (correlation among dependent
variables) is not a problem.
4-29
-------
Evaluation of levels of water quality variables (or changes in water quality) associated with
potential ecological effects was based on water quality criteria values (total lead, ammonia, total
phosphorus, nitrate, dissolved oxygen, temperature, conductivity, and turbidity) or on critical effect
levels derived from the literature (color, turbidity, total phosphorus, and conductivity). Individual
states are still in the process of modifying narrative and numeric surface water quality criteria for
application to wetlands. According to guidance provided to individual states by the EPA Office of
Wetlands Protection, initial narrative and numeric water quality standards for wetlands should be
developed or modified using existing information as much as possible, with a longer-term goal of
developing biocriteria for wetlands (U.S. EPA, 1991). Relevant water quality standards associated
with designated (protected) uses for surface waters in the State of Minnesota are listed in table 4-4.
Changes in color, turbidity (suspended solids), and trophic state (total phosphorus [TP]) can
be evaluated based on their effects on wetland transparency and on the potential for successful
growth of submerged macrophyte communities. Relatively few data have been published on light
requirements for submerged aquatic plants. Data have been compiled for sea grasses (Dennison et
al., 1993) and for submerged macrophytes in the littoral zone of northern lakes (Chambers and
Kalff, 1985). Chambers and Kalff report an average minimal light requirement for freshwater
angiosperms in Canadian lakes corresponding to 21.4 ± 2.4 percent of surface light levels.
Corresponding color, turbidity, and chlorophyll a levels, which would reduce light at the bottom of
a type 3, type 4, or type 5 wetland to 21.4 percent of surface illumination, can be calculated (table
4-5).
Calculations were based on the following relationships:
Equation 4-1 (Wetzel, 1975):
Kd = extinction coefficient
= In (Vy x 1/z
= In (1/0.214) x
where \ = incident radiation, \ = radiation at depth z
SeccM depth = 1.65/K,, (Dennison et al., 1993)
Equation 4-2 (Brezonik, 1978):
1/S.D. = 0.106 + 0.128 (turbidity, nephelometric turbidity units [NTU]) + 0.0025 (color,
platinum cobalt units [PCU])
Equation 4-3:
log10(Secchi depth, cm) = 2.07 - 0.13 Iog10(total P, ug/L)
(Derived from data for colored lakes with average depth <2.4 m, in Beaver and Crisman,
1991.)
4-30
-------
Table 4-4. State of Minnesota Water Qualify Criteria for Surface Waters by Designated Use
(U.S. EPA, 1988c)
Water Quality
Variable
Total phosphorus
NO3 + NO2
NH4
Surficial dissolved
oxygen
Surficial water
temperature
Units
UgP/L
mgN/L
mgN/L
mg/L
degC
State of MN* or Other Numeric Water Quality Criteria
Recreation." 40 pg P/L for North Central Hardwood Forest
ecoregion
Consumption. >10 mgN/L
For NH3-N. Fisheries & Recreation. A: >0.016 mg/L; B: >0.04
mg/L
Fisheries & Recreation. 2A: >7 mg/L at all times. 2B, 2C:S5
mg/L at all times
Fisheries & Recreation. A: no material increase, 30°C max.
8B, C: no increase in monthly avg. of max. daily temp. >1.7°C in
Total extractable lead pg/L
Specific conductivity
Turbidity
umhos/cm
NTUd
lakes, 35°C max.
4-day average. >1.3,3.2, or 7.7 pg/L at 50,100,200 mg CaCO3/L
hardness
Agr.& Wildlife. A: >700 mg/L TDSC
Fisheries & Recreation. A: >10; B, C: >25
"Depending on attainable use (Fisheries & Recreation or Agriculture & Wildlife). Class A = associated with
salmonid fisheries, Class B = supporting cool- and warm-water sport or commercial fisheries and associated
aquatic community, Class C = supporting indigenous fish and associated aquatic community.
bBased on attainable lake trophic state for North Central Hardwood Forest ecoregion.
cApprox. 1,094 pmhos/cm. TDS = total dissolved solids.
dNTU = nephelometric turbidity units.
Table 4-5. Water Quality Values Associated With Mean Light Requirements of 21.4
Percent Incident Radiation for Submerged Aquatic Vegetation in Northern
Lakes (Chambers and Kalff, 1985)
Wetland
Type
3
.4
5
Depth Range,
cm
15-60
60-120
120-240
K.m"
2.57
1.28
0.64
Secchi Depth,
meters
0.64
1.29
2.58
Turbidity,
FTUa
11.4
5.2
2.2
Color,
PCU
583
268
113
Total P, ug/ L
107
N/Ab
N/Ab
aFTU = formazin turbidify units.
"Secchi depths outside of range of observations used in deriving equation.
4-31
-------
Minnesota's standard of 40 pg P/L for TP in lakes within the North Central Hardwood
Forest is lower than the estimated requirement of 107 pg P/L to maintain sensitive submerged
aquatic macrophyte communities in type 3 wetlands. The lower standard is ecoregion based and is
designed to minimize the frequency of nuisance algal blooms.
During predisturbance conditions, mean specific conductivity values for surface water (440
pmhos/cm) and ground water (610 pmhos/cm) were near the upper end of the range associated with
freshwater vegetation in the glaciated prairie region. Stewart and Kantrud (1971) list a range of
<40-500 pmhos/cm for normal climatic conditions and a range of <40-700 pmhos/cm for extreme
(drought) conditions. Thus, 700 pmhos/cm was considered a threshold value for specific
conductivity for these wetlands.
Investigators had to apply dissolved oxygen criteria derived for other surface waters to study
area wetlands because of a lack of better literature values. These criteria, however, were probably
overprotective since wetlands in the study area typically contain no fish or fish species extremely
tolerant of low dissolved oxygen (e.g., common carp [Cyprinus carpio], fathead minnow
[Pimephales promelas], and brook stickleback [Culaea inconstans]). However, investigators
believed that the average level and diurnal fluctuations in both dissolved oxygen and temperature
could be critical in determining acceptable spawning habitat or refugia for amphibians.
4.3.4.2. Ecosystem Response Analyses
Investigators could not use MANOVA to test categorical effects with the full complement of
study sites because data matrices were complete for a subset of sites; the power of these tests was
more limited than for regression analyses. However, MANOVAs did demonstrate a significant
fivefold increase in soluble reactive phosphorus (SRP) and a threefold increase in dissolved
phosphorus (DP) at the peak of storm-sewer disturbance activities (after storm sewers were
connected and during watershed construction activity; table 4-6). Threefold increases in SRP and
DP were still evident in storm water-impacted sites at 6 to 12 months and at 12 to 24 months
following peak disturbance. Nitrate levels were strongly elevated, fortyfold in dredged or
impounded sites and eightfold in storm water-impacted sites during the peak of disturbance activity,
but no significant (categorical) increases were observed during subsequent time intervals (table 4-6).
All increases were significant at a probability level (a) of 0.05, some at a probability level of 0.001.
To assess further the long-term impacts of construction and residential development
surrounding wetlands in urbanizing areas, nonparametric Kruskal-Wallis tests were used to compare
different categories of wetlands. Wetlands in nondeveloping watersheds experienced declines in
dissolved nitrogen between predisturbance and recovery periods, while those wetlands in developing
watersheds experienced no change or a slight increase in dissolved nitrogen. Construction activity
in the watershed was associated with increased In (total suspended solids [TSS]) within wetlands in
the second year following disturbance (Detenbeck et al., 1992).
Comparison of water quality changes among wetlands with or without vegetated buffers in
watersheds with or without construction activity showed significant effects (p<0.05) only with
respect to the initial impact period at the peak or immediately following disturbance. Wetlands
4-32
-------
Table 4-6. Summary of Results off MANOVAs Testing for Significant Difference in Water
Quality Change Among Disturbance Classes for Each of Four Time Periods
Average In Ratio (post/pre)a
(geometric mean)
{back-transformed 95% CI)
Variable
Aln soluble reactive
phosphorus
Aln dissolved
phosphorus
Aln nitrate
Aln dissolved
phosphorus
Aln soluble reactive
phosphorus
Aln dissolved
phosphorus
Aln dissolved
phosphorus
Time ;
Period1" N Depth Change Wetland Fill
1 19 -0.038 -O090
1 19 -0.37 -0.081
1 19 3.68d -0.30
(39.6)
(10.4 - 151)
3 16e 0.24 -6732~
4 14e 0.17 -0.44
4 14" 0.11 -0.48
4 21 0.16 -0.39
Storm Water
1.65C
(5.2)
(1.7-16)
1.21*
(3-4)
(1.6-7.2)
2.02d
(7.5)
(2.9 -19.7)
1 15°
(3.2)
(1.3 - 7.3)
1.09°
(3.0)
(1.2 - 7.6)
1.15°
(3.2)
(1.4-7.2)
0.71°.
(2.0)
(.1.1 -4.8)
"Differences in the change in water quality between disturbance classes were tested by Tukey's test to
control for experimentwise error. Categories not significantly different from each other are indicated
by a line. Only variables demonstrating a. significant change in water quality and significant
differences in response among disturbance regime categories are included here.
(Notes continued on next page)
4-33
-------
bl — Peak-disturbance vs. predisturbance.
2 = 0-6 months postdisturbance vs. predisturbance.
3 = 6-12 months postdisturbance vs. predisturbance.
4 » 12-24 months postdisturbance vs. predisturbance.
•Different from zero: p<0.05.
•"Different from zero: p<0.01. .
'Only data for which both dissolved and total constituents were available (surface water samples) were included
in the analysis.
4-34
-------
without vegetated buffers in watersheds with construction activity had greater SRP levels than
either wetlands associated with construction activity but surrounded by vegetated buffers or
wetlands in watersheds without new construction activity. There was no significant difference
between peak or predisturbance SRP for wetlands in watersheds without new construction activity
and those surrounded by vegetated buffers. Nitrate levels followed the same pattern as SRP levels
(Detenbeck et al., 1992).
DP was least in wetlands with no construction activity in the surrounding watershed but did
not show significant differences between buffered and nonbuffered wetlands. Dissolved nitrogen
and water color were greater in nonbuffered wetlands than in watersheds without construction
activity. Longer-term effects of buffers were not detected for growing season averages of water
quality variables in the first year following disturbance (p>0.05).
4.3.4.3. Analyses Relating Measurement and Assessment Endpoints
The impact of physical or hydrologic disturbance on mid-wetland water quality depends on
the ecological significance of the observed magnitude of change. In addition, potential impacts on
biota of downstream surface waters must be considered. The impact of changes on water quality
variables for which numeric water quality criteria exist or threshold values have been derived can
be evaluated by assessing the incidence of criteria or threshold value exceedance.
4.3.4.4. Stressor-Response Profile
Investigators used stepwise multiple regression analysis to assess the effects of the intensity
of physical or hydrologic disturbances on mid-wetland water quality. The change in water quality
between pre- and postdisturbance time periods was the dependent variable. Although numerous
statistically significant relationships were found, this case study reports on a subset focusing on
water quality variables for which threshold values were derived or criteria were available (table
Tmi'i E^0nS W6re reP°rted in gfeater detail by Detenbeck and colleagues in earlier reports
(1991a, 1992).
Storm-water inputs, construction, dredging or impoundment, wetland fill, and increases in
watershed area or area of urban or residential land use had a significant effect on mid-wetland
water quality. Construction, particularly within the buffer zone surrounding wetlands was
correlated with increased concentrations of suspended solids, total lead, and nitrate in wetlands
during the first year following disturbance. Increased urbanization relative to wetland area tended
to increase total nutrient levels and the fraction of nutrients associated with paniculate matter
Paniculate nitrogen and phosphorus tended to increase as the area of wetland fill increased
Storm-water inputs (quantified as an increase in impervious surface area) tended to decrease
dissolved nutrient levels in wetlands, probably by decreasing the water retention time of the
wetlands (Detenbeck et al., 1992; Brown, 1985). When predisturbance mid-wetland water quality
data were compared among sites, wetlands in isolated basins had significantly higher nutrient
dissolved organic carbon, and color values than did wetlands with intermittent or continuous flow
In effect, connecting previously isolated wetlands with a storm-water sewer network can flush
nutrients downstream. Deepening wetlands by impoundment or dredging tended to lessen some
4-35
-------
H3
•*•*
O,
1
CD
3
§
1
ep CS
fiJDi-4
CO C7\
6 %
.5 *-
Ind
Var
« Pi
x"
RB
in
d
i
X
S
C
X
x- +
o
+
oo
X X
tn **
i i
oo c*>
-f 9
II II
Z P Q
£
& 3
X
oo
o
9
II
£2 K;
E >
S
C
i g>
il
j
-
5; S
C-1 •* 00
-
d d d O
oo r-
Xs
x
vq
d
II
EDIFF
X
9
II
£
c-l
_ d
9 9
II II
s
co
do
.S
M
E-a
"SI
,33'
a B<
« jb
il-8
r r^ r*
.
1 I f
a, a.
2, S -»
, j.
i
>£?
4-36
-------
changes in mid-wetland water quality, probably by increasing retention time and sedimentation
efficiency.
Regression analyses also were used to determine the effect of disturbance intensity on
recovery during the second year following disturbance. By that time, neither area of wetland filled
nor percentage of wetland filled nor change in wetland type (water depth) had a discernible effect
on mid-wetland water quality. However, changes in watershed land use relative to either
watershed area or wetland area produced long-term effects on wetland water quality. Increases in
urban or residential land use were associated with increases in dissolved nutrients (SRP DP N
and DOC) and TP and decreases in surface dissolved oxygen. An increase in the percentage of
watershed developed (percentage urban and residential) was associated with a long-term increase in
TP and decrease in surface water temperature. Construction activity in the watershed was
corrected with decreased SRP and increased levels of suspended solids. As relative watershed
area (and flushing rate) increased, dissolved N and P and TP tended to decrease, offsetting some of
the increase due to accelerated nutrient loading.
Comments on Characterization of Ecological Effects
Strengths of the case study include:
• A statistical analysis of ecosystem responses was conducted, allowing estimates
of response along a gradient of disturbance as well as categorical response.
Uncertainties due to Type I errors can be quantified.
Limitations include:
• Expected impacts due to water quality changes were derived in pan from
criteria developed for surface waters other than wetlands.
• Habitat impacts were not measured directly.
• Only static endpoints were used; ecological processes were not measured.
General reviewer comments:
• EPA is developing biocriteria to address ecological effects, but these criteria
are not yet available for wetlands.
• Indices of abiotic ecological quality, e.g., habitat destruction, should be linked
to biota.
4-37
-------
4.3.5. Risk Characterization
43.5.1. Risk Estimation
Investigators did not detect significant changes in mid-wetland specific conductivity,
temperature, dissolved oxygen, or ammonium levels between pre- and postdisturbance periods, nor
were any significant differences in response noted among disturbance categories by MANOVA.
Peak-, postdisturbance-, or recovery-period-specific conductivity levels exceeded the threshold value
of 700 urnhos-cm"1 in three cases, and temperature levels exceeded the absolute criteria for
Minnesota lakes (35°Q in one case (table 4-8). The change in surface water temperature (as
evidenced by the upward shift in minimum, mean, and maximum values between pre- and
postdisturbance conditions) was >1.7°C for most cases, but differences in temperature increase were
not noted among disturbance classes. This level of temperature variability may be natural for
shallow wetland systems that are less resistant to temperature than are lakes.
Surficial dissolved oxygen levels were occasionally below the lower limit criteria for Class
B and C waters for both predisturbance conditions (7 percent) and postdisturbance conditions (23
percent; table 4-8). Under the range of pH values measured for similar wetlands in the
metropolitan area (pH 6-8; Detenbeck et al., 1991a) and the temperature range observed for these
wetlands (7-36°C), approximately 2 to 11 percent of total ammonia plus ammonium would be
present in the toxic (un-ionized) form (Thurston et al., 1974). Levels of total ammonium in
wetlands during the predisturbance period were high enough to exceed Minnesota's water quality
criteria for Class B and C fisheries and recreation surface waters (0.04 mg NH3-N/L; U.S. EPA,
1988a) at approximately 60 percent of sites under the highest pH and temperature conditions
observed. The proportion of sites at potential risk declined over the next 2 years to 5 to 15 percent
following disturbance.
Mid-wetland nitrate levels increased significantly immediately following depth changes due
to impoundment or dredging (39.6x) or immediately following storm-water inputs (7.5x). In no
instance did nitrate levels exceed water quality criteria for drinking water standards; in general,
nitrate levels were 1 to 3 orders of magnitude below the criteria of 10 mg N/L. However, if these
wetlands are nitrogen-limited, a sevenfold to fortyfold increase in nitrate could be expected to
stimulate productivity dramatically.
For water quality variables showing a significant categorical response to disturbance, the
change in risk to wetland water quality or potential water quality function can be expressed as a
frequency of criteria exceedance for pre- and postdisturbance populations. The majority of wetlands
studied had predisturbance turbidity levels exceeding target levels for type 3 wetlands (64 percent
>10 NTU) and type 4 wetlands (21 percent >5 NTU; figure 4-5a). At the peak of disturbance, the
frequency of sites exceeding target levels for type 3 wetlands decreased slightly for wetlands with
no construction activity in the watershed (to 58 percent) and increased slightly for wetlands with
construction activity in the watershed (to 66 percent; figure 4-5a). However, in the first year
following disturbance, average turbidity levels exceeded target levels for type 3 wetlands for 37
percent of sites with no construction activity, and turbidity levels exceeded the target level of 10
NTU for 84 percent of sites exposed to construction activity (figure 4-5b).
4-38
-------
1
Q
1
"§
§
3
_!>
•i
£>
1
o*
&
I
•o
I
1
c
W5
1
O
'S
I
•a
g
i'
00
4
_«
1
ij
's ||
II
"* 1
fl tU
ii
u £
"E
Growing Season Peal
) Disturbance Mean (r«
!
l|
50 "S
fi
u £
I
!>
1
^s
II
1
ill
E^^
ascRpj
N m *H CN)
tn
EeB§
BS 28
-* IS «v> J
F 690 (110-1,712)
0312(42-2,426)
8182(32-511)
M 132 (45-292)
f
OO
fuT
1
«
se
§
11
§m S
333
S S. S.
SS S3
d 0° d d
ffss
dodo
S 3 3 8
ssse
O Ov
oC «' t-: *
F 9.1 (8.4-10.6)
04.5(0.6-8.8)
89.5(2.7-13.6)
M 6.8 (5.0-9.6)
m*
i
5
t
-lie
11?
•
S-S'S-
9 »• m
** CN w
2 d d>
o ; •
t-; a a
•*— ' «**• N^
TT « 0 0
ss?is
i^-s-^?
||li
eeSS
^ oo o e*-
O\ od o t^
f* r~* £*$ f^
F 18.8 (14.5-24.0)
020.4(14.0-29.3)
S 22.3 (16.2-36)
M 17. 1(10. 8-21)
^ ' •
01
si.
1
o
8
$££
III
&a.;i
S3S $S
I
Il
g«or-.c-
III
4-39
-------
I
HI
LLI
DC
LL
I
1
i
8
PRE-DISTUR 3ANCE
JSTRN
CONST AUCTION
10 15
TURBIDITY, NTU
25
Figure 4-5a. Cumulative frequency distribution of average mid-wetland turbidity values over
the growing season. Predisturbance turbidity distribution (P) is compared to
peak-disturbance distributions for wetlands in watersheds with (C) and without
(NC) construction activity. Proportions of the urban wetland population with
turbidity at or below target levels of 5 or 10 NTU are indicated by the dotted
lines.
4-40
-------
TURBIDITY, NTU
Figure 4-5b. Cumulative frequency distribution of average mid-wetland turbidity values over
the growing season. Predisturbance turbidity distribution (P) is compared to
first-year postdisturbaince distributions for wetlands in watersheds with (C) and
without (NC) construction activity. Proportions of the urban wetland
population with turbidity at or below target levels of 5 or 10 NTU are indicated
by the dotted lines.
4-41
-------
Total extractable lead levels were quite high in the dry predisturbance period, possibly
because of increased availability of lead following oxidation of lead sulfides or increased chelation
by elevated dissolved organic carbon levels. The predisturbance average for surface water was 9 ug
Pb/L, well above the criteria of 7.7 pg/L for chronic toxicity at a hardness level of 200 mg
CaCOa/L. Maximum mid-wetland lead levels declined between the pre- and postdisturbance
periods, possibly due to increased precipitation and a lowering of redox levels. However, the
magnitude of the inter-annual decrease in total lead decreased as a function of construction activity
in the period immediately following storm-water inputs (figure 4-6).
The level of construction activity in the watershed associated with exceedance of the water
quality criteria for chronic toxicity was calculated as a function of initial lead levels. Figure 4-6
shows the level of the In (construction/wetland area) ratio associated with exceedance of the criteria
for chronic toxicity (3.2 ug/L at 100 mg/L CaCO3 hardness) corresponding to the 95th percentile,
75th percentile, and 50th percentile (median) values of predisturbance lead levels. Four of six sites
had construction activity greater than that associated with criteria exceedance for the upper 5
percent of predisturbance values, three sites had levels associated with criteria exceedance for the
upper 25 percent of predisturbance values, and two cases had construction activity (In ratio >2.9,
ratio >18.2) higher than that associated with exceedance for sites in the upper half of the
predisturbance distribution. Similar predictions can be derived for other initial lead distributions.
Predictions are based on mean response (In [post/pre] Pb); the actual response is expected to fall
within the 95 percent confidence interval for the regression.
Mid-wetland TP levels exceeded target levels derived to protect clarity of type 3 wetlands
(107 ppb P) and criteria for Minnesota lakes in the North Central Hardwood Forest ecoregion (40
ppb P) in the majority of cases for predisturbance, peak-disturbance, postdisturbance, and recovery
periods (70 to 80 percent >107 ppb P, 94 to 100 percent >40 ppb P). Based on results of
regression analyses, potential increases in TP for storm water-impacted sites related to urbanization
were offset by increased flushing rate related to increased watershed/wetland area ratios.
Investigators used regression equations for mid-wetiand TP to predict stressor levels
associated with criteria or threshold value exceedance for different percentiles of the population of
predisturbance wetland conditions (figures 4-7a-d). For wetlands associated with the lower 50
percent of predisturbance TP levels, springtime TP levels were expected to exceed target levels of
40 ppb P and 107 ppb P for wetlands experiencing depth changes of <2.4 or <1.4 units, respectively
(figure 4-7a). For the upper 75th percentile of sites, a target level of 40 ppb P could be achieved in
cases of littie or no construction activity (construction/wetiand area <5) and a net decrease in
watershed/wetland area (-13 to -30; figure 4-7b). The lower 50th percentile of cases were expected
to remain below the target level of 107 ppb P only for cases of limited construction activity (ratio
of 0 to 8) and no change or a decrease in watershed/wetland area (figure 4-7c). For wetlands
affected by storm water, the lowest 50th percentile could achieve the target level of 40 ppb P
following disturbance for increases in the urbanization ratio of up to 8, but only if relative
watershed size (and flushing rate) was increased proportionately (figure 4-7d).
Color levels were elevated during predisturbance (drought) conditions corresponding to
high dissolved organic carbon levels (Detenbeck et al., 1991a). However, only 2 percent of sites
had color levels exceeding the target of 583 PCU for type 3 wetlands, approximately 20 percent
4-42
-------
Stormwater, Spring, Pre vs. During and Post-Disturbance
Y - -1 .8 + 0.46 ln(Constrn/Wtld area), r2=0.80
Target = 2.6 ppb Pb
2.7
so
i a-
•o
.JT
0.37 $
Q.
I
B-
so
_0.14
0.05
-1 0 1 2 3
Ln(Constrn/Wtid area)
Figure 4-6. Regression line and 95 percent confidence interval for relationship between In
(construction/wetland area) and springtime In (peak+postdisturbance
Pb/predisturbance Pb) for urban wetlands affected by storm-water additions.
Construction activity levels corresponding to protection of wetlands in the 95th
percentile, 75th percentile, and 50th percentile (median) of the full
predisturbance mid-wetland lead distribution are shown. Protection of mid-
wetland water quality here is defined as a target level of <;3.2 jig Pb/L.
4-43
-------
SPRINGTIME MIDWETLAND TOTAL P
DEPTH CHANGE CASES
Q
"O
1-
0.5-
0
TARGET=40ppbP
TARGET=107ppbP
50%
To 5
dWSHD/WTLD
20
Figure 4-7a. Isopleths for mid-wetland TP threshold values of 40 jig P/L or 107 /tg P/L;
predictions for springtime mid-wetland TP, depth change cases. Isopleths
define the combinations of two stressors predicted to yield the target value of
mid-wetland TP based on the 25th percentile, 50th percentile (median), or 75th
percentile of predisturbance TP distributions.
4-44
-------
GROWING SEASON MIDWETLAND TOTAL P
DEPTH CHANGE CASES
TARGET = 40 PPBP
-35
0 -26 -22 -18 -U -iQ T -2 ' £
d WSHDM/TLD
10 14 18 2226
Figure 4-7b. Isopleths for mid-wetland TP threshold values of 40 /*g P/L or 107 /tg P/L-
predictions for growing season mid-wetland TP, depth change cases, target level
of 40 fig P/L. Isopleths define the combinations of two stressors predicted to
yield the target value of mid-wetland TP based on the 25th percentile, 50th
percentile (median), or 75th percentile of predisturbance TP distributions
4-45
-------
GROWING SEASON MIDWETLAND TOTAL P
DEPTH CHANGE CASES
-25;
25%
TARGET =107 ppbP
0 -26 -22 -te -14 --10 -<3 -2 ' £ §
dWSHD/WTLD
26
Figure 4-7c. Isopleths for mid-wetland TP threshold values of 40 ng P/L or 107 fig P/L;
predictions for growing season mid-wetland TP, depth change cases, target level
of 107 fig P/L. Isopleths define the combinations of two stressors predicted to
yield the target value of mid-wetland TP based on the 25th percentile, 50th
percentile (median), or 75th percentile of predisturbance TP distributions.
4-46
-------
GROWING SEASON MIDWETLAND TOTAL P
STORMWATER CASES
15-
10-
-20
0 -26 -22 -18 -14 -10 -§ -2 ' £§ 1'0 1'4 1'8 2!2
d WSHD/WTLD
Figure 4-7d. Isopleths for mid-wetland TP threshold values of 40 jig P/L or 107 jig P/L;
predictions for growing season mid-wetland TP, storm-water cases. Isopleths
define the combinations of two stressors predicted to yield the target value of
mid-wetland TP based on the 25th percentile, 50th percentile (median), or 75th
percentile of predisturbance TP distributions.
4-47
-------
of wetlands had color >268 (target for type 4 wetlands), and 50 percent had color >113 PCU (target
for type 5 wetlands) during the predisturbance period. Color alone probably is not limiting
submerged macrophyte production in these systems. Maximum color levels for wetlands in fill and
storm-water disturbance categories decreased following disturbance activities but increased between
peak-disturbance and recovery periods for dredge or impoundment cases in response to increased
watershed/wetland area ratios. Isopleth plots showing combinations of depth change and change in
watershed/wetland for which target color levels could be achieved show that color levels are
relatively insensitive to changing watershed/wetland area ratios and that the lowest target level for
color is achievable for the lower 50th percentile given a small increase in depth (figure 4-8).
4.3.5.2. Uncertainty
Sources of uncertainty in this risk assessment include both qualitative errors (e.g., errors in
assumptions) and quantitative errors (e.g., measurement or prediction errors). Table 4-9 lists the
main sources of uncertainty in each phase of the risk assessment, along with an estimate of the
magnitude of uncertainties.
Information gaps related to quantifying the total wetland resource in TCMA and the true
frequencies of physical or hydrologic disturbances to wetlands contributed to overestimation and
underestimation, respectively, of the true incidence of anthropogenic disturbance. Information gaps
on direct habitat loss or conversion rates and lack of (tested) water quality criteria specific to
wetlands limited the investigators' ability to create a balanced assessment of impacts to the full
wetland ecosystem as compared with impacts to downstream surface waters.
4
Sources of uncertainty in the empirical field study on impacts and recovery included the
interaction of effects of climatic variability between pre- and postdisturbance periods with effects of
anthropogenic physical or hydrologic disturbance. By using the paired comparison regression
approach, investigators were able to factor out potential additive effects of climatic differences
between years but were not necessarily able to factor out interactive (e.g., multiplicative) effects.
The design of the study would have been improved by including information from paired
comparisons of undisturbed reference sites. Finally, the risk assessment could be improved by a
separate field validation of regression predictions based on a separate set of study sites. ,
4.3.5.3. Risk Description: Summary and Interpretation of Ecological Significance
Table 4-10 compares the risk to urban wetland water quality and water quality improvement
function from physical or hydrologic disturbance to potential loss or conversions of wetland habitat.
While neither dredging nor impoundment activity (water-depth change) caused many significant
long-term changes in mid-wetland water quality, these activities probably had the greatest effect on
wetland habitat. Wetland habitat is permanently removed by wetland fill activity and severely
modified by dredging operations. Although emergent vegetation began to recover at disturbed
wetland sites within 1 year following disturbance, the recovery of submerged vegetation appeared to
be delayed by more than 2 years, especially where organic substrates had been removed (Detenbeck
et al., 1992). Similarly, storm-water additions create a significant long-term shift in hydrologic
regime, which may affect vegetation succession patterns and spawning habitat for amphibians.
4-48
-------
0.2
0.1-
o-
LU
-0.2-
-0.3-
-0.4-
•0.5
SPRINGTIME MIDWETUMSID COLOR
DEPTH CHANGE CASES
TARGET=113PCU
TARGET=268 PCU
TARGET=583 PCU
50%
50%
50%
-26 -£2 -18 -U -iO ^6 -2 ' ^ 6 10 14 18 22 26
d WSHD/\AH"LD
Figure 4-8. Isopleths for spring mid-wetland color target levels of 113 PCU, 268 PCU, and
583 PCU, based on predicted response of median predisturbance values and
combinations of two stressors, depth change and change in watershed/wetland
area ratios
4-49
-------
Table 4-9. Uncertainties Affecting Measurement of Risk to Urban Wetland Water Quality
Status and Water Quality Improvement Function Related to Physical or
Hydrologic Disturbance
Phase of Risk Assessment
Level/Measure of Uncertainty
Characterization of Exposure
Total area, number of wetlands in metropolitan area
Incidence of physical or hydrologic disturbances to
metropolitan area wetlands over time
Intensity of physical disturbances to wetlands in
metropolitan region
Conversion factors for wetlands in metropolitan
region
Unknown certainty; no quantitative updated
inventory available
Unknown, especially for unregulated activities
Range, distribution of measured values
Unknown extrapolation error from nationwide trend
analysis
Characterization of Ecological Effects
Selection of threshold values or pertinent water
quality criteria
Unknown certainty: (a) surface water quality criteria
derived for clearwater lakes and streams, not
wetlands; (b) water quality threshold values to
protect transparency based on relationships derived
for colored lakes and macrophyte depth distributions
for relatively clearwater lakes
Estimate of relative risk due to habitat loss vs. water Direct effects of habitat loss or conversion on
quality degradation threatened or endangered species not measured
Measurement extrapolations Temperature and dissolved oxygen min./max. values
not recorded
Precision/accuracy of water quality measurements
Probability of Type I error in identifying significant
changes in water quality
Stressor-response analysis
Loadings to downstream surface waters not directly
measured
Relative error generally < 10 percent
p ^0.05
Type I error ^0.05; uncertainty of predicted
response indicated by regression r2 values, 95
percent confidence intervals
4-50
-------
3-
•8
Ex
l
y5
fii
III
a
0
E?S.
'
A
CO
B
*^
8
'C \/\
°
— ^ o "
8
M
|
O
CL.
S
E
I
o
o
U
3
2
Q
4-51
-------
Although the immediate effects of wetland fill on surface water quality are limited, the
long-term cumulative effect of the loss or conversion of wetland area must be considered in
determining risk to aquatic resources in this region. Earlier studies demonstrated a relationship
between the extent of wetlands and low total lead or high color in downstream lakes, and between
proximal wetlands and lowered trophic status in downstream lakes, or lowered suspended solids,
fecal coliform, nitrate, and flow-weighted NH4 or TP in streams of the TCMA region (Johnston et
al., 1990; Detenbeck et al., 1991b, 1993).
Given the high level of total extractable lead in wetlands during the predisturbance period,
any increase in lead would be considered detrimental to both wetland biota and biota of
downstream surface waters. However, in the absence of disturbance activity, average total lead
levels were predicted to decrease by 84 percent due to interannual climatic variation alone to levels
just above detection limits. There is a high degree of uncertainty as to the actual impact of total
extractable lead in wetland systems for two reasons. Surface water quality criteria were derived
under standard testing conditions of low dissolved oxygen content, which may affect the availability
of lead to biota. Second, much of the lead trapped in wetlands is associated with paniculate
matter, so that sediment concentrations and the potential for bioaccumulation need to be assessed
(Stockdale, 1991).
Long-term categorical impacts on mid-wetland water quality were observed in response to
construction activity and storm-water inputs (increased total and volatile suspended solids) or in
response to residential development in the watershed (increased dissolved nitrogen). The
proportion of wetlands with turbidity greater than identified thresholds for protection of submerged
macrophyte communities increased over the first year following construction activity. The
ecological significance of increased dissolved nitrogen in these systems is unknown at this point but
could be very important if this change is an indicator of disruption of nitrogen cycling (Detenbeck
etal., 1992).
Changes in land use (residential and urban development) and watershed area relative to
wetland area were associated with statistically significant impacts on nutrients and water color in
the first and second years following disturbance. However, it is clear that the trophic status of
these wetlands is high due to prior loading. For fully or partially impounded wetlands, cumulative
effects of wetland eutrophication may occur over time as loadings continue, but longer term studies
are needed to assess these effects (Kadlec, 1985). Increased loadings of SRP or TP to wetlands
converted from isolated potholes to components of storm water networks that experience
intermittent or continuous flow probably create greater risks to downstream surface waters than to
the wetlands themselves. The inverse relationship between watershed/wetland area and mid-
wetland phosphorus concentrations for storm-water wetlands suggests that increased nutrient loads
are being flushed downstream (Detenbeck et al., 1992). Given the high proportion of eutrophic
and phosphorus-limited lakes in the TCMA, any additional inputs of phosphorus to downstream
lakes are likely to be detrimental to these systems (Metropolitan Council, 1981).
Best management practices, such as the use of vegetated buffers, were only partially
protective of mid-wetland water quality. Storm water represents a point-source input and is not
filtered by vegetated zones surrounding wetlands. Vegetated buffers were associated with lower
4-52
-------
SRP and nitrate in wetlands with construction activity in the surrounding watershed, but this
moderating effect was only temporary.
Comments on Risk Characterization
Strengths of the case study include:
Risk to wetland water quality is described both as a function of initial conditions
(predisturbance water quality values) and as a junction of the intensity of
disturbance. Aspects of both temporal and spatial variability are addressed as
they affect uncertainty estimates in risk analysis.
A key feature of this case study is its predictive component: a stress-response
tool developed as an empirical statistical model Additional discussion is
needed, however, regarding the representativeness of this data set for
application to others.
Limitations include:
Although quantitative estimates are provided for some elements of uncertainty
(e.g., probability of Type I errors, experimental error values expressed as
percent variance explained in regression analyses), most of the descriptions of
uncertainty are qualitative. A rigorous quantitative analysis of overall
uncertainty is not possible given the level of available information.
A discussion of the larger issues associated with wetland assessment (e.g.,
landscape and wildlife aspects) is missing and could be included as a "lessons
learned" section. .
Effects on organisms, especially mammals, are not discussed.
The focus is on water quality impacts, while habitat destruction is glossed over.
The potential forecasting use of the case study was not portrayed clearly and
should be emphasized. Whether the study area wetlands are typical of those
found in the area should be noted. Empirical models can be misused if
differences between the study area and a new area are not understood.
General reviewer comment:
With regard to mitigation, it is necessary to realize that virtually all wetlands
were previously impacted, thus rendering it much less likely that perturbations
of the kind reported here will result in farther extinctions.
4-53
-------
4.4. REFERENCES
Adamus, P. (1989) Wetlands and water quality: EPA's research and monitoring implementation
plan for the years 1989-1994. Environmental Research Laboratory, Office of Research and
Development, U.S. Environmental Protection Agency, Corvallis, OR.
Anderson, J.P.; Craig, W.J. (1984) Growing energy crops on Minnesota's wetlands: the land use
perspective. Center for Urban and Regional Affairs, University of Minnesota, Minneapolis,
MN.
Ayers, M.A.; Brown, R.G.; Oberts, G.L. (1985) Runoff and chemical loading in small
watersheds in the Twin Cities metropolitan area, Minnesota. U.S. Geological Survey,
Water Resources Investigation Report 85-4122, St. Paul, MN.
Beaver, J.R.; Crisman, T.L. (1991) Importance of latitude and organic color on phytoplankton
productivity in Florida lakes. Can. J. Fish. Aquat. Sci. 48(7): 1145-1150.
Brezonik, P.L. (1978) Effect of organic color and turbidity on Secchi disc transparency. /. Fish.
Res. Board Can. 35:1410-1416.
Brown, R.G. (1984) Atmospheric deposition of selected chemicals and their effect on nonpoint-
source pollution in the Twin Cities metropolitan area, Minnesota. U.S. Geological Survey,
Water Resources Investigation Report 83-4195, St. Paul, MN.
Brown, R.G. (1985) Effects of wetlands on quality of runoff entering lakes in the Twin Cities
metropolitan area, Minnesota. U.S. Geological Survey, Water Resources Investigation
Report 85-4170, St. Paul, MN.
Chambers, P.A.; Kalff, J. (1985) Depth distribution and biomass of submersed aquatic macrophyte
communities in relation to Secchi depth. Can. J. Fish. Aquat. Sci. 42:701-709.
Cowardin, L.M.; Carter, V.; Golet, F.C.; LaRoe, E.T. (1979) Classification of wetlands and
deepwater habitats of the United States. Biological Services Program, U.S. Fish and
Wildlife Service. FWS/OBS-79-31.
Dahl, T.E.; Johnson, C.E. (1991) Status and trends of wetlands in the conterminous United States,
mid-1970's to mid-1980's. U.S. Department of the Interior, Fish and Wildlife Service,
Washington, DC.
Dennison, W.C.; Orth, R.J.; Moore, K.A.; Stevenson, J.C.; Carter, V.; Kollar, S.; Bergstrom,
P.W.; Batiuk, R.A. (1993) Assessing water quality with submersed aquatic vegetation.
Bioscience 43(2): 86-94.
4-54
-------
Detenbeck, N.E.; Johnston, C.A.; Taylor, D.L.; Lima, A.; Hagley, C.A.; Bamford, S. (1991a)
Effects of disturbance on water quality junctions of wetlands: final report to the U.S. EPA.
Environmental Research Laboratory-Duluth. Natural Resources Research Institute,
University of Minnesota and AScI Corporation, Duluth, MN.
Detenbeck, N.E.; Johnston, C.A.; Niemi, G.J. (1991b) Use of a geographic information system to
assess the effect of wetlands on lake water quality in the Minneapolis/St. Paul metropolitan
area. In: Proc. of the MNLake Management Conference. Brainerd, MN, October 7-9
1990, pp. 81-85.
Detenbeck, N.E.; Taylor, D.L.; Lima, A. (1992) Assessing recovery of freshwater wetland water
quality from disturbance. Final report to U.S. EPA, Environmental Research Laboratory
Duluth, MN.
Detenbeck, N.E.; Johnston, C.A.; Niemi, G.J. (1993) Wetland effects on lake water quality in the
Minneapolis/St. Paul metropolitan area. Landscape Ecol. 8:39-61.
Eggers, S.D.; Reed, D.M. (1987) Wetland plants and plant communities of Minnesota and
Wisconsin. U.S. Army Corps of Engineers, St. Paul District.
Galinato, M. (1985) Seed germination studies of dominant wetland species of the Delta Marsh.
M.S. thesis, Dept. of Botany, Iowa State University, Ames, IA.
Gosselink, J.G.; Lee, L.C. (1989) Cumulative impact assessment in bottomland hardwood forests.
Wetlands, vol. 9 (special issue), 174 pp.
Hagley, C.A.; Taylor, D.L. (1991) An approach for evaluating numeric water quality criteria for
wetlands protection. Report to the U.S. EPA, Environmental Research Laboratory,
Duluth, MN.
Heiskary, S.A.; Wilson, C.B. (1990) Regional patterns in lake water quality in Minnesota. Lake
Line 10(6):26-30.
Johnston, C.A.; Detenbeck, N.E.; Niemi, G.J. (1990) The cumulative effect of wetlands on stream
water quality and quantity: a landscape approach. Biogeochemistry 10:105-141.
Kadlec, R.H. (1985) Aging phenomena in wastewater wetlands. In: Godfrey, P.J.; Kaynor, E.R.;
Pelczarski, S.; Benforado, J., eds. Ecological considerations in wetlands treatment of
municipal wastewaters. Chapter 23. New York, NY: Van Nostrand Reinhold Company.
Kadlec, J.A.; Wentz, W.A. (1974) State-of-the-art survey and evaluation of marsh plant
establishment techniques: induced and natural. Volume I: Report of research. U.S. Army
Corps of Engineers Waterways Experiment Station. Vicksburg, MS.
4-55
-------
Leibowitz, S.; Preston, E.M.; Arnaut, L.Y.; Detenbeck, N.E.; Hagley, C.A.; Kentula, M.E.;
Olson, R.K.; Sanville, W.D.; Sumner, R.R. (1992) Wetland research plan FY92-96: an
integrated risk-based approach. U.S. EPA, Environmental Research Laboratory, Corvallis,
OR. EPA/600/R-92/060.
Leslie, M.; Clark, E.H. II. (1990) Perspectives on wetlands loss and alterations. In: Bingham, G.;
Clark, E.H. II; Haygood, L.V.; Leslie, M., eds. Issues in wetlands protection: background
papers prepared for the National Wetlands Policy Forum. Chapter 1. Washington, DC: The
Conservation Foundation.
Metropolitan Council. (1981) A 1980 study of the water quality of 60 lakes in the Twin Cities
metropolitan area. Publication 01-81-047. Metropolitan Council, St. Paul, MN.
Minnesota Department of Natural Resources (MN DNR). (1967) Metropolitan lake inventory.
Bulletin 45. Minnesota Dept. of Natural Resources, St. Paul, MN.
Minnesota Department of Natural Resources (MN DNR). (1984) Minnesota official list of
endangered, threatened, and special concern plants and animals. Minnesota Dept. of
Natural Resources, St. Paul, MN.
Minnesota Pollution Control Agency (MPCA). (1990) Minnesota water quality: water years 1988-
1989. 1990 Report to Congress. Minnesota Pollution Control Agency, St. Paul, MN.
Niering, W.A. (1985) The Audubon Society nature guides: wetlands. New York, NY: Alfred A.
Knopf, Inc.
Omernik, J.M. (1986) Ecoregions of the conterminous United States map (scale 1:7,500,000). U.S.
Environmental Protection Agency, Environmental Research Laboratory, Corvallis, OR.
Osborne, L.L.; Wiley, M.J. (1988) Empirical relationships between land use/cover and stream
water quality in an agricultural watershed. /. Environ. Mgmt. 26:9-27.
Owens, T.; Meyer, M. (1978) A wetlands survey of the Twin Cities 7-county metropolitan area-
east half. IAFHE FSL Research Report 78-2. University of Minnesota, St. Paul, MN.
Rushton, B. (1991) Variation in field parameters in a native wetland used for storm water
treatment. In: Rushton, B.; Cunningham, J.; Dye, C., eds. Statewide stormwater
management workshop, Octobers, 1991. Brooksville, FL: South Florida Water
Management District, pp. 75-84.
Shaw, S.P.; Fredine, C.G. (1956) Wetlands of the United States. Circular 39. U.S. Dept. of the
Interior, Fish and Wildlife Service. Washington, DC: U.S. Government Printing Office,
1971 (reissue).
Sokal, R.R.; Rohlf, F.J. (1981) Biometry: the principles and practice of statistics in biological
research. 2nd ed. San Francisco, CA: W.H. Freeman and Company.
4-56
-------
Stewart, R.E.; Kantrud, H.A. (1971) Classification of natural ponds, and lakes in the glaciated
prairie region. Resource Publication 92. U.S. Dept. of the Interior, Fish and Wildlife
Service. 57 pp.
Stockdale, E.G. (1991) Freshwater wetlands, urban stormwater, and nonpoint source pollution
control.- a literature review and annotated bibliography. Washington State Department of
Ecology.
Taylor, D.L.; Detenbeck, N.E.; Lima, A. (1992) Patterns of wetland disturbance in the
Minneapolis-St. Paul metropolitan area: implications for policy makers. Unpublished
manuscript.
Thurston, R.V.; Russo, R.C.; Emerson, K. (1974) Aqueous ammonia equilibrium calculations
Technical Report Number 74-1. Fisheries Bioassay Laboratory, Montana State University
Bozeman, MT. 18 pp.
Tiner, R.W., Jr. (1984) Wetlands of the United States: current status and recent trends. U S Dept
of the Interior, Fish and Wildlife Service, Washington, DC.
U.S. Army Corps of Engineers. (1992) Public Notice, 14 February 1992. U.S Army Corps of
Engineers, St. Paul District.
U.S. Environmental Protection Agency. (1986) Hazard evaluation division standard evaluation
procedure: ecological risk assessment. Office of Pesticide Programs, Washington, DC.
~
U.S. Environmental Protection Agency. (1988a) Nitrogen-ammonia/nitrate/nitrite: water quality
standards criteria summaries: a compilation of state/federal criteria. EPA 440/5-88/029.
U.S. Environmental Protection Agency. (1988b) State water quality standards summaries. Office
of Water Regulations and Standards, Washington, DC. EPA 440/5-88/031.
U.S. Environmental Protection Agency. (1991) National guidance: water quality standards for
wetlands. Office of Water Regulations and Standards, Office of Wetlands Protection
Washington, DC. EPA 440/S-90-011.
U.S. Environmental Protection Agency. (1992) Framework for ecological risk assessment. Risk
Assessment Forum, Washington, DC. EPA/630/R-92/001.
U.S. Weather Service. (1991) Minneapolis-St. Paul Airport, MN. Personal communication.
Walker, W. (1987) Phosphorus removal by urban runoff detention basins. Lake and Reservoir
Management 3:314-326.
4-57
-------
Werth, J.; Meyer, M.; Brooks, K. (1977) A wetlands survey of the Twin Cities 7-cowty
metropolitan area-west half. IAFHE RSL Research Report 77-10. University of Minnesota,
St. Paul, MN.
Wetzel, R.G. (1975) Limnology. Philadelphia, PA: W.B. Saunders Company.
4-58
-------
SECTION FIVE
ECOLOGICAL RISK ASSESSMENT CASE STUDY:
THE ROLE OF MONITORING IN ECOLOGICAL RISK ASSESSMENT:
AN EMAP EXAMPLE
-------
AUTHORS AND REVIEWERS
AUTHORS
John H. Gentile
Environmental Research Laboratory - Narragansett
U.S. Environmental Protection Agency
Narragansett, RI
K. John Scott
Science Applications International Corporation
Narragansett, RI
John F. Paul
Environmental Research Laboratory - Narragansett
U.S. Environmental Protection Agency
Narragansett, RI
Rick A. Linthurst
Atmospheric Research and Exposure Assessment Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC
REVIEWERS
Robert J. Huggett (Lead Reviewer)
Virginia Institute of Marine Science
The College of William and Mary
Gloucester Point, VA
Gregory R. Biddinger
Exxon Biomedical Sciences, Inc.
East Millstone, NJ
i>
Joel S. Brown
University of Illinois at Chicago
Chicago, IL
Richard E. Purdy
Environmental Laboratory
3-M Company
St. Paul, MN
Frieda B. Taub
School of Fisheries
University of Washington
Seattle, WA
Richard Weigert
Department of Zoology
University of Georgia
Athens, GA
5-2
-------
CONTENTS
ABSTRACT 5-6
5.1. RISK ASSESSMENT APPROACH . . . . •'.'. . . . 5-8
5.2. STATUTORY AND REGULATORY BACKGROUND . 5-8
5.3. CASE STUDY DESCRIPTION 5-10
5.3.1. Problem Formulation ,. 5-12
5.3.1.1. Background 5-12
5.3.1.2. Site Description 5-12
5.3.1.3. Ecosystem Classification . . . 5-13
5.3.1.4. Sampling Design . . . . . 5-13
5.3.1.5. Ecological Indicators , 5-14
5.3.2. Conceptual Model Development 5-17
5.3.2.1. Ecological Effects . 5-17
5.3.2.2. Exposure 5-21
5.3.2.3. Exposure-Response Associations 5-24
5.3.2.4. Estuarine Class Conceptual Models 5-26
5.3.2.5. Problem Formulation Summary . 5-28
5.3.3. EMAP and Regional Risk Assessments . . . . . . 5-30
5.3.3.1. Regional Risk Assessment: Problem Formulation 5-31
5.3.3.2. Regional Risk Assessment: Analysis Phase 5-31
5.3.3.3. Regional Risk Assessment: Risk Characterization 5-32
5.4. REFERENCES 5-37
5-3
-------
LIST OF FIGURES
Figure 5-1. Structure of assessment for EMAP Virginian Province 5-9
Figure 5-2. EMAP Virginian Province base stations for 1990-1991 5-11
Figure 5-3. Contribution of EMAP data to problem formulation 5-18
Figure 5-4. Summary of Virginian Province response and exposure indicator values for the
entire province, large tidal rivers, large estuaries, and small estuarine systems
for 1990 and 1991 data, individually and combined 5-22
Figure 5-5. EMAP estuaries indicator relationships, 1990 and 1991 base stations 5-29
LIST OF TABLES
Table 5-1. Summary of EMAP Response and Exposure Indicator Data for 1990-1991 . . . 5-19
Table 5-2. Indicator Associations: The Percent Area of Degraded Benthos Co-occurring
With Low Dissolved Oxygen and Sediment Toxicity for EMAP Base Stations
for 1990-1991 5'25
COMMENT BOX
Comments on Problem Formulation, Conceptual Model
Development, and Regional Risk Assessments 5-33
5-4
-------
LIST OF ACRONYMS
EMAP Environmental Monitoring and Assessment Program
EPA Environmental Protection Agency
ER-M Effects Range-Median
GIS geographic information system
HEP Harbor Estuary Program
NEP National Estuary Program
NOAA National Oceanic and Atmospheric Administration
NPDES National Pollution Discharge Elimination System
NRC National Research Council
NS&T National Status and Trends Program
ORD Office of Research and Development
OTA Office of Technology Assessment
R-EMAP Regional EMAP
5-5
-------
ABSTRACT
Using data collected from the Environmental Monitoring and Assessment Program's
(EMAP's) Near Coastal program in the Virginian Biogeographic Province during July through
September 1990-1991, this paper describes the role and specific contributions of monitoring data in
the ecological risk assessment process. This case study suggests that EMAP monitoring data can:
• contribute to the problem formulation phase of an ecological risk assessment;
• characterize areal and spatial extent of ecological resources;
« identify regional resources potentially at risk (e.g., degraded benthos); and
• provide initial information on the role of exposure and habitat characteristics.
EMAP data were collected using a systematic probability-based sampling design that
facilitates detection of spatially distributed patterns but does not estimate intra-annual variability or
short-term episodic events. The EMAP information was then used to develop a conceptual model
that described the areal extent of ecological resources at risk, their spatial distribution, and
associated exposure and habitat information. The assessment endpoint was benthic community
integrity. Resource condition, measured using a province-wide benthic index, was operationally
defined in terms of one or more anthropogenic stressors. Currently, resource condition does not
discriminate anthropogenic from natural physical stress.
In this case study, large estuaries exhibited the lowest areal extent of degraded benthos,
16 ±7 percent; low dissolved oxygen was the exposure indicator most closely associated with
degradation. In small estuarine systems, 24±10 percent of the area exhibited degraded benthic
condition, nearly half (48 percent) of which was associated with sediment toxicity. For large tidal
rivers, 41 ±24 percent of the sampled area was degraded, and 45 percent of this degradation co-
occurred with low dissolved oxygen. Co-occurrence of degradation and low dissolved oxygen was
confined to the mouths of the Potomac and Rappahannock Rivers. Although these associations
imply neither causality nor direct anthropogenic stress, they could, along with other evidence, be
used to direct further study. In this regard, on a province basis more than half of the area of
degraded benthos was not associated with any of the exposure indicators discussed.
Data on spatial distribution indicated that degradation of benthic resources occurred mainly
in the upper Chesapeake Bay, the oligo-mesohaline portions of the five tidal river systems (e.g.,
Hudson-Raritan), and the associated small bays. These bays are areas of intense demographic
pressure and extensive urban development.
Although useful in identifying regional areas of concern, EMAP province-scale data are not
sufficient for conducting a complete risk assessment at the regional scale. Where local monitoring
data are too heterogeneous (relative to spatial, temporal, and ecological scales and methodologies)
to be usable in regional ecological risk assessments, investigators may need to acquire additional
data through:
5-6
-------
• an appropriately scaled monitoring program employing a random sampling design,
such as the Regional EMAP [R-EMAP] program in EPA Region II;
• selection of the appropriate response, exposure, and habitat indicators to
characterize the spatial extent of ecological problems and associated exposures; and
• incorporation of extant data (e.g., National Oceanic and Atmospheric
Administration's [NOAA's] National Status and Trends [NS&T] Program, National
Estuary Program [NEP], states, etc.).
Monitoring data alone cannot establish the causal relationships necessary to develop a
complete analysis of ecological risk. Therefore, ecological risk assessments should include
laboratory exposure-response information (e.g., ecotoxicity), effects of multiple stressors, and
measures of contaminant bioavailability to provide evidence for postulating potential causes of risk
to the region or to specific watersheds. Risks to specific watersheds can be examined initially by
using geographic information system (GIS) and landscape methods that describe the spatial
relationships and distribution of response, exposure, and habitat indicators (stressor-specific,
whenever possible). This information can then be overlaid with landscape information on
anthropogenic stressors and hydrologic features (e.g., transport and fate) in the surrounding
watershed. Establishing causal relationships between sources and effects provides the basis for •
instituting appropriate control strategies. Ongoing local compliance (e.g., National Pollution
Discharge Elimination System [NPDES], states, municipalities) and watershed assessment (e.g.,
R-EMAP, EMAP, NS&T, NEP) monitoring programs can evaluate the effectiveness of the control
strategy.
5-7
-------
5.1. RISK ASSESSMENT APPROACH
The U.S. Environmental Protection Agency's (EPA's) implementation of a risk-based
assessment, monitoring, and decision-making strategy requires the integration of the Office of
Research and Development's (ORD's) ecological risk assessment framework (U.S. EPA, 1992);
research, monitoring, and assessment programs under ORD's Environmental Monitoring and
Assessment Program (EMAP); and ORD's ecological risk assessment research programs.
Successfully implementing a risk-based approach for decision making for adoption throughout EPA
requires the integration of these three programs. The framework and process for conducting
ecological risk assessment must not be separated from monitoring programs responsible for data
acquisition and verification nor from research programs responsible for developing the needed
methods and models. The combination of these programs provides the template for all ecological
risk research, irrespective of specific programmatic applications, while ensuring that EPA can
respond directly to the full spectrum of ecological risk assessment needs.
This case study illustrates the roles and contributions of EMAP's Near Coastal Program to
the ecological risk assessment process as described by EPA's Framework for Ecological Risk
Assessment. The case study also examines the use of monitoring data to identify potential problems
for estuarine resources and the potential use of biogeographic province-scale information in
regional assessments. Since EMAP and other monitoring programs typically are not designed to
generate all the information required for a complete ecological risk assessment, this paper focuses
specifically on the use of monitoring data (e.g., EMAP Virginian Biogeographic Province data
from 1990 to 1991) in the problem formulation stage of the risk assessment process (figure 5-1).
The areal extent and spatial patterns of ecological resources for the Virginian Province identify
specific regional areas potentially at risk. The case study uses the Hudson-Raritan estuary as an
example to illustrate the types of information needed for a complete ecological risk assessment.
5.2. STATUTORY AND REGULATORY BACKGROUND
The EPA, U.S. Congress, and private environmental organizations have long recognized
the need to improve our ability to document the condition of our environment and specifically our
ecological resources (National Research Council [NRC], 1990). Federal, state, and local agencies;
waste dischargers; and researchers all conduct marine environmental monitoring. Five federal
agencies conduct environmental quality monitoring activities in the coastal ocean. Each agency's
programs focus on different spatial scales, ranging from effluent discharges from individual sources
(e.g., EPA's National Pollutant Discharge and Elimination System [NPDES] Program) to
measuring far-field, long-term effects of discharges from multiple sources (e.g., the National
Oceanic and Atmospheric Administration's [NOAA's] National Status and Trends [NS&T]
program, EPA's National Estuary Program [NEP]). However, these programs do not, either
individually or taken together, constitute a comprehensive national status and trends monitoring
program focused on contributing information for identifying the potential risks to coastal
environmental resources (NRC, 1990). Congressional hearings on the Monitoring Improvement
Act in 1984 (U.S. House of Representatives, 1984) concluded that, despite considerable
expenditures on monitoring, federal agencies could assess neither the status of ecological resources
nor the overall progress toward legally mandated goals of mitigating or preventing adverse
ecological effects. In 1988, the EPA Science Advisory Board (U.S. EPA, 1988), affirming the
5-8
-------
PROBLEM FORMULATION
Stressors: Low dissolved oxygen levels, metals, and organic chemicals in
sediments
Ecosystem(s) at Risk: Large and small estuaries and large tidal rivers in the
Virginian Biogeographic Province
Ecological Components: Benthic macroinvertebrates
Endpoints: The assessment endpoint is benthic community integrity;
measurement endpoints include the five indicators of benthic community
structure that best distinguished between degraded and reference sites.
ANALYSIS and RISK CHARACTERIZATION
EMAP province-scale monitoring data alone cannot be used to complete the
analysis and risk characterization phases of a risk assessment. However data
needs and approaches for conducting such regional/watershed risk
assessments are discussed.
Figure 5-1. Structure of assessment for EMAP Virginian Province
5-9
-------
existence of major gaps in environmental data and recognizing the broad base of support for better
environmental monitoring, recommended that EPA initiate a program to monitor ecological status
and trends of the nation's ecological resources. EMAP is EPA's response to these
recommendations. This case study illustrates EMAP's contribution to the risk-based assessment
framework that is the cornerstone of EPA's decision-making process.
5.3. CASE STUDY DESCRIPTION
This case study describes the role and contribution of monitoring data in the ecological risk
assessment process. The EMAP response, exposure, and habitat indicator data presented in this
case study were collected from the estuarine waters of the Virginian Biogeographic Province,
which extends from Cape Cod, Massachusetts, to Cape Henry, Virginia, at the mouth of the
Chesapeake Bay (figure 5-2).
Information from response, exposure, and habitat indicators constitutes the data acquisition
component of ORD's Framework for Ecological Risk Assessment and contributes directly to the
problem formulation stage of the risk assessment process. The monitoring data specifically
contribute to the development of a conceptual model that delineates the spatial, temporal, and
ecological boundaries of the problem; the specific ecosystems and ecological components
potentially at risk; and the potential exposure pathways and co-occurrence with ecosystem
attributes/resources of concern.
This case study analyzed data collected during 1990-1991 by determining the cumulative
percent area (i.e., cumulative distribution function) for each ecological response and exposure
indicator for the entire province and its component resource classes (large estuaries, small estuarine
systems, and large tidal rivers). Because the EMAP sampling design is based upon a 4-year
sampling cycle, the areal estimates based on 2 years of data reported in this case study for the
response and exposure indicators must be viewed as examples of how the data can be used and
should not be construed as the most complete or accurate reflection of the power of the EMAP
sampling design. Since EMAP uses a probability-based design, the results from 2 years of
sampling are likely representative of the remaining 2 years. However, the additional data will
improve the estimates of central tendency, decrease uncertainty, and increase the power to detect
change.
Analyses examined the associations between response and exposure indicators to explore
the potential reasons for the observed changes in ecological condition. The areal extent of resource
change that co-occurred with the exposure indicators was determined for the province as a whole
and for each resource class. Information on exposure-response associations focused attention on
specific regional areas, such as the Delaware Bay and the Hudson-Raritan estuary. For these
areas a full ecological risk assessment-a reiteration of problem formulation, the analysis of causal
relationships, and the characterization of risks-can be conducted if the data warrant. Although
these types of analyses are straightforward, their interpretation deserves discussion.
A typical assumption implicit in interpreting results such as these is that changes in
resource status (e.g., degraded or subnominal condition) result from anthropogenic stress. One
must view such interpretations with caution since these data are not designed to provide definitive
5-10
-------
&
t
I
O
35
I
5-11
-------
information on causality or to separate anthropogenic from natural stressors. Rather they provide a
"weight of evidence" approach, suggesting the direction for additional data acquisition and
research. These data also are the basis for developing testable hypotheses to explain observations
regarding the status of ecological resources. For example, low dissolved oxygen and physical
alterations of habitats may not have anthropogenic origin in certain situations; therefore, they
should not be associated with degradation, as defined by EMAP. For this reason, EMAP primarily
seeks to determine the status of ecological resources. Although a useful and important part of the
program, understanding the reasons for changes in status is secondary.
Although not explicitly part of this case study, a regional scale risk assessment could use
both historical data (e.g., NEP, states, EPA Regions, academia) and new data (e.g., Regional-
EMAP, EMAP) to characterize the magnitude and extent of the problem at the regional scale. In
addition, changes in ecological resources can be coupled to specific stressors. Using geographic
information system (GIS) and landscape methods, these stressors can be linked to potential sources
associated with land-based activities. The overall effectiveness of control strategies applied to point
and nonpoint sources could then be evaluated by both compliance (e.g., NPDES, states,
dischargers) and long-term monitoring programs (e.g., R-EMAP, EMAP, states). This case study
illustrates the application of a risk-based assessment and monitoring strategy that provides direct
and indirect evidence for inferring causal associations between the observed ecological effects and
specific stressors, thus enabling the manager to plan and evaluate remedial control strategy options.
5.3.1. Problem Formulation
5.3.1.1. Background
Problem formulation, the initial phase of the ecological risk assessment process, consists of
the following components: stressor and ecological effects characterization, identification of
ecosystems potentially at risk, selection of assessment and measurement endpoints, and
development of a conceptual model (U.S. EPA, 1992). The conceptual model synthesizes the
information in each of these components to describe the potential stressors and exposure pathways;
their co-occurrence, direct and indirect links with specific ecosystems and assessment endpoints of
concern; the spatial, temporal, and ecological boundaries of the risk assessment; and inferences as
to potential causal associations between stressors and ecological effects. In this case study, the
conceptual model describes (1) the areal extent of degraded benthic resources, (2) the areal extent
of exposure to specific categories of stressors, (3) the relationship between the areal extent of
degraded benthic resources and exposure to categories of stressors, (4) the relative importance of
different stressors in each estuarine ecosystem, and (5) specific regional estuarine systems with
degraded benthic resources that could become the subject of detailed regional risk assessments.
The following sections describe the EMAP indicators and analyze and interpret data for 1990 and
1991 to provide information for the conceptual model.
5.3.1.2. Site Description
The data were collected from the estuarine waters of the Virginian Biogeographic Province,
which extends from Cape Cod, Massachusetts, to Cape Henry, Virginia, at the mouth of the
Chesapeake Bay. Covering approximately 23,573 km2, the province includes several large
5-12
-------
estuarine systems (e.g., Chesapeake Bay, Delaware Bay, and Long Island Sound) as well as a
substantial number of small estuarine systems and large tidal rivers (Holland, 1990). Both the
Labrador Current and the Gulf Stream affect the Virginian Province, which has a
continental/subtropical climate. Estuarine resources vary widely in size, shape, and ecological
characteristics. Many estuaries, like Chesapeake Bay, are large, continuously distributed resources
that consist of expansive regions with a broad variety of habitat types. Other estuaries consist of
relatively discrete resources composed predominantly of one habitat type. For sampling design
purposes, the estuarine waters of the Virginian Province were classified into three categories:
large estuarine systems, large tidal rivers, and small estuarine systems.
5.3.1.3. Ecosystem Classification
Large estuarine systems are defined as systems having surface areas greater than 260 km2
and aspect ratios (length/average width) less than 20. Application of these criteria to the Virginian
Province resulted in the identification of 12 large estuarine systems with a total surface area of
16,096 km2, or 70 percent of the province's estuarine area. Large tidal rivers were defined as
systems having surface areas greater than 260 km2 and aspect ratios greater than 20. These criteria
resulted in the identification of five large tidal rivers—Hudson, Potomac, James, Delaware, and
Rappahannock Rivers—with a total surface area of 2,840 km2, or 13 percent of the total province
area. Small estuarine systems were defined as systems having surface areas less than 260 km2 but
greater than or equal to 2.6 km2. Application of these criteria to the Virginian Province resulted in
the identification of 137 small estuarine systems with a total surface area of 4,279 km2, or 17
percent of the province.
The classification process categorized estuaries into classes (strata) for which a common
sampling design can be used. The process also ensured that selected components of estuarine
resources were sampled sufficiently in different systems. Further, the classification process
facilitated the synthesis and integration of data into assessments for evaluating the effectiveness of
management actions (Holland, 1990).
5.3.1.4. Sampling Design
The EMAP sampling design provides unbiased estimates of the status and trends in
indicators of ecological condition with known confidence. There are four essential features of the
EMAP sampling design as applied to estuaries: regionalization, classification, statistical sampling,
and index period. A regionalization scheme partitions the estuarine and coastal resources of the
United States into geographical areas with similar ecological properties. The classification scheme
defines certain populations of interest (e.g., large estuaries, small estuarine systems, etc.) within
large geographical areas that are functionally similar and can be sampled using a common
approach. The value of the EMAP sampling design is that it is both systematic in areal coverage
yet probabilistic relative to the sampling strategy (Overton et al., 1991). This design, therefore,
can determine areal extent (with confidence intervals) and the spatial patterns of response,
exposure, and habitat indicators irrespective of the characteristics of their statistical distributions.
The statistical sampling provides for the determination of unbiased estimates of the status and
trends of the estuarine ecological resource classes. When fully implemented, EMAP will base its
status assessments on data collected over a 4-year baseline (Holland, 1990). This multiyear cycle
5-13
-------
was chosen to dampen the year-to-year variability resulting from natural phenomena such as
extremely dry or wet years and hurricanes. A consistent, probability-based sampling design is
employed within each EMAP resource group to facilitate future integrated assessments among
EMAP resource groups (e.g., estuaries, surface waters, forests).
Fully characterizing natural seasonal variability or assessing status for all seasons is beyond
the scope of EMAP. Because intra-annual variability is thought generally to exceed interannual
variability, an index period (July to September) was chosen to represent that portion of the year
when the measured parameters are expected to show the maximum response to pollutant stress
(Council and Miller, 1984; Sprague, 1985; Mayer et al., 1989), dissolved oxygen concentrations
are lowest (Holland et al., 1987; U.S. EPA, 1984; Officer et al., 1984), fauna and flora are most
abundant, and within-season variability is expected to be minimal. This sampling design may fail
to detect short-term, episodic events. However, persistent unexplained degradation identified by
EMAP would certainly stimulate additional research in the area of concern. This approach is
consistent with EMAP's goals of determining the long-term status and trends of ecological
resources, with the status and trends then being used as the basis for intensive site-specific research
to understand the reasons for the observed problems.
Sampling sites in the large estuarine class were selected using a randomly placed systematic
grid. The distance between the systematically spaced sampling points on the grid was
approximately 18 km. The grid is an extension of the systematic grid proposed for use by all
EMAP resource groups (Overton et al., 1991). For the Virginian Province, 54 sample sites were
identified for the large estuaries for 1990, and 48 sites in 1991. Sampling sites were limited to
waters >2 meters in depth; as a result of this limitation, investigators were unable to sample ~5
percent of the province area. In all cases, the entire large estuarine resource is sampled each year
during the index period. A linear analogue of the above design was used for sampling site
selection in the large tidal rivers. A systematic linear grid was used to define the spine of the five
large tidal rivers in the Virginian Province. Randomly selected transects were placed along the
spine of the river within sequential 25-km segments, starting at the mouth of the river and ending
at the head of the tide. A total of 49 sample sites were selected for large tidal rivers in the
Virginian Province in 1990 and 1991. The 137 small estuarine systems in the Virginian Province
were randomly sampled from the entire list frame of small systems. They were ordered from
north to south by combining adjacent estuaries into groups of four. One system was selected
randomly from each group without replacement for each sampling year, yielding 62 sample sites
for 1990 and 1991 in the Virginian Province. The location of the sample within each selected
small system was randomly selected. Details of the design can be found in Holland (1990).
5.3.1.5. Ecological Indicators
EMAP defines and uses three types of ecological indicators: response, exposure, and
habitat (Hunsaker and Carpenter, 1990). Ecological response indicators quantify the integrated
response of ecological resources to individual or multiple stressors. Examples include
measurements of the condition of individuals (e.g., frequency of tumors), populations (e.g.,
abundance, biomass), and communities (e.g., species composition, diversity). Because benthic
communities play an important role in estuarine ecosystems (Holland et al., 1987, 1988; Rhoads et
5-14
-------
al., 1978; Pearson and Rosenberg, 1978; Sanders et al., 1980; Boesch and Rosenberg, 1981), this
case study uses the condition of benthic assemblages as its only response indicator.
Characteristics of benthic assemblages have been used to measure and describe ecological
status and trends of marine and estuarine environments for several decades (Sanders, 1956, 1960;
Boesch, 1973; Pearson and Rosenberg, 1978; Holland et al., 1988). This literature has identified a
diverse array of benthic assemblage attributes that can characterize ecological status and trends,
including (1) measurements of biodiversity/species richness, (2) changes in species composition, (3)
changes in the relative abundance or productivity of functional groups, (4) changes in relative
abundance and productivity of "key" species, (5) changes in biomass, and (6) relative size of biota
(Weisbergetal., 1993).
EMAP has operationally defined "degraded" or "subnominal" to classify the status of
benthic resources. Three variables are used to characterize sites as degraded: sediment toxicity,
sediment contaminants, and dissolved oxygen (Weisberg et al., 1993). Fifty-eight different
attributes of benthic assemblages were evaluated and used to develop a "benthic index" to measure
ecological status and trends in the Virginian Province. Of these, 28 benthic measurements differed
significantly between degraded and reference sites and were candidates for the discriminant
analyses that led to the development of a benthic index. While the operational definition of
"degraded," as used by EMAP, assumes the presence of anthropogenic stress, alterations in benthic
communities also can result from naturally occurring physical stresses and low dissolved oxygen.
Since EMAP does not have an exposure indicator for eutrophication or physical stressors, the
current benthic index may not always discriminate between natural and anthropogenic effects. This
limitation suggests a need for additional exposure indicators. Finally, the term "degraded" also
assumes some unique property or characteristic of benthic assemblages when, in fact, stressed
communities reflect changes in successional status.
Using the 1990 data, five benthic measures (proportion of salinity-normalized expected
number of species, number of amphipods, percent of total abundance as bivalves, number of
capitellids, and average weight per individual polychaete) correctly differentiated reference sites
from degraded sites with about 90 percent certainty (Weisberg et al., 1993). This version of the
benthic index was specifically developed for the entire Virginian Province from 1990 data and may
not be applicable outside the province or in other years. However, the important point is not the
specific composition of the current index but rather the process of using discriminant analyses to
identify .combinations of candidate benthic measurements (measurement endpoints) that reliably
distinguish between degraded and reference sites. This approach resulted in the development of a
benthic index for 1991 data in the Louisianian Province that is analogous to the index for the
Virginian Province (Summers et al., 1993).
Exposure indicators are physical, chemical, or biological measurements that quantify
pollutant exposure, habitat degradation, or other causes of degraded ecological condition.
Exposure indicators include direct measurements of contaminant or dissolved oxygen concentration
in the water and sediments, contaminant concentrations in biological tissues, biomarkers, and acute
toxicity of sediments. The Virginian Province study used three types of exposure indicators to
infer changes observed in EMAP response indicators: metals and organic contaminant
concentrations in sediments, sediment toxicity, and bottom dissolved oxygen. Clearly, these are
5-15
-------
not the only exposure indicators that are operative in estuarine systems and potentially responsible
for ecological effects.
Metals and organic chemicals from freshwater inflows and from point and nonpoint sources
concentrate in estuaries and accumulate in bottom sediments (Turekian, 1977; Forstner and
Wittmann, 1981; Schubel and Carter, 1984; Nixon et al., 1986). These bottom sediments often
are contaminated to the point that they represent a threat to humans and ecological components
(Weaver, 1984; Office of Technology Assessment [OTA], 1987; NRC, 1989). While the extent
and magnitude of sediment contamination is only now becoming well described (NRC, 1989), it is
a potentially important exposure indicator.
Whereas chemical measures of contaminant concentrations indicate the potential for
ecological effects, sediment toxicity tests provide an indirect measure of contaminant
bioavailability. A commonly used amphipod sediment toxicity test is well established and has been
employed in a variety of monitoring and testing programs (Swartz, 1987, 1989; Chapman, 1988;
Scott and Redmond, 1989; Scott et al., 1990).
Dissolved oxygen concentration is an important exposure indicator to both pelagic and
benthic marine biota. Low dissolved oxygen is one of the more important factors contributing to
fish and shellfish mortality in estuarine and coastal waters. Prolonged exposure to waters at less
than 60 percent saturation can result in altered behavior, reduced growth, adverse reproductive
effects, and mortality (Reish and Barnard, 1960; Vernberg, 1972). Excessive nutrient input can
bring about low dissolved oxygen by stimulating phytoplankton blooms. Important as this indicator
is to EMAP, its measurement presents special problems because of the wide diurnal and tidal
fluctuations in concentrations. To address this problem, continuous and point sampling techniques
currently are being evaluated (Holland, 1990).
Habitat indicators are physical, chemical, and biological measurements that provide
information about the conditions (e.g., water depth, temperature, sediment characteristics, salinity)
necessary to support ecological processes in the absence of pollutants. In estuaries, salinity and
temperature are among the most dominant factors controlling the distribution of flora and fauna and
the functioning of ecological processes (Remane and Schlieper, 1971). Sediment grain size has a
role in regulating benthic community composition, while organic carbon affects the bioavailability
of contaminants. Water depth itself can influence the temperature regime, salinity distribution, and
dissolved oxygen concentration. These habitat variables are important for normalizing the
responses of the response and exposure indicators and for defining subpopulations (e.g., fine vs.
coarse-grained sediment, low vs. high salinity) for further analysis. In addition, these habitat
indicators can be used to postclassify indicator data for a variety of analyses. For example,
sediment toxicity data could be postclassified according to grain size or total organic carbon, both
of which are known to affect contaminant bioavailability. Grain size also affects benthic
assemblages in that benthos occupying sandy substrates are different from those dominated by silt-
clay. EMAP's Virginian Province 1990 Demonstration Project Report presents discussions and
examples of postclassification (Weisberg et al., 1993).
5-16
-------
5.3.2. Conceptual Model Development
The goal of the problem formulation phase is the development of a conceptual model that
identifies the potential relationships between valued ecosystem attributes (e.g., biotic integrity) and
human or natural attributes, functions!, or activities that are causes for concern (e.g., population
density, deforestation, sea level rise, volcanic eruption). In the initial stages, problem formulation
focuses on defining the two ends of a conceptual model: ecological responses in ecosystems
potentially at risk and exposure to one or more stressors. The conceptual model identifies the
potential exposure pathways by which stressors and ecosystem attributes may be connected to
define the spatial, temporal, and ecological boundaries of the assessment and the ecosystems that
are potentially at risk.
Figure 5-3 illustrates how monitoring data from the Virginian Province contributes to the
components of problem formulation and the development of the conceptual model. As shown, the
assessment endpoint is benthic community integrity; measurement endpoints include five specific
benthic community metrics.
Analysis of areal extent for the status of each indicator represents only the area sampled
during 1990-1991 and is not scaled to the total 4-year area. Presenting annual data provides a
picture of year-to-year variability. In addition, unless otherwise noted, all data are presented as
mean estimates within the bounds of the 95 percent confidence limits. Weisberg et al. (1993)
provide details on these calculations.
5.3.2.1. Ecological Effects
This case study characterizes ecological effects by determining the areal distribution of
degraded benthos using the assessment and measurement endpoints described above. Weisberg et
al. (1993) describe the algorithm and rationale for calculating numerical values for the benthic
index and the numerical cutpoint of <3.4 used to distinguish degraded from reference benthic
condition. Cumulative distribution functions of benthic index values estimated the percent area of
degraded benthos (Weisberg et al., 1993). Benthic index data from 1990 and 1991 were analyzed
individually and then combined for the Virginian Province and for large estuaries, small estuarine
systems, and tidal rivers (table 5-1).
• Virginian Province: The stations sampled in 1990 and 1991 represented 40 percent
of the provincial area. Degraded benthic assemblages occurred in 19+6 percent of
the province for the combined years and for each individual year (with slightly
larger estimates of uncertainty).
• Large Estuaries: The stations sampled in 1990 and 1991 represented 40 percent of
the large estuarine area. The study identified degraded benthic assemblages in
16+7 percent of the sampled area; there was little difference between years 1990
and 1991 (15±10 percent in 1990 vs. 17±10 percent in 1991).
• Small Estuaries: Thirty-nine percent of the area found in small estuarine systems
was sampled in the 2 years. Of this area, 24+10 percent exhibited degraded
5-17
-------
EMAP VIRGINIAN
FRAMEWORK
P
tt
0
a
L
er
M
EXPOSUREJMUCAIQB&
Sediment metals
Sediment organic*
Sediment toxicity
Dissolved oxygen
Add volatile su!fkfes(AVS)
Organic carbon
Grain size-percent silt day
EEEEGISJMD1CAIDBSL
• Spedas abundance
• Spedes composition
• Spedas blomasa
• Large Estuaries
» Small Estuaries
* Large Tidal Rivers
Assessment Enpdnts:
• Benthte community Integrity
Measurement Endpoints:
• Proportion of expected number
of spedss
• Number of amphlpods
• Percent abundance as bivalves
• Number of capJtalllda
• Mean weight per individual
polychaete
• Large Tide! Rivera: Degraded benthos associated with
severe hypoxia. Toxksty/conJaminantion minimal.
• Small Estuaries: Sediment toxidty and chemical exceedences
of EHMhighest and associated with degraded benthos
• Large Estuaries: Low sediment toxidty; localized low
dissolved oxygen associated with degraded benthos
ESTUARINE CLASS AT RISK:
Example: Small Estuarine Systems
REGIONAL AREA OF CONCERN:
Exampje: Hudson-Raritan Estuary
SCALE
Figure 5-3. Contribution of EMAP data to problem formulation
5-18
-------
Table 5-1. Summary of EMAP Response and Exposure Indicator Data for 1990-1991
Estuarine Classes
Indicators
Benthic index (1990-1991)
Number of stations
Sampled area (km2)
% Degraded area (B.I. < 3. 4)
Benthic index 1990
Number of stations
Sampled area (km2)
% Degraded area (B.I. < 3. 4)
Benthic index 1991
Number of stations
Sampled area (km2)
% Degraded area (B.I. < 3.4)
Dissolved oxygen (1990-1991)
Number of stations
Sampled area (km2)
% Degraded area (D.O.<2.0
ppm)
Dissolved oxygen (1990)
Number of stations
Sampled area (km2)
% Degraded area (D.O.<2.0
ppm)
Dissolved oxygen (1991)
Number of stations
Sampled area (km2)
% Degraded area (D.O. <2.0
ppm)
Province
(23,573 km2)
206
9,546
19±6
105
4,931
19+9
101
4,615
19 + 8
198
9,299
6+4
97
4,683
7±6
101
4,616
4±4
Large
(16,889 km2)
96
6,720
16+7
48
3,360
15+10
48
3,360
17+10
94
6,580
5+4
46
3,220
6+7
48
3,360
4±5
Small
(4,875 km2)
61
1,927
24±10
32
1,050
22+17
29
877
25+16
59
1,910
<1.0
30
1,032
<1.0
29
878
1+2
Tidal
(2,602 km2)
49
899
41+24
25
521
57+40
24
378
19+13
45
809
26+26
21
431
37+42
24
378
15 ±27
5-19
-------
Table 5-1. Summary of EMAP Response and Exposure Indicator Data for 1990-1991
(continued)
Estuarine Classes
Indicators
Sediment toxicity (1990-1991)
Number of stations
Sampled area (km2)
% Degraded area (<80%
survival)
Sediment toxicity 1990
Number of stations
Sampled area (km2)
% Degraded area (<80%
survival)
Sediment toxicity 1991
Number of stations
Sampled area (km2)
% Degraded area (<80%
survival)
Sediment chemistry (1990-1991)
Number of stations
Sampled area (km2)
% Degraded area (>ERM)
Sediment chemistry (1990)
Number of stations
Sampled area (km2)
% Degraded area (>ERM)
Sediment chemistry (1991)
Number of stations
Sampled area (km2)
% Degraded area (>ERM)
Province
(23,573 km2)
172
7,832
17+6
84
3,716
10+7
88
4,116
22±10
202
9,450
7
104
4,908
8
98
4,542
6
Large
(16,889 km2)
76
5,320
14±8
34
2,380
3+5
42
2,940
24±13
96
6,720
4
48
3,360
4
48
3,360
4
Small
(4,875 km2)
52
1,661
28+13
26
820
38±25
26
820
19±14
59
1,861
16
332
1,050
23
27
811
8
Tidal
(2,602 km2)
44
852
8+7
24
516
6+11
20
335
10+7
47
869
13
24
498
5
23
371
24
5-20
-------
benthos; again the difference between years was small (22±17 percent in 1990 vs.
25±16 percent in 1991).
• Large Tidal Rivers: The 2-year sampling accounted for 34 percent of the tidal river
area in the province. Forty-one (±24) percent of the sampled area exhibited
degraded benthos. The estimates of degraded condition showed large differences for
the 2 years: 57±40 percent of the area in 1990 was degraded compared with 19±13
percent in 1991.
This case study used the bentbic index to classify the areal extent of degraded benthic
assemblages in the Virginian Province and its component resources classes. Figure 5-4 shows a
pattern of increase in the percent area of degraded benthos across resource categories (1990-1991):
16 percent for the large estuaries, 24 percent for the small estuarine systems, and 41 percent for
tidal rivers. Uncertainty estimates for area! extent of degraded benthos were within 6 percent for
the province, 7 percent for large estuaries, 10 percent for small estuarine systems, and 24 percent
for large tidal rivers.
Although the benthic index used in this case study appears to work well for distinguishing
sites of differing environmental quality, other indices also may be effective. First, covariance
among many of the candidate measurements was high, suggesting that several alternative
combinations could produce comparable results. Second, index development was based on only 33
indicator testing sites that, although representative, did not represent all possible conditions. Third,
the stepwise discriminate analysis may not have included important measurements of the benthic
assemblage. Indicator development needs to be a flexible process: as other studies or the analysis
of large data bases suggest increased confidence in selected measurements, they can be incorporated
into the developing index through forced stepwise discriminate analysis.
5.3.2.2. Exposure
The case study characterized exposure by determining the area! distribution of each
exposure indicator: low dissolved oxygen, sediment toxicity, and metals and organic contaminates
in the sediments. Data for each exposure indicator, collected during the index period (My to
September), were analyzed individually and then combined for the Virginian Province and for large
estuaries, small estuarine systems, and large tidal rivers. Critical values were selected for each
indicator: dissolved oxygen <2 pprn; sediment toxicity £80 percent control survival; and sediment
chemistry values > Effects Range-Median (ER-M) (Long and Morgan, 1990). The case study did
not include estimates of bioavailability based on total organic carbon and acid volatile sulfides plus
simultaneously extractable metals (Di Toro et al., 1991, 1992). Cumulative distribution functions
were used to calculate the percent area for exposure indicator values. Note that it is not the intent
of EMAP to characterize naturally occurring seasonal variability or to assess status for all seasons.
Table 5-1 summarizes data for dissolved oxygen, sediment toxicity, and sediment chemistry for
1990 and 1991 individually and for 1990-1991 combined.
• Virginian Province: Bottom dissolved oxygen concentrations lower than 2.0 ppm
occurred in 6+4 percent of the area of the province. The extent of area affected in
1990 was similar (7±6 percent) to that in 1991 (4±4 percent). Toxic sediments
5-21
-------
05
O -r- O
O O
o> o> o>
O
o
O
00
O
CD
i
o
CM
DC
UJ
A
•••»
8
cp
O T- O
O) O) O)
CD O> O)
O
O
O
00
O
CO
i 'i
O O
•<*• CM
E9JV lU90J9d
5-22
-------
occurred in 17 ±6 percent of the estuarine area, and more estuarine area showed
toxicity in 1991 (22+10 percent) than in 1990 (10+7 percent). Sediment
contaminant concentrations exceeding the ER-M values of Long and Morgan (1990)
were found in 7 percent of the estuarine area sampled over the 2 years; the extent
of area exhibiting exceedances in each year was similar (8 percent in 1990, 6
percent in 1991).
• Large Estuaries: Low bottom dissolved oxygen occurred in 5 ±4 percent of the
sampled area of large estuaries; the 2 years had similar estimates for affected area:
6 ±7 percent in 1990 and 4+5 percent in 1991. The amount of large estuarine area
exhibiting toxic sediments was 14±8 percent. An eightfold difference occurred in
the extent of toxic sediments between 1990 (3+5 percent) and 1991 (24±13
percent). The interannual difference in extent of sediment toxicity was not reflected
in Long and Morgan exceedances in chemical concentrations. The 2-year and
single-year estimates of area affected by contaminant exceedances were all 4
percent.
• Small Estuaries: For small estuaries, the area with low dissolved oxygen did not
exceed 1 percent for the 2-year or either of the single-year samples. Conversely,
toxic sediments were much more prevalent in small estuaries. Twenty-eight (±13)
percent of the area exhibited toxic sediments over the 2 years. Nearly twice the
sampled area was affected by toxic sediments in 1990 (38+25 percent) than in 1991
(19±14 percent). This pattern also was found for the extent of exceedances in
contaminant concentrations, where 23 percent of the area in 1990 exhibited elevated
contaminants compared with only 8 percent of the area in 1991. The estimate for
the affected area in the 2-year composite was 16 percent.
• Large Tidal Rivers: Low dissolved oxygen was most widespread in the large tidal
rivers with 26 ±26 percent of the area exhibiting dissolved oxygen concentrations
< 2 ppm. For 1990, the extent of area with low dissolved oxygen was over two
times the value for 1991 (37±42 percent vs. 15±27 percent). Toxic sediments
occurred in 8±7 percent of the tidal river area over the 2 years; the affected area in
either year did not surpass 10 percent. In contrast to small estuaries, the 1991
value for the percent area having elevated contaminant concentrations exceeded the
1990 value: 24 percent for 1991, as compared with 4 percent for 1990. Overall, 13
percent of the tidal river area was degraded relative to this indicator.
The percent area of low dissolved oxygen (<2 ppm) ranged from 1 percent (13 km2) in the
small estuarine systems to 4 to 6 percent (140 to 211 km2) in large estuaries and 15 to 37 percent
(56 to 159 km2) for large tidal rivers. These data suggest that, based on percent area, low
dissolved oxygen presents a greater problem in tidal rivers than in any other estuarine class (figure
5-4). However, when compared on the basis of absolute area, tidal rivers and large estuaries
appear quite similar. In contrast, the percent area of sediment toxicity was consistently greater in
small systems, 28 percent (465 km2), than in large estuaries, 14 percent (745 km2), or tidal rivers,
8 percent (68 km2). However, when compared on the basis of absolute area, the area of sediment
toxicity was almost twice as extensive in large estuaries than in small systems. The sediment
5-23
-------
chemistry data for 1990-1991 indicate that the small estuarine systems are at the greatest risk.
However, this type of degradation does not show consistent distribution between the 2 years (table
5-2). These data reinforce the need to use the entire 4-year data set to minimize uncertainty in the
description of estuarine condition.
5.3.2.3. Exposure-Response Associations
An important aspect of the conceptual model is the development of qualitative and
quantitative associations or co-occurrences between exposure information and ecological effects
information. Such associations lead to the development of hypotheses that can explain the observed
changes in ecological responses and that can direct analyses in subsequent phases of the framework
and further research. Because of the uncertainty inherent in this stage of the risk assessment
process (Layard and Silvers, 1989), these hypotheses may not indicate causality. In fact, a
definitive statement of causality is not a prerequisite for a risk assessment (U.S. EPA, 1992). Four
areas of interest involve associations of benthic degradation with (1) low dissolved oxygen, (2)
sediment toxicity, (3) both of the exposure indicators, and (4) neither of the exposure indicators.
A separate analysis compares the co-occurrence of degraded benthos with the percent area for one
or more sediment contaminants exceeding the ER-M values of Long and Morgan (1990). Table
5-2 presents these analyses, conducted for the Virginian Province, large estuaries, small estuarine
systems, and tidal rivers.
» Virginian Province: In the Virginian Province, 20 percent of the total area has
degraded benthos. Of the 1,844 km2 with degraded benthic condition, 17 percent
co-occurs with sediment toxicity, 21 percent co-occurs with low dissolved oxygen,
< 1 percent have both, and 62 percent is not associated with either toxicity or low
dissolved oxygen. These data suggest that low dissolved oxygen and sediment
toxicity are almost equally associated with the area of degraded benthos in the
province and together co-occur with 40 percent of the degraded area. The
remaining 60 percent of degraded benthos is not associated with either exposure
indicator. The percent area of degraded benthos that co-occurred with ER-M
exceedances was 16 percent for 1990-1991 combined, 24 percent for 1990, and 7
percent for 1991.
• Large Estuaries: In large estuaries, 16 percent of the total area has degraded
benthos: 7 percent co-occurs with sediment toxicity, 20 percent co-occurs with low
dissolved oxygen, and there is no overlap in co-occurrence with both exposure
indicators. These data indicate that 73 percent of the degraded benthos in large
estuaries results from stressors other than sediment toxicity and low dissolved
oxygen. Low dissolved oxygen did co-occur with degraded benthos in 20 percent
of the area of large estuaries, principally in sections of Chesapeake Bay and Long
Island Sound. The percent of degraded benthos that co-occurred with ER-M
exceedances was 7 percent for 1990-1991.
• Small Estuaries: Small estuarine systems present a somewhat different picture, with
24 percent of their total area exhibiting degraded benthos. Forty-eight percent of
the area with degraded benthos co-occurs with sediment toxicity, 3 percent co-
5-24
-------
1
i.
s?
0
T3
>
"3
Q
1
1
M
'I
|
i
U
1
B
CU a
« S
li ^
fc ^
Q >M
CM CM
° a
oJJ o
^^ CO
B
OJ pQ
JS ^^
.. W
C o
0 eg
•S &^
.25 S
Table 5-2. Indicator Asso
Sediment Toxi
1
5
^
1
s
tS
B
!LJ
^
03
."S
p
•i
1
w
1
CO
S
•g
m
W
4)
S?
2
y^
sZV
£
"^
1
y
-
"O O m
rj- — i O Tf OO
oo o\
-H cJ <-<
O co
tN t- O C-- t^
0 0
-HO C~- O
C^
i—^
CNI ^ V ^o i2
VO
O WO £2 >O
00 -H CO - ON
CO CO *— !• i-t CN
X
§ 1 ! &
fl ^ t-< -»-»
« 5a O S2
o o x c g
^" 1 1 q |
" x + 1 1?
Q | Q 5 .§ gj
Q ' eg Q 1 1 6
i
T3
§
O
S.
o
42-
f
o
1
1
M
o
m
o
a
^
a
J .
I
•8
T3
d
bH
-------
occurs with low dissolved oxygen, 3 percent co-occurs with both, and 47 percent of
the area with degraded benthos is not associated with either low dissolved oxygen
or sediment toxicity. These data illustrate a stronger relationship between sediment
toxicity and degraded benthos in small estuarine systems. The data also suggest
that dissolved oxygen is a less important factor. The percent degraded benthos
associated with ER-M exceedances was 50 percent for 1990-1991 combined, 67
percent for 1990, and 25 percent for 1991.
• Tidal Rivers: Tidal rivers have the highest areal extent of degraded benthos, 40
percent of the total class. In contrast to small estuaries, only 10 percent of the
degraded benthic area co-occurs with sediment toxicity. However, 45 percent of
the degraded benthos co-occurs with low dissolved oxygen, zero percent co-occurs
with both exposure indicators, and the remaining 45 percent of the degraded benthic
area in the tidal rivers is not associated with either exposure indicator. The percent
degraded benthos for the tidal rivers associated with ER-M exceedances was 8
percent for 1990-1991 combined, 5 percent for 1990, and 20 percent for 1991.
The above approach represents one way of conducting analyses for associations. Other
techniques are being explored (see Summers et al., 1993).
5.3.2.4. Estuarine Class Conceptual Models
In this case study, we have used only EMAP Virginian Province monitoring data for
postulating potential risks for each estuarine class. Because only 2 years of data (1990 and 1991)
are available, one must be cautious in their interpretation. The systematic, probabilistic sampling
design includes 4 years of data collection to achieve complete coverage of the province and
estuarine classes. Consequently, the areal estimates reported for both response and exposure
indicators represent examples of how the data can be used and are not complete or accurate
reflections of the power of the EMAP sampling design. However, even though designed around a
4-year sampling cycle, the estimates calculated from 2 years of data are representative of what
would be expected for the whole province after 4 years of sampling. With additional years of data,
the uncertainty will decrease, increasing the power to detect changes in areal extent.
In addition to being a monitoring program, EMAP is also a research program.
Consequently, the choices of both response and exposure indicators must be viewed within the
context of testable hypotheses. For example, data analyzed in this case study suggest that the
algorithm used for the benthic index may require modification. However, since the benthic metrics
(measurement endpoints) represent a consensus of what benthic ecologists deem important,
variations in the index can be evaluated from the existing data bases. In fact, EMAP's indicator
program is examining several other indices (Holland, 1990).
The exposure-response associations examined in this case study do not imply direct
causality. For example, low dissolved oxygen and sediment toxicity are indicators of an aggregate
of stressors from potentially a variety of causes and sources. Likewise, the sediment chemistry
values, which were not normalized for bioavailability, provide only circumstantial evidence for
ecological effects. The indicators used in this case study were never intended to assign causality.
5-26
-------
Rather, they provide preliminary information from a weight-of-evidence perspective. In
conjunction with knowledge of other system properties (e.g., grain size, organic carbon, etc.),
information from a weight-of-evidence perspective can identify potential problems.
The following summarizes our understanding, to date, regarding the potential problems in
the Virginian Province and its three estuarine classes.
• Virginian Province: The assessment endpoint used in these analyses, benthic
integrity/condition, was represented using a benthic index metric designed to
discriminate "degraded" sites from reference sites. The data from 1990-1991
indicated that approximately 19 percent of the benthic area of the province was
degraded according to the criteria established for the benthic index. Data from
exposure indicators show that 6 percent of the province area experienced dissolved
oxygen values <2 ppm, while 15 percent of the province area had toxic sediments.
Seventeen percent of the degraded benthic area co-occurred with sediment toxicity
(<80 percent control survival), while 20 percent co-occurred with low dissolved
oxygen and 62 percent of the degraded benthic area was not associated, with either
indicator.
• Large. Estuaries: For the most part, large estuarine systems are the downstream
repositories of the stressor inputs entering from both the large tidal rivers and small
estuarine systems. Approximately 16 percent of the area of large estuaries in the
Virginian Province (1990-1991) exhibited degraded benthos. Not unexpectedly, the
magnitude of sediment toxicity co-occurring with degraded benthos was only 7
percent. Twenty percent of the area of degraded benthos co-occurred with low
dissolved oxygen. This area was restricted to the main stem of the Chesapeake Bay
north of the Potomac River. In no areas of degraded benthos did low dissolved
oxygen and sediment toxicity co-occur.
• Small Estuaries: The areal extent of degraded benthic communities in small
systems for 1990-1991 was 24 percent. Only 3 percent of the area of the small
estuarine systems with degraded benthos experienced hypoxic stress. In contrast, of
the 24 percent of small estuarine area experiencing degraded benthos, 48 percent
co-occurred with sediment toxicity. Approximately 50 percent of the area of small
estuarine systems experiencing degraded benthos also had one or more sediment
contaminant values exceeding the ER-M. Thus, a close correspondence exists in
the annual patterns of sediment toxicity and sediment chemistry (>ER-M) in the
small estuarine systems.
• Tidal Rivers: Just under one-half of the estuarine area in the large tidal rivers (40
percent) had degraded benthos in 1990-1991. Toxicity and hypoxic stressors rarely
co-occurred at stations in the Virginian Province, including those in the large tidal
river systems. Only 10 percent of the area with degraded benthos co-occurred with
sediment toxicity, which was restricted spatially to the oligohaline headwaters
(<0.5 ppt) of the Rappahannock, Delaware, and Hudson Rivers. In contrast, areas
of low dissolved oxygen (45 percent) occurred primarily in the lower, mesohaline
5-27
-------
portions of the Potomac and Rappahannock Rivers. These data support current
understanding of sediment contaminant distributions in urbanized waterways and of
existing dissolved oxygen problems in the main stem of Chesapeake Bay.
However, none of the five tidal rivers in the Virginian Province have areas of co-
occurrence of both sediment toxicity and low dissolved oxygen.
5.3.2.5. Problem Formulation Summary
Of the three estuarine classes examined in this case study, large estuarine systems exhibited
the lowest percent area of degraded benthos (16+7 percent), followed by the small estuarine
systems (24+10 percent) and the tidal rivers (41+24 percent). Although areal extent of
degradation is important, the spatial pattern (geographic distribution) of resource degradation is
particularly important for identifying specific regional ecosystems at risk (figure 5-5). These data
clearly suggest that much of the degradation of benthic resources is closely associated with the five
tidal river systems and their associated small estuaries. The co-occurrence of exposure information
on sediment chemistry, sediment toxicity, and dissolved oxygen was used to formulate hypotheses
to suggest possible explanations for the observed spatial patterns of degraded benthos. Co-
occurrence of low dissolved oxygen can be postulated as an explanation for 20 percent of the
degraded benthos in large systems, but only for 3 percent of the degraded benthos in small systems
and for more than 45 percent of the degraded benthos in tidal rivers. Conversely, co-occurrence
of sediment toxicity can explain only 7 percent of the degraded benthos in large systems, 10
percent in tidal rivers, and 48 percent in small estuaries.
In addition, since hypoxia and toxicity co-occur infrequently (<5 percent), one might
expect them to represent differing system and source characteristics. For example, toxicity was
more prevalent in the lower salinity portions of these systems (mesohaline and oligohaline) than
was low dissolved oxygen, suggesting a potential association with urban point sources in the upper
reaches of estuaries. Chemistry data on the exceedances of ER-M values for one or more chemical
contaminants support this interpretation. Analysis of these data suggests that toxicity problems
within small estuarine systems are localized in small tidal rivers and small embayments bordered
by heavily industrialized urban areas.
Hypoxia can result from municipal discharges in portions of tidal river systems independent
of industrial discharges or from nutrient enrichment in those small systems deeper and more open
to larger embayments. Poorly flushed small systems with high carbon loads characteristic of
sewage discharges would lead to high sediment oxygen demand and hypoxia. Nonpoint runoff
from agricultural land bordering small estuaries and coastal lagoons also may result in nutrient
enrichment, subsequent algal blooms, and hypoxia. However, numerous and extensive studies
focus on explanations for the low dissolved oxygen in the large estuaries, especially in the main
stem of the Chesapeake Bay.
The results from this case study indicate that, of the three exposure indicators, sediment
contamination and toxicity are the primary risks in small estuarine systems while low dissolved
oxygen presents the primary risk in large systems and, particularly, the tidal rivers. These
exposure data do not identify specific contaminant stressors, nor do the data imply that these are
the only stressors of concern. This conclusion is supported by the fact that more than 50 percent
5-28
-------
ill*
a
M)
.1
«
3
1
5-29
-------
of the area of degraded benthos was not associated with any of the exposure indicators used in this
case study. Other unmeasured contaminants could cause the observed toxicity. While it is not the
intent of this case study to conduct an evaluation of the EMAP sampling design and indicator
programs, the analysis of data used in this case study has resulted in several observations on its
utility in the ecological risk assessment process (comment box).
5.3.3. EMAP and Regional Risk Assessments
The data presented above could lead to the formulation of several hypotheses regarding
ecological condition at the provincial (biogeographic) scale and potential causes of degraded
conditions. For example, some hypotheses might address the relative effects of contaminants in
small estuarine systems versus those due to low dissolved oxygen in large tidal rivers. The
provincial scale of EMAP sampling does not allow for adequate testing of hypotheses associating
environmental exposure with ecological effect. Thus, finer-scale regional studies are necessary to
refine and focus EMAP-generated hypotheses in a way that will lead to the development of more
definitive cause-effect data. In addition to evaluating EMAP hypotheses, these assessments also
should lead to more informed management decisions at the regional level. The following section
presents an example of such an assessment.
Having demonstrated the use of EMAP province-scale information in the problem
formulation phase of the risk assessment process, the next step would examine how the hypotheses
developed at the province scale can be used to assess the regional risks to specific estuarine areas.
The data and analyses presented above have focused on two types of information: (1) the
distribution of benthic resources over large biogeographic areas (i.e., the Virginian Province) and
(2) the relationship of those benthic resources to specific categories of exposure indicators. EMAP
uses this information for characterizing and comparing the status of resources across provinces and
within classes of estuaries. However, ecological resource data at the province scale have limited
regulatory value; such data are not readily coupled to political boundaries and a control strategy via
specific categories of stressors and defensible causal inferences. To optimize regulatory
applicability, province-scale data must be placed within the context of regional assessments; that is,
integrated into a risk-based decision framework that identifies the potential causal relationships
between ecological resources and specific stressors and links the relationships to land-based
activities amenable to source control.
EMAP data can identify the status of estuarine resources (represented in this case study by
benthic resources) and, more importantly, the spatial patterns and extent of resource degradation
within the province. These province-scale patterns can identify the types, spatial extent, and
possible reasons for problems within various regional settings. Figure 5-5 illustrates the spatial
distribution of degraded benthos within the Virginian Province after 2 years of sampling and the
use of province-scale data to identify potential areas for regional assessments. Degradation
generally is focused in the upper Chesapeake Bay, within the five tidal river systems and their
associated small bays. These are areas of intense demographic pressure, extensive urban
development, and the source of anthropogenic stress. Considerable benthic degradation occurs
throughout the Hudson, East, and Raritan Rivers. This degradation is associated with sediment
toxicity and elevated sediment chemistry values (figure 5-5). Data suggest that EMAP information
can help identify regional areas of degraded resources and provide preliminary associations with
5-30
-------
exposure type. The probabilistic nature of the EMAP design also permits a determination of the
relative magnitude of degradation, thereby focusing attention on areas with potentially the greatest
problems. Using the Hudson-Raritan. estuary and watershed as an example, the following sections
briefly describe one approach for conducting a regional assessment that uses the EMAP design,
indicator, and assessment concepts.
5.3.3.1. Regional Risk Assessment: Problem Formulation
i •
While useful in identifying regional problem areas, EMAP province-scale data are not
collected in sufficient detail for conducting a complete regional risk assessment. Although,extant
local monitoring data are usually available, they often are heterogeneous relative to spatial,
temporal, and ecological scale and methodologies (e.g., type of sampling gear, analytical methods,
etc.). Within the Hudson-Raritan basin, decades of monitoring data are available from NOAA,
states, and more recently the Harbor Estuary Program (HEP). However, each of these programs
has its own problem-oriented objectives and sampling and analysis goals. This heterogeneity in
objectives makes it difficult, if not impossible, to satisfy the information needs of problem
formulation and fully characterize the type and spatial extent of the ecological problems at a
regional scale.
The first step, then, in the regional risk assessment involves revisiting the problem
formulation phase of the risk assessment process to characterize the spatial extent of degraded
resources and associated measures of exposure. Data for this purpose can be acquired through (1)
an appropriately scaled monitoring program employing a random sampling design (e.g., EPA
Region II, R-EMAP); (2) selection of the appropriate response, exposure, and habitat indicators to
characterize the spatial extent of ecological problems and associated exposures; (3) the
incorporation of extant data, where possible, into a probabilistic sampling design analogous to that
used by EMAP; or (4) through a combination of all three approaches. The conduct of problem
formulation at the regional scale will provide a detailed description and spatial representation of the
types, magnitude, spatial distribution, and areal extent of ecological problems. These ecological
effects can then be associated more closely with specific exposure and habitat indicators and
stressors, leading to the development of one or more conceptual models for the region or specific
watershed within the region. Currently, ORD, in cooperation with EPA Region II, is conducting a
Regional-EMAP project in the Hudson-Raritan estuary to develop just such a series of conceptual
models for various areas within the estuary.
5.3.3.2. Regional Risk Assessment: Analysis Phase
The analysis phase of the ecological risk assessment process involves the development of
detailed models describing the spatial and temporal patterns of exposure and stressor-response
models that illustrate the change in status of ecological response as a function of incremental
changes in exposure. Monitoring programs may collect some types of data that are relevant to a
detailed analysis of ecological risks; however, they do not normally collect the full spectrum of
necessary data, nor do monitoring data provide the necessary uniform spatial coverage for the area
of concern. Within a regional setting like the Hudson-Raritan, where sediment toxicity and
contaminated sediments are known to be associated with degraded benthic resources, the risk
assessor would likely synthesize extant data from the ORD research laboratories, Region II, HEP,
5-31
-------
NOAA, states, private sector, etc., to develop the causal relationships necessary to fully
characterize the regional risks.
Extant data for this area can prove particularly important in identifying possible causes for
observed resource degradation. For example, there is a history of PCS contamination in the
Hudson River, dioxins hi the Raritan River, petroleum contamination in the Arthur Kill River, and
organic enrichment in Jamaica Bay. In addition, during the last several years NOAA has
synthesized data on benthic community structure, sediment toxicity, and metal and organic
contaminants in sediments and water in this area. Although not sampled probabilistically, these
data help identify spatial patterns of problems and their potential causes in various sections of the
estuary. A regional risk assessment can use these data, along with laboratory toxicity information
and measures of contaminant bioavailability, to develop evidence needed for postulating causal
inferences for the region as a whole or for a specific watershed. The causal relationships may be
quantitative or inferential, relying on weight-of-evidence and professional judgment.
The contribution of the EMAP design to the Hudson-Raritan basin study, conducted by
EPA Region II, will significantly strengthen inferences of risk within this watershed (National
Governors Association, 1993). In addition, this R-EMAP project also will examine methods for
incorporating extant data into the probabilistic EMAP design, further enhancing its utility. Most
likely, monitoring data alone will prove insufficient for establishing the causal relationships
necessary for developing a complete risk assessment. Nevertheless, the intent is to develop
multiple, converging lines of evidence for linking observed ecological effects to one or more
specific stressors or to stressor categories that are amenable to remediation. The extant data in the
Hudson-Raritan basin suggest that different stressor-response relationships may emerge for different
watersheds. This conclusion would lead to different source control management strategies for each
watershed.
5.3.3.3. Regional Risk Assessment: Risk Characterization
The risk characterization phase of the framework describes three methods for integrating
exposure and effects information into a statement of the likelihood of risk with associated
uncertainties: point comparisons, distributional comparisons, and modeling. Depending on the
type of data, any one or a combination of these approaches can be used with the types of
monitoring data presented here. GIS and landscape methods can provide initial descriptions of
risks to specific watersheds. These methods can describe the spatial relationships and distribution
of response, exposure, and habitat indicators (stressor-specific whenever possible). These
descriptions can then be overlaid with landscape information on hydrologic features (e.g., transport
and fate) in the surrounding watershed. Descriptive approaches, using GIS and landscape methods,
can integrate field data describing the spatial extent, magnitude, and degree of association between
response and exposure indicators. However, descriptive approaches do not establish functional
exposure-response relationships, Establishment of functional relationships requires the
decomposition of measurements of "aggregate exposure" (e.g., sediment toxicity-related bioeffects
from multiple stressors) into specific stressors using diagnostic biomarkers, fractionation protocols,
and laboratory ecotoxicity tests.
5-32
-------
For example, overlays of response and exposure indicators indicate that there is a high
degree of co-occurrence of benthic community degradation with sediment toxicity and specific
sediment organic contaminants (e.g., dioxins and dibenzofurans) in the Raritan River. This
example suggests the potential for a strong causal relationship between specific stressors and
ecological effects. Literature data, additional in situ field testing along a gradient, or laboratory
testing can evaluate the hypothesis. As a clearer picture of the specific stressors emerges, GIS and
landscape methods can integrate (1) information on the spatial distribution of specific contaminants,
(2) areas of degraded benthos, (3) information on the discharges from land-based activities, and (4)
hydrologic information from the surrounding watersheds. Using available data and converging
lines of evidence, a series of inferences can be developed regarding causal associations from
response to exposure to stressors to sources. Supporting these initial inferences requires additional
analyses such as site-specific studies on organism-residue relationships, contaminant "spiked"
laboratory sediment-residue and toxicity analyses, and site-specific field studies using natural
contaminant gradients. Together, these studies would focus on quantifying functional and causal
relationships and the uncertainties associated with each phase of this process.
In summary, spatial models describing response-exposure-stressor-hydrologic relationships
can be coupled with landscape models describing specific watershed activities that are sources of
anthropogenic inputs. The establishment of the appropriate causal relationships between sources
and effects provides the basis for the manager to institute appropriate control strategies. Existing
local compliance (e.g., NPDES, states, municipalities) and watershed assessment (R-EMAP,
EMAP, NS&T) monitoring programs can evaluate the effectiveness of the control strategy.
Comments on Problem Formulation, Conceptual Model Development, and Regional Risk
Assessments
General reviewer comments:
• The case study's introduction and the background do a good job of setting the
stage for the problem formulation and of explaining the benefits and limitations
of the EMAP program. The authors refer to the use of EMAP in this fashion as
a "weight-of-evidence" approach. Perhaps it would be more accurate to call it
a screening approach, because "weight-of-evidence" has a toxicological
interpretation that implies real knowledge of cause and effect for a stressor and
organisms. In using the term "weight-of-evidence," EMAP is suggesting such a
relationship.
5-33
-------
Comments on Problem Formulation, Conceptual Model Development, and Regional Risk
Assessments (continued)
• The percent co-occurrence of degraded benthos with low dissolved oxygen does
not give the percent of degradation that can be attributed to low oxygen. As
stated, these co-occurrence data represent a contingency table that tests
association. Consider the large estuaries where 7 percent of degraded benthos
is associated with sediment toxicity, despite the fact that 14 percent of the
estuaries have sediment toxicity. The conclusion is (if significant) that sediment
toxicity tends to be associated with undegraded benthos. For oxygen, the
respective figures are 20 percent and 5 percent,- hence, low oxygen appears to
be quite strongly associated with degraded benthos.
There are a number of ways of analyzing for associations in such data. If one
can assume that samples are independent, then a log-linear model of frequency
data might be appropriate. In this case, sample size permitting, there could be
four levels: riverine type, benthos condition (degraded vs. undegraded),
sediment toxicity, and ER-M exceedance. Such an analysis would determine
differences among riverine types in various conditions, associations of exposure
measures with effect measures, and associations among the different exposures.
• The section on regional risk assessment refers to the association of degraded
areas with areas of "intense demographic pressure, extensive urban
development, and the source of anthropogenic stress. " Should these be
considered as pan of the exposure characterization? Could sampling data be
further stratified by stressed areas within waterways and by stressed and
unstressed waterways?
Authors' comments:
EMAP Sdmplins Desien
Strengths of the case study include:
Quantifies areal extent of indicator values.
Describes the spatial patterns and distribution of ecological resources and
associated habitat and exposure indicators.
Permits the estimation of uncertainties for indicator values.
Quantifies postremediation changes in areal extent of resources and exposures.
Scalable to regions and specific sites (e.g., bays, estuaries).
5-34
-------
Comments on Problem Formulation (Continued)
Limitations include:
• Limiting sampling to index period (e.g., once per year) fails to address
seasonality and episodic events.
• Sampling design currently does not capture local spatial scale and short-term
temporal scale events.
• Incorporation of nonprobabilistic extant data with the EMAP'sprobabilistic
sampling design is currently not feasible and is a major limitation for risk
assessment applications.
EMAP Indicators
Strengths of the case study include:
Sites of exposure and habitat indicators are measured simultaneously with
response indicator.
Response indicator is hierarchical in design, with clear links between assessment
endpoints, measurement endpoints, and metrics. :
Habitat indicators are directly related, facilitating the interpretation of response
and exposure indicator information.
Limitations include:
Currently, EMAP has no response or exposure indicators for nutrient or carbon
enrichment (eutrophication).
Response indicators have been developed and applied only for benthic
resources.
Exposure indicators for physical stressors are lacking.
There is currently no systematic program for validating existing indicators.
Accurate measures of bioavailability are needed for interpreting contaminant
exposure indicators.
The benthic index metric, sediment toxicity, and bioavailability indicators
require evaluation, validation, and revision.
5-35
-------
U.S. Environmental Protection Agency. (1992) Framework for ecological risk assessment. U.S.
Environmental Protection Agency, Risk Assessment Forum, Washington, DC. EPA/630/R-
92/001.
U.S. Environmental Protection Agency. (1988) Future risk: research strategies of the 1990s. U.S.
Environmental Protection Agency, Science Advisory Board, Washington, DC. SAB-EC-88-
040.
U.S. Environmental Protection Agency. (1984) Chesapeake Bay: a framework for action. Prepared
for the U.S. Congress by the Environmental Protection Agency, Region 3, Philadelphia,
PA.
U.S. House of Representatives. (1984) Environmental monitoring and improvement act. Hearings
before the Subcommittee on Natural Resources, Agricultural Research, and the
Environment of the Committee on Science and Technology. March 28, 1984. Washington,
DC: U.S. Government Printing Office.
Vernberg, F.S. (1972) Dissolved gases. In: Kinne, O., ed. Marine ecology. Vol. I, Part 3. New
York, NY: Wiley-Interscience, pp. 1491-1526.
Weaver, G. (1984) PCB contamination in and around New Bedford, Mass. Envir. Sci. Tech.
18:22A-27A.
Weisberg, S.B.; Frithsen, J.B.; Holland, A.F.; Paul, J.F.; Scott, K.J.; Summers, J.K.; Wilson,
H.T.; Heimbuch, D.G.; Gerritsen, J.; Schimmel, S.C.; Latimer, R.W. (1993) Virginian
province demonstration project report, EMAP-estuaries: 1990. U.S. Environmental
Protection Agency, Office of Research and Development, Washington, DC. EPA/620/R-
93/006.
• U.S. GOVEWWEHT PRINTING OFFICE: 1 995-650-006/22040
5-40
------- |