ft EA~United States
..-~ Environmental Protection
~.. Agency
Conceptual Models and Methods
to Guide Diagnostic Research
-
-

-------
This page left blank intentionally

-------
EP A 600/R-06/024
AED-04-022
May 2006
Conceptual Models.and Metho
to Guide Diagnostic Research
u.s. Environmental Protection Agency
Office of Research and Development
National Health and Environmental Effects Research Laboratory
Research Triangle Park, NC 27711

-------
Notice
The research reported in this document has been funded wholly by the United States
Environmental Protection Agency (USEP A). This document was prepared jointly by
co-authors employed by the Office of Research and Development (ORD), National Health and
Environmental Effec~s Research Laboratory (NHEERL), Atlantic Ecology Division (AED),
Mid-Continent Ecology Division, and the Gulf Ecology Division. It has been subjected to review
by NHEERL and approved for publication. This approval does not signify that the contents
reflect the views of the USEP A, nor does the mention of trade names or commercial products
constitute an endorsement or recommendation for their use. This report is AED Contribution
# AED-04-022.
Acknowledgments
We thank the reviewers of this document, whose suggestions and comments have greatly
improved its organization and content. Ed Dettmann and Tim Gleason of AED were internal
reviewers, Dave Mount of MED served as an external reviewer, and Marilyn ten Brink and
Wayne Munns provided careful reviews of issues related to science and policy, respectively.
Publication support was provided by Patricia DeCastro of the Computer Science Corporation.
II

-------
Conceptual Models and Methods
List of Authors and Affiliations
Daniel E. Campbell I
Kay T. Hol
Janis C. Kurtz3
Virginia D. Engle3
Marguerite C. Pelletier 1
Naomi E. Detenbeck2, .
Brian H. Hill2
I
Robert M. Burgess
Kenneth T. Perez 1
Virginia Snarski2
United States Environmental Protection Agency, Office of Research and Development,
National Health and Environmental Effects Research Laboratory, IAtiantic Ecology Divi.sion,
Narragansett, RI, 2Mid-Continent Ecology Division, Duluth, MN, 3Gulf Ecology Division,
. Gulf Breeze, FL
1Il

-------
Conceptual Models and Methods
Table of Contents
Notice / Acknowledgments ........... ..................... ....................... ........................... .................... ........ ii

Exec':ltive. Summary ..................... ................. ......... ............... ................... ......:.... ......................... viii
1.0 Introduction[[[l

1.1. Present Context for Aquatic Ecosystem Protection in the United States ........................3
1.2. Use of Conceptual Models in Monitoring and Assessing Ecosystems............................5
2.0 Methods for Developing Conceptual Models ........................................ ......... .................. .........7
2.1 The Context for Diagnosing Impairment to Aquatic Ecosystems ....................................7
2.1.1 Aquatic Ecosystems and the Hydrologic Cycle[[[7
2.1.2 Diagnosis and Management at the Watershed Scale ............................................8
2.1.3 Stress in Aquatic Ecosystems [[[11
2.2 Development of Diagnostic Tools [[[11
2.2.1 Causal Webs: Bottom-up and Top-down................ ............. ............ .......... .........12
2.2.2 Canonical Models[[[ .13
2.2.3 Watershed Scale for Diagnosis and Control[[[19
2.2.4 Pollutant Identification and Evaluation ........."............. .......... ............ ......... .......20
2.2.5 Modeling Methods and the Allocation of Effects...............................................28
2.2.6 Response-Based Classification .................................... ................ ................. ......29
3.0 Development of Detailed Conceptual Models ................................ ................... .................. ....32
3.1 Narrative Descriptions of the Stressors [[[ ..32

3.1.1 Nutrients[[[ ............................ ............ ....... ........32

3.1.2 Toxic Chemicals ..............,............................. .................,..... ............ ..................34

3.1.3 Suspended and Bedded Sediments[[[36

3.1.4 Habitat.......... ................................. ...... ........ .......................... ..............................37

3.1.5 Special Characteristics of Aquatic Ecosystems Applicable to all Stressors.......39
3.2 Generic Conceptual Models of the Four Classes of Aquatic Stressors ..........................41
3.2.1 A Generic Energy Systems Model for Habitat Alteration ..................................42
3.2.2 A Generic Energy Systems Model for Suspended and Bedded Sediments ........49
3.2.3 A Generic Energy Systems Model for Toxic Chemicals.........'...........................55
3.2.4 A Generic Energy Systems Model for Nutrients ................................................60
3.2.5 Caveat on the Detailed Models ofPollutants[[[64

-------
Conceptual Models and Methods
List of Figures
Figure ES 1.0rganization of the elements of the Diagnostics Research Program showing how............... ix
conceptual models, classification, and Pollutant Identification Evaluation (PIE) methods
are used together to make an integrated diagnosis ofthe causes of impairment. Information
on the results of implementing a TMDL feeds back to Phase IV in the PIE module, where
it is evaluated to confirm or deny the original diagnosis
Figure ES2 Conceptual model shown as an Energy Systems Language module of the ...........................x
three primary factors controlling the action of pollutants in aquatic ecosystems
Figure I. Conceptual model showing the links between the listing process in the ..................................2
monitoring and assessment report and the process for establishing a Total
Maximum Daily Load (TMDL) for watershed management (USEPA 2002a).
Figure 2. The environmental system of a region showing the chain of connections ................................4
linking economic activity with environmental degradation.
Figure 3. Definitions for the basic symbols of the Energy Systems Language ........................................8
(modified from Odum 1994).
Figure 4. Aquatic ecosystems storing water within the hydrologic cycle. ................................................9
Figure 5. Networks of ecosystem units arranged on the landscape within................................................9
fundamental watersheds A, S, C, and D each of which contains a
river systems that empties into the sea. Dashed lines indicate watershed boundaries.
Figure 6.
Aquatic ecosystems are arranged on the landscape in a longitudinal series of .......................10
processing units dominated by production (P) or respiration (R). Excess production
in one system sets the stage for excess consumption in the next and vice versa.
Figure 7. Illustration of the top-down and bottom up approaches to tracing .........................................13
causal pathways (Ziemer 2004).
Figure 8.
(a) Conceptual model of the bottom-up approach to tracing causality................................... 14
through a web of interactions. In this example, the observation of a species
decline initiates model construction. (b) Conceptual model of the top-down
approach to tracing causal pathways thro':lgh a web. In this example, the
observation of watershed clearing initiates model construction.
Figure 9.
(a) Model of the factors controlling the action of stressors in aquatic ecosystems ........... 16-18
within a water storage unit with unidirectional flow (b) Model of the factors controlling
the action of stressors in aquatic ecosystems within water storage units with bi-directional
flows (c) Model of the factors controlling the action of stressors in aquatic ecosystems within
water storage units with unidirectional horizontal flow and two different processing capacities,
e.g., stratified systems or sediments and water column. (d) Model of the factors controlling the
action of stressors in aquatic ecosystems within water storage units with bidirectional flows
and two processing capacities, e.g., stratified systems or sediments and water column.
Figure 10. Two hypothetical exposure-effect relationships for two different pollutants. .........................18
v

-------
Conceptual Models and Methods
Figure 11. Summary diagram of the principle factors that control habitat of a species. ...........................38
and its alteration
Figure 12. Energy systems model of the effects of habitat alteration on a species, .................................44
Q, in an aquatic ecosystem. Forcing functions, components, and pathway flows
are defined in Table 4:
Figure 13. Summary diagram showing the factors that control the effects of ..........................................51
suspended and bedded sediments in aquatic ecosystems.
Figure 14. An energy systems model of the effects of suspended and bedded sediments .......................51
on aquatic ecosystems. Forcing functions, components, and pathway flows
are defined in Table 5.
Figure 15. (a) Factors that affect the availability of toxic chemicals in freshwater .................................54
and marine systems. (b) Factors that affect the availability of toxic chemicals
in the water column and in sediments.
Figure 16. An energy systems model for the effects of a toxic chemical, T, on an aquatic .....................56
ecosystem. Forcing functions, components, and pathway flows are defined in Table 6.
Figure 17. Factors that control the effects of nutrients in aquatic ecosystems. .........................................59
Figure 18. An energy systems model of the effects of excess nutrients on an .........................................61
aquatic ecosystem. Forcing functions, storages, and flows are defined in Table 7.
Figure 19. Canonical energy systems model of excess nutrients in an estuary ........................................67
along with the equations that describe the behavior of this system. Forcing functions,
components, and pathways in the model are defined in Table 8. To is the
initial temperature where the rate function was evaluated and z is the depth.
Figure 20. A preliminary classification tree that groups estuaries by effective .......................................69
exposure regimes based on our conceptual model of the factors that control biological
impact. This classification based on effective exposure could be applied to any of the
stressors with modifications for the particular properties of a given stressor.
Figure 21. Normalization of the exposure-effect relationship by adjusting both .....................................70
the x and y axis. (1) A type B modifying factor shifts the response variable
on the exposure axis. (2) At a standard exposure, different levels of the biological effect
variables characterize aquatic ecosystems of different kinds.
VI

-------
Conceptual Models and Methods
List of Tables
Table I. Proposed diagnostic tools used in the development of Pollutant Identification. .........................24
and Evaluation (PIE) methods for determining the causes of impairment to the
aquatic ecosystems in estuaries. Phase 1 is screening and verification. Phase 2 is the
identification of source, stressor, and effect. Phase 3 is establishing the linkage between
source and stressor and stressor and effect. Phase 4 is confirmation of the diagnosis.
Table 2. A Decision Table for evaluating outcomes in Phase 2 of a PIE. For each. .................................27
stressor considered in Phase 2, there should be evidence of a source, the presence
of a stressor, and an effect to move to Phase 3. An X indicates presence
Table 3. Possible outcomes of Phase 3 showing linkages between source, stressor and effect.................28
Table 4. Definition of the forcing functions, components, and pathways in the generic energy...............45
systems model to evaluate the effects of habitat alteration on aquatic ecosystems.
Table 5. Definition of the forcing functions, components, and pathways in the generic energy...............52
systems model to evaluate the effects of suspended and bedded sediments on
aquatic ecosystems.
Table 6. Definition of the forcing functions, components, and pathways in the generic energy...............57
systems model designed to evaluate the effects of toxic chemicals on aquatic ecosystems.
Table 7. Definition of the forcing functions, components, and pathways in the generic...........................62
energy systems model to evaluate the effects of nutrient loading on aquatic ecosystems.
Table 8. Definition of the forcing functions, components, and pathways in the canonical.......................65
energy systems model of excess nutrient in an estuary.
List of Text Boxes
(A) The Inherent Contlict Between Economic Prosperity and Environmental Quality .............................5
(B) Definitions of Ecological Stress and Emergy [[[1 0
(C) Energy and Emergy Signatures: An Alternative Method for Classifying Ecosystems .......................30

-------
Conceptual Models and Methods
Executive Summary
This report contains conceptual methods and.
models to guide research on and development of
tools for diagnosing the causes of biological
impairment within the aquatic ecosystems of the
United States. It was produced to satisfy
requirements in the U.S. Environmental Protection
Agency Aquatic Stressors Framework (USEPA
2002b). The goal of the National Health and
Environmental Effects Research Laboratory's
(NHEERL) Diagnostics Research Program is (1)
to provide tools to diagnose the causes of .
biological impairment in aquatic ecosystems, (2)
to develop a classification system that simplifies
the process of developing Total Maximum Daily
Loads (TMDLs) or other regulatory programs for
the myriad of water bodies requiring them, and (3)
to support the States and Tribes in determining the
causes of impairment of water bodies to be placed
on their 303(d) lists. To accomplish these goals,
NHEERL convened a Diagnostic Research Work-
group. This workgroup developed an overview
conceptual model of the factors controlling the
action of pollutants and detailed conceptual
models for four aquatic stressors: nutrients,
suspended and bedded sediments, toxic chemicals
and altered habitat that were identified as the
stressors of major concern in the Aquatic Stressors
Framework (USEPA, 2002b). Four canonical
forms of the overview conceptual model were used
as the framework for developing detailed
conceptual models of the four stressors and as a
basis for classifying aquatic systems according to
their sensitivity to each stressor of concern. The
proposed classification framework is designed to
simplify the development ofTMDLs by grouping
aquatic systems according to similarity in their
response to a particular stressor, In addition,
classification may enable a more refined approach
for quantifying stressor-response relationships for
criteria development (USEPA 2004).
The approach to determining the causes of
impairment used here consists of a linked set of
hierarchical, modular methods and models and a
proposed classification scheme for aquatic
ecosystems. In addition, an Energy Systems
Theory (Odum 1994) framework for developing
causal network models of stressor action in aquatic
ecosystems is presented in a series of text boxes as
a parallel discussion. These two parallel
approaches are brought together in the final
section of the paper where detailed energy systems
models of the main factors controlling the action
of the four major aquatic stressors are given. The
end result of our research will be guidelines for
diagnosing the causes of biological impairment
within aquatic ecosystems of the United States.
The suite of methods and tools under
development (Figure ES-1) bridges the critical link
between the assignment of a water body to the
303(d) list and the initiation of a TMDL (USEPA
2002a). This link includes the determination of a
definitive cause or causes for the observed
impairment that placed the water body on the
303(d) list in the first place, as well as, the
development of tools that will allow us to
understand how to restore the impaired ecosystem.
To forge this link we identified six critical
research elements that need to be accomplished to
make a definitive diagnosis of the cause of
impairment and to ensure the development of an
effective TMDL or other regulatory program that
will successfully control a pollutant. These
research tasks are:
(1) Link sources, which result from human
activities, to stressors and biological effects that
occur in receiving water bodies in a manner
that supports the development of effective
corrective action.
(2) Formulate a set of simple, standardized models
(canonical models) that incorporate the
fundamental causal mechanisms determining
biological condition and the observed
impairment. Canonical models serve as a
starting point for classification and as a frame
for the development of detailed models of
stressor action, which in turn may serve as a
guide to the development of integrative
diagnostic indicators.
(3) Development methods that use the hierarchical
structure of watersheds and human activities
when determining the scope of control needed
for a successful TMDL.
V III

-------
Conceptual Models and Methods
PIE Methods
Integrated Diagnosis
A. Causes of Impairment
B. Scale of Control
Figure ES-l. Organization of the elements of the Diagnostics Research Program showing how conceptual models,
classification, and Pollutant Identification Evaluation (PIE) methods are used together to make an
integrated diagnosis of the causes of impairment. Information on the results of implementing a TMDL
. feeds back to Phase IV in the PIE module, where it is evaluated to confirm or deny the original
diagnosis.
(4) Extend the logical methods of deduction and
elimination used to develop techniques in
Toxicity Identification and Evaluation, TIE,
(Burgess 2000, Ho et at. 2002) so that they
can be applied to other pollutants. We have
named this method, now under development,
Pollutant Identification Evaluation (PIE).
These tools, when fully developed, will allow
the States and Tribes to make a definitive
diagnosis of the causes of impairment to
aquatic ecosystems. The general logical
methods of causal analysis and the approach to
problem solving described in the Stressor
Identification Guidance Document (USEPA
2000a) are used in the development of PIE
tools and methods.
(5) Construct detailed energy systems models of
the action of single and multiple stressors
within aquatic ecosystems to serve as the
precursor of simulation models that will
predict system behavior and allow the
allocation of observed effects among
multiple demonstrated causes.
(6) Classify ecosystems based on their response
to stressors to simplify the process of
diagnosis and the subsequent development of
regulatory programs.
When used together, the methods and
models developed within these six research
elements will allow us to diagnose biological
impairment in a simpler and more accurate
manner, thereby accomplishing our goal of
taking a water body from its 303( d) listing to the
successful implementation of a TMDL.
The overview conceptual model that was
used as a starting point for all of our research
consists of three factors that we hypothesize
control the action of pollutants in aquatic
ecosystems.
IX

-------
Conceptual Models and Methods
Residence Time
(
A
'\
Inflow
Water Storage
Outflow
P = Pollutant
M = Modifying Factors
Type A and B
J,-
Figure ES-2. Conceptual model shown as an Energy Systems Language module of the three primary factors
controIling the action of poIlutants in aquatic ecosystems.
They are:
(1) the residence time of water and the pollutant
in the system,
(2) the natural processing capacity of the system
for the pollutant including the pathways that
decompose, take-up, or sequester the
material, and .
(3) ancillary factors that modify the form of a
pollutant, (i.e., the rate of processing) or the
kind of action the pollutant exerts within the
ecosystem (Figure £8-2).
These three factors are evaluated in a manner
that quantitatively determines the effective dose of
a pollutant for different types of ecosystems. Also,
in this report, we hypothesize that characteristic
properties of aquatic systems related to residence
time, processing capacity, and modifying factors
can be used to differentiate classes of ecosystems
that develop different biologically effective
concentrations of a material when loaded with a
given quantity of a pollutant. The classification
problem may be further simplified by grouping
pollutants according to their mode of action such
. that an ecosystem processes all members of a class
in a similar manner. In this case, we can express
the bioeffective concentration in aggregate units
(e.g., standard toxicity units). The conceptual
models and metQods presented in this document
need to be mathematically formulated and
evaluated with data from laboratory experiments
and field studies before their veracity can be
demonstrated and their full potential realized.
The development of PIE tools and methods
as outlined in this document holds considerable
promise for the diagnosis of the causes of
impairment to estuaries and other coastal
ecosystem. The diagnosis of the causes of
. impairment to coastal ecosystems is particularly
difficult because of the complex flow regime in
estuaries and the presence of multiple stressors
in the marine environment, which commonly
. receives pollutants from many sources.
The overview conceptual model and its
canonical forms can serve as a guide to the
design of a stressor-based classification system,
the development of scale independent indicators
that give the expected condition of a system
based on its class, and the identification of the
scale of control needed for successful regulation
of a given pollutant (a database tool to
accomplish these tasks is under development).
In addition, the overview conceptual model and
x

-------
Conceptual Models and Methods
its canonical forms serve as a frame for the
development of detailed energy systems models
for the individual stressors and their commonly
occurring co-stressors. When quantitatively
evaluated and programmed on a computer, these
detailed conceptual models can simulate various
scenarios of stressor-loading, giving results that
can be used to allocate an observed effect among
multiple active stressors. This document may be
of immediate use to scientists in the process of
developing initial problem formulations for risk
assessments in support of the development of a
TMDL or other regulatory program for one of the
four classes of aquatic stressors covered by our
research.
XI

-------
Conceptual Models and Methods
1.0 Introduction
In this report, we propose methods and
conceptual models to guide research on the
development of tools for diagnosing the causes of
biological impairment within aquatic ecosystems
of the United States. The approach to determining
the causes of impairment used here consists of a
linked set of hierarchical, modular methods and
models and a proposed classification scheme for
aquatic ecosystems. In addition, an energy
systems framework for developing causal network
models of stressor action in aquatic ecosystems is
presented as a parallel discussion in a series of
text boxes. These two parallel approaches are
brought together in the final section of the paper
where detailed energy systems models of the main
factors controlling the action of four classes of
aquatic stressors, nutrients, suspended and bedded
sediments (SABS), toxic chemicals, and habitat
alteration, are given. These four classes of aquatic
stressors were identified as high priority areas for
research by the Aquatic Stressors Framework
(USEPA 2002b). A simple model of the main
factors controlling the action of the three classes
of aquatic stressors that are pollutants (nutrients,
toxic chemicals, and SABS ) was proposed as the
linchpin to hold together the proposed methods
and models needed to diagnose impairment to
aquatic ecosystems of all kinds. This model also
serves as the basis for determining the scale of
management needed to develop an effective total
maximum daily load (TMDL) for a pollutant at
the scale of watersheds as well as water bodies. A
detailed conceptual model of habitat alternation
was also developed and related to the simple
conceptual model controlling pollutants. A
preliminary classification framework for coastal
systems is presented in another report (US EP A
2004); however, our thinking on conceptual
models and classification proceeded in parallel;
therefore, the connection between the need to
classify aquatic ecosystems and the development
of conceptual models is reported here. When fully.
developed we believe these tools will simplify and
improve the accuracy of water body evaluations
currently being carried out by the states under
Sections 305(b) and 303(d) of the Clean Water
Act. Ultimately, our goal is to ensure the success
of water body restoration programs such as those
that control the TMDL of a pollutant (USEPA
2002a)
The models and methods presented in this
paper are diagnostic in that they are all concerned
with or support the determination of the cause 'or
nature of a particular phenomenon B an observed
impairment of ecological condition in an aquatic
ecosystem. These methods include the use of
characteristic signs and symptoms which suggest a
cause; however, this evidence is strengthened by
the use of definitive tests of causality whenever
possible. The goal of diagnosis is to establish
cause as unambiguously as possible under any
given setof circumstances. Classification schemes
are also diagnostic in the broad sense because the
classification of systems based on their behavior
under stress allows us to use the logical process of
inference to impute causality and simplify the
diagnostic process. Imputed causes of impairment
can be verified using the definitive diagnostic
tools that we are developing. Another diagnostic
method is to focus on specific organisms that have
been shown to be sensitive to or tolerant of a
particular stressor. Specific diagnostic indicators
are not discussed in this report, but they may be
developed in the course of our future research. A
final method of determining cause is to construct
and analyze models of causal networks based on
demonstrated and hypothesized mechanisms of
interaction that are linked to the transformation of
energy (Odum 1994). Whenever possible and.
especially in difficult or ambiguous cases, more
than one diagnostic method should be used to
determine active cause(s) of an observed
impairment. When the results of several methods
agree, we have greater confidence in the resulting
diagnosis.
Our research program bridges the critical link
(Figure 1) between assigning a water body to the.
303(d) list and the initiation of a TMDL (USEPA
2002a). This key link includes the determination
of a definitive cause or causes for the observed
impairment that placed the water body on the
303(d) list in the first place,.as well as, the
development of tools that will allow us to under-
stand how to restore the impaired ecosystem. To

-------
Conceptual Models and Methods
/' Priority ranking ~

Develop
schedule
Identify impainsd and ,
thnsatened watera Listing Process of
needing TMDLs Integrated Monltor,lng
\ and Assessment Report
Submit
Identify was !!Itlainment' 303(d) list with 305(b)
status all waters(305b) as part of Integrated Report .

'\.. 303(d.) liS( Update
~'::~~~"r ff.ti~~Prov:~ ~~ng

....----t:> . .. c cte

-YES /' ~ Dba_osm troews

EPA ... tabU.h.. Problem/pollutant Tools, metbDds and models
or ;:~;.. Identification . to dita:CD'OSO impafr.IQ.ont
,
TMDL submitlal
Target analysis
(amaunt by which pellutant must
be nsduced to meet WaS)
J

Source
assessment
,
Linkage of
--- sources and target
Update
next
listing
cycte
\

Monitoring schedule

"
TMDL
Establishment
Proc..s (Watershed
Management)
Allocation to
sources or
categories
of sources
~x~iWD
~~
~>UU
Figure 1. Conceptual model showing the links between the listing process in the monitoring and assessment report
and the process for establishing a Total Maximum Daily Load (TMDL) for watershed management
(USEPA) 2002a)
forge this link we identified six critical research
objectives that need to be accomplished to make a
definitive diagnosis of the cause of impairment and
to ensure the development of an effective TMDL
or other regulatory program that will successfully
control a pollutant. These research tasks are
(1) Link sources, which result from human
activities, to stressors and biological effects that
occur in receiving water bodies in a manner that
supports the development of effective'
corrective action.
(2) Formulate a set of simple, standardized models
(canonical models) that incorporate the funda-
mental causal mechanisms determining ecosys-
tem condition and observed impairment to
serve as a basis for classification, a guide to the
development of integrative diagnostic indica-
tors and as the starting point for the develop-
ment of detailed models of stressor action.
(3) Develop methods to account for the
hierarchical structure of watersheds and human
activities when determining the scope of a
TMDL.
(4) Extend the logical methods of deduction and
elimination used to develop techniques in TIE,
Toxic Identification and Evaluation (Burgess
2000, Ho et al. 2002) to other pollutants to
allow the states to make a definitive diagnosis
of the causes of impairment to aquatic
ecosystems. We have named this method, now
under development, Pollutant Identification and
Evaluation (PIE). The general logical methods
of causal analysis and the approach to problem
solving described in the Stressor Identification
Guidance Document (USEPA 2000a) provide a
logical framework for these methods.
(5) Construct detailed simulation models of the
action of single and multiple stressors within
aquatic ecosystems to predict system behavior
2

-------
Conceptual Models and Methods
and to allocate observed effects among multiple
demonstrated causes.
(6) Classify ecosystems based on their response to
stressors to simplify the process of diagnosis
and the subsequent development of a TMDL.
When used together, the methods and models
developed within these six research areas will
'allow us to diagnose biological impairment in a
simpler and more accurate manner, thereby
accomplishing our goal of taking a water body
from 303(d) listing to the successful implementa-
tion of a TMDL (Figure 1). For example,
development of the PIE diagnostic methods (4)
. and the development of detailed models of
stressor action (5) provide a means for
identifying, evaluating and interpreting cause and
effect relationships by relying first on PIE
methods that use field data, experiment, and logic
to definitively demonstrate the presence of causal
links between source, stressor and effect within
an aquatic ecosystem. When source, stressor and
effect are shown to exist and the linkages between
source and stressor and stressor and effect have
been demonstrated the pollutant is designated as a
demonstrated cause ofthe impairment. Next the
detailed conceptual models for the four stressors,
i.e., nutrients, toxic chemicals, suspended and
bedded sediments and habitat alteration (2), are
used to create simulation models (5) to predict
impairment and allocate observed effects among
multiple demonstrated causes, thereby allowing
us to set priorities for restoration. The
development of tools to link stressors and effects
in watersheds (3) will allow us to trace the causal
links between watershed activities (sources) and
individual pollutants (1) OV6r at least three scales
of watershed organization (e.g., stream orders and
receiving water bodies within their nested
watersheds) to determine the appropriate spatial
scale at which a TMDL needs to be implemented
to control the concentrations of a particular
pollutant in hierarchically organized aquatic and
human systems. Together, the canonical models
(2) and the classification system (6), now under
development, will be used to simplify and
improve the accuracy of impairment decisions by
grouping aquatic systems according to similarities
in their behavior under stress. Classification
categories will identify systems that are
functionally similar and this similarity may serve
as both a basis to impute causality across the class
and as a guide for restoration of individual
systems within the class.
In this report, we have used the principles and
methods of systems ecology, specifically Energy
Systems Theory, EST (Odum 1983, 1994), to
develop and trace pathways of causality in
ecological networks and to synthesize
information and develop models for the action of
stressors in aquatic ecosystems. This method is an
integrating thread that flows through this work
and it provides either the primary method or an
alternative method for accomplishing all but one
of the six research tasks. However, other
approaches and methods were drawn upon to
design our research, including the methods of
Toxic Identification and Evaluation and Stressor
Identification. Many readers will be unfamiliar
with energy systems concepts and for this reason
supporting information is provided in a series of
text boxes that present topics that are relevant to
accomplishing our research objectives.
1.1. Present Context for Aquatic Ecosystem
Protection in the United States
The dilemma faced by modern industrial
societies that accounts for the need to protect
natural environments is discussed in the text box
(A) and diagramed in Figure 2 using the Energy
Systems Language (Odum 1971, 1983). The
United States government has tried to solve this
dilemma by promulgating environmental laws
and regulations aimed at limiting environmental
degradation. The Clean Water Act (CW A) is
foremost among the laws passed to protect
aquatic ecosystems. Among other things, this law
charges the states with periodically assessing the
condition of all fresh and salt waters within the
state's boundaries (Section 305 [b]) and with
reporting those water bodies that do not meet
water quality standards (Section 303 [d]). This
assessment is performed within the context of a
regulatory structure (water quality standards,
criteria, designated uses, etc.) that is intended to
3

-------
Conceptual Models and Methods
____1-
I
C, Land
Conversion
-----'"
Figure 2. Diagram of the environmental system of a region, in which lines with arrows show the chain of
connections that link economic activity (circles on the right) with environmental degradation
(see Figure 3).
ensure that the quality of aquatic environments is
maintained within limits that are acceptable to
society .
Currently, the states and the federal
government deal with the difficult tradeoff
between economic prosperity and environmental
quality by assigning designated uses to the various
water bodies. The capacity to choose the best use
for a particular water body makes it possible to set
physical, chemical, and biological water quality
standards that correspond to those expected for a
certain degree of desired economic development.
However, the law does not allow designated uses
to be created only to support economic
development, in fact, economic feasibility and the
water body's pre-existing condition at the time the
Clean Water Act was passed are only two of many
considerations. The anti-degradation clause of the
CW A is intended to prevent backsliding.
However, all these legal safe-guards may not be
sufficient to counter the tendency.for economic
priorities to drive society's expectations for the
condition of the environment.
The periodic monitoring and assessment of the
condition of all water bodies under Clean Water
Act Provision 305(b) is intended to ascertain
whether a given water body has attained the
desired status for all applicable water quality
standards. If applicable water quality standards are
achieved, no further work must be done. In the
event that a physical, chemical, or biological water
quality standard has been consistently violated, a
4

-------
Conceptual Models and Methods
(A) The Inherent Conflict Between Economic Prosperity and Environmental Quality
The ecological and economic context for society's practical need to establish regulatory mechanisms
is illustrated by a simple energy systems model shown in Figure 2 (Campbell 2001 a). This model
illustrates the dilemma faced by modern society on every scale of human activity: A fundamental
conflict within society arises because economic prosperity (1) is not obtained without the use of
resources (2) which inevitably leads to the creation of wastes (3) and degradation of the natural
environment (4). Furthermore, resource use promotes the expansion of the human population which
requires more support area (5) causing formerly natural lands to be appropriated for human use (6). As
natural lands (7) are converted into farmlands and cities and the functioning of the remaining natural
areas is compromised by pollution, the critical life-supporting services that the environment provides
to individuals and society (8) are gradually lost. When total system productivity begins to decline as a
consequence of this lost support capacity, people perceive that their standard of living is declining and
they look for ways to mitigate environmental degradation without sacrificing economic productivity.
Establishing Total Maximum Daily Loads. for a pollutant and other regulatory mechanisms are ways
that society has chosen to try to restore a balance between human and natural use of resources. More
permanent solutions may be found by changing system designs to make human systems more closely
reflect natural systems in their structure and function (Holmgren 2002).
process is set in motion to restore conditions to
meet the standard. This process includes
designating the water body as impaired by placing
it on the 303 (d) list, diagnosing the cause of
impairment, allocating observed impairments
among multiple demonstrated causes, acting on
the diagnosis by establishing a TMDL or other
regulatory program to control the stressor or
stressors of concern, and monitoring the
mitigation of adverse effects, which may confirm
the diagnosis. Continued monitoring of state
waters is used to determine if the actions taken are
sufficient to restore the water body to the water
quality standards that accompany a given
designated use.
1.2 Use of Conceptual Models in Monitoring
and Assessing Ecosystems
An understanding of the interrelationships
between ecosystem components and processes and
environmental conditions is critical for success-
fully linking stressor sources to biological effects
in water bodies receiving a pollutant. In addition,
knowledge of the cause and effect relationships
operating within an ecosystem is a critical element
of successful diagnosis and of successful monitor-
ing and assessment programs. It may be helpful to
examine the usefulness of conceptual model
building in guiding monitoring research as a
means for understanding how we might use
conceptual models to guide diagnostic research.
Manley et ai. (2000) argued that conceptual
models are the foundation for scientifically-based,
ecologically-focused monitoring programs.
Conceptual models are viewed as "working
hypotheses about ecosystem form and function"
(Manley et ai. 2000), and as "qualitative or
quantitative statements concerning the nature of
ecological risk" (Gentile et ai. 2001). Conceptual
models are used to inform resource managers and
scientists about the critical elements of ecosystems
and their relationships to environmental stressors,
and they are applied to guide the scope and scale
of monitoring programs. Grant et ai. (1997) and
Gentile et ai. (2001) list several stages of
conceptual model development: defining goals
and objectives, delineating spatial, temporal and
ecological scales and boundaries, identifying
sources of natural and anthropogenic stressors,
5

-------
Conceptual Models and Methods
identifying critical ecosystem components at risk
from the stressors, identifying causal relationships
between system components and stressors,
developing a graphical model, and describing
expected patterns of model behavior.
Ecosystems have been defined on the basis of
both structure and function. Ecologists are
somewhat artificially divided into two schools of
thought regarding ecosystems: those who view
ecosystems as collections of populations or
communities and those who view ecosystems as
collections of processes and functions (O'Neill et
al. 1986). The population-community approach
views .ecosystems as networks of interacting
populations. The process-function approach views
ecosystems as the sum of the physical, chemical,
and biological processes active within a space-
time unit. While it is tempting to emphasize one
approach over the other, there is danger in
assuming that relevant system dynamics can be
understood through only one approach.
Manley et al. (2000) addressed the structure
vs. function problem in developing conceptual
models for the Sierra Nevada Ecosystems Project.
Their approach clearly emphasizes ecosystem
processes, because of the abil ity of energy and
material flows to integrate system components
through space and time, while also including
broad structural components. Gentile et al. (2001)
used a risk-based approach to develop conceptual
models for managing South Florida ecosystems.
Their approach, while considering ecosystem
structure and function, also emphasizes ecological
values (ecosystem sustainability), endpoints
(attributes of ecological and/or societal
importance), and measures (stressor-response
relationships). The goal of Gentile et al.'s risk-
based approach is to "establish a parsimonious set
of endpoints for each ecosystem of concern, such
that any change in the structure or function would
be manifested as.a change in one or more of the
endpoints" (Gentile et al. 2001). Similarly, our
approach to developing conceptual models will
incorporate both structural and functional
elements and endpoints.
6

-------
Conceptual Models and Methods
2.0 Methods for Developing Conceptual Models
Because energy transformation is the
underlying cause of all phenomena, the structural
and functional elements of ecosystems can be
integrated and related to external forcing
functions and stressors by constructing networks
of energy flow using the Energy Systems
Language (Odum 1994). For this reason, one
approach that we used to develop conceptual
models of stressor action within ecosystems
began by first considering the. energy
transformations that are the proximal cause of the
behavior of the system variables that we want to
understand and predict (e.g., the metabolic
processing of materials by organisms affects in
situ concentrations of nitrogen, carbon,
phosphorus, and oxygen in a water body). Next,
we consider energy transformations of the larger
system, which often account for the origin of
stressors. The larger system provides the context
for understanding and interpreting stressor action
on the ecosystems of a water body. First, the
larger context for the water body of concern is
established by characterizing the natural
processes and human activities of the next larger
system (usually the watershed); then we
formulate specific methods and models to address
diagnostic research needs and determine the
causes of impairment within those water bodies.
These research needs (i.e., methods, models and
other tools) are defined from a study of the
requirements of existing water quality regulations
and the current methods that states, regions, and
tribes use to comply with these regulations.
In this study, model development was
initially performed using narrative descriptions
and simple summary diagrams. These verbal
descriptions were formulated as detailed concep-
tual models using the Energy Systems Language,
ESL (Odum 1983, 1994). The ESL (Figure 3) is a
symbolic language that is ideal for building
conceptual models of ecologicai and environ-
mental systems because the symbolic objects of
the language correspond to components and
processes found in ecological and economic
systems and are mathematically defined. Concep-
tual models created in ESL are easily converted
into quantitative mathematical models for
simulation and prediction of ecological variables.
2.1 The Context for Diagnosing Impairment I
to Aquatic Ecosystems
The key concepts needed to gain an overview
of the diagnostic process were derived directly by
considering the implications ofthe words in our
charge given in the Aquatic Stressors Framework
(USEPA 2002b), which was to understand and
predict the actions of stressors on aquatic
ecosystems. In addition to sorting out the effects
. of multiple stressors and determining the cause of
observed ecological impairments, we recognized
that the states and tribes needed a way to simplify
the process of determining the causes of impair-
ment to efficiently deal with an overwhelming
number of impaired water bodies. We propose to
. accomplish this end by classifying ecosystems
based on differences in their response to stressors.
The need to produce a stressor-based classifica-
tion system was a primary consideration in the
development of our conceptual models, thus, the
conceptual overview described here provides the
basis for developing and testing our classification
system as well as a context for the development
of diagnostic methods and models. The ideas that
provide a context for our research are discussed in
more detail below.
2.1.1 Aquatic Ecosystems and the Hydrologic
Cycle
. All systems of concern in this document are
aquatic ecosystems and as such, they are the
result of water flows in the global hydrological
cycle (Figure 4). Aquatic ecosystems can be
visualized as a network of water storage units of
different kinds and sizes (e.g., streams, wetlands,
lakes, estuaries) that are interconnected by water
flows. Aquatic ecosystem units are arrayed over
the surface of the continents from pole to pole
and from sea level to the highest mountains
(Bailey 1998). Stream flow is unidirectional and
each storage unit is connected to the one below it
in the network (Figure 4). Aquatic system units
are also linked by bidirectional water flow in
estuaries seaward of the head of tide. The
dissipation of the energy carried by the flow of
water over the landscape creates a hierarchical
7

-------
----.
Conceptual Models and Methods
0-
-q
-r-
?
X
+9
9-
-r}
Energy circuit A pathway whose flow is proportional to the storage or
source upstream.
Source A forcing function or outside source of energy that delivers forces
according to a program controlled from outside.
Tank A compartment or state variable within the system that stores a quantity
as the balance of inflows and outflows.
Heat sink Dispersion of potential energy into heat accompanies all real
transformation processes and storages. The energy leaving the system
at this point is no longer usable by the system.

Interaction Interactive intersection of two pathways that are coupled
to produce an outflow in proportion to a function of both; a work gate.
Consumer An autocatalytic unit that transforms energy, stores it and feeds
it back to improve inflow.
Producer Unit that collects and transforms low-quality energy under the control
of high quality flows.
Box Miscellaneous A symbol to use for whatever unit or function is needed.
Switching Action A symbol that indicates one or more switching actions
'controlled by a logic program.
Figure 3. Definitions for the basic symbols of the Energy Systems Language (modified from Odum 1994).
network of interconnected storage units and flows
that each have their own unique characteristics for
transporting and processing energy and materials.
2.1.2 Diagnosis and Management at the
Watershed Scale
Water flows organize the natural landscape
hierarchically so that aquatic ecosystems are
distributed in space and linked by converging
water flow (Figure 5). Water flows that connect
one ecosystem unit to another on the landscape
transfer materials in excess of each ecosystem's
processing capacity from one system to the next
converging materials and creating a three
dimensional hierarchy of landform and energy
and material flows. In Figure 6, this complex
process is visualized as a longitudinal series of
connected water bodies, in which the excess
nutrients supplied to an ecosystem cause
production, P, to exceed respiration, R, producing
an excess of fixed carbon over that which the
ecosystem can process. This excess is transferred
by water flows to the next system in the series
where it stimulates additional respiration, thereby
increasing nutrient remineralization. Any excess
nutrients are in turn transferred to the next system
in the longitudinal series, where they may
stimulate increased production (Odum 1971). The
transfer of materials is more complicated in
systems with bidirectional flow, but the same
model will apply wherever there is a net transfer
of nutrients. In summary, aquatic ecosystem
system functions are linked across multiple scales
in the nested hierarchy of watershed organization,
and this condition must be considered when
diagnosing the causes of impairment within a
particular water body and when promulgating an
effective TMDL for a pollutant with
hierarchically structured sources and flows.
8

-------
Conceptual Models and Methods
Bog
I , I I I , ,
I I I , , , ,
I , I , , , ,

... ~ / " " / "

I , , I ,
~ I I I I
J... ' , ,
,. ~ I I

;.~
Rain
~.
. ...
~
Evaporation
. .. . .
.
.
Freshwater Marsh
Lake
Floodplain Forest
River
Estuary
Salt Marsh
Continental Shelf
Figure 4. Aquatic ecosystems storing water within the hydrologic cycle.
A
4"""'"''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''............~
Land
Sea
D
Figure 5. Networks of ecosystem units arranged on the landscape within fundamental
watersheds A, B, C, and D each of which contains a river system that empties into
the sea. Dashed lines indicate watershed boundaries.
9

-------
Conceptual Models and Methods
P>R
R>P
P>R
Figure 6. Aquatic ecosystems are arranged on the landscape in a longitudinal series of processing units
dominated by production (P) or respiration (R). Excess production in one system sets the
stage for excess consumption in the next and vice versa. .
(B) Definitions of Ecological Stress and Emergy.

Stress is an energy drain caused by an injury or impairment to an ecosystem that results trom the
overuse of one or more ecosystem components or processes compared to a typical or original
functional state (Odum 1968). In general, stress is the result of a change or perturbation in the long
term or "normal" emergy signature (see Text Box C) of a place. In many places, the long term emergy
signature is primarily comprised of inputs from the natural environment. The change in external
forcing is ofteri a product of human activities and it invariably results in a change in the "normal" or
expected functioning of the ecosystem under the original signature.
Emergy is the available energy of one kind, previously used up, directly and indirectly, to make a
product or service (Odum 1996). Its unit is the emjoule. Em- is an acronym for energy memory that
indicates the energy associated with a quantity has been used in the past (Scienceman 1987). Emergy
can use any kind of energy as the base, for example coal joules, solar joules, etc. However, in
evaluating environmental systems, we commonly use solar energy as the base unit. Solar emergy is the
available solar energy previously used up to make a product or service. Its unit is the solar emjoule
(abbreviated sej). Available energy is energy with the capacity to do work, sometimes called exergy.
A vail able energies in different products and processes do work of different kinds when used in a
system network. In calculating the emergy of an item, the available energy of different kinds must be
converted to a single kind to make them comparable. These transformed values are the emergy
contributions of each required input to the production process and they can be summed to determine
the emergy of the item. Transformities are the coefficients by which the available energies input to a
production process are multiplied to get emergy. Solar transformity is the solar energy required to
make one joule of a product or service. Its units are solar emjoules per joule (sej/J). The transformity
of a product is its solar emergy divided by its available energy. Thus, the fundamental equation of
emergy analysis is:
Emergy = Transformity X Available Energy (exergy).
The change in empower (emergy per unit time) that occurs in an ecosystem under stress is a measure
of the ecological cost or benefit that results from the change in forcing inputs. This change can take
various forms depending on the frequency, magnitude, and duration of the change in' forcing functions
and the properties of the system under stress (Campbell 2000, Holling 1986).
10

-------
Conceptual Models and Methods
2.1.3 Stress in Aquatic Ecosystems
One common anthropogenic cause of stress
(see text boxes A and B) in ecosystems is the
production of wastes of various kinds. Wastes are
produced as a consequence of human activities on
the landscape (Figure 2) and transported through
aquatic environments (Figure 4) which are
changed as a result. Three of the four stressors
that we are examining, i.e., suspended and bedded
sediments, nutrients, and toxic chemicals, are
wastes resulting from human activities, which
often have stressful effects on the biota when
introduced into aquatic ecosystems at concentra-
tions greater than those the system is able to
process. A conceptual representation of the
impact pathway that results in stress in an
ecosystem is a simple chain of cause and effect:
Human activities ~ pollutant sources ~ presence
of the stressor in the environment (e.g., the
concentration of a pollutant) ~ observed effect
(e.g., a biological impact).
Given this causal chain and the state of our
present knowledge of aquatic ecosystems
(USEPA 1998, 2000b), we asked two questions,
"What tools do we need to make a definitive
diagnosis ofthe cause of stress observed in
aquatic ecosystems?" and "What do we need to
know to develop effective management plans to
control the impact of stressors on aquatic
ecosystems?" These questions and an
examination of the process that the states and
tribes currently use' to answer them (GAO 2002)
led us to propose the development of five
diagnostic tools and a classification scheme.
2.2 Development of Diagnostic Tools
Diagnostic tools need to be developed that
will allow us to meet the research objectives. A
brief description of the diagnostic tools proposed
to address these tasks is given below.
(I) Causal webs: Restoration of aquatic
ecosystems and the mitigation of observed effects
require us to demonstrate the link between
sources of a pollutant in human activities on the
landscape and concentrations of that pollutant in
the environment. The construction of causal webs
from both top-down and bottom-up perspectives
was proposed as a way to understand and trace
the links between human activities in the environ-
ment, pollutant sources, stressor concentrations
and observed effects. .
(2) Canonical models: A canonical model is a
simple diagrammatic and/or mathematical model
that captures the key characteristics of a system.
For example, the canonical model of bidirectional
flow describes an estuary, whereas, unidirectional
flow describes a stream. Where sources of pollu-
tants and their effects are present in aquatic
ecosystems, we need to determine the mechan-
isms of stressor action and control. To address
this need we developed an overview conceptual
model and a series of standard models with
simple properties (canonical models) that were
used to guide the development of detailed models
and a classification system and to estimate the
strength of links within a spatially distributed
network of aquatic systems having different
residence times and processing capacities for the
pollutant.
(3)Watershed scale for diagnosis and control:
When sources of pollutants and their effects are
present in watersheds and in their associated
aquatic ecosystems distributed over a landscape,
there is a need to determine the appropriate scale
of control (water body, local watershed, next
larger watershed, etc) that is required to develop
an effectiv.e regulatory program to control a given
stressor or combination of stressors. To address
this need, we developed a conceptual overview of
the way that aquatic ecosystems are linked on the
landscape and proposed a method for determining
system connectivity and the scale of regulation
needed to implement a successful TMDL using a
linked network of canonical models.
(4) Pollutant Identification and Evaluation (PIE): .
A definitive diagnosis of the cause of an observed
impairment requires that we demonstrate the
operation of the impact pathway at the site of the
impairment. This is accomplished by establishing
the presence of source, stressor and effect and
then demonstrating the link between stressor and
effect by showing that the observed concentration
of the pollutant can produce the observed
11

-------
Conceptual Models and Methods
biological effect under the pr~vailing conditions
at a given location. This need will be answered
for estuaries by the PIE field and laboratory
methods under development.
(5) Models for the allocation of effects: The
canonical models serve as a starting point for
constructing more detailed quantitative models
for further development, evaluation and
simulation of particular indicators. In this study,
detailed conceptual models of the action of
individual stressors within an ecosystem were
constructed using Energy Systems Language
(Odum 1971, 1983, 1994). Where multiple
stressors are shown to be causing an observed
impairment, quantitative versions of these more
detailed energy systems models for computer
simulation can be developed to allocate observed
effects among several active causes, i.e., stressors
for which source, stressor and effect and the links
between source and stressor and stressor and
effect have been demonstrated using PIE or other
methods.
(6) Response-based classification: The overview
conceptual model of the factors controlling the
action of stressors in aquatic ecosystems and the
series of canonical models based upon it will
serve as the basis for defining classification
criteria and developing and testing model-based
and statistically-determined classification
systems.
. 2.2.1 Causal Webs: Bottom-Up and Top-Down
Linkages within a network can be shown by
using the weight of evidence approach, which
relies on the specification and documentation of
verified impact pathway models to trace the
production of a stressor or stressful condition
(Le., an observed impairment to an aquatic
ecosystem) to human activities. These conceptual
models may be built using either a top-down or
bottom-up approach and include transport
mechanisms and other factors linking human
activities, sources, stressors, and effects on the
landscape. Figure 7presents a conceptual model
showing how these impact pathways may be
delineated from both the top-down and from the
bottom-up perspectives. For any given analysis of
landscape patterns, each link in the chain
connecting the stressor with the effect must be
shown to be present and capable of producing the
hypothesized action. For example, lawn chemical
application must be sufficient in a certain ur.ban
area to account for the observed concentrations of
pesticide in sediments when transport and
microbial break down of the substance have been
accounted for. Because resources and their
concomitant human activity patterns are often
clustered as by-products of an underlying process
of organization, multiple impact pathways often
operate at the same time and this complicates the
diagnostic process. Where more than one impact
pathway can be shown to be potentially active the
PIE diagnostic methods can be applied to a
random selection of sites to determine the active
agents of impairment. If any step in the impact
pathway model can be shown, on the basis of
data, to be insufficient to result in the observed
distribution of a stressor; additional research is
begun to find sources that can account for the
observed distributions.
For understanding the causes of impairment
in ecosystems, one may use either inductive
(bottom-up, diagnostic) or deductive (top-down,
predictive) reasoning (Ziemer 2004).
Environmental monitoring has traditionally used
the inductive, diagnostic approach based on
observation and correlation. For example, the loss
of species abundance in streams is noted and
appears to be related to the loss of habitat used by
that species. Habitat loss is correlated with
increased sediment loads in streams. The
sediment load appears to be related to increased
erosion from watersheds, which have been
logged. Using an inductive approach and this
chain of evidence, we can conclude that declines
in species abundance or condition are the result of
timber harvesting (Figure 8a).
Causality can also be investigated by
reversing the usual diagnostic path, which is
inductive. The top-down, deductive approach
attempts to predict the impacts to ecosystems
based on known or anticipated responses of the
systems components to stressors. Using the
timber harvesting example above, we predict that
loss of forest cover in the watershed due to clear-
cutting will result in exposure of the soils to
weathering, leadil1g to increased losses of soil
from the watershed. This soil enters the streams
12

-------
Conceptual Models and Methods
Land Use
(logging, farming,
urbanization)
~
o
-c
~
o
~
=
On-site Change
(soil, vegetation, etc)
t=
o
.....
.....
o
a
~
-c
Altered Watershed Products
(water, sediment, nutrients, heat)
I Transport Processes I

t .

I Off-Site Impacts I
Adapted from Robert Ziemer, USDA FS RSL)
Figure 7. Illustration ofthe top-down and bottom up approaches to tracing causal pathways (Ziemer 2004).
where it accumulates altering stream hydraulics
and habitat that is necessary for the biota, leading
to a decline in species abundance or condition
(Figure 8b).
Either approach may illuminate the cause of
impairment, if ecosystem processes are
adequately understood and sufficient stressor-
response data are available. In practice, the level
of understanding and adequacy of data are rarely
met and neither approach should be used
exclusively. This is especially true when multiple
stressors and complex systems are involved. Our
method uses both the top-down and the bottom-
up perspectives, and in so doing combines the
strengths of predictive modeling based on first
principles with the empirical verification of
mechanisms upon which accurate diagnoses
depend.
2.2.2 Canonical Models
In general, we can conceptualize the
connection between contiguous waters using
. canonical energy systems models (Figure 9). A
canonical model captures a process or represents
a system of interactions in a simple form. The
canonical models given in this paper are used to
represent the storage units of aquatic systems on
the landscape and the links between them. Source,
stressor and effect are dynamically linked
together by energy systems models in their fully
developed form. If a canonical or other model is
applied to every grid square of a spatial field, or
in every element of a connected network, it is
termed a unit model (Odum 1994). The network
of ecosystems described above may be simply
modeled as an interconnected series of canonical
energy systems models that are inserted into each
cell, such that the model components and flows
can assume different values in every cell based on
the ecosystem characteristics and the forcing
functions at the boundaries. In its simplest form
the canonical model for a stressor acting on a
stream reach in a single unit of the landscape
accounts for inputs of the pollutant to the stream
reach from upstream, the addition of the pollutant
over the length of the stream segment from its
watershed, and the assimilation or break down of
the stressor by chemical and biological processes
taking place within the water body (Fig. 9a). The
13

-------
Conceptual Models and Methods
Bottom-Up
Approach
Watershed
Clearing
~ ~.. ?

...,..................",............,
Decreased
vegetative cover
.

I Increased erosion
+

I Sedimentation I
.
I Habitat loss
Non-point
I Permit violators I . sources

1 pOint~,s 1/

I", /1 Toxicity I


I Species declines I
Figure Sa. Conceptual model of the bottom-up approach to tracing causality through a web
of interactions. In this example, the observation of a species decline initiates model
construction.
Top-down
Approach
Watershed
Clearing
Decreased
vegetative cover

+
/
~
Increased soil
compaction
+

I Increased runoff
.
Channel modifications
Figure Sb. Conceptual model of the top-down approach to tracing causal pathways through
a web. In this example, the observation of watershed clearing initiates model
construction.
14

-------
Conceptual Models and Methods
output variable for this basic model is a flow of
the pollutant moving into the next .stream segment
or downstream water body.
For estuaries, the rules that govern transport
will be more complicated, but the unit model will
be similar in structure. The common and
distinguishing properties of the four canonical
models used to represent aquatic ecosystems with
different hydrological properties are shown in
Figure 9. Three prominent factors are derived by
inspection from the structure of energy systems
models of aquatic ecosystems and these are
hypothesized to be the primary factors controlling
the stressful actions of pollutants in aquatic
ecosystems. They are (1) the residence time of
water and pollutant in the system, (2) the natural
processing capacity of the system for the
pollutant including the pathways that decompose,
take-up, or sequester the material, and thereby
determine its turnover time, and (3) additional
factors that modify the form of the exposure-
effect relationship or the kind or intensity of
action that the pollutant exerts within the
ecosystem. Processing capacity can be altered by
the presence or absence of other materials
changing the biologically effective concentration
of the pollutant. These factors are designated as
type A modifiers, i.e., they alter the biologically
available concentration ofthe pollutant. Other
additional factors (3) change the form or character
of the relationship between the biologically
available concentration and the observed effect.
They are designated as type B modifiers, e.g.,
turbidity is a type B modifier of nitrogen
availability, because it alters the primary
production obtained by adding an additional unit
of nitrogen.
These three factors can be evaluated in a
manner that quantitatively determines the
effective exposure to a pollutant that is
experienced by an ecosystem. For example, the
product of the average ambient biologically
effective concentration of a pollutant and the
average time that an entity is exposed to a
molecule of that material determines the dose
received by the entity, e.g., an ecosystem or one
of its components. We hypothesize that different
ecosystems will have characteristic properties
related to residence time, processing capacity, and
modifying factors, which can be used to
differentiate classes of ecosystems, and that that
these classes may be distinguished by the
different effective exposures that will occur when
the systems are loaded with a given quantity of
the pollutant. The problem of diagnosis can be
further simplified if pollutants can be grouped or
classified according to their activity such that an
ecosystem type processes all members of a class
of pollutant in a similar manner. In this case," the
bioavailable concentration might be expressed in
aggregate units, i.e., standard toxicity units for a
given class of material. A consideration when
using source, stressor, and effect within a
classification scheme is that the average
bioavailable concentration and residence times for
the pollutant must match the time periods that are
relevant to producing an effect. For example,
readily available data from standard monitoring
programs may not be sufficient to account for the
effects of an extremely toxic pollutant buried in
the sediments, if its concentration is suddenly
increased due to roiling of the sediment during a
storm. In this case, the relevant time period for
determining source, stressor and biological effects
might be hours rather than weeks or months. In
streams, episodic events such as floods often
govern ecological responses.
When quantitatively evaluated, residence
time and processing will give an expression for
the exposure of the ecosystem to biologically
active concentrations of a particular stressor:
Residence Time (days)* Bioavailable
Concentration (g m-3) = Exposure (g m-3 - days)
In addition, Type B modifying factors shift
the relationship along the exposure axis, causing a
different effect than that expected based on the
typical stressor-response relationship. The
application of Type B modifying factors to the
calculated exposures gives an estimate of effective
exposure, which is the variable that drives model-
based classification.
We recognized the need for two configura-
tions of the basic model to account for uni and bi-
directional water flow and two additional .
configurations to represent systems of each flow
type where processing capacity and/or residence
time is not homogenous. These four canonical
models are unidirectional flow (Fig. 9a),
bidirectional flow (Fig. 9b), unidirectional flow,
stratified (Fig. 9c), and bidirectional flow,
15

-------
Conceptual Models and Methods
{
Inflow
P = Pollutant
M = Modifying Factors
Type A and B
Residence Time
A
,
Outflow
Figure 9a. Model of the factors controlling the action of stressors in aquatic ecosystems
.within a water storage unit with unidirectional flow.
Inflow
-
P = Pollutant
M = ModifYing Factors
Type A and B
Water Storage
Residence Time
(
1'0
Net Export or
Import
Figure 9b. Model of the factors controlling the action of stressors in aquatic ecosystems within
water storage units with bi-directional flows
Water Storage
16

-------
Conceptual Models and Methods
stratified (Fig. 9d). Additional or other internal
differences in flow an.d/or processing capacity can
be accommodated by adding compartments to the
stratified modules. For example, sediment and
water might be the two relevant compartments in
the processing of toxic chemicals, or in broad
estuaries lateral flows can be considered by
adding adjacent units. When quantitatively
evaluated, these canonical models will provide, as
output, the effective dose of a pollutant that we
expect to be related to the observations of
biological effects and ecosystem condition
(Figure 10).
The four canonical models given above were
applied to;develop more detailed models that
include multiple stressors (Section 3). Transfers
from one canonical module to another can be
used to determine the appropriate scale for
effective management of a pollutant as described
. below. In taking this approach, we assume and
will try to verify that observed effects are a
function of the effective exposure of the biota to a
pollutant according to a relationship that is
characteristic for an individual stressor (Figure
10). Also, we assume that a molecule of a
material will be processed through similar
pathways regardless of the ecosystem in which it
is found, e.g., denitrifying bacteria will always
reduce nitrate to nitrous oxide a:nd diatomic
nitrogen gas. A known or demonstrated
relationship between effective exposure and an
observed biological effect allows us to use
calculations of effective exposure as a
classification variable to determine groupings.
Other factors in the canonical models like the
amount of a pollutant that is processed annually
might also prove to be good general classification
variables.
Residence Time
{
A
,
Inflow
Water Storage
Outflow
-
P = Pollutant
M = Modifying Factors
Type A and B
Residence Time 2
Figure 9c. Model of the factors controlling the action of stressors in aquatic ecosystems
within water storage units with unidirectional horizontal flow and two different
processing capacities, e.g., stratified systems or sediments and water column.
17

-------
Conceptual Models and Methods
Residence Time
r
"
Inflo
Water Storage
-
Residence Time 2
P = Pollutant
M = Modifying
Factors
Figure 9d. Model of the factors controlling the action of stressors in aquatic ecosystems within water
storage units with bidirectional flows and two processing capacities, e.g., stratified systems
or sediments and water column.
Pollutant A
Effect Measure
(arbitrary units)
Pollutant B
Exposure
Figure 10. Two hypothetical exposure-effect relationships for two different pollutants.
18

-------
Conceptual Models and Methods
2.2.3 Watershed Scale/or Diagnosis and
Control
Often watersheds, resource distributions, and
human activity patterns do not follow political
boundaries. Therefore, it is important that the
states use a consistent protocol and method for
the collection and analysis of data when
compiling their 305(b) and 303(d) lists, so that
comparable states of impairment can be
determined, shared water bodies accurately
assessed, and TMDLs or other regulatory
programs implemented using boundaries
determined by the controlling processes rather
than political divisions (GAO 2002, USEPA
2002c). Consistent data on the condition of state
waters displayed in a geographic format allows
the national distribution of the condition of
aquatic resources (water bodies, streams,
wetlands), human activities, water quality
variables, biological variables, etc. to be seen and
studied (for example, see www.epa.gov/waters).
The accurate assessment of shared water bodies
requires that the states coordinate their
assessment activities with adjacent states and
share the relevant 305(b) data, so that
distributions of the condition variables may be
plotted for an entire watershed, wherever
watershed boundaries extend across the borders
of two or more states. When the distributions of
stressors and/or effects are clustered together on
the landscape, underlying factors may be
responsible for their convergence, and there is a
need for a broader analysis of the underlying
factors, pollutant tra~sport, and the connections.
between tributary stream segments and the
receiving water bodies.
The fate and transport of a pollutant within.
the aquatic ecosystems of hierarchically nested
watersheds is the key factor in determining if a
TMDL must be implemented at a scale larger than
the individual water body. The evaluation of a
linked series of fate and transport modules
provides the conceptual basis for a method to
determine the appropriate boundaries over which
TMDLs must be developed for a connected
network of water bodies. In the unit model given
above, the two primary factors that determine the
residence time of the stressor within the system
are the ecosystem's processing capacity for the
pollutant in the water body and the transport of
the pollutant through the system. If pollutant
sources from a larger system contribute more than
a certain percentage of the loading to a water
body, a TMDL being developed for the water
body would need to consider the pollutant loading
from the larger system to be effective.
Existing water quality analysis programs,
such as the Office of Water's BASINS (see
http://www.epa. gov/OST/BASINSI) and its
associated models, Qual2e and Hydrological
Simulation Program-Fortran (HSPF) are able to
make detailed analyses of the transport and
assimilative capacity of streams and to define the
areas that an effective TMDL must consider.
Coupling these watershed models with a 2 or 3
dimensional estuary model allows seaward
sources of the polluting materials to be
considered. One problem with such models is that
they are data intensive and may require
considerable research effort and expense
gathering the needed data, modeling and
analyzing the results. This makes an accurate
diagnosis of the causes of impairment imperative
before large amounts of money are spent on a
detailed model analysis to set a TMDL. A
coupled set of the simple canonical models
presented above may allow us to make a quick
first order determination of the watershed scale
necessary for the regulation and control of a
particular pollutant and help focus diagnostic
. research on the primary cause of impaifl11ent.
Further analyses using detailed numerical
modeling programs can be employed to develop
or refine TMDLs for stressors shown to be the
primary cause of impairment.
The problem of diagnosing the causes of
impairment at the watershed scale can be
approached by first compiling a data base for
stream reaches and other water bodies covering
the area within a fundamental watershed.
Fundamental watersheds are defined here as
networks of water bodies and their associated
aquatic ecosystems that are linked by flows of
water and have a terminal connection to the open
sea or to one of the Great Lakes (Figure 5). If
pollutant inputs from the open sea and the
atmosphere are small relative to watershed fluxes,
fundamental watersheds represent the largest area
within which wetlands, stream segments, lakes,
estuaries and their watersheds must be managed
to ensure that limits established for pollutants and
habitat alteration will be effective. The data on
19

-------
Conceptual Models and Methods
land use, pollutant sources, mass transport,
pollutant concentrations, and biological
conditions in the water bodies and stream reaches
within a fundamental watershed can be mapped to
show the distributions of the pollutant in the
individual water bodies, the transport of the
pollutant from one system to the next, and the
association between human use patterns, pollutant
sources and observed impairments. The evidence
supporting the presence of a causal chain linking
source, stressor, and effect can be assembled by
evaluating the degree of overlap between overlays
of the spatial plots, showing each link in the
impact chain. The water bodies influenced by
point and non,point sources of a pollutant can be
determined using simple fate and transport
models. Once zones of association among the
three links in the causal chain have been
established using the assembled data, a diagnosis
of the watershed management scale needed to
control the pollutant can be made based on the
coincidence of patterns of source, stressor, and
effect and the magnitude of the pollutant flows
from one system to the next. This diagnosis can
be checked by identifying random. locations
within areas of coincidence in the individual
systems and then performing PIE diagnostic
methods (described in the next section) on
samples from these stations and at these locations
in the water bodies of concern. A pollutant shown
to be responsible for impairment over a portion of
the area of the. fundamental watershed would have
a TMDL determined for the affected portion of
the watershed, which defines the scale of
effective management for that pollutant.
This method can also be used to make
preliminary diagnoses in more complicated cases
where multiple stressors are involved, although
the probability that PIE methods (see section
2.2.4) will be needed to make a definitive
diagnosis is higher than in the simpler single
stressor case. For example, the hypothetical
stressors of acid mine drainage and nutrient
enrichment are active in a West Virginia.
watershed. Assume that old coal mines dot the
northwest quarter of the watershed, where many
stream reaches have low pH. This region drains
into the lower watershed. The entire northeast
quarter of the watershed is forested. Acidity is a
problem in the lower half of the watershed
although no mines drain from that area. Nutrient
enrichment is also a problem in the lower
watershed where several municipal sewage
treatment plants discharge into the river.
Biological impairment of stream biota occurs
throughout the northwest quarter and in the lower
watershed but not in the northeast quarter. No
other stressors have been measured in the region;
however, there is a nearby atmospheric
monitoring station that shows elevated S04 in rain
fall. Applying the spatial coincidence method
proposed above and the rules of logic proposed
by Hill (1965) and USEPA (2000a), we observe
from our overlapping maps that either acid mine
drainage (AM D) and/or acid rain might be
responsible for biological impairments in the
northwest quarter. The fact that no impacts are
observed in streams in the Northeast quarter
allows us to eliminate acid deposition pointing to
AMD as the most probable cause of biological
impairment in the streams of the Northwest
quarter. In this case, the overlapping patterns
show the presence of two possible sources, a
stressor, and an effect, but one source could be
eliminated based on the evidence of co-
occurrence. The information given is not
sufficient to make a similar diagnosis for the
cause of impairment in the lower watershed,
because we are unable to separate nutrient
enrichment from acidity, both of which have
source, stressor,. and effect present. In this case,
direct cause and effect tools that link source and
stressor and stressor and effect such as those
under development in PIE methods can be
applied to try and sort out the cause of biological
impairment in the lower watershed.
2.2.4 Pollutant Identification and Evaluation
Pollutant Identification and Evaluation is a
method for the diagnosis of the causes of
impairment to aquatic ecosystems in estuaries and
other water bodies. PIE is based on the
experimental methods and logical rules of
analysis used to demonstrate the causes of
toxicity that have been developed over the past 15
years in the Toxic Identification and Evaluation,
TIE, research program. Recently, a general
logical approach to determine the most probable
cause of an observed biological impairment in an
aquatic ecosystem has been promulgated by the
USEPA. Stressor Identification, SI, (USEPA
20

-------
Conceptual Models and Methods
2000a) provides a framework for determining the
cause of biological impairments observed in
aquatic ecosystems. Currently, it has been applied
primarily to fresh water streams; however, the
same logical techniques, e.g., elimination,
association, and weight of evidence, can be
applied to determine the causes of impairment in
other water bodies such as lakes and estuaries.
2.2.4.1 Methodsfor Discerning Cause and Effect
The basic principles used to identify the most
probable cause of a condition from observed
associations are elucidated by Hill (1965) for
problems in epidemiology. Such an approach
forms the basis for demonstrating probable cause
in many scientific disciplines. SI, TIE and PIE
use the elements of Hill's causal criteria in
arguing for causation from observations of
association. PIE uses. these logical rules to
determine the most probable cause of an observed
condition;. therefore, it is compatible with SI and
operates within its logical framework, but it also
goes beyond the methods used in SI by
developing laboratory and field methods to
directly demonstrate the presence of causal
mechanisms.
Energy Systems Models approach the
problem of causality somewhat differently by
constructing causal networks based on proven or
hypothesized relationships among system
variables. The fundamental notion underlying this
approach is that the transformation of energy is
the proximal cause of all action. These models
can be further analyzed to rank causes and
allocate effects among multiple causes. Hill
(1965) stated that his rules to make the case for
causation from observed correlation are a second
best, though practical, alternative to the detailed
scientific research and experimentation that is
often necessary to demonstrate the mechanisms of
cause and effect. Modeling causal networks and
developing tools to demonstrate causal.
mechanisms deal directly with causation. Hill's
postulates are logically consistent with such direct
research and make a contribution where such
direct causal analyses do not exist or are difficult
to implement. lt is our position that the USEPA .
needs both kinds of research and tools to .
effectively deal with cause and effect in
environmental systems. Both SI and PIE are
needed and can be used to support the 303(d)
listing process and the development of water
quality regulatory programs such as TMDLs.
Energy systems models of causal networks are
perhaps best used in the allocation of effects
among multiple causes. SI, PIE, and energy
systems models are not identical approaches and'
all three provide unique tools needed in the cause
and effect toolbox. Some of the characteristics of
and differences between Sl and PIE are given in
the following paragraph along with the ways in
. which PIE and SI augment and support each
other.
PIE uses a succinct logical sequence, the
impact pathway, to demonstrate causality and it is
focused on the development of particular tools for
the identification of the causes of impairment in
estuaries, where multiple active causes are the
rule rather than the exception. SI summarizes and
codifies the logical rules of causal analysis in a
method that can be used in the application of PIE
and other analyses. SI relies largely, but not
exclusively, on existing information and the
weight-of-evidence to determine probable cause
based on observed associations. The development
of particular tools to fit within the framework is
encouraged although not a part of SI per se. An
SI analysis is triggered by the observation of a
biological impairment, whereas, a PIE can be
triggered by any violation of a water quality
standard. This is important because the pathway
that a state or local government follows into a
causal analysis is not always through the
observation of a biological disturbance, e.g., the
largest category of impairment on the 303( d) lists
is unknown. SI is designed to be most effective in
determining the cause of impairment when the
weight of evidence points to a single most
probable cause among many candidate causes,
whereas, PIE focuses on developing tools that
will allow diagnoses of the causes of impairment
in estuaries where multiple stressors are the rule
and where several stressors may be actively
causing biological impairment in most systems.
PIE and SI researchers have met several times
over the past two years to compare approaches
and exchange information. For example, the PIE
method considers the sources of stress early in the
diagnostic process and SI adopted early
consideration of source, because it proved useful.
PIE complements SI through the development of
specific diagnostic tools for estuaries that can be
used within the context of the SI approach. At
21

-------
Conceptual Models and Methods
present, SI has been applied primarily to
determine the cause of biological impairments in
fresh water stream systems, where it has proved
to be an effective vehicle for communication and
diagnosis. Currently, PIE tools and methods are
being developed and tested in case studies of two
estuaries and in a retrospective case study of New
Bedford Harbor, MA.
2.2.4.2 Description of the PIE Method
The Pollutant Identification and Evaluation
method has four phases with costs and
complexity increasing from the first to the fourth:
Phase 1: Screening in which we compile existing
data for verification and diagnosis, Phase 2:
Identification in which we establish the presence
of source, stressor, and effect, Phase 3: Linkages
in which we demonstrate the connections between
source and stressor and stressor and effect, and
Phase 4: Confirmation, in which time and
circumstance are relied upon to demonstrate the
accuracy of our diagnoses. Pollutant
Identification and Evaluation uses a phased
approach where a number of candidate causes are
possible and specific tools are used to prove or
disprove logical links and/or evidence is used to
develop a cause and effect relationship or show
the lack of one.
In most cases, the water bodies under
consideration are on the 303(d) list; therefore, the
screening and verification phase should be able to
find evidence that confirms the earlier assessment
of impairment performed by the State. For the
purposes of our diagnostic research, an 'effect' is
defined as an actual biological or ecological
impairment observed at some level of
organization in the water body under
investigation. Therefore, for us the presence of
the stressor alone is not sufficient cause for
concluding an effect is occurring or likely.
Because State and Regional practices sometimes
assign water bodies to the 3.03(d) list based on the
presence of a stressor (e.g., exceeding a Water
Quality Criteria) and without observing a
biological effect, our method could find that no
biological effects are present at a 303(d) listed
site. We believe a scientifically defensible method
must be able to link directly the presence of a
stressor to an effect before making a diagnosis.
The only exception to this definition is when
pathogens are the suspected stressor. In this case,
the presence of pathogens in numbers known to
represent a risk to human health is sufficient
evidence to conclude that an effect is present or
likely. Thus, we have assumed that an effect on
aquatic biota must be detectable for us to make a
diagnosis of the cause of impairment, but we do
not assume that this affect will necessarily be
known before the analysis is carried out.
Phase 1 analysis is triggered once a water
body has been placed on the 303(d) list for a
water quality standard violation, an observed
biological impairment, or other observed
impairment. In Phase 1, data on the water body,
its immediate watershed, and the next larger
system, of which the watershed is a part, are
compiled and used to evaluate the system and, if
possible, make a diagnosis of the cause of the
observed impairments (see Phase 1 in Table 1),
when the evidence is strong enough (See SI
guidance, USEPA 2000a). Screening provides
information on human activities in the watershed
and may provide clues to show how these
activities are linked to the production of
pollutants and to physical alteration of the water
bodies. Phase 1 data may be organized into causal
webs using the top-down, bottom-up or energy
systems approaches to evaluate the interactions
that link human activities with stressor sources
and stressor concentrations within watersheds. If
possible, Phase 1 verifies the probable cause of
impairment by using existing data to demonstrate
that sources of a pollutant are present and that
related effects have been observed in the 303(d)
listed water body in question. If a biological
effect cannot be verified from the existing data,
subsequent phases (2&3) of the PIE method
provide a framework for gathering evidence and
demonstrating that an effect is or is not present
(Table 1). On one hand, if the existing records are
complete and extensive and demonstrate'
impairment, a given water body might go directly
into Phase 3 based on Phase 1 results. On the
other hand, if such complete records show no
impairment is present, the analysis can stop after
Phase 1. A final task in the Screening Phase is to
collect available information on the cross
boundary flows of the stressors. This information
will be used as described above to determine if
diagnosis must be pursued on a scale larger than
the water body to result in an effective TMDL.
22

-------
Conceptual Models and Methods
The compiled data for an individual water body
and its watershed is represented spatially to help
develop sampling plans for the work done in
Phase 2 and 3 and to look for patterns as
described in the preceding section on watershed
diagnosis. The screening process may also
identify data gaps to focus on in Phase 2 and 3.
The list of diagnostic tools shown in Table 1 are
under development for use in Phases 1, 2, and 3
of a PIE; however, this list should not be viewed
as a comprehensive listing of all tools that may be
needed or prove to be useful in establishing the
causes of an ecological impairment. Other tools
may be added to this list to be developed and
tested in the future.
Phase 2 seeks to identify biological effects in a
water body "and the stressors that might be
responsible for the observed impairments. Phase 2
tools (Table 1) provide data to show the presence
of source, stressor, and effect along the causal
pathway discussed above. In addition, Phase 2
tools can corroborate evidence from the state and
regional assessments and the screening and
verification phase. Table 2 identifies the possible
outcomes of Phase 2 and it contains guidance on
how to proceed given these outcomes. Generally,
we recommend that evidence for all three
components of the impact pathway be present to
advance to Phase 3, but we also recommend
courses of action when a specific line of evidence
is absent. In the case where both a source and a
stressor are evident, but no effect can be found,
the stressor might still be moved into Phase 3 due
to the preliminary nature of the search for effects
performed in Phase 2. In all cases the strength of
the evidence should be used along with the
logical rules of causal analysis in eliminating a
stressor in Phase 2 or in moving it ahead to
Phase 3. "
In Phase 3, diagnosis moves from developing
lines of evidence for source, stressor, and effect, "
to a clear and testable demonstration of cause and
effect links between these lines of evidence.
Phase 3 tools (Table 1) seek to provide a "
definitive demonstration that (a) the causal chain
between the source of a stressor and the observed
effect is unbroken and (b) "the stressor is capable
of producing the observed effect under controlled"
conditions and under the prevailing conditions
present in the water body under analysis. Table 2
gives the conditions that may occur at the
beginning of Phase 3 and suggests actions to
confirm or eliminate the stressor. If (a) is false
and no other cause of the impairment can be
found, the stressor is designated for further
research. If (b) is false, the stressor is eliminated
from further consideration. A stressor is
considered to be an active, but not necessarily the
only, cause of an observed effect if the above two
conditions linking source, stressor and effect can
be demonstrated to be true. Table 3 gives possible
results at the end of Phase 3 and the suggested
actions that follow from each sequence of events. "
The strength of the evidence from the Phase 3
tests should be used along with the logical rules
of causal analysis in the elimination of a stressor
or in the diagnosis of a stressor as an active cause
of impairment. "
Phase 4 confirms that the diagnosis of the cause
or causes of impairment is correct. PIE methods
are used to make a diagnosis of impairment for a
water body or watershed based on tests carried
out in the field and in the laboratory. Unlike a
controlled experiment, which can be duplicated in
the laboratory and where the results of a treatment
can be statistically confirmed, ascertaining the
results of manipulating ecological systems in the
fi~ld may be difficult and involve relatively long
time delays before the results of the manipulation
are known. Generally, confirmation of an initial
diagnosis cannot be made until after a TMDL has
been performed and the water body monitored for
some time to determine if the anticipated recovery
takes place. In most cases, it will require several
years of follow-up monitoring to demonstrate that
an initial diagnosis was correct. Successful
classification holds some hope for hastening the
confirmation phase. If water bodies can be
classified into groups with similar behavior under
loading with a pollutant, the results of past
TMDLs could be used to establish an expectation
of confirmation for systems with similar
parameters. In this case, similar circumstances
allow an early peek into the probable outcome of
the confirmation phase. A final confirmation can
only be made after the system has been monitored
for a sufficient time to demonstrate recovery of
the biological variables originally judged to be
impaired.
23

-------
Conceptual Models and Methods
Table 1. Proposed diagnostic tools used in the development of Pollutant Identification and Evaluation
(PIE) methods for determining the causes of impairment to the aquatic ecosystems in estuaries.
Phase 1 is screening and verification. Phase 2 is the identification of source, stressor, and effect.
Phase 3 is establishing the linkage between source and stressor and stressor and effect. Phase 4 is
confirmation of the diagnosis.
Phase Stressor Diagnostic Tool Diagnostic Information Status of Tool 
 All GIS maps. Possible sources. GIS maps and data need to be
  High quality Determine reference sites. checked for accuracy and
  environmental Plot data on GIS maps to give compatibility of data.
  data sets, e.g., spatial relationships for 
  NA WQA, EMAP. sources and stressors. 
  The scientific Identify data gaps. 
  literature.  
2 Nutrients Grain size Indicator of eutrophic Needs to be validated. May
  normalized TOC. conditions (nutrient loading also give information on clean
   with C, N, P). sediment and toxic chemicals.
  River Indicator of normal or Relationships need to be
  DINIDOC/PIN excessive DIN/PIN loadings. developed between DIN, PIN
  and residence  and benthic effects.
  time analysis.  
  Analysis of Indicator of organic enrichment Sampling must be complete to
  dissolved oxygen resulting in eutrophic avoid erroneous conclusions
  concentrations. conditions. due to natural variability
  Dawn-dusk and Effect measure for N, P, and C. 
  continuous.  
 Toxic Toxicity testing Integrator for all toxic May not detect toxicants that
 Chemicals and limited chemicals. have a chronic effect, dietary
  chemical  route of exposure, or for
  analyses.  which organisms may not be
    sensitive (e.g. Hg, dioxins)
    Increasing classes of species
    tested or exposure duration
    may increase sensitivity.
 Habitat Reduction- Effect measure for nutrients, This measure is based upon
 Alteration. oxidation potential suspended and bedded Pearson and Rosenberg (1970)
 by pollutants. (RPD) assessment. sediments, toxic chemicals or 'better' conditions have a
   physical change. deeper RPD than impacted
   Oxygen availability in areas. Some variability is
   sediments. inherent in measurements.
    Validation of habitat quality
    based on RPD depths for
    different grain size habitats
    needs to be performed.
 Clean River particulate Measure total suspended Criteria for suspended and
 Sediments load analysis. particulate inflow bedded sediments in marine
  Turbidity.  ecosystems have not been
    established. Turbidity may
    measure biological particles as
    well as sediment. Use dry
    weight/ash weight
24

-------
Conceptual Models and Methods  
Phase Stressor Diagnostic Tool Diagnostic Information Status of Tool 
Table 1  Grain size Measurement of grain size can Current measurements may be
cont'd  analysis. act as a surrogate measure of needed to verify flow
   energy regime. conditions.
 Pathogens Bacteria Indicator of the presence of Does not indicate whether the
  measurements. bacteria from vertebrate source of bacteria is natural
  For example: E. sources. (wildlife) or from
  coli  human/domestic or agricultural
    activities.
 AIl stressors Benthic General state of benthic Reference systems may be
 (Nutrients community community. May validate difficult to find. Does not
 Habitat analysis. impairment assessment relative currently diagnose a single
 alteration *   to an appropriate reference stressor. Relationships
 Toxic  system. between benthic community
 chemical)   condition and 'natural' forcing
    factors need to be developed.
3 AIl stressors Models Provide information on Specific to the stressor
   mechanistic, organism and modeled.
   ecological pathways that 
   support linkages between 
   sources, stressors and effects 
 Nutrients Measurement of Provides temporal information Correct interpretation and
  N, P, chlorophyIl on the amounts of nutrients quality control of data is
  a and DO over present. critical to avoid false positives.
  appropriate time  Reference estuarine systems
  periods.  may be difficult to find.
    Nutrient criteria for marine
  Stable isotope Linkages to source. systems have not been
  analysis. established.
 Toxic Toxicity Determines which toxic Methods require further
 Chemicals Identification chemicals are causing observed development and validation.
  Evaluations (TIEs) toxic effects Because these methods use
    toxicity tests, all the potential
    issues of toxicity tests are
    inherent in TIEs. See Phase 1
  Physical and Provides information on the Correct interpretation and
  chemical amounts ofbioavailable toxic quality control of data is
  measurements of chemicals present in water and critical to avoid false positives.
  suspected toxic sediments. May provide links New analytes need method
  chemicals. between source, stressor, and development.
  Calculate effect. 
  bioavailability  
  using models.  
 Habitat Physical and Provides information on the Interpretation of data relative
 Alteration* chemical amounts of nutrients, toxic to a reference, baseline or
  measurements chemicals, suspended and threshold measures is critical
   bedded sediments and the to avoid false positives.
  Grain size, salinity physical properties of water and 
  and Toe sediment samples. 
  analyses.  
25

-------
Conceptual Models and Methods
Phase

Table 1
cont'd
Stressor
Suspended
and bedded
sediments
Pathogens
Diagnostic Tool
Calculate and/or
measure energy
regime from grain
size distributions
or current meter
data.
Deposition
analysis.
Bacteria source
analysis.
Genetic finger
printing.
Diagnostic Information

Grain size distribution may
provide information about
energy regime. May provide
historical change in type of
deposition as a result of
agricultural/construction
activities.
Rate of sedimentation.
May link stressor with source.
Distinguish between human and
non-human E.coli, as the
presence of the bacterium often
automatically indicates a
violation of many 'fishable' and
'swimmable' criteria.
Status of Tool
Reference or historical
deposition rates and associated
grain size may not be
available. Criteria for
suspended and bedded
sediments in marine
ecosystems have not been
established.
Determining deposition is
difficult. Criteria for clean
sediment reference sites in
marine ecosystems have not
been established.
Determine which tests are
accurate and available for
extramural purchase.
*Habitat Alteration The stress from habitat alteration occurs in two forms: direct and indirect. Examples of direct
effects include anthropogenic activities that result in physical changes to the habitat like damming, dredging, paving
and filling. Indirect habitat alteration results from the adverse effects of a pollutant, such as, excess toxic chemicals,
suspended and bedded sediments, or nutrients. Both direct and indirect habitat alterations affect benthiC
communities. However, the PIE method is only concerned with diagnosing habitat alteration that results from the
effects of pollutants. The direct effects of habitat alteration on benthic communities will be removed through the
choice of reference systems.
26

-------
Conceptual Models and Methods
Table 2. A Decision Table for evaluating outcomes in Phase 2 of a PIE. For each stressor considered in
Phase 2, there should be evidence of a source, the presence of a stressor, and an effect to move
to Phase 3. An X indicates presence.
Source Stressor Effect Guidelines for Action
   Eliminate stressor from consideration
X X X Move stressor ahead to Phase 3
 X X Look for source, look to a larger system to ensure all possible sources
   are considered. Look for a precursor to the stressor that may have a
   source. Repeat Phase 2. If the source is not found, one may want to
   move the stressor to Phase 3 to determine if a strong linkage can be
   developed between the stressor and the effect. If a strong linkage can
   be developed, one may want to invest more resources to determine if a
   source exists.
X  X May be the result of temporal loading, repeat Phase 2 with attention to
   temporal scales. The source may emit a different stressor with a
   similar effect. Repeat Phase 2 with attention to all possible stressors.
   The source may be spatially linked to a proximal stressor, if one is
   certain the proximal stressor is in the causal pathway, e.g., monitoring
   low DO or chlorophyll events in the absence of nutrient 
   concentrations. If ultimately no evidence of a stressor or proximal
   stressor is found, eliminate stressor from consideration.
X X  Need to ensure that the endpoint measured is a sensitive effect
   endpoint and/or the stressor is bioavailable. Given the preliminary
   nature of Phase 2 effect endpoints, one may want to move this stressor
   forward to Phase 3. Ifno effect is found, eliminate stressor from
   consideration.
X   Eliminate stressor from consideration. Continue assessment of other
   stressors.
 X  Need to ensure that the endpoint measured is a sensitive effect
   endpoint. Need to consider possible lag time for the stressor to have a
   measurable effect, or if the stressor is ephemeral. Need to look at the
   larger system to ensure all possible sources are considered. Look at the
   assimilative capacity and residence time of the pollutant in the system.
   Repeat Phase 2.
  X Consider other stressors that may have similar effect endpoints.
   Consider possible lag time for the stressor to have a measurable effect,
   or if the stressor is ephemeral. Also, look to a larger system to ensure
   all possible sources are considered. Repeat Phase 2.
27

-------
Conceptual Models and Methods
Table 3. A Decision Table for possible outcomes of Phase 3 showing the presence or absence of linkages
between source, stressor and effect. In all cases, the strength ofthe weight-of-evidence should be used
along with professional judgment in elimination or diagnosis of a stressor. An X represents an affirmative
result.
Source and
Stressor
Linked
Stressor and
Effect Linked
Guidelines for Action
x
x
Eliminate stressor from consideration.
Links established between source, stressor and effect. Weight of evidence may'
include two independent lines of evidence that point to the same stressor, and
laboratory and field measures. Diagnosis completed.

Look for source, need to look to larger systems (increase spatial scale) to ensure
all possible sources are considered. Repeat Phase 3

May be a result of temporal loading, repeat Phase 3 with attention to temporal
scales. Source may emit several different stressors, repeat Phase 3 with attention
to all possible stressors. If ultimately no link between a stressor and an effect can
be established, eliminate stressor from consideration.
x
x
2.2.5 Modeling Methods and the Allocation 0/
Effects .
A systematic and exact framework based on
physical laws, e.g., the laws of thermodynamics,
is helpful in modeling, understanding, and.
predicting patterns and processes in the
environment. In this paper, we used such a
framework for model development. Energy
Systems Theory (Odum 1971, 1983, 1994) was
chosen to guide the development of conceptual
models of the action of stressors in aquatic
ecosystems (See Text Box D). Energy Systems
Language is a vehicle for model-building that
allows the symbolic and mathematical representa-
tion 'of the energetic and kinetic aspects of
stressor-response relationships in an integrated
manner. Energy transformation is the underlying
cause of all action and can be used to trace causal
pathways within a network organization. Thus, a
consideration of energy transformations was a
key factor in gaining an integrated understanding
of those factors controlling the fate and transport
of pollutants within an ecosystem, including the
metabolic processing of these materials, and the
action of modifying factors in determining
bioavailable concentrations and the biological
impacts of stressor action.
We developed a hierarchical, modular
framework for constructing conceptual models.
At the highest level of aggregation we designed a
set of four canonical models which can be used to
describe the links between aquatic ecosystems on
the landscape. The more detailed generic
conceptual models for a partieular pollutant,
which are described below, can be plugged into
the processing box of the appropriate canonical
model to represent a particular system in more
detail. The appropriate forcing functions must be
added to accurately represent these more complex
systems. The canonical models serve as a frame
within which more complex pieces can be
inserted to evaluate the details of a particular
pollutant's action within a particular ecosystem
type.
In this paper, we used a progressive process
for the development of detailed conceptual
models. The first step in this process was writing
narrative descriptions of the stressors identified in
the Aquatic Stressors Implementation Plan
(USEPA 2002b). This process allowed everyone
to contribute to the conceptualization process and
it drew upon the combined expertise of the group
in assembling knowledge about each stressor.
These narrative descriptions were produced in a
standard format that facilitated comparison and
the development of overview figures of the
factors controlling each stressor. The information
from the narrative descriptions and overview
figures was then used to construct detailed models
28

-------
Conceptual Models and Methods
for each stressor. We used ESL to develop the
detailed conceptual models for each stressor
including the physical forces and flows that
control the residence time of pollutants in the
system, the biological and chemical factors that
determinepro~essing capacity (type A modifiers)
and the modifying forcing functions and materials
(type B modifiers) that alter stressor action.
The detailed conceptual models serve as a
guide to building quantitative simulation models
for the allocation of effects among several causes
when more than one cause is active. In addition,
the detailed conceptual models can be used as a
checklist of the important processes that might
affect the action of a stressor in an aquatic
ecosystem. These more complex models include
the factors that we believe to be important in .
controlling the action of the four stressors, but
they should not be construed to be a comprehen-
sive representation of the effects of a particular
pollutant in a particular aquatic ecosystem.
Rather, they are a guide to thinking about such
problems. As always, good quantitative models
and analysis will depend on the perspicacity of
the investigators. Nevertheless, with some
practice and the detailed examples as a guide, we
hope that many investigators will be able to use
this model development process and
simplification through functional aggregation
(Odum 1976) to develop their own conceptual
models to guide diagnostic research. The
conversion of simplified versions of these
detailed models of stressor action into computer
programs for simulation will make it possible to
predict the behavior of particular ecosystem
variables under various levels of pollutant
loading. The special character of habitat alteration
as a modifYing factor on the action of other
stressors was recognized and addressed in its
conceptual model. A sensitivity analysis of the
effects of increasing concentrations of the several
stressors that are demonstrated causes of
impairment in a place allows observed biological
effects to be allocated among the stressors.
.2.2.6 Response-Based Classification
Our work on model-based classification is in
its inchoate stages; therefore, only our research
strategy is reported here. A report on the
statistical classification of coastal ecosystems of
the 48 contiguous states using physical variables
alone has been completed (USEPA 2004). To
simplify the problem of diagnosing the causes of
impairment, a classification scheme must be
based on similarities and differences in the way
that aquatic ecosystems respond to the loading of
a pollutant. If all systems respond to a particular
concentration of a pollutant in the same manner,
then all systems are in a single class and classi-
fication does not help. However, if ecosystem
types correspond to coherent groupings of
effective exposures, aquatic ecosystems may be
partitioned into stressor-based classes that will
aid the process of diagnosis and restoration.
Responses to habitat alterations, which are not
the result of the addition of a pollutant, are
considered separately.
We hypothesize that many measurable
ecosystem variables can be quantitatively related
to three key system characteristics (residence
time, processing capacity, and modifYing factors)
that control the effective exposure of an
ecosystem to a pollutant. One way to identify
classes is to list all the variables that are related to
each one of the three key factors controlling'
effective exposure, and then group these variables
based on quantitative or pseudo-quantitative
relationships to ecosystem condition variables. A
brief description of how we will use models to
develop a classification scheme follows: (1) One
of the four canonical models representing a water
body type will be used as a template for
evaluating the activity of individual pollutants
within the water body. (2) For a given water body
or a relevant functional division of that water
body, we will evaluate each of the 3 factors in our
model to determine effective exposure. (3) We
will then look for differences in the classification'
variables (obtained above) that can characterize
water body types associated with ranges of
effective exposure. (4) We hope to define
stressor-based classes of aquatic systems from the
data on residence time, processing capacity, and.
additional factors by looking for patterns and
differences in the relationship between effective
exposure to a stressor and the observed behavior
of condition variables within the aquatic
ecosystem.
As mentioned above, exposure is a function
of the biologically effective concentration of
pollutant molecules and the time that those
29

-------
Conceptual Models and Methods
molecules reside in the ecosystem. Residence
time of the pollutant in turn depends on the
residence time of the water and the capacity of
that ecosystem to process the polluting material.
To verify the assumed relationship between
exposure and biological effects, we will use
existing data, wherever possible, to demonstrate
that exposure to the pollutant is directly related to
ecological effects through a quantified dose-
response relationship. The efficacy of this model
in classifying ecosystems according to different
effective exposure regimes verified by observed
differences in response is currently being tested.
The eco-energetic system of ad urn et at.
(1974) is an alternative classification method for
coastal water bodies that will be updated using
emergy signatures (see text box C) determined
from our database. By documenting the anthropic
and natural parts of the emergy signatures of
coastal water bodies (adum et at. 1977, Campbell
2000) along with the ecological organization
found in those systems, we may be able to
establish a characteristic relationship between the
qualitative and/or quantitative features of the
emergy signature and the ecological organization
to be expected in the coastal system receiving that
signature. Stressor-based and non-stressor-based
classes of coastal systems may be identified in
this way and compared with the classes of coastal
systems developed by statistical and model based
methods.
(C) Energy and Emergy Signatures: An Alternative Method for Classifying Ecosystems

If the external forcing functions of a system are arranged in categories along the abscissa of a graph in
order of increasing transformity and their magnitudes converted to energy and plotted on the ordinate,
the plot is called the energy signature of the system (adum et at. 1977). If these energy values are
then multiplied by their appropriate transformities and the results plotted as before, the resulting graph
is the emergy signature of the system (adum 1996). Developing a library of energy and emergy
signatures provides a new approach to the old problem of predicting the ecosystems that develop
under the influence of known environmental forcing functions. Assembling data on the forcing
functions from many systems and grouping them into categories based on the emergy supplied may
serve as a robust means for predicting ecological structure and function. Emergy signatures can also
provide a unique method for characterizing ecosystems using the quantitative and qualitative
differences in the signatures. Campbell (2000) demonstrated that the emergy signatures of a fluvial, a
lagoon, and a macrotidal estuary were different. If the ecological processes and species which can
utilize the dominant energies of the signature are also different, the features of the energy signature
may serve as a means for predicting ecological organization.
The "natural" or expected emergy signature of a system is a baseline for estimating potential impacts.
An emergy deficit or excess in any category in the signature gives a quantitative measure of the stress
on a system adapted to the original inputs. The total of subsidizing (+) and stressful (-) emergies may
be an integrated measure of overall stress on an ecosystem. Also, the impact of stress on ecosystems
may be measured by changes in ecological organization and network power flow where both are
30

-------
Conceptual Models and Methods
(D) The Energy Systems Language
The Energy Systems Language, ESL (Figure 3), is particularly suited for developing diagnostic
models because the transformation of available energy is used to trace causality in the ecosystem upon
which both diagnosis and prediction depend. ESL can be used in both a general form for building
conceptual models and in a specific form which allows complete mathematical expression of the
model and simulation (Odum and Odum 2000). The development of simulation models facilitates the
allocation of observed effects among multiple demonstrated causes through a sensitivity analysis of
the models. ESL is very rich allowing, the modeler to represent the complex energetic and kinetic
aspects of a system in a parsimonious manner. Campbell and Wroblewski (1986) showed that box
diagrams (an alternative model system) required considerable written explanation to approximate the
system description that was achieved with models constructed using ESL. A disadvantage of using
ESL is that it requires some training for a scientist to become fluent in the specific form of the
language. For conceptual model development, a familiarity with the general meaning of the symbols
and the rules of syntax for building diagrams is sufficient, because translation of the network into
mathematical expressions and equations is done later. Understanding the more complex conceptual
models does require some patience in tracing interaction pathways on the diagrams and using labels to
key the numbers on the pathways to the associated tables.
The ELS symbols and their mathematical definitions are widely available in the literature (Odum,
1971, 1983, 1994), thus only a brief description of the language and symbols is presented here (Figure
3). Systems are viewed as being composed of components and processes that are intercoimected by
energy flows and interact within a network that is influenced by one or more external forcing
functions. External forcing functions (circles) describe energy inflows to the system (both matter and
information have energy equivalents); ecosystem components are producers (bullets), consumers
(hexagons), storages (tanks); processes are shown as interaction symbols (the unidirectional double-
pointed rectangular arrow) or work gates where energy is being transformed and where some energy is
degraded into unusable form; longitudinal dispersion/vertical mixing (a rectangular arrow with points
in opposite directions); logic programs (symbol with four concave sides) handle logic and switching
functions; pathways carry energy, matter, or information (lines with arrows) and connect forcing
functions and components directly or through the work gates. The symbol with an arrow going to
ground is the heat sink carrying energy that is no longer available for use within the system. A
rectangular box is used to show system boundaries or to delineate subsystems.
31

-------
Conceptual Models and Methods
3.0 Development of Detailed Conceptual Models
Detailed conceptual models were developed
for each of the four classes of aquatic stressors.
Each detailed model considers: (1) information
available on the action of the primary stressor
within an ecosystem, (2) the major stressors
interacting with the primary stressor, and (3)
the network of stressor interactions affecting
biological components and flows in the aquatic
ecosystem. These detailed models are the starting
point for developing computer simulation models
capable of allocating biological effects among
multiple demonstrated causes of impairment. As a
first step in developing these detailed conceptual
models, we wrote narrative descriptions of the
four stressor classes listed in the Aquatic
Stressors Framework (USEPA 2002b). In this
section, the narrative descriptions of the stressors
are presented using a similar outline for each,
along with the translation of these narratives into
detailed ESL models for each stressor.
3.1 Narrative Descriptions of the Stressors
Narrative descriptions and summary diagrams
for the four classes of aquatic stressors that are of
particular concemin the Aquatic Stressors
Framework are presented in this section. The
stressors considered are as follows: (1) nutrients,
(2) metal and organic toxicants, (3) suspended
and bedded sediments, and (4) altered habitat.
3.1.1 Nutrients
Definition: Nutrients are nourishing elements
required by all living systems for normal
functioning. The primary elements associated
with nutrient over enrichment and cultural
eutrophication are nitrogen and phosphorous.
Nitrogen is important in causing and controlling
eutrophication in marine systems, while
phosphorous is generally recognized as the
controlling element for nutrient enrichment in
most freshwater systems. The concurrent addition
of nitrogen and phosphorous to ecosystems can
. have synergistic effects, stimulating the growth of
organisms at the base of the trophic web, such as
phytoplankton. At times, other elements, like
silica and iron, may regulate phytoplankton
growth through their role as limiting factors (e.g.,
the occurrence of diatom and toxic dinoflagellate
blooms in coastal waters) but their importance
with respect to nutrient enrichment is often
secondary to nitrogen. Characteristic features
often observed in eutrophic ecosystems are the
accumulation of organic carbon in sediments,
increased abundance of certain algal and
macroalgal species, e.g., blue greens and Viva,
overgrowth of other species that can tolerate the
new conditions, and low dissolved oxygen in the
sediments and in the overlying water.
Natural occurrence: Living systems require
many elements for their normal functioning and
over time they have adapted to ensure that as far
as possible (within the constraints imposed by
climate and the geochemistry of the substrate)
optimum supplies of these materials will be
available for use. In the sea, phytoplankton
require approximately 16 moles of N for every
mole of P they assimilate, the Redfield ratio
(Redfield, 1958). If the ratio of available N to P is
less than 16:1, primary production in the ocean
will be limited by N; if the ratio is higher,
production will be P limited. The availability of
nutrients is affected by Nand P inputs, storage
and recycling (denitrification) and nitrogen
fixation. Because living systems adapt to
prevailing conditions, we find a range of nutrient
states that are characteristic variations within
many kinds of ecosystems, e.g., lakes, streams,
wetlands, etc. In general these variations manifest
as different structural and functional states of the
ecosystems, which are determined by adaptation
to the different levels of nutrient supply. These
states are termed oligotrophic, for low nutrient
inflow, mesotrophic for moderate inflow, and
eutrophic for high inflow. In the absence of
human a«tivities ecosystems are expected to
manifest a range of characteristic structural and
functional conditions that correspond to the
natural range of nutrient inflows into the system.
Changes caused by human activities: One of
the chief characteristics of human economic and
cultural use of natural resources is to disrupt the
natural cycles followed by the chemical elements,
32

-------
Conceptual Models and Methods
including nutrients (Odum et al. 20(0). The
addition of excess nutrients also disturbs the
expected pattern of production (P) and
consumption (R) within ecosystems (see dis-
cussion on P and R, Fig. 6). In a process called
cultural eutrophication, human economic and
social activities concentrate nutrient elements on
the landscape and then redistribute them in
patterns that differ from those that were formerly
established by the natural biogeochemical cycles.
Most often, the ecosystems that receive these
concentrated nutrient materials are not adapted to
process them efficiently. Those animals and
plants that can process the excess nutrient inflows
flourish, moving the structural and functional
characteristics of the original ecosystem toward
those of a more eutrophic system.
Mechanisms of stressor action: The general
mechanism of action for excess nutrient is to
over-stimulate plant growth. If sunlight, nutrients
or any other environmental factor is inadequate to
support further growth of organisms capable of
primary production, the condition is said to be
"limiting growth". When nutrients are unlimited,
photosynthetic organisms grow and multiply until
a new limitation is encountered. Given adequate
nutrients, phytoplankton will multiply until their
density limits further growth by self-shading
(light limitation). In general, those plants that are
best adapted for nutrient uptake are the ones that
flourish under conditions of nutrient enrichment,
e.g., duckweed on the surface of standing water in
a wetland receiving sewage. The actual effects on
plant growth of any nutrient in excess will also
depend on the other factors that control plant
production. For example, if light or a required
nutrient is not present in sufficient quantity, the
increased growth of plants, which is expected to
follow from an excess of the nutrient may not be
observed. In addition, the spatial pattern of excess
plant growth follows from the geometry of the
system that determines the availability of light.
For example, in shallow aquatic systems, light
strikes the surface of the water first. If the water is
torpid, a surface dwelling plant such as duckweed
may proliferate and shade out all plants that might
grow below the surface. If the water is in motion
breaking-up surface films, light next extends into
the water column supporting the growth Of
phytoplankton, which in turn can attain such high
. dersities that rooted aquatic plants are cut off
from the light and decline. The long term effect of
excess plant production is the accumulation of
organic matter on the bottom and the overgrowth
of the benthic animals that are best adapted to
process this excess organic material. Prolonged
nutrient enrichment or extremely high nutrient
loads cause systemic problems for ecosystems as
a result of the overproduction of organic matter
and a subsequent increase in oxygen consumption
by microbes. The conditions that accompany the
hypoxia and then anoxia that follows can make
usually beneficial nutrients, such as ammonia,
toxic. Even though a range of sediment oxidation
levels is expected based on the different ability of
oxygenated water to penetrate sediments of
different grain size, increased organic matter
supply to the sediments increases the utilization
of oxygen by microorganisms there and lowers
the oxygen penetration expected for an~
particular grain-size, which results in the loss of
organisms and the decay of sediment structure. A
definitive determination of cultural eutrophication
can be difficult, because reference systems are
often tare or nonexistent.
Expected scale of stressor induced effects:
The location and intensity of cultural
eutrophication on the landscape is almost entirely
due to the form, extent and intensity of human
activities. The space time boundaries for
establishing a successful TMDL or other
regulatory program can be determined by
examining the distribution of processes that are
concentrating and distributing nutrients over an
entire watershed connected to the open sea.
Where external nutrient supplies from human
activities account for more than a small fraction
ofthe total nutrient supplied to an ecosystem, we
can expect to observe the effects of nutrient
enrichment. Internal stores of nutrients and
organic matter in the sediments, which are the
result of past loading patterns, will, in part,
determine the scale and intensity of
eutrophication in a water body.
Factors to consider in making a diagnosis: It
is often difficult to decide if an ecosystem is
impaired by excess nutrient inputs, because
nutrients are ubiquitous constituents of living
systems naturally present in varying amounts that
lead to different structural and functional states.
Extreme cases of nutrient loading in aquatic
33

-------
Conceptual Models and Methods
environments can result in chronically low
dissolved oxygen concentrations and the
complete deterioration of the conditions
necessary to support aerobic forms of life. Even
though areas of hypoxia and anoxia develop
naturally in some ecosystems; increasing the
intensity, extent or frequency of hypoxic or
anoxic conditions in an area or finding these
conditions where they are not expected, are
events often associated with human alteration of
the environment. The effects of carbon loading on
aquatic environments are often apparent in the
increased organic matter found in sediments of a
given grain size; however, it is more difficult to
determine with certainty the origin of the carbon.
increase. Often a chain of intermediate results
separate nutrient loading from an observed effect, .
such as a low oxygen event or a change in benthic
species distributions. In making a diagnosis that
nutrients are the cause of any given impairment,
the investigator should ascertain that the impact
pathway is as completely documented as possible,
using the weight of evidence to decide if nutrients
are the most probable cause of an observed effect.
3.1.2 Toxic Chemicals
Definition: Toxic chemicals are defined as
compounds or elements that elicit a dose-
dependant toxic response in a biological system
(Rand and Petrocelli 1985, Manahan 1989). The
affected biological system may range over
organizational scales from the sub-cellular to the
ecosystem. Toxic response endpoints range .from
necrosis of cells, through mortality of individual
organisms, to significant changes in community
composition. Toxic chemicals may' be anthropo-
genic (e.g., polychorinated biphenyls or PCBs,
dioxins, pesticides) or natural (copper, cadmium,
nickel, hydrogen sulfide, ammonia) in origin.
Some toxic chemicals like the metals copper,
nickel and lead are naturally-occurring but their
concentrations in the environment are greatly
elevated by human activity, e.g., mining
(O'Neill 1985).
Natural occurrence: Many potentially toxic
chemicals occur naturally in aquatic environments
and are only toxic when their concentration is
increased beyond a threshold. For example, all
metals have a geological source, while organic
compounds such as polycyclic aromatic
hydrocarbons (PAHs) are produced both through
natural and anthropogenic processes (O'Neill
1985, Burgess et a!. 2003). In nature, these
chemicals are not often found at high enough
concentrations to be toxic. Exceptions include
localized oil seeps or easily eroded mineral
deposits. However, under these circumstances, it
is likely that local ecosystems have adapted over
thousands of years to the presence of these
otherwise toxic chemicals, just as populations of
organisms have been shown to adapt to more
recent contamination (Nacci .et a!. 1999, 2002).
Some toxicants such as copper or ammonia are
required by living systems at low concentrations
(O'Neill 1985). However, these same chemicals
at elevated concentrations can cause toxicity. As
Paracelsus (1493-1541) said, "therapeutic and
toxic properties often differ by just the dose"
(Doull et a!., 1980). Toxic chemicals can not be
easily grouped into one large stressor category,
because of their common natural occurrence and
their ability to be both therapeutic and toxic
depending on dose; therefore, they are best
considered one at a time or in functionally similar
groups or subcategories (Doull et a!. 1980).
Changes caused by human activities: Human
activities have two major effects on the
occurrence of toxic chemicals in the environment:
disruption of the natural cycles of some toxic
chemicals and synthesis of 'new' toxic
substances. First, a chief characteristic of the
human economic and cultural use of natural
resources is to extract raw materials and in so
doing disrupt the natural material cycles.
Concomitant waste production and disposal
practices often result in the addition of elevated
concentrations of many elemerits (e.g., Cu, Pb,
Hg) and compounds (e.g., ammonia, hydrogen
sulfide) in the environment. Furthermore, human
activities cause the distribution of these materials
into patterns that differ from those established by
natural biogeochemical processes. Often, eco-
systems that receive concentrated doses of toxic
chemicals (e.g., mine tailings, manufacturing
waste) have not adapted to effectively process
them resulting in potentially severe disruption of
ecosystem structure and function. Second, over
the last 100 years human activities, especially
those related to the chemical and pharmaceutical
industries, have created many new synthetic
34

-------
Conceptual Models and Methods
toxicants (Colborn et al. 1996, Carson, 1962).
Some of these synthetic toxic chemicals have
been designed purposefully to be toxic (e.g.,
herbicides, pesticides) while others were designed
for other uses (e.g., PCBs as electrical component
insulators) but have had environmentally deleter-
ious effects (Ho et al. 1997 , West et al. 2001).
Mechanisms of stressor action: Toxic
chemicals may have many adverse effects that
result from a variety of mechanisms; therefore, it
is difficult to develop a simple scheme for
accurately categorizing their impacts. The most
commonly measured effect in aquatic toxicology
is mortality; in addition, sublethal effects include
reproductive changes, alteration of growth,
competition and behavior, e.g., being an efficient
predator or an elusive prey (Rand and Petrocelli
1985). Beyond effects on individual organisms,
ecological effects include changes in communities
such as alterations of diversity, biomass,
functional processes, and species composition and
in ecosystem properties such as the health and
integrity of systems and changes in spatial
(habitat heterogeneity) and temporal (nutrient
regeneration rates) dynamics (Rand and Petrocelli
1985). Current research at NHEERL's Atlantic
Ecology Division is focused on extrapolating
from the individual to population, community and
ecosystem endpoints (Kuhn et al. 2000, 2002).
Broad classes of mechanisms of toxic action
that produce stressful effects include: narcosis a
reversible effect that reduces metabolic activity
(van Wezel and Opperhuizen 1995), alteration of
chemicals to more toxic forms via the mixed
function oxidases (MFa) enzyme system (Doull
et aI., 1980, Coakley et aI., 2001), exceeding the
detoxification capacity of metallothionien
proteins (Kagi and Schaffer, 1988), and the broad
class of mechanisms that promote and produce
cancer (Ashby and Tennant, 1988, Lijinsky, .
1989). Regardless of the mechanism, toxic
chemicals damage the metabolic structure and
function of the organism. All organisms have
repair mechanisms; however, if adverse effects
are manifested, the repair mechanisms have been
overwhelmed.
Expected scale of stressor induced effects:
Anthropogenic activities have distributed toxic
chemicals globally (LaFlamme and Hites 1978,
Valette-Silver 1993, Muir et al. 2000); however,
because most toxic chemical discharges are
concentrated at point sources, toxic chemicals
most commonly manifest adverse effects
radiating out from the point of discharge within
an individual water body. The majority of toxic
sediments are found in close proximity to
industrial, municipal or agricultural sources
(Long 1992, Daskalakis and O'Connor 1995,
Long et al. 1996); however, one should not rule
out widely-distributed areas of contamination
simply because concentrations are not extremely
elevated or areas of elevated contamination not
directly associated with obvious sources (e.g.,
. illegal disposal of hazardous wastes). Examples
of toxic pollutants that have wide distributions
include mercury released into the atmosphere by
coal burning power plants (Renner 2002) and the
atmospheric transport and deposition of PCBs
that through trophic transfer and biomagnification
appear in high concentrations in the fat and milk
of high-latitude mammals including indigenous
peoples (Muir et al. 2000, Swain, 1988).
Factors to consider in maldng a diagnosis:
Extensive data are available on the toxicity of
many classes of chemicals and our understanding
of the toxic mechanisms of many chemi<;als
allows prediction of their bioavailability and
effects. Toxicity testing is a valuable tool but to
date it has focused on organisms as endpoints,
while many of the effects considered in
developing a TMDL include impacts on
populations, communities and ecosystems. Also,
traditional toxicity testing and endpoints may lack
sensitivity to certain toxic chemicals that
bioaccumulate (e.g., dioxins, mercury). In this
case, tissue residues are a better measure of the
stressor's potential effects. Instrumental analysis
of environmental samples for toxic chemical
concentrations represents a viable secondary
approach for determining the exposure of
organisms to these chemicals. However, such
analyses can be prohibitively expensive, are not
capable of detecting all toxic chemicals, and may
erroneously over- or underestimate the bioavail-
ability of target toxic chemicals. Linking toxicity
testing and instrumental analysis in a TIE .
approach captures the strengths of both methods.
35

-------
3.1.3 Suspended and Bedded Sediments
Conceptual Models and Methods
Definition: Suspended and bedded sediments
act as a stressor in the environment when they
have negative effects on organisms and lor
habitats due to their presence in excess quantity,
i.e., they are present in quantities that the
ecosystem is not adapted to process. Suspended
and bedded sediments fall into two broad
categories: suspended sediments and bedload.
Suspended sediments are those particles that are
suspended in the water at any given time because
the turbulent energy of the water is sufficient to
prevent them from settling to the bottom under
the influence of gravity. Bedload describes those
particles moving through the aquatic system via
sliding, rolling or saltating on or near the bed
(channel, or lake bottom). Whether a particle is
moved as suspended sediment or bedload depends
on the complex interactions of particle size and
shear stress, with shear stress being a function of
the interaction of channel slope, gravity, bottom
roughness, and current speed. The direct
measurement of bedload transport is difficult,
therefore patterns of mud, sand, gravel, and
current speed are commonly used to estimate its
magnitude and occurrence.
Natural occurrence: The most common
source of sediment entering streams is the
weathering (by wind and precipitation) of parent
material in the catchment, which is subsequently
moved down slope by gravity into the receiving
waters. This process may be accelerated by
natural disturbances (e.g. storm events) within the
catchment. Thus, sediment in the receiving waters
is an expected result of catchment evolution and
weathering processes. Background concentrations
of sediments in receiving waters are variable,
depending on the age, geologic composition and
history of the catchment. The balance of erosion
and deposition resulting from the interactions of
physical forces and conditions changes the nature
of subsequent interactions, so that sediment
features on the bottom grow and decline over
time. The quality and quantity of sediment
movements is closely tied to natural episodic
events such as floods, tsunamis, and hurricanes.
Changes caused by human activities:
Anthropogenic activities have greatly accelerated
the rate of sediment movement from catchments
to receiving waters because of land use practices
that increase erosion and runoff. In general, the
conversion of natural lands within a catchment to
human uses will result in increased sediment
transport and deposition in the receiving waters of
the systems. Specifically, row crop agriculture,
livestock grazing, forestry, mining, and urban
development have all been linked to increases in
sediments in streams, lakes, reservoirs, wetlands
and estuaries.
Mechanisms of stressor action: Suspended
and bedload sediments have two major avenues of
action in aquatic systems: '1) direct effects on
biota and 2) direct effects on physical habitat,
which results in indirect effects on biota.
Examples of direct effects on biota include.
suppression of photosynthesis by shading primary
producers; increased drifting of, and predation on,
benthic invertebrates; and shifts to turbidity-
tolerant fish communities. Indirect effects on
biota will occur as the biotic assemblages that
rely upon aquatic habitat for reproduction,
feeding, and cover are adversely affected by
habitat loss or degradation.
Expected scale of stressor induced effects:
The effects of suspended and bedload sediments
span the scales of biological organization from
individuals to ecosystems. The biological .
responses to this stressor at a site are related to
site-specific effects (turbidity shading, su~strate
embeddedness) and to the cumulative loadings of
sediments from the catchment above the site. In
addition, the cumulative effects of these
biological responses (failed reproduction or
reduced habitat) at sites are additive over the
entire catchment, so that catchment-wide stressor
impacts are possible based on the cumulative
nature of the stressor.
Factors to consider in making a diagnosis:
The stress of excess suspended and bedded
sediments on receiving systems rarely acts alone.
Other stressors associated with accelerated clean
sediment loadings are increased temperature,
which usually results from riparian canopy losses,
and increased nutrients inflows from clear cutting
and agriculture in the catchment. Often, these
various stressors are the result of the same
anthropogenic activities, e.g., land conversion and
human disturbance.
36

-------
Conceptual Models and Methods
3.1.4 Habitat
Definition: Habitat describes the environment
in which particular organisms reside and includes
both physical attributes (e.g., structure, tempera-
ture, hydrological regime) and chemical attributes
(e.g., dissolved oxygen, salinity). The physical
structure of habitat includes both abiotic
components (e.g., geologic substrate, temperature,
oxygen, etc.) and biotic components (e.g.,
vegetative structure, oyster shells, worm tubes,
etc.) and thus its extent can be measured in two or
three dimensions (area or volume, respectively)
and its quality. ascertained by a wide variety of
measures. The scale and components of habitat
must be defined in relation to a particular
population or group of populations. For the
purposes of illustration, habitat will be defined
with respect to fish populations in the conceptual
model discussed here.
The words "habitat" and "suitable habitat" are
often used interchangeably, giving rise to the
concept that habitat for a population must
encompass a minimum set of conditions that are
associated with a finite probability of survival.
Habitat can be examined at a range of scales,
from micro-habitats (at the scale of the organism)
to habitat patches (homogeneous units related to
populations) to habitat mosaics (a collection of
habitat patches, often dispersed in a backgroun.d
matrix of unsuitable habitat). Habitat quality, or
the probability that habitat will support a self-
sustaining population, can also be defined at each
of these scales. Factors affecting temporal and
spatial scales of habitat variability are described
in Figure 11.
Natural occurrence: In general, aquatic habitat is
of natural origin, although it is possible to create
artificial habitat (e.g., artificial reefs, sewage
treatment systems). Natural regimes (spatial and
temporal patterns) exist for temperature, water,
geologic substrate, biological structure, dissolved
oxygen, and salinity, and these differ over
ecosystem types and geographic regions. The
hierarchical structure of biological and ecological
organization must be considered to understand the
role of natural habitat in the survival of indivi-
duals and populations. If the set of energy inputs'
to an ecosystem provides the available energy to
organize its abiotic and biotic structures and
functions upon wpich habitat depends; habitats,
whether natural or produced by human activities,
may be effectively distinguished by the incoming
emergy signature (see text box C) of the place.
Changes caused by human activities: Both
the quality and quantity of natural habitat can be
altered by human activity. Potential changes
include 1) shifts in the spatial and temporal
patterns of water level regimes or flow,
temperature and dissolved oxygen fluctuations, or
gradients of salinity, 2) changes in the physical
structure of habitat, both biotic (vegetation) and
abiotic (substrate particle size, organic matter
content), 3) loss of area of suitable habitat, and
4) changes in the spatial configuration of suitable
habitats on the landscape. In addition, human
activities are changing the natural emergy
signatures of many places on earth, and indeed of
the earth as a whole (Brown and Ulgiati 1999).
Many new aquatic habitats are being created by
altering the natural energy and material flows and
storage units and by creating new materials and
discharging them to the environment in sufficient
quantities to change the character of a place, and
thereby the unique suite of species that the
ecosystem can support there. For example, Odum
et al. (1974) mention emerging new systems
associated with pulp mill wastes, thermal
pollution, sugar cane wastes, oil shores, brine
pollution, etc.
Mechanisms 01 stressor action: We will
distinguish between attributes of habitat that can
act as stressors on individual organisms
(temperature, hydrologic regime, dissolved
oxygen, salinity) and those attributes required for
physical structure. The physical structure of
habitat can provide a substrate for attachment,
substrates for growth of primary producers or
prey items, shelter from disturbance such as wave
action, and/or a refuge from predators. As such,
habitat structure can modify the availability of
food resources or reduce population loss rates
from predation or physical damage. Vegetative
habitat can also modify the chemical character-
istics of surrounding waters through photo-
synthesis or through limiting turbulence and
exchange with the atmosphere or with flowing
water.
37

-------
Conceptual Models and Methods
1  2  3 4
Stream River  Estuary Coastal!
 l'IIearshore
~ i    
5     
Lake  6  
   -  
7
PhysICO chemical habitat
Air temperature
Flow velocity/turbulence
Runoff:groundwater
Evapotranspiration
9
runoff
potential
9
slope/
substrate
9
vegetation
(seasonality)
9
climatic
regime
seasons,
storms,
wet/dry cycles
8
Air temperature
Flow velocity
Runoff:groundwater
ET
Tidal prism volume:discharge
10
tidal range
11

estuarine/
lacustuarine
morphometry
 6 
 Biophysical habitat 
7  7
Hydrologic regime  Hydrologic regime
 Vegetative structure
Vegetative structure  Substrate
Substrate  Tidal regime
Figure 11. Summary diagram of the principle factors that control habit!!t of a species and its alteration.
Mechanisms of stressor action associated
with temperatures outside of the metabolically-
suitable tolerance range of a species include direct
mortality, altered growth, reproductive changes
and emigration or movement out of the formerly
. suitable habitat. Altered temperature regimes will
be manifested directly as structural changes
within a population and compositional changes
within the fish community. Indirectly, altered
temperatures can affect fish through changes in
prey availability and changes in interspecies
competition based on differences in optimum
temperature or tolerance ranges.
Direct mechanisms of action associated with
altered hydrologic regimes include physical
disturbance affecting individuals (scouring,
increased drift), and populations (desiccation, and
loss of suitable instream habitat volume). Indirect
mechanisms of effect associated with altered
hydrologic regimes that apply to both individuals
and populations include: 1) changes in vegetative
structure based on selection for plants with
different growth forms and life history strategies,
2) changes in sediment characteristics (e.g.,
siltation, embeddedness) due to changes in flow,
3) changes in thermal regime related to the
balance between groundwater and surface water
inputs, 4) decreases in dissolved oxygen
associated with stagnation and reduced
turbulence, 5) changes in salinity gradients and
stratification patterns in estuaries, and 6) changes
in retention time and associated processing rates
that affect the concentrations of nutrients and
toxins. Tertiary effects include changes in
competitive ability related to changing physical
and chemical conditions, in some cases selecting
for invasive species.
38

-------
Conceptual Models and Methods
Mechanisms of stressor action associated
with altered dissolved oxygen regimes include
massive mortality, altered growth, and avoidance
of areas with unsuitable habitat. As with other
habitat stressors, effects will be observed at the
individual, population, and community levels.
Mechanisms of action associated with altered
salinity gradients include loss or gain of suitable
habitat with resulting effects on growth and
survival, emigration or immigration resulting in
changes in community structure and composition.
Expected scale of stressor-induced effects:
Habitat quality can be affected at the scale of
organisms (microhabitat), at the scale of habitat
patches, or at the scale of landscapes. The total
area, spatial configuration and density of habitat
patches can all affect the survivorship of popula-
tions. For example, the density of habitat patches
could affect the rate of colonization and the
number of populations experiencing periodic
recruitment failures. Spatial configuration of
habitat patches can be particularly important
where different habitat types are critical for
different life stages of an organism. The scale to
be considered depends on the range of the species
of interest, both within a year, and between
different life. history stages.
Factors to consider in making a diagnosis: A
reference condition for habitat structure, habitat
extent, and the distribution of habitat types must
be established to determine whether changes have
occurred. Historic changes in habitat type, area,
and distribution can be measured or inferred
through use of mapped inventories (e.g., National
Wetlands Inventory, http://www.nwi.fws.gov/),
aerial photographs, or through the use of indica-
tors such as the association of hydric soil com-
plexes with different wetland vegetation types.
Information on the historic range and biogeogra-
phic constraints for individual species may help
distinguish between the effects of habitat loss and
change in the quality of habitat. The effects. of
changes in habitat area can be inferred through
. development of empirical species-area curves, or
. production-area relationships. Responses to loss
of suitable habitat area are more likely to be a
function of organism requirements than of habitat
features. The effects of changes in the mosaic of
habitat structure can be examined through pattern
descriptors such as patch density or diversity,
patch cohesion, dispersion, and perimeter: area
ratios, and the development of relationships
between these measures and biotic endpoints. In
addition, the effect of changes in the mosaic
structure of habitats can be simulated through use
of spatially-explicit population models.
Species are adapted to utilize variations in
naturally occurring ecological conditions
(niches), such as different thermal, dissolved
oxygen, and hydrologic regimes. The presence
and absence of species often can be directly
related to their requirements for a specific habitat.
Foi-example, guilds offish species adapted to
different physical regimes have been identified
based on commonalities in physiological
adaptations, behavior, or life history traits, and
they can be used to infer that changes in these
parameters are associated with community-level
shifts. For temperature and dissolved oxygen,
limits associated with mortality have been
determined in laboratory tests, while sublethal
effects may be predicted through bioenergetics
models. More accurate limits on fish species,
distributions in streams can be derived through
development of empirical associations between
parameters that describe properties of these
regimes (e.g., 7-day low flow, 21-day average
maximum temperature) and fish
presence/absence.
3.1.5 Special Characteristics 0/ Aquatic
Ecosystems Applicable to all Stressors
Several aspects of aquatic ecosystems are
particularly relevant to understanding habitat
alteration, but they also apply to the other three
stressors and should be considered in developing
and analyzing conceptual models of stressor
action. These general factors will be discussed in
this section with reference to habitat alteration,
because this stressor is the most complex. Similar
considerations might be applied to the other three
stressors (toxic chemicals, nutrients, and
suspended and bedded sediments). The general
characteristics considered here are as follows:
1) Biological components, stressors, and modify-
ing factors are all linked in a single interacting
network. 2) Spatial and temporal variation of
stressor actions and effects is common, i.e.,
stressor interactions and effects are dynamic in
39

-------
Conceptual Models and Methods
space and time. In addition, the subject,
classification and assessment of habitats, is
briefly considered in this section.
Factors modifying altered habitat are linked:
Many variables determining suitable habitat are
linked and, thus cannot be examined in isolation.
Each of the factors that alter habitat can be
modified by the state of the other modifying
factors. Some examples of such interactions
follow: a) increased heat inflow to an estuary will
cause a larger or smaller change in the average
water temperature depending on water flow and
vertical mixing in the estuary; b) the response of
an organism to a habitat variable can also change
in the presence of a pollutant, which can be
considered as a modifying factor on habitat (for
example, increased sedimentation in a stream
ecosystem can alter the response of the biota to
changes in flow regime); c) shifts in the spatial
and temporal patterns of water level, regimes or
flow, temperature, and dissolved oxygen
fluctuations, or gradients of salinity can each
affect the biological consequences of a change in
one or more of the others; and d) the structure of
habitat itself might alter the biological effects Of a
change in one of the modifying factors as
discussed below.
Spatial and temporal variations in habitat
quality: Over short time periods at the micro-
habitat scale, the effects of shifts in the mean or
the temporal patterns of water level, flow, temper-
ature, dissolved oxygen, or salinity on organisms
should depend only on the mean and variance of
conditions in the system without impacts, i.e.,
those conditions to which resident organisms are
adapted. In this case, the micro-habitat scale is
defined relative to the spatial grain or heterogen-
eity in those environmental parameters defining
suitable versus unsuitable conditions (leading to
death or to a net loss of energy). The temporal
scale ("short" time period) is defined relative to
the time required for an organism to move from
an unsuitable habitat to a suitable habitat.
Over longer time periods and at spatial scales
consistent with the range of an organism,
organism sensitivity to changes in the mean or in
the spatiotemporal variation of environmental
parameters will depend on the spatial or temporal
distribution ofrefugia (i.e., suitable habitat). An
example of refuge from scouring flows would be
the hyporheic zone or backwater habitat in a
stream. Gradients in the physical structure of
habitat, both biotic (vegetation) and abiotic
(substrate particle size, organic matter content),
organize habitats in space. In general, the spatial
and temporal variability of habitat under normal
disturbance regimes will determine the sensitivity
of populations within the habitat to change. For
example, some ecosystems naturally go through
stages of succession, or periodic shifts in vegeta-
tion structure related to wet and dry periods (van
der Valk 1981). These ecosystems are more likely
to have seed banks for vegetation adapted to
different phases of the wet-dry cycle, resting
stages of other organisms (e.g., ephipia) adapted
to changing conditions, and/or to a relatively high
proportion of colonizing species, which allow
populations to recover following periods of stress.
The effects of spatial and temporal variability
in habitat quality on landscape-scale population
dynamics have been generalized by Turner et al.
(1993; see Figure 2). Turner and colleagues
simulated a terrestrial landscape with 8 seral
stages of succession based on the assumption that
seed sources would not be limiting and captured
the results in a state-space diagram. Landscapes
can be compared by scaling the disturbance
interval to the recovery time, and scaling distur-
bance extent to landscape extent. To judge the
general effects of spatial and temporal variability
in habitat on animal population stability, this type
of simulation would be repeated and the results
summarized with disturbance interval scaled to
the time to recolonize patches and disturbance
extent scaled to organism's home range on the
landscape. Turner's simulations were based on
the spatial and temporal dynamics associated with
terrestrial succession. In an aquatic landscape, the
concept could be extended to cover dynamics of
daily migrations (between suitable and unsuitable
microhabitat based on diurnal variations),
movement over a breeding season (annual
. migrations), and movement between breeding
seasons. An upper limit to regional population
viability could be expressed as a function of
disturbance extent relative to critical dispersal
distance and disturbance interval relative to
organism longevity or length of viability of
resting stages (e.g, egg bank) where appropriate
(See Turner et al. 1993, Figure 3).
40

-------
Conceptual Models and Methods
Meta-population models or spatially-explicit
population models can be used to predict the
sensitivity of different systems to changes in the
spatial configuration of suitable habitats on the
landscape. Sensitivity to change will depend on
organism life history traits and the ability to
migrate across the landscape as well as the
density of suitable habitat patches (Gibbs 1993).
Island biogeography theory (MacArthur and
Wilson 19q3) suggests that recovery of popula-
tions will occur more rapidly when suitable
habitat patches are less fragmented or dispersed
(Gibbs 1993). Due to a lack of long-term data,
this hypothesis has been tested for relatively few
aquatic systems (Niemi et al. 1990, Detenbeck et
al. 1992). Recovery time for aquatic populations
depends on the distance of refugia from the point
of other stressor impacts (Detenbeck et al. 1992).
Simulation models have been developed that can.
mimic the movement of organisms across
landscapes or potentially across riparian zones
(Gardner et al. 1992). These models predict that a
critical change in landscape structure occurs as
the ratio of inner to outer edges (or fractal
dimension of a landscape) changes. In general,
the predictions of these models have not been
verified against actual data. .
Classification and Habitat Assessment:
Existing habitat classification schemes such as
Cowardin's scheme for wetlands and deepwater
habitat tend to categorize habitat into patches of
similar environmental conditions and substrate
type, and thus would not necessarily predict
sensitivity to a change in environmental variables
(Coward in et al. 1979). Classification schemes or
parts of classification schemes describing the
normal or reference range of variation in abiotic
environmental variables and biotic structure
(vegetation) might provide a framework for
predicting sensitivity to change in habitat quality.
Examples of such classification schemes include
use of the hydrology modifier in Cowardin's
classification scheme for wetlands or Poffs
concept of flow regimes for streams (Poff and
Ward 1989). For lakes, temporal and spatial
variability of dissolved oxygen and temperature
regimes has been described as a function of lake
morphometry (surface area, maximum depth) and
trophic state (Stefan et ill. 1995, 1996).
Classification schemes describing variation in
the hydrologic regime are likely to explain the
relative sensitivity of different aquatic ecosystems
to changes in the biotic component of physical
habitat, whether natural or unnatural. For systems
in which gradients of physical and chemical
conditions exist (e.g., estuaries), the shift in these
gradients relative to suitable biophysical habitat
such as emergent vegetation will determine
sensitivity to change (Sklar and Browder 1989).
Thus overall sensitivity will depend both on the
steepness of biophysical gradients and estuarine
morphometry; more gently sloping systems will
probably have a broader band of suitable habitat
types such as emergent vegetation. In general,
spatial variability will be a function of source
characteristics, topography or bathymetry, and
mixing factors (surface area/depth, salinity
ratios).
3.2 Generic Conceptual Models of the Four
Classes of Aquatic Stressors
This section presents generic models for the
four classes of stressors; however, these stressors
do not exist within aquatic ecosystem alone, but
in the company of one or more of the other
stressors, which may affect the action of the
stressor being modeled. For example, suspended
and bedded sediments and nutrients are natural
components of all aquatic ecosystems and will co-
occur with each other and with toxic materials
such as anthropogenic organic chemicals. To
meet this challenge, an energy system perspective
(Odum 1994) was used in developing conceptual
models of the individual stressors. Therefore,
each stressor was represented with the complex
of factors that determine its actions within the
ecosystem, including factors that might be
considered stressors in their own right. Thus, the
detailed conceptual model for a single stressor
includes information on other stressors, where
they are important modifiers on the actions of the
stressor under study. In this manner, both single
and multiple stressor problems will be addressed
with a unified approach. In cases where additional
stressors are important, the single stressor models
may be modified or coupled with other single
stressor models to examine joint effects.
Modeling each stressor within the context of
the larger aquatic ecosystem is important, because
the result of performing a PIE or other method of
41

-------
Conceptual Models and Methods
causal analysis in marine ecosystems often
presents us with situations where more than one
stressor is shown to be actively causing impair-
ment to the biota. In this situation, we need
models that incorporate accurate information on
causal mechanisms to allocate observed effects
among multiple stressors. Sensitivity analyses of
ecosystem simulation models can be used to set
priorities for mitigating a stressor by determining
the pollutant to which the ecosystem output
variables (measurement endpoints) are most
sensitive. In addition, these models can serve
other useful purposes, such as predicting the
threshold for an effect and determining joint or
bundled criteria for combinations of stressors
interacting within an ecosystem.
The first step in producing pollutant-specific,
quantitative models for eventual conversion into
computer simulation models is the construction of
a conceptual model of stressor action. This
process is the means for putting our mental
models of stressor action into a concrete form. In
this section, the narrative descriptions of the
major classes of stressors given in Section 3. I are
used to create ESL diagrams showing stressor.
action within the context of an ecosystem. The
starting point for developing more complex
models for the individual stressors is to apply the
narrative descriptions within the appropriate
canonical model described in Section 2.2.2. Once
created, a detailed conceptual model can be
written as a set of simultaneous differential
equations, and then translated into finite.
difference equations and programmed for
computer simulation. The outcome of these
simulations is the prediction of stressor behavior
in the ecosystem These models are specific for an
individual pollutant, so the processing pathways
and modifying factors may be somewhat different
for each material and ecosystem type; however,
the water flows governing residence time will be
similar within each of the canonical models.
3.2.1 A Generic Energy Systems Modelfor
Habitat Alteration
The mechanisms of habitat alteration are
somewhatdifferent from and more complex than
the other three stressors considered in this paper.
For this reason and because habitat alteration has
just been discussed, the generic energy systems
model of habitat is examined first. Nutrients,
suspended and bedded sediments, and toxic
substances are all materials that act as pollutants,
when excess amounts are added to an ecosystem.
Habitat alteration can be caused by a pollutant,
such as suspended and bedded sediments or a
toxicant, but it can also result from a physical or
biological change in the environment. For
example, diminished water flow, temperature
change, and oxygen depletion are physical
changes in a water body that may affect the
suitable habitat available to a particular species.
Habitat is always defined in relation to the needs
of a particular species or group of species. The
biological space, or conditions (habitat) available
to support that species is also a function of its
competitors and predators, thus, the introduction
or invasion of a new species may alter the habitat
available to established species. In general, the
many factors that alter habitat can be viewed as
modifying factors that affect the actions of a
particular pollutant in the ecosystem.
Studies of habitat change are customarily
focused at the population level of biological
organization and are usually performed with the
enhancement or preservation of a particular
species in mind. The health of a particular
population will depend on the existence of habitat
of sufficient quality to support growth and repro-
duction. From an energy systems perspective,
analyses that focus only on the habitat needs of a
single population or species are necessary, but
may not be sufficient to answer questions related
to the long term health and survival of that
population. This is true because the ecological
requirements of species are interconnected and
knowledge of the effects of habitat alteration on
one population or species may not reflect the
cumulative effects of habitat change in the whole
system. For these reasons single species studies
are not sufficient to answer the important
questions about the consequences of broad scale
habitat alteration that confront society. A better
understanding of the ultimate fate of an individual
species or population can be found by examining
change in the unique emergy signatures support-
ing the various required habitats for that species
and its coevolved prey, predators, and compete-
tors. The emergy signatures of habitats within the
system are determined by changes in the system
42

-------
Conceptual Models and Methods
forcing functions, which in turn are driven by the
dynamics ofthe next larger system. Therefore,
analysis at one level of organization in a hier-
archical system is seldom able to achieve an
understanding of the ultimate causes and
remedies for phenomena observed at that scale of
organization. In this model, we consider the
effects of habitat alteration within an ecosystem
context, viewing multiple levels of organization
and using the principles of energy systems theory
to trace causality.
3.2.1.1 A Comprehensive Measure of Impairment
The word "habitat" refers to the area and type of
environment in which a particular organism or
population normally lives. Therefore, the concept
of habitat is tied to a space (area or volume)
within an ecosystem. The emergy that enters this
area through its forcing functions determines the
ecological organization that can develop there,
and may be an integrative measure of habitat
quality for the species that can thrive under the
prevailing conditions. The change in the empower
density developed by a particular species within
an ecosystem that results from a change in
environmental forcing may be an integrative
measure for evaluating the overall effects of
habitat change on that species. Establishing that a
habitat has been altered is a prerequisite to
determining the cause of that alteration. The
degree of habitat alternation can be assessed by
measuring the empower flowing through the
habitat area including the species of interest and
comparing it to the empower of a reference state.
Empower ofthe habitat is an integrated measure
of ecological functioning that gives an estimate of
the overall effects of alteration on the species of
concern and on the ecosystem as a whole. Areas
of the landscape are or were suitable habitat
because they receive or have received in the past
a suite of forcing functions (an emergy signature)
that establishes ranges of variables and/or a suite
of energy flows within toe area (niches) which are
suitable for the survival and reproduction of the
various species that are found there. Over greater
or lesser periods of time, the operation of the
same or different external forcing functions have
been responsible for creating the stored emergy in
structures, e.g., vegetation, stream morphology,
etc., found in the area and necessary for the
persistence of a particular species. Habitat change
occurs when the emergy signature changes as a
result of changes in anthropogenic or natural
forces impinging on the area causing a loss of
structural components (e.g. species) or an
alteration of processes and thus a change in
empower of the system.
3.2.1.2 Model Description
An energy systems model to examine the
effects of habitat change was constructed (Figure
12), beginning with the narrative description of
altered habitat as a stressor and the summary
diagram presented in Figure 11. All forcing
functions, components and pathways in Figure 12
are defined in Table 4. The boundaries of the
system to be evaluated are shown as a large box
that contains the ecosystem components,
including the population or populations utilizing
the habitat. The choice of the system boundary
establishes the spatial scale (e.g., an estuary or
stream reach) that is needed to examine the
important factors controlling the growth and
survival of a species or group of species. The
choice of boundaries also establishes the
geometry of the system (i.e., average depth,
volume, etc.) and the habitats present within it.
Usually, state variables within the ecosystem are
evaluated per unit area or on the basis of the
system as a whole. We set the system boundaries
in Fig.12 to correspond with those of a stream
reach used by the states in 305 (b) assessments to
illustrate the process of model development and
to show how habitat alteration might affect an
example species, Qi, found within this system.
This model assumes that the ecosystem contains
habitat features (e.g., plant structure, stream form,
prey items) supporting species Qi; however, these
features can be changed as needed to represent
the particular species and ecosystem and
generalized to consider any number of species
(i =1 to n) and the competitors and predators of
that species, Cij. (i =1 to n, J = 1 to m). The
processes and components of the stream eco-
system modeled here can be viewed as a generic
representation of the classes of processes and
43

-------
Conceptual Models and Methods
Jw
Figure 12. Energy systems model of the effects of habitat alteration on a species, Qj, in an aquatic
ecosystem. Forcing functions, components, and pathway flows are defined in Table 4.
components that we believe are important in
determining the effects of habitat alteration in
most aquatic ecosystems. For example, to convert
this model for use in an estuary set it within the
canonical module for bidirectional flow and add
of subtract the appropriate state variables (e.g.,
add salinity and subtract stream form) and their
interactions.
Once the system and its boundaries are
chosen, the next step in building the model is to
specify the important external forcing functions.
Forcing functions include the energy, material,
and information sources that drive trends in the
internal components or state variables of the
system. In general, these external forcing
functions are state variables of a larger system
that can be understood through a diagrammatic
representation of the larger system at its level of
organization. The important forcing functions for
the example (Figure 12) are solar radiation, S,
temperature, T p, runoff, Jwo, groundwater base
flow, Jw\, wind, Wd, nutrient concentration in
runoff, No, oxygen concentration in runoff, 00,
suspended mineral solids concentration in runoff,
Sdo, the suspe.nded detritus (organic matter) in the
runoff, Do, the oxygen concentration at saturation
in the overlying water, OA, and animal immigra-
tion from and emigration to downstream eco-
systems, OS. The external forcing functions
comprise the energy and emergy signatures of the
system; they are arranged around the outside of
the box that delineates ecosystem boundaries and
are shown from left to right in order of increasing
transformity: In general, all the habitats used by a
species during the stages of its life cycle should
be included in the model, if the aggregate
reproductive success of a species is of concern.
Where critical habitats are widely separated in
space or are of very different character, modeling
and the determination of the energy flows
supporting the species, Qi, become more
complicated. In such cases, a linked series of
models should be used to evaluate the species of
concern.
44

-------
Conceptual Models and Methods
Table 4. Definition of the forcing functions, components, and pathways in the generic energy systems
model to evaluate the effects of habitat alteration on aquatic ecosystems.

Symbol Definition

Forcing Functions
S

JR
Tp
W
No

00
Sdo

Do
GW

Jwo

JW1
ET
A
Wd
OA
OS
Components
N
P
M
Ps
Sd
Sf
R
o
o
B
Q
C
Pathways and Flows
Jw
kl
k2
k3
k4
ks
k6
k7
ks
k9
Solar insolation as a time series
Solar radiation that remains unused (albedo)
Temperature as a time series
Watershed (time series of water flow)
Nutrient concentration in the runoff (can be a time series)
Oxygen concentration in the runoff (can be a time series)
Sediment concentration in the runoff (can be a time series)
Detritus (organic matter) concentration in runoff (can be a time series)
Groundwater Aquifer and its characteristics
Water flowing in as runoff
Groundwater base flow
Water evapotranspired in the system
Atmosphere system
Wind as a time series
Concentration of oxygen in the air
Downstream ecosystems
Nutrient
Phytoplankton
Macrophytes
Plant structure (critical to a species and used by competitors)
Sediment in the stream
Stream form (physical structure of the stream bed used by a species and/or its
competitors) .
Rock in the stream bed
Concentration of oxygen in the stream
Detritus on the stream bottom
Benthic bacteria
A species for which the effects of habitat change is to be determined
Competitors and predators of Q, the species of interest.
Water flowing out of the system
Light used by phytoplankton (the producer symbol implies GPP, NPP respiration.)
Light used by macrophytes (the producer symbol implies GPP, NPP respiration.)
Light attenuated by turbidity in the water
Nutrient inflow in runoff
Oxygen inflowing in runoff.
Suspended particulate matter inflowing in runoff.
Suspended particulate matter flowing out in streamflow.
Oxygen flowing out in streamflow.
Nutrients flowing out in streamflow.
45

-------
Conceptual Models and Methods
Symbol
klO
kit
kl2
kl3
kl4
kl5
kl6
kl7
kl8
'k19
k20
k21
k22
k23
k24
k25
k26
k27
k28
k29
k30
k31
k32
k33
k34
k35
k36
k37
k38
k39
k40
k41
k42
k43
~4
k45
~6
Definition
Immigration of competitors and predators from downstream ecosystems
Emigration of competitors and predators to downstream ecosystems
Immigration of species Q ITom downstream ecosystems
Emigration of species Q to downstream ecosystems
Nutrient uptake by phytoplankton
Nutrient uptake by macrophytes
Phytoplankton eaten by competitors
Phytoplankton death to detritus
Contribution of macrophytes to plant structure
Plant structure lost as a result of macrophyte losses
Oxygen produced by phytoplankton
Oxygen produced by macrophytes
Macrophytes eaten by species, Q
Macrophyte death to detritus
Macrophytes eaten by competitors and predators
Detritus eaten by species Q
Detritus eaten by competitors
Plant structure effects on species Q
Plant structure effects on competitors
Oxygen used by benthic bacteria.
Oxygen used by species Q.
Oxygen used by competitors and predators.
Nutrients recycled by competitors and predators.
Nutrients recycled by species Q.
Nutrients recycled by benthic bacteria.
Exchange of oxygen with the atmosphere.
Species Q consumed by predators.
Temperature effects on bacterial respiration.
Sediment building streamform.
Nutrient dissolved from rocks.
Rock dissolved by the water currents.
Currents building streamform from rock.
Stream structure lost by sedimentation (burial).
Effects of streamform on survival and growth of species Q.
Effects of stream form on survival and growth of competitors.
Detritus consumed by bacteria
Detritus (organic matter) supplied ITom outside the system
46

-------
Conceptual Models and Methods
After specifying the forcing functions, the
internal system components are specified. The
system components should include all the internal
state variables and processes that are necessary
for the survival and reproduction of the relevant
population or populations of concern and their
competitors and predators.
In Figure 12, the system components are
substrate including rock, R, bedded sediments,
Sd, stream form, Sf, water flow, Jw, vegetation as
phytoplankton, P, and aquatic macrophytes, M, .
plant structure, Ps, oxygen, 0, nutrient, N,
detritus, D, and bacteria, B, the species of interest,
Qi; and the competitors and predators of the
species of interest including all other important
species in the ecosystem aggregated according to
similar function, Cij. All components are
measured in appropriate units, usually mass or
energy. Structural metrics like streamform and
plant structure are expressed as measurable
quantities or attributes that correspond to the
properties of these variables that are important
aspects of habitat for the species of concern
and/or its competitors and predators.
In Figure 12, the network of interactions that
controls the growth and survival of species, Qi, is
shown by the lines connecting components and
processes. Each line represents a flow of energy,
material or information and is identified by a
pathway coefficient, ki, as defined inTable 4.
Briefly, the network shows runoff carrying in
flows of nutrient, ~, oxygen, k5, and suspended
sediment, k6, according to the streamflow and the
material concentrations that result from activities
in the surrounding watershed. Concentrations of
these same materials within the system are
removed by the stream outflow, Jw, 'as fluxes of
sediments, k7, oxygen, k8, and nutrients, k9. In
this model, suspended sediments, in addition to
those entering in inflowing water, are assumed to
be resuspended in proportion to the sediments
accumulated on the bottom. The light attenuated
by suspended sediments is shown on the pathway
designated, k3. The light absorbed by phytoplank-
ton and aquatic macrophytes is designated by
pathways, k1 and k2, respectively. The internal
structure of the producers and consumers, i.e.,
pathways of gross and net production and
respiration, are included in the definition of the
hierarchical producer and consumer symbols, but
they are not shown explicitly in the model.
Pathways klO through k13 show the emigration and
immigration of species Qi and the species
included in Cij to and from downstream
ecosystems. The uptake of nutrients by
phytoplankton and aquatic macrophytes is shown
by pathways k14 and k15, respectively.
Phytoplankton is consumed by competitors of
species Qi on pathway kl6 and phytoplankton die
and sink to the bottom on k17 forming detritus.
Pathways k20 and k21 represent the oxygen
produced by phytoplankton and aquatic
macrophytes, respectively. The increase in plant
structure occurs on pathway, k18, and plant
structure is lost as vegetation is consumed or dies
to become detritus, according to the sum of
pathways k22, k23, k24. Detritus consumed by
species Qi and its competitors is shown on
pathways, k25 and k26, respectively. The role of
plant structure in promoting growth and survival
of species Qi and its competitors is shown on
pathways, k27 and k28, respectively. The oxygen
consumed by bacteria, species Qi, and
competitors and predators is given on pathways,
k29, k30, and k3I. respectively. The nutrients
recycled by the metabolism of the Cij's, Qi, and B
are shown on pathways, k32, k33 and k34. The
exchange of oxygen with the atmosphere occurs
along pathway, k35, and is driven by the oxygen
concentration gradient between air and water and
the mixing energy supplied by wind and current.
The amount of species Qi consumed by its
predators is shown on pathway k36 and the effects
of temperature driving bacterial action is
governed by the pathway coefficient, k37.
Sediments are worked into stream structure along
pathway, k38, and rock dissolves, ~o, to yield
nutrients and form stream structure on pathways
k39 and ~ I. respectively. The structure of the
stream is lost on pathway, ~2, as sediments
accumulate on the bottom. The effects of stream-
form providiDg habitat that promotes the growth
and survival of species Qi and the competitors and
predators of species Qi are represented by
pathways, ~3 and ~4, respectively. Competitors
and predators might be separated in this model by
adding state variables. Detritus is consumed by
bacteria on pathway ~5' The biological oxygen
demand added to the system in the runoff is
represented by the flow on pathway ~6'
47

-------
Conceptual Models and Methods
If Qi is a particular species of fish whose
habitat is controlled by temperature and water
flow, this model could be considerably reduced in
complexity. In this example, the effects of
temperature on species, Qi, and its competitors,
Cij, is explicitly included but the affects of flow
regime are modulated through other factors such
as streamform, plant structure, and oxygen
concentration. If streamflow itself accounts for
most of the control action, a much simpler model
could be built and plugged into the canonical
model of unidirectional flow (Fig. 9a). In this
case, the effects of stream form, plant structure
and oxygen would be aggregated into the single
variable streamflow and the simpler model
evaluated against existing data.
The generic energy systems models that we
have developed in this paper are intended to
capture the most complexity that will be needed
to describe the action of a stressor in the majority
of cases. In this way, these models can serve as a
checklist for the completeness of an analysis and
as a guide to simplification through functional
aggregation. We envision that this generic model
might be used to guide analysis of the effects of
altered habitat in any aquatic ecosystem. Any'
modifications required to address a particular
question or system can be accomplished using the
methods and overview models presented above
and the expert knowledge of local scientists about
the particular system to be evaluated. An example
of an evaluated energy systems model for an
estuary can be found in Campbell (2005).
3.2.1.3 Evaluating Temporal Aspects of Habitat
Alteration
Species may have requirements for special
conditions in time as well as in space. The
temporal dimension of an energy systems model
is described by the time series of values used to
specifY the forcing functions. For example, the
complete definition of a forcing function entering
a system includes a time series of values whose
length determines the maximum time of the
simulation. The time series of a forcing variable
may have different frequencies or pseudo-
frequencies of oscillation. The highest frequency
of oscillation that can be investigated in the
model will be determined by the measurement
interval of the observations. A change in the
frequency of forcing events, e.g., the frequency of
floods of a given magnitude, may constitute a
change in the suitable habitat for species whose
life cycles are interrupted by the alteration of
inundation regimes. In general, the dynamic
properties of a system in time can be most easily
investigated using model simulations validated
with temporal data or with data gathered in space
for time substitutions. The relationship between
the frequency of a disturbance and the time
needed to recover from that disturbance has been
proposed as a sensitive indicator of the risk to a
given species from sporadic or repeated habitat
change (Turner et al. 1993). Simulation models
are ideal tools for investigating the temporal
dimension of habitat change.
3.2.1.4 Examples of Habitat Change
The critical parameters of a habitat may be
altered by: 1) A qualitative change in the emergy
signature of the place caused by the addition or
removal of an emergy source. For example, a
battery factory built on the edge of a small
estuary or wetland suddenly adds the pollutants
Pb, Cd, Zn, and Ni to the system (Odum et al.
2000), or a hurricane or large storm moves
enough sediment to close the breach way to a
coastal pond, ther~by removing the tidal forcing
supplied from the sea as seen in some of Rhode
Island's coastal ponds. The stored structures, e.g.,
the salt content of the waters, sand bars, etc. built
by the work of energy sources from the sea will
remain for varying times depending on the unique
turnover characteristics of the structure and
subsequent catastrophic events. 2) Habitat may be
altered by a change in magnitude of a forcing
function which carries a system variable outside
of the range suitable for a species. For example,
natural or anthropogenic climate change may'
result in sufficient warming in the Gulf of Maine
to preclude successful reproduction of cold water
species 'such as the sea scallop, Placopectin
magellanicus, over much of its area, conversely,
warmer waters can result in increased survival for
species like Homarus american us that are near
the northern limits of their range (Dow 1977).3)
Structural changes in a system might result from
direct or indirect affects of a qualitative or
quantitative change in the suite of forcing
functions. For example, road or bridge
construction in the Florida Keys could increase
48

-------
Conceptual Models and Methods
turbidity to a point where sea grass habitat would
be destroyed, or the nitrogen loading to an estuary
from a new municipal treatment plant could in-
crease primary production to the point where the
microbial consumption of fixed carbon settling to
the bottom depletes the oxygen concentration to
lethal levels. (4) Habitat may also be altered by a
regular or persistent shift in the timing or
frequency of forcing events. For example, climate
change might result in a change in the frequency
of occurrence of 100 year hurricanes in Florida
and Louisiana, which would result in increased
risk for people and property residing in coastal
habitats and lowland areas. Warmer winters in the
Gulf of Maine could more consistently support .
development of zooplankton populations early in
the year which would decrease the frequency of
large pulses of phytoplankton biomass delivered
to the benthos, and there by, diminish the number
of dominant year classes in demersal fish
populations of the region (Townsend and
Cammen 1988, Townsend et al. 1994).
In general, an existing habitat will be altered
by the introduction of a material across its
boundaries, when that material is concentrated
above a threshold for action. Once the material
exceeds this threshold, it becomes an additional
source of available energy for the system and will
have an effect. The available energy will either be
used by existing organisms or by organisms that
are carried to the system from other places, or it
will act as an energy drain (a stress) that exacts a
metabolic cost from the organisms that process it
and also from those that fail to process it. When
the emergy sources of an incoming signature are
balanced (Campbell 2000) and when the ecosys-
tem with its concomitant species and populations
have had sufficient time to adapt to these inputs,
usually many types of organisms are present and
their numbers are fairly well-balanced. When a
new energy source with either a positive or
negative effect is added, the growth of some
species are favored over others, the number of
species is often diminished, and those species best
adapted to use or resist the effects of the new
source overgrow the others and appear in greater
numbers in the ecosystem (Yount 1956). When a
stress is large enough to be close to. the edge of
the niche space for all organisms few survive and
both numbers and species richness will be
diminished.
3.2.2 A Generic Energy Systems Modelfor
Suspended and Bedded Sediments
An energy systems model of the action of
suspended and bedded sediments !n a stream
ecosystem was constructed (Figure 14) by
beginning with the narrative description of the
characteristics of suspended and bedded
sediments as a stressor and using the overview
diagram presented in Figure 13. All forcing
functions, components and pathways in Figure 14
are defined in Table 5. An area corresponding to
the boundaries of the system to be evaluated was
delineated, in this case a stream reach. The
system boundary is shown as a box enclosing the
ecosystem components including the populations
and stream features affected by suspended and
bedded sediments (Figure 14). The processes and
components of the fresh water ecosystem
modeled here can be viewed as a generic
representation ofthe classes of processes and
components that are important in determining the
effects of suspended and bedded sediments in any
aquatic ecosystem. In large rivers and. estuaries
dredging is a major source of problems related to
the deposition of suspended and bedded.
sediments and this process could be added to the
models for those systems. This generic model can
be used to guide the allocation of biological
effects to suspended and bedded sediments in any
aquatic system with appropriate modifications,
guided by the canonical models presented above,
and the expert knowledge of local scientists about
the particular system to be evaluated. For.
example, to evaluate a shallow estuary the
canonical model for bidirectional flow (Figure
9b) would be used to structure water flows and
turnover times of the system rather than the
unidirectional flow model used here and clean
sediment processes of concern, such as dredging
and/or the disposal of dredged materials would be
added to the model.
The important forcing functions for the
stream reach shown in Figure 14 are solar
radiation, S, temperature, T p, runoff, Jwo,
groundwater base flow, JwJ, wind, Wd, nutrient
concentration in runoff, No, bedload transport into
the system by streamflow, Sbo, suspended
sediment concentration in runoff, SSo. The
immigration of pelagic and benthic fish from
downstream ecosystems, OS, into the stream
49

-------
Conceptual Models and Methods
reach is shown as proportional to the downstream
populations, BFD and PFD. By convention, the
external forcing functions are arranged around the
outside of the box delineating ecosystem
boundaries in order of increasing transformity.
The system components should include all
the internal storage units that are important in
determining the effects of suspended and bedded
sediments in the stream. The important internal
system components in the clean sediment model
are shown on Figure 14 and identified in Table 5.
The system components are substrate including
rock, R, bedded sediments, Sb, streamform, Sf,
streamflow, Jw, kinetic energy in the water, Wke,
suspended sediments, Ss, vegetation as
phytoplankton, P, aquatic macrophytes, M,
nutrient, N, detritus, D, pelagic invertebrates, PI,
benthic invertebrates, BI, pelagic fish, PF, and
benthic fish, BF.
The network of interactions is represented by
the lines connecting the components and forcing
functions through processes. Each line represents
a flow of energy, material or information and is
identified by a pathway coefficient, kj, in Figure
14. The network includes runoff carrying inflows
of nutrient, kt, bedload sediment, k5, and
suspended sediment, ~, according to the concen-
trations supplied by activities in the surrounding
watershed. These same concentrations are
removed by the stream outflow, Jw, as fluxes of
suspended sediments, k7, nutrients, k8, and
bedload sediments k9. In this model some kinetic
energy of the water goes into vertical mixing, k39,
mediating the resuspension and settling of
suspended sediment along pathway k25. The light
attenuated by suspended sediments is shown on
the pathway designated, k3. The light absorbed by
phytoplankton and aquatic macrophytes is
designated on pathways k(, and k2, respectively.
Pathways klO through k13, show the emigration
and immigration of benthic and pelagic fish to
and from downstream ecosystems. The uptake of
nutrients by phytoplankton and aquatic .
macrophytes is shown by pathways kl4 and k15,
respectively. Phytoplankton is consumed by
benthic invertebrates on pathway k\6 and kl7
shows the phytoplankton consumed by pelagic
invertebrates. On pathway k18. phytoplankton dies
and sinks to the bottom forming detritus. Pathway
kl9 shows the negative effects of abrasion and
scour on aquatic macrophytes and pathway k20
carries the dead macrophyte biomass into the
detritus pool. The positive support given to
macrophyte growth by well-developed
stream form acts on pathway k22. Pathway k23
shows the detritus consumed by benthic
invertebrates and pathway k24 represents the
detritus lost by burial. Additional pathways of
external supply and downstream washout can be
added if loading with organic material is an
important factor accompanying loading with
clean sediment. Pathway k25 shows the net
vertical flux of suspended sediments governed by
the turbulent mixing energy in the water. Pathway
k26 represents the action of water column energy
building streamform. Note that this process is
modeled as a push-pull interaction (Odum 1994)
and that the kinetic energy in the water can result
in a decrease in streamform, if it exceeds the
optimum amount needed to build form in a
particular stream. Pathways k27 and k28 represent
the positive effects of the expected stream form on
the growth and survival of benthic fish and
invertebrates, respectively. The effect of bottom
roughness on the production of turbulent energy
in the water is represented by pathway k29.
Pathway k30 represents the loss of stream
structure as a result of burial by and accumulation
of excess clean sediment. Note that many
ecosystem components, including macrophytes,
benthic invertebrates, benthic fish, and detritus
will be affected negatively by the loss of
stream form. Pathway k3\ carries the nutrient
dissolved from rock by the flowing waters and
pathway k32 represents the stream form created as
a result of this action. The negative effects of
scour and abrasion on benthic invertebrates are
shown on pathway k33. The benthic invertebrates
eaten by benthic fish and pelagic fish are shown
on pathways k34 and k35, respectively. The pelagic
invertebrates eaten by benthic fish and pelagic
fish are shown on pathways k36 and k37,
respectively. Pathway k38 represents the effects of
abrasion on pelagic invertebrates. The production
of turbulent kinetic energy in the water is shown
on pathway k39 and the kinetic energy used to
drive vertical circulation and resuspension is
shown on pathway kto. The action of the wind in
generating kinetic energy in the water is
represented on pathway kt(, and the
50

-------
Conceptual Models and Methods
Climate
~=::::::---------~------
-----. .
Benthic
Invertebrates
Figure 13. Summary diagram showing the factors that control the effects of suspended
and bedded sediments in aquatic ecosystems.
Jw
1
Figure 14. An energy systems model of the effects of suspended and bedded sediments on aquatic ecosystems.
Forcing functions, components, and pathway flows are defined in Table 5.
51

-------
Concept~al Models and Methods
Table 5. Definition of the forcing functions,. components, and pathways in the generic energy systems
model to evaluate the effects of suspended and bedded sediments on aquatic ecosystems. .

Symbol Definition

Forcing Functions
S
JR
Tp
W

SSo
Sbo

No
GW
Jwo
Jw,
ET
A
Wd
DS
Components
N
P
M
Wke
Ss
Sf
Sb
R
D
PI
PF
BI
BF
Pathways and flows
Jw
k,
k2
k3
kt

ks
k6
k7
ks
k9
klO
Solar insolation as a time series
Solar radiation that remains unused (albedo)
Temperature as a time series
Watershed (time series of water flow)
Suspended sediments in the runoff (can be a time series)
Bedload sediments trom upstream (can be a time series)
Nutrient concentration in the runoff (can be a time series)
Groundwater aquifer and is characteristics
Water flowing in as runoff .
Groundwater base flow
Water evapotranspired
Atmosphere system
Wind as a time series
Downstream ecosystems
Nutrient
Phytoplankton
Macrophytes
Kinetic energy in the water
Suspended sediment in the stream
Streamform (physical structure of the stream bed used by a species and/or its
competitors)
Bedded sediments
Rock in the stream bed
Detritus on the stream bottom
Pelagic invertebrates
Pelagic fish
Benthic invertebrates
Benthic fish
Water flowing out of the system .
Light used by phytoplankton (the producer symbol implies GPP, NPP respiration.)
Light used by macrophytes (the producer symbol implies GPP, NPP respiration.)
Light attenuated by turbidity in the water (shading)
Nutrient inflow in runoff
Bedload transport into the system.
Suspended sediments inflowing in runoff.
Suspended sediments flowing out in streamflow.
Nutrients flowing out in streamflow.
Bedload transport out of the system.
Immigration of benthic fish from downstream ecosystems
52

-------
Conceptual Models and Methods
Symbol.
kll
kJ2
k\3
kl4
kl5
kl6
kl7
kl8
kl9
k20
k21
k22
k23
k24
k25
k26
k27
k28
k29
k30
k31
k32
k33
k34
k35
k36
k37
k38
k39
k40
~I
k42
~3
~4
k45
k46
k47
~8
Definition
Emigration of benthic fish to downstream ecosystems
Immigration of pelagic fish from downstream ecosystems
Emigration of pelagic fish to downstream ecosystems
Nutrient uptake by phytoplankton
Nutrient uptake by macrophytes
Phytoplankton eaten by benthic invertebrates
Phytoplankton eaten by pelagic invertebrates
Phytoplankton settling to detritus
Scour and abrasive effect of suspended sediments on macrophytes
Injured and dead macrophyte bioma~s going to detritus
Macrophyte biomass eaten by benthic invertebrates
Positive effect of streamform on macrophytes
Detritus eaten by benthic invertebrates
Detritus buried by sediments
Resuspension of bedded sediments and settling of suspended solids
Kinetic energy of water building streamform from sediments
Positive effect of streamform on benthic fish
Positive effect of streamform on benthic invertebrates
Roughness of streamform creating turbulence
Loss of streamform by burial
Nutrients dissolved from rock
Streamform built by kinetic energy acting on rock in the stream bed
Scour and abrasive effects of suspended sediments on benthic invertebrates
Benthic invertebrates eaten by benthic fish
Benthic invertebrates eaten by pelagic fish
Pelagic invertebrates eaten by benthic fish
Pelagic invertebrates eaten by pelagic fish
Scour and abrasion effects of suspended sediments on pelagic invertebrates.
KinetiC energy transported into and generated by fluid flow in the system.
Kinetic energy dissipated in mixing.
Kinetic energy generated by wind.
Kinetic energy transported out of the system by streamflow.
Benthic fish eaten by pelagic fish.
Pelagic fish eaten by benthic fish.
Nitrogen recycled by benthic fish.
Nitrogen recycled by pelagic fish.
Nitrogen recycled by benthic invertebrates.
Nitrogen recycled by pelagic invertebrates.
53

-------
Conceptual Models and Methods
Stream ---+ River ---+ lake---+
Estuary ---+ Coastal
............YY.~~~r..I3.~~J9.~.Q~~.::n.m~..~ff~.~~~..~!!.IQ.~i.~.g.f:1.~f.Q!~~.I..~~p..9.~!-:J.r~.~.....~
Metals
[hardness, pH, AVS, DOC, TOC]
[salinity, AVS, DOC, TOC, pH]
Nonionic Organics
[DOC, TOC]
[DOC, TOC, salinity]
Ammonia

[ pH, temperature]
[pH, salinity, temperature]
Freshwater
Marine
Figure 15a. Factors that affect the bioavailability of toxic chemicals
in fTeshwater and marine systems.
Freshwater, Marine and Estuarine Water Bodies

~dine: factors - Type A modifiers
TOC/DOC- Nonionic Organics, Metals

Mitie:atine: factors - Type B Modifiers

PHH d }- Metals, Ammonia
ar ness
Temperature-Ammonia, N. Organics, Metals
Water
. Sediment
Kow
[T w] Solubility
Bioturbation
Cone. Gradient Flow
Temperature
Hydrolysis
[T s] Photo-degradation
Bacterial degradation
Binding factors - Type A Modifiers
TOC- Nonionic Organics
A VS - Metals, Carbon
I ::;tigating}factors -Type B Modifiers

I Hardness Metals, Ammonia
. eH-redox
Temperature- Ammonia, N. Organics, Metals
Figure 15 b. Factors that affect the availability of toxic chemicals in the water
column and in sediments. T is the concentration of bioavailable toxin.
54

-------
Conceptual Models and Methods
removal of kinetic energy in streamflow is shown
as pathway ktz. Pelagic fish eat benthic fish, kt3,
and in turn benthic fish eat pelagic fish, kt4.
Nutrients are recycled by the metabolism of the
benthic fish, pelagic fish, benthic invertebrates,
and pelagic invertebrates, as shown on pathways,
kt5, kt6, kt7, and ktg, respectively. The reader
should also examine Kaufmann et al. (1999) for
guidance on developing clean sediment models
for stream ecosystems. .
3.2.3 A Generic Energy Systems Modelfor
Toxic Chemicals
An energy systems model of the actions of
toxic chemicals in aquatic ecosystems was
constructed (Figure 16) by beginning with the
narrative description of the characteristics of toxic
chemicals as a stressor and using the conceptual
summary presented in Figure 15a and b. The
construction of a generic energy systems model
that covered all toxic chemicals was a challenge,
because toxic actions are diverse and the number
of toxic chemicals is large. Nevertheless, at a high
level of abstraction, we believe that all toxicants
demonstrate some basic similarities in behavior.
The narrative description and Figure 15
summarize the salient factors controlling toxic
action in fresh water and marine systems. We
propose that most toxic chemicals can be modeled
as a subsystem containing state variables for
available toxin and unavailable toxin (Figure 16).
An available toxicant is capable of causing toxic
effects to organisms, populations, and the
ecosystem. An unavailable toxic chemical has
been sequestered by one or more possible
mechanisms that depend on the particular
chemical. For example, organic toxic chemicals
like PCBs partition into organic material adsorbed
to particulate matter, and positively charged metal
ions are neutralized by sulfates and other
negatively charged ligands. Our generic model of
toxic action in an ecosystem is centered on the
processes controlling the availability of the toxic
chemical in sediments and the overlying water.
We also hypothesize that despite the diversity of
mechanisms of toxic action all forms of toxicity
act as an energy drain on biological organization
(Odum 1968). The'se energy drains can manifest
as direct mortality or they can impose an
additional metabolic cost on organisms that
diminishes growth and reproduction. These
common characteristics of toxic substances and
toxic action are represented in the energy systems
model presented in Figure 16. The details of the
model constructed for a particular pollutant will
be different depending on the chemical and its
properties, but we believe that this highly
aggregated conceptual model can serve as a
useful guide in organizing thinking and
structuring analyses of the effects of toxic
chemicals on aquatic ecosystems.
All forcing functions, components and
pathways in Figure 16 are defined in Table 6. An
area corresponding to the boundaries of the
system to be evaluated is delineated along with its
accompanying geometric characteristics, in this
case the area, average length, width, depth, etc. of
the stream reach. The system boundary is shown
as a box enclosing the ecosystem components
including the plant and animal populations
affected by toxic loadings. The formulations
given in this model are more precise than those
given for the other three stressors, because the
specifics of some component interactions are
shown within the aggregate symbols for
producers and consumers. This amount of detail
allows us to write a set of simultaneous
differential equations describing the network of
interactions directly from the energy circuit
diagram. We have not given the equations
because the purpose of these models is to provide
a conceptual overview and not the mathematical
formulations that would be used in simulation.
The rate processes governing the equilibrium
between bioavailable and sequestered toxic
chemical are given by pathway coefficients for
the toxic chemical subsystem in the water
column. Similar rate processes are diagramed in
the sediments, but have not been explicitly
identified with coefficients. However, these
processes mirror those given for the water column
and the reader can easily identify parallel
processes in the sediment diagram. This generic
model can guide the allocation of biological
effects to toxic chemicals in any aquatic system
with appropriate modifications, guided by the
canonical models and the expert knowledge of
55

-------
Conceptual Models and Methods
ET
Jw
Figure 16. An energy systems model for the effects of a toxic chemical, T, on an aquatic ecosystem.
Forcing functions, components, and pathway flows are defined in Table 6.
local scientists about the particular system to
be evaluated.
The important forcing functions for the
model of the effects of toxic chemicals on a
stream reach (Figure 16) are solar radiation, S,
temperature, T p, runoff, Jwo, and waste water
inflow, JWI. All of these forcing functions are
given as a time series of values when fully
specified. Each of the water inflows can carry
concentrations of nutrients, chemicals, and other
materials determined by the characteristics of
upstream flow and land use in the immediate
watershed of the reach, W, or by the
characteristics of the waste stream from waste
water treatment facilities, WTF. The nutrient
concentration, No, the concentration of toxic
chemical, To, and the concentrations of
modifying chemicals, Xo, in streamflow and
immediate runoff enter the stream reach. The
nutrient, toxic chemical, and modifying
chemical'loadings in the waste stream are given
as the concentration of a single nutrient, N 1,
toxic chemical, TI, and modifying chemical, XI,
respectively. Where more than one pollutant is
important in the process under evaluation, the
model can be expanded to consider these
additional materials. A final forcing function is
the immigration and emigration of consumers
from downstream ecosystems, DS, into the
stream reach under evaluation. Consumer
movements are proportional to the downstream
populations, CDS, and seasonal or other
behavioral programming, MP. The program
controlling migration was made more complex
by adding temperature to provide a temporal cue
for migration and by using the burden of toxic
chemicals carried by the organisms, XTc, to
make the probability of migration more or less
likely by affecting the organism's behavior. The
program controlling the migratory behavior of
consumers can be duplicated on the upstream
boundary where this interaction is important.
The system components should include all
internal state variables that are thought to be
56

-------
Conceptual Models and Methods
Table 6. Definition of the forcing functions, components, and pathways in the generic energy systems
model designed to evaluate the effects of toxic chemicals on aquatic ecosystems.

Symbol Definition

Forcing Functions
S

JR
Tp
W
Jwo

No

To

Xo
WTF
JW1
NI
TI
XI
ET
OS

CDS
MP
Components
N
P
X
Tw
ATw
XTw
C
XTc
Sd
B
Ts
ATs
XTs
Xs
Pathways
Jw
kl
k2
k3
~
ks
k6
k7
kg
Solar insolation as a time series
Solar radiation that remains unused (albedo)
Temperature as a time series
Watershed (time series of water flow)
Water flowing in as runoff
Nutrient concentration in the runoff (can be a time series)
Concentration of the toxic chemical in the runoff (can be a time series)
Concentration of chemicals that modify toxicity (can be time series)
Outflow trom waste water treatment facilities (can be a time series)
Waste water inflow .
Nutrient concentration in the waste water (can be a time series)
Concentration of the toxic chemical in was~e water (can be a time series)
Concentration of modifying chemicals in waste water (can be a time series)
Water evapotranspired in the system
Downstream ecosystems
Downstream populations of migrating consumers
Behavioral programs controlling animal migrations
Nutrient
Primary producers
Modifying chemicals in the water
Toxic chemical subsystem
A vailable toxin in the water
Unavailable toxin in the water (i.e., chemically bound or neutralized)
Consumers
Toxic chemicals in the tissue of consumers
Sediments
Bacteria
Toxic chemical subsystem in the sediments
Available toxin in the sediment pore water
Unavailable toxin in the sediments (i.e., chemically bound or neutralized)
Modifying chemicals in the sediment
Water flowing out of the system
Light used by primary producers
Nutrient used by the primary producers
Gross primary production
Nutrient inflow in runoff
Toxic chemical inflowing in runoff. .
Modifying chemicals inflowing in runoff.
Nutrient added in waste water.
Toxic chemical added in waste water.
57

-------
Conceptual Models and Methods
Symbol
k9
klO
kll
kl2
k13
kl4
k\5
k\6
kl7
k\8
k\9
k20
k2\
k22
k23
k24
k25
k26
k27
k28
k29
k30
k3\
k32
k33
k34
k35
k36
k37
k38
k39
~o
~\
~2
k43
~4
k45
~6
~7
Definition
Modifying chemicals added in waste water.
Modifying chemicals flowing out in water leaving the system.
Toxic chemical flowing out in water leaving the system.
Nutrients flowing out in water leaving the system.
Immigration of consumers from downstream or seaward ecosystems
Emigration of consumers to downstream or seaward ecosystems
Available toxin used up in decreasing plant growth.
A vailable toxin used up in causing direct mortality of plants
Consumption of primary producers by consumers.
Modifying chemicals produced by plants
Plant death from direct mortality caused by available toxin
Plant death from other causes
Plant respiration
Toxic chemical bound and precipitated to the bottom
Modifying chemicals used to bind the toxin
Bound toxin settling to the bottom
A vail able toxin in the sediment used in decreasing consumer growth
A vail able toxin in the sediment used to cause direct mortality of consumers
A vail able toxin in the water used in decreasing consumer growth
A vail able toxin in the water used to cause direct mortality of consumers
Resuspension and settling of the toxic chemical subsystem
Resuspension and settling of the modifying chemicals
Bacterial consumption of modifying chemicals if appropriate
Bacterial decomposition and release of bound toxin from the sediment
Bacterial release of modifying chemicals from the sediment
Toxic chemicals eaten by consumers
Toxic chemical incorporated into consumer biomass
Carbon assimilated by consumers
Toxic chemical and carbon metabolized by consumers
Loss of biomass and toxins in mortality caused by available toxin.
Modifying chemicals interacting with available toxin in the water
Available toxin bound by modifying chemicals in the water
Bound toxin produced in the water
Bound toxin decomposed in the water column .
Available toxin generated in the water by decomposition of bound toxin
Modifying chemicals generated ~y the decomposition of bound toxin
Nutrients recycled by consumer respiration.
Nutrients recycled by bacterial respiration.
Nutrients recycled by plant respiration.
58

-------
Conceptual Models and Methods
I R;,,, H
I
Estuary
~
Coastal
Boundary
Shelf
Physico-chemical factors
Sources:
Atmospheric Deposition
WatershedlLand use NPS input
Point-source and Riverine input
Reactive substances:
TN, TP, DIN, DIC,
DOC, POP, DIP, DO,
DOC, TSS, DOM
Modifying factors:
Dissolved & Particulate
Material, DO,
Temperature, Hardness,
Hypsography, etc.
Residence Time:
Freshwater Inflow
Surface Area
DCPIPRE
Biological components
Producers:
Bacterial Processes:
Denitrification,
N fixation, etc.
Macrophytes
Phytoplankton
Consumers:
I
Benthic suspension feeders,
deposit and interface feeders.
Benthivorous Fish
Zooplankton
Planktivorous Fish
I
Top Predators
Figure 17. Factors that control the effects of nutrients in aquatic ecosystems.
important in determining the effects of the toxic
chemical in the stream. Where new knowledge
or additional research indicates that other
components are important they must be added to
the model structure. In energy systems models,
components are combined according to function
as shown by the broadly aggregated producer
and consumer components in Figure 16. Where
analysis shows that there are relevant differences
in behavior within an aggregate group, it can be
broken down according to functional differences
to provide the additional detail needed to explain
observations. Each toxic chemical and
modifying chemical combination will require the
particular details ofthe relationship to be
substituted into the general form of the model
given here.
The ecosystem components found in the
water are nutrients, N, the toxic chemical
subsystem, T w, including bioavailable toxic
chemical, AT w, and bound or sequestered toxiC
chemical, XT w; modifying chemicals, X,
primary producers, P, and consumers, C. In
addition, the model includes a sediment phase
containing sediments, Sd, the toxic chemical
subsystem in the sediments, Ts, including
bioavailable toxic ch~mical, AT s, and bound or
sequestered toxic chemical, XT s, modifying
chemicals, Xs, and bacteria, B.
The network of interactions is represented
by the lines connecting the components and
forcing functions through processes as shown in
Figure 16. Each line represents a flow of energy,
. material or information and is identified by a
pathway coefficient, the kjs. The light absorbed
by phytoplankton and aquatic macrophytes is
designated by pathway k]. Pathway k2 shows the
nutrient taken up in primary production. The
gross primary production, GPP, of the plants is
given on the pathway designated h
Temperature drives the metabolic processes of
plants, consumers, and bacteria in the ecosystem
using an Arrhenius formulation (ekT) to control
the rate. Runoff carries in flows of nutrients, ~,
toxic chemicals, k5, and modifying chemicals,
kt" according to the concentrations supplied by
59

-------
Conceptual Models and Methods
activities in the surrounding watershed.
Nutrients, toxic chemicals, and modifying
chemicals (type A modifiers) can also enter the
system in the waste water stream as shown,
respectively, on pathways, k7, k8, and k9. These
same concentrations are removed by the stream
outflow, Jw, as fluxes of modifying chemicals,
klO, toxic chemicals, kll, and nutrients, k12.
Pathways kl3 and kl4 show the immigration and
emigration of consumers to and from
downstream ecosystems under the control of
behavioral migration programs that are modified
by temperature and the accumulation of toxic
chemical in animal tissue.
The negative effects of available toxic
chemical on primary production are shown on
pathway kls. The available toxin taken up in
causing direct mortality to plants is represented
by pathway k16. The plants eaten by consumers
are shown on pathway kl7 and the modifying
chemicals, e.g., dissolved organic carbon, DOC,
released by plants are given on pathway k18.
Pathway kl9 shows plant biomass that dies from
toxic exposure and sinks to the bottom forming
detritus, while pathway k20 represents plants that
die from natural causes. Plant respiration, shown
on pathway k21, is a function of temperature.
Pathways k22, k23, and k24, show the interaction
of toxic chemicals and modifying chemicals in
the water to make the toxic chemical unavail-
able. Pathway k2S shows the adverse effects of
available toxic chemicals in the sediment on the
growth of consumers and pathway k26 represents
direct mortality of animals caused by available
toxins in the sediment. Similarly, pathways k27
and k28 show the decreased growth and direct
mortality of consumers caused by available
toxicant in the water.
Mixing in this stream ecosystem is assumed
to be proportional to water flow and the
concentration gradient between sediments and
the water column. The net resuspension or
settling of materials in the toxic chemical
subsystem is shown on pathway k29. A similar
balance of resuspension and settling for the
modifying chemicals is given on pathway k30.
The modifying chemicals consumed by bacteria
are shown on pathway k31. Bacteria breakdown
unavailable toxic chemical bound in the
sediment and recycle it to the water column on
pathways k32 and k33. The consumption of bound
toxic chemical and modifying chemicals from
the sediment by consumers is given on pathway
k34. Consumer biomass growth is shown on
pathway k36 and the concomitant incorporation
of toxic chemicals into animal tissue is given on
pathway k3S. Pathway k37 shows the respiration
of consumers and pathway k38 gives the
consumer biomass that dies and returns to the
sediments with its body burden of toxic
chemical. The transitions between available and
unavailable forms of the toxic chemical within
the toxic chemical subsystem in the water
column are given by the next six pathway
coefficients (Table 6). A similar set of
coefficients governs these transitions in the
sediment and they can be easily derived by
following the pattern used for the water column.
Pathway k39 is the quantity of modifying
chemical, X, that interacts with a quantity of
available toxic chemical, ~o, to form
unavailable toxicant in the water. Conversely,
a quantity of unavailable toxic chemical in the
water, ~2, decomposes both biologically and
chemically under the influence of temperature to
form a quantity of available toxic chemical on
pathway, ~3, and modifying chemicals on
pathway ~4. Nutrients are recycled into the
water by consumers, ~s, bacteria, ~6, and
plants, ~7'
(/
3.2.4 A Generic Energy Systems Modelfor
Nutrients
An energy systems model of the actions of
nutrient enrichment in aquatic ecosystems was
constructed (Figure 18) using the narrative
description of nutrient enrichment and the
overview diagram presented in Figure 17. Figure
18 like the generic energy systems models for
the other three stressors is structured using a
fresh water stream ecosystem as the example.
The intent is that these generic models serve as a
useful guide to understanding the stressor's
action over a range of aquatic ecosystems (see
the spectrum of system types shown at the top
of Figures 11, 13, 15a, and 17). As pointed out
in the narrative description, phosphorus is often
the limiting nutrient in fresh water and terrestrial
ecosystems, whereas, nitrogen is more
frequently limiting in coastal and marine waters.
60

-------
Conceptual Models and Methods
Jw
Figure 18. An energy systems model of the effects of excess nutrients on an aquatic
ecosystem. Forcing functions, storages, and flows are defined in Table 7.
Ecosystem processes controlling nitrogen are
included in the generic model for nutrient
enrichment, because nitrogen can be limiting in
fresh water stream systems and the pathways are
needed to model eutrophication in coastal and
marine ecosystems. All forcing functions,
components and pathways in Figure 18 are
defined in Table 7.
First, an area corresponding to the
boundaries of the system to be evaluated is
delineated. The system boundary for a stream is
shown as a box enclosing the ecosystem
components including the populations and
stream features affected by nutrient loading
(Figure 18). This generic model can be used as a
guide for allocating observed biological effects
to nutrient loading in any aquatic system with
appropriate modifications, guided by the
canonical models and expert knowledge of
scientists about the particular system to be
evaluated. For example, to evaluate the effects
of nutrient loading on a shallow estuary the
canonical model for bidirectional flow (Figure
9b) would be used to represent water flow and
turnover time of the system rather than the
unidirectional flow model used here. For a
phosphorus-limited, deep lake, the nitrogen
pathways shown in Figure 18 might be omitted
and other appropriate pathways added using the
canonical model in Figure 9c. The important
forcing functions for the stream reach shown in
Figure 18 are solar radiation, S, temperature, T p,
runoff, Jwo;waste water flow, JwJ, wind, Wd.
All of these forcing functions are given as a time
series of values when fully specified. Each of the
water inflows can carry concentrations of
nutrients determined by the characteristics of
land use in the watershed, W, or by the
characteristics ofthe waste stream from the
waste water treatment facility, WTF. The
nutrient concentration in runoff, No, the turbidity
in runoff and streamflow, To, and oxygen in
streamflow enter the stream reach from
61

-------
Conceptual Models and Methods
Table 7. Definition of the forcing functions, components, and pathways in the generic energy systems
model to evaluate the effects of nutrient loading on aquatic ecosystems.

Symbol Definition

Forcing Functions
S

JR
Tp
W
Jwo
No
To
00
WTF

JW1
N]
ET
A
Wd
OA
NA
.NR
DS

CDS
MP
Components
N
P
M
T
o
Sd
D
B
Np
C
Pathways

Jw
kl
k2
k3
~
ks
k6
k7
Solar insolation as a time series
Solar radiation that remains unused (albedo)
Temperature as a time series
Watershed (time series of water flow)
Water flowing in as runoff
Nutrient concentration in the runoff(can be a time series)
Turbidity in the runoff (can be a time series)
Oxygen concentration in the runoff (can be a time series)
Outflow from waste water treatment facilities (can be a time series)
Waste water inflow
Nutrient concentration in the waste water.(can be a time series)
Water evapotranspired in the system
Atmosphere system
Wind as a time series
Concentration of oxygen in the air
Diatomic nitrogen concentration in the atmosphere
Nitrogen in rain (can be an important input to estuaries)
Downstream ecosystems
Downstream populations of migrating consumers
Behavioral programs controlling animal migrations
Nutrient
Phytoplankton
Macrophytes
Turbidity in the water
Concentration of oxygen in the stream
Sediment on the bottom
Detritus
Bacteria
Nitrogen pool in the sediment
Consumers
Water flowing out of the system
Light used by phytoplankton (the producer symbol implies GPP, NPP respiration.)
Light used by macrophytes (the producer symbol implies GPP, NPP respiration.)
Light attenuated by turbidity in the water
Nutrient inflow in runoff
Oxygen inflowing in runoff.
Turbidity inflowing in runoff.
Nutrient added in waste water.
62

-------
Conceptual Models and Methods
Symbol
ks
k9
klO
kll
kl2
kJ3
kl4
kl5
kl6
kl7
kls
kl9
k20
k2!
k22
k23
k24
k25
k26
k27
k2S
k29
k30
k31
k32
Definition
Nutrient supplied in rainfall
Oxygen flowing out in water leaving the system.
Turbidity flowing out in water leaving the system.
Nutrients flowing out in water leaving the system.
Immigration of consumers from downstream or seaward ecosystems
Emigration of consumers to downstream or seaward ecosystems
Nutrient uptake by phytoplankton
Nutrient uptake by macrophytes
Phytoplankton eaten by consumers
Phytoplankton death to detritus
Oxygen produced by phytoplankton
Oxygen produced by macrophytes
Oxygen used by consumers.
Oxygen used by bacteria
Oxygen exchange with the atmosphere
Macrophytes eaten by consumers
Macrophyte death to detritus
Detritus consumed by bacteria
Detritus eaten by consumers
Nitrogen fixation
Denitrification
Nitrogen used by bacteria.
Nutrients recycled by bacterial metabolism
Nutrients recycled by the metabolism of consumers
Settling and resuspension of turbidity
upstream. Nutrient loading in the waste stream is
given as the concentration of a single pollutant,
N1. Where more than one material is important in
the eutrophication process the material inputs and
the model storages and interactions can be
expanded to consider additional materials. Odum
(1994) gives model formulations for two limiting
nutrients. Inputs from and interactions with the
atmosphere, A, may be needed to model the
effects of nutrient enrichment. Wind energy, Wd,
along with streamflow, Jw, drive the exchange of
oxygen across the water surface in proportion to
the difference between the concentration of
oxygen at saturation in the atmosphere, OA, and
oxygen concentration in the water, O. In addition,
for lakes, wide estuaries, and coastal and shelf
waters, the input of nutrient concentrations in
rainfall, NR, can be important (the time series of
rainfall might be added to the model where this
input is large). Diatomic nitrogen gas in the
atmosphere, NA, serves as a source of nitrogen for
bacterial processes fixing nitrogen and receives
the nitrogen that results from bacterial
denitrification. A final forcing function is the
movement of consumers in the stream reach into
and out of downstream ecosystems, OS. The
movement of consumers into the reach is
proportional to the downstream populations, CDS,
controlled by seasonal and/or other behavioral
programming, MP.
The system components should include all
internal state variables or stored quantities that are
thought to be important in determining the effects
of nutrient loading in the stream. When the
analysis shows that there are differences in .
function or behavior within an aggregate group,
the group can be disaggregated to provide the
additional detail needed to explain the observed
data. The ecosystem components are nutrient, N,
63

-------
Conceptual Models and Methods
turbidity, T, oxygen, 0, phytoplankton, P, aquatic
macrophytes, M, detritus, D, sediments, Sd,
bacteria, B, the sediment nitrogen pool, Np, and
consumers, C, including a broad range of .
organisms (Figure 18).
The network of interactions is represented by
the lines connecting the components and forcing
functions through processes as shown in Figure
18. Each line represents a flow of energy,
material or information and is identified by a
pathway coefficient, the kj's on Figure 18. The
light absorbed by phytoplankton and aquatic
macrophytes is designated by pathways kl and k2,
respectively. The light attenuated by turbidity in
the water is shown on the pathway designated, k3.
Runoff carries in flows of nutrient, ~, oxygen, ks,
and turbidity, k6, according to the concentrations
supplied by activities in the surrounding
watershed. Nutrients also enter the system in the
waste water stream, k7, and in rainfalI, ks. These
same concentrations are removed by the stream
outflow, Jw, as fluxes of oxygen, k9, turbidity, klO, .
and nutrient kll. Pathways kl2 and k13 show the
immigration and emigration of consumers to and
from downstream ecosystems under the control of
temperature driven behavioral migration
programs. Migratory movements are controlIed
by temperature thresholds, Tpt. The uptakes of
nutrients by phytoplankton and aquatic
macrophytes are shown on pathways kl4 and k1S,
respectively' Phytoplankton eaten by consumers is
shown on pathway kl6 and on pathway kl7
phytoplankton die and sink to the bottom forming
detritus. The oxygen produced by phytoplankton
and macroalgae is shown on pathways kls and k19,
respectively. The oxygen used by the consumers
and bacteria is given on pathways k20 and k21. The
oxygen exchanged with the atmosphere is shown
on pathway k22. Benthic macrophytes eaten by
consumers are shown on pathway k23 and
pathway k24 shows the macrophytes that die and
falI to the bottom as detritus. The detritus
consumed by bacteria is represented by pathway
k2S and the detritus eaten by consumers is shown
as pathway k26. The rates of nitrogen fixation, k27,
and denitrification, k2S, are mediated by bacterial
processes in the sediment. The nitrogen from the
sediment nitrogen pool that is processed by
bacteria is shown as pathway k29. Nutrients are
recycled into the water by bacteria, k30, and
consumers, k31. In the model, streamflow
determines the balance between settling and
resuspension of turbidity in the water along
pathway k32.
3.2.5 Caveat on the Detailed Models 0/
Pollutants
The descriptions of the network of
interactions in aquatic stream ecosystems
controlIing suspended and bedded sediments,
toxic chemicals and nutrients are not complete
listings of alI factors that may prove to be
important in determining the effects of these
stressors on aquatic ecosystems. The detailed
models are simply guides to thinking about the
particular problem whose details may include
some or alI of the factors diagramed here or other
factors that may prove important in understanding
the particular problem. Models are best used as
dynamic tools to be continualIy revised as the
circumstances of analysis change or as more
knowledge is gained. These conceptual models
can be further developed and evaluated to create
simulation models suitable for the alIocation of
biological effects among multiple stressors. The
detailed models in their present form may be
useful as tools to stimulate thinking about the
processes controlling the biological effects of
suspended and bedded sediments, toxic
chemicals, and on aquatic ecosystems from fresh
water streams to the coastal shelf.
64

-------
Conceptual Models and Methods
Table 8. Definition of the forcing functions, components, and pathways in the canonical energy
systems model of excess nutrient in an estuary.

Symbol.

Forcing Functions

Tp
AR

NIA
R
NIR

NOR
T
NOB
NIB
Components
No
N1
Pathway flows
JIA
J1R
JOR
Ira
In
Jp
JRM
JD
h
Definition
Temperature as a time series.
Rainfall from the atmosphere (data as a time series)
Inorganic nitrogen in rainfall (can be a time series)
River inflow (time series of water flow)
Inorganic nitrogen concentration in river inflow (can be a time series)
Organic nitrogen in river inflow(can be a time series)
Tidal exchange of waters
Organic nitrogen in the coastal water(can be a time series)
Inorganic nitrogen in the coastal water(can be a time series)
Organic nitrogen in the estuary (state variable)
Inorganic nitrogen in the estuary (state variable)
Inorganic nitrogen flux in wet and dry deposition
Inorganic nitrogen flux in river inflow
Organic nitrogen flux in river inflow
Organic nitrogen flux in tidal exchange
Inorganic nitrogen flux in tidal exchange
Organic nitrogen production in the estuary
Organic nitrogen remineralized in the estuary
Denitrification
Nitrogen fixation
65

-------
Conceptual Models and Methods
4.0 Development of Quantitative Models
The detailed conceptual models for the four
stressor classes will be used as a starting point for
developing simple quantitative models for each
stressor (Figure 19). These models are highly
aggregated versions of the detailed models
presented above that nevertheless are able to
capture the key features of the stressor's action
within an aquatic ecosystem. An advantage to
these simpler models is that they have fewer
parameters and are therefore easier to evaluate.
Figure 19 shows the diagram and equations for a
quantitative canonical energy systems model of
the action of nutrients within an estuary with two
equations and one condition giving the
mathematical formulation of the model. In this
model the ecosystem is represented simply as an
order-disorder loop (Odum 1994) that is loaded
with nutrients in both ordered (organic) and
disordered (inorganic) form from the land,
atmosphere, and sea. N( is the observed quantity
of inorganic nutrient in the water and No is the
nutrient in complex organic form while N-F N, +
No is the total nutrient in the system. Production
is represented by Jp and respiration by JRM in the
model. Temperature, Tp, is a factor that modifies
the rates of both catabolic and anabolic
processing. Loading of inorganic nutrient is given
by J(R + J'A + h(N1B -N() and the net rate of
processing (JRM -Jp) inorganic nutrient by
knnNoek(TpoTpl) - kpN,ek(TP-TPO). Denitrification and
nitrogen fixation, respectively augment catabolic
and anabolic processing and are included as rate
processes driven by temperature. Calibration
temperatures for the various processes are given
as TO, T1, T2, and 1'3 in the model equations
shown in Figure19. This model allows us to
determine an important but difficult to measure
variable, the processing capacity of aquatic
ecosystems for a particular material (pollutant), by
using commonly measured values. We
hypothesize that processing capacity will be an
important variable in explaining stressor-response
relationships. Canonical models such as the one
presented in this section can be used to develop
and test indicators of whole system function. For
example, a persistent imbalance in the nutrient
used in anabolic versus catabolic processes over
several annual cycles may be an indicator of
nutrient stress in the ecosystem. An accumulation
of NT in the system or an increase in the total rate
of nutrient processing might be early signs of
stress.
The development of detailed conceptual models
for the individual stressors is a first step toward
constructing simulation models to be used in the
analysis and prediction of the combined effects of
multiple stressors in aquatic ecosystems. The next
step in this process is to construct detailed
mathematical formulations of stressor action.
These models are operationally defined
(Bridgman 1928) so that every element of the
model (sources, state variables, and flows) is
evaluated with a measured quantity. Model
evaluation entails specifying values for all
components, processes and forcing functions at
an initial time, to, when the simulation begins. In
addition, the forcing functions must be specified
for the time period of the simulation. Data on
output functions is needed at points in time over
the period of simulation for comparison with
model predictions. Model coefficients are
determined from data on the initial conditions.
Key processes and effects in the model are
calibrated using empirical measures'or theoretical
relations that can predict them. Model predictions
of the values for state variables are verified by
checking them against the data available for the
output variables. A verified model is able to
predict the observed values for all output
variables with reasonable accuracy. Verified
models can be used in sensitivity analyses to
determine the degree to which output variables of
interest are influenced by changes in the forcing
functions. The sensitivity of model outputs to the
values chosen for model parameters can also be
tested. When model output is verified with an
independent data set using the coefficients from
the original model and the input functions from
the new test case, the model is said to be
validated. Our confidence in the original model
increases with each independent data set
66

-------
Conceptual Models and Methods
successfully validated. The end result of model
development will be the ability to forecast
ecological conditions in aquatic ecosystems given
known loadings of a stressor into the system.
Equations:
dNJdt = JOR + kpN1ek(TP-TO) - knnNoek(Tp.TO)+ JT(NoB - (NJz))
dNldt = JIR + J1A - ~N1ek(Tp.TO)- kdN~k(Tp.T2) + knnNoek(TP'TI)+ kfNoek(TP-TJ)
JINIB - (N1/z))
Figure 19.Canonical energy systems model of excess nutrients in an estuary along with the equations that
describe the behavior of this system. Forcing functions, components, and pathways in the model are
defined in Table 8. Tj is the initial temperature where the rate function was evaluated and z is the depth.
67

-------
Conceptual Models and Methods
5.0 Applications
The conceptual models presented in this
document need to be mathematically formulated
and evaluated with data from existing studies and
field work before their veracity can be demonstra-
ted and their full potential realized. Nonetheless,
the present state of our scientific knowledge makes
us confident that we are on the right track toward
developing a set of useful tools to definitively
diagnose the causes of impairment in aquatic
ecosystems and to allocate effects among multiple
causes when more than one has been demonstrated
to be active. This document may be of immediate
use to scientists in the process of developing initial
problem formulations for risk assessments or other
analyses in support of the development of a
TMDL or other regulatory program for one of the
four classes of aquatic stressors covered by our
research. Also we have proposed a plan for
applying these conceptual models in the
development of a stressor-based classification
system for use in the regulatory programs needed
to address violations of the water quality
standards.
5.1 Use ofthe Conceptual Models in
Classification
We will apply our conceptual models and
expand the database used in a preliminary
classification of estuaries (USEPA 2004) to
develop and test stressor-based classification
systems (this database tool is under development).
We illustrate a possible approach to model-based.
classification in Figure 20. The aquatic systems to
be classified include water bodies and their
watersheds. Any system defined in this way can be
classified, e.g., a stream reach and its watershed,
an estuary and its drainage area, a lake and its
watershed. The classes will be based primarily on
water body and associated watershed character-
istics and they will be developed specifically for
particular stressors. Basic information for the
pollutant (stressor) will be determined along with
the loading rate from the adjacent watershed,
watersheds upstream, the atmosphere, and the
ocean for estuaries. In addition, we will determine
the stored quantity of the pollutant presently
residing in the system.
The classification process is first applied fqr
a unit load of pollutant and the expected
biologically effective concentrations of the
material are predicted for different classes of
aquatic systems. Our initial approach to
developing a model based classification scheme
is as follows: The first step in our stressor-based
method for classification of aquatic systems is to
place the system to be classified into one of the
four canonical models controlling residence time
(Figure 9). Once this is accomplished, we will
divide the systems into ranges of average
temperature and two classes (continuous or
discontinuous) based on the effect of seasonality
on the way materials are processed. Thus, we
will distinguish between boreal, temperate, and
tropical systems at this step. If temperature
determines the overall rate of metabolic
processing of the pollutant within the system, we
can use the information on residence time and
the temporal pattern of pollutant processing to
determine the average residence time ofthe
material in the system and its relevant range of
temporal variation. If residence time of the
pollutant varies markedly over the area of the
system under study, we will divide the system
into subsystems and analyze each subsystem
separately. We will separate systems into
residence time classes (Abdelrhman 2005) based
on. the chronic dose-response characteristics of
the particular stressor. Knowing the temporal
pattern of processing and the variation of
turnover with time will allow us to partition a
system into residence time classes, if necessary.
Once we have divided systems into classes based
on residence time of the pollutant, we may split
the classes using secondary factors that control
processing capacity ,e.g., the ratio of wetland
area to water body area or volume,
concentrations of DOC and A VS and other
stressor specific measures.
Wetlands have been characterized as nature's
kidneys, because they filter impurities from the.
waters flowing through them. We hypothesize
that the presence or absence of wetlands will be a
factor of importance in processing pollutants.
Also, we will consider other processing factors at
this stage based on the particular pollutant being
68

-------
Conceptual Models and Methods
A Classification Scheme Starting with the Canonical Models
Freshwater well-mixed
Freshwater stratified
~--
~

..~~I
Estuary, well-mixed
Estuary, stratified
I~c-~I
. - ;;;~ ~~_.'
. ~ 6' ......
~
..~::L~f::'-'

...- ,I,
Temperature
Residence Time
Processing
I -
Exposure (Bioeffective cone. X Residence time)

+ I Modifying Factors
. EEz Effective Exposure
EEx
EEy
Figure 20. A preliminary classification tree that groups estuaries by effective exposure regimes based on
our conceptual model of the factors that control biological impact. This classification based
on effective exposure could be applied to any of the stressors with modifications for the
particular properties of a given stressor.
evaluated. The result of our analysis of processing
capacity will be to estimate the bioeffective
concentration of pollutant expected in the class.
This value multiplied by the residence time gives
the exposure. We will determine the expected
exposure-effect relationship for the pollutant from
past studies in the literature or from the new data
base tool presently under development and predict
the effects on biological output variables from the
exposures determined for each class.
Next, we will consider the effects of type B
modifying factors on the biological impacts
expected in particular classes. We will group these
modifying factors according to their effects on the
pollutant. Those with a positive effect (decrease
response) and those with negative effects (increase
response) will be combined to estimate the net
effect on the biological response. Once we have
determined a positive or negative effect, we will
adjust the exposures calculated above to determine
effective exposures in the system containing
modifying factors. If no type B modifying factors
are present, the exposure value determined above is
the effective exposure and it passes directly to the
bottom line in Figure 20. We hypothesize that
effective exposure will characterize sets of
aquatic systems where similar biological effects
will be observed for a unit load of pollutant.
Classes will be derived by examining the
effective exposure groups to see what'
combinations of properties are represented by
systems in the group.
Next we will apply the actual loads
entering the aquatic systems and determine the
effective exposures. We will plot the observed
values for the biological output variables
against the exposures to construct an exposure-
effect curve for the pollutant. We will compare
this relationship to the one expected from past
laboratory and field studies. We expect eco-
system classes to plot as a family of curves on
the exposure-effect plane or as a single curve
on the effective exposure-effect plane, Le., after
type B modifying factors have been considered
and the exposure-effect relationship has been
normalized. We may also need to consider
factors that shift a particular response variable
on the ordinate to adjust the y-axis values and
further tighten the effective exposure-response
relationship. Figure 21 is a hypothetical
69

-------
Conceptual Models and Methods
-
()
@
~
-a
()
'@)
o
]
I:Q
------------............

,
,
,
,
"
"
"
BEl
BE2
BE3
EEl
"
"
"
"
"
,
"" System A
"
,
,
"
" .
"
"
"
"
--...._----------
System B
EI
E2
Exposure
Figure 21. Nonnalization of the exposure-effect relationship by adjusting both the x and y axis. (I) A type B
modifYing factor shifts the response variable on the exposure axis. (2) At a standard exposure, different
levels of the biological effect variables characterize aquatic ecosystems of different kinds.
example showing how normalization procedures
might work. Modifying factors that mitigate the
biological response shift the exposure-effect
relationship toward lower effective exposure
(EE I) for a given level of effect. If the factors
controlling the biological response variable can be
determined, all effects can be expressed relative to
the same baseline exposure-effect relationship (2
in Figure 21). If these relationships can be
demonstrated, managers might allow greater
loading in a class of aquatic systems that is less
sensitive to the pollutant and still attain a given
level of condition that is deemed to be acceptable
by society
5.2 Use of the Conceptual Models in Risk
Assessment Problem Formulation
The general conceptualization of the factors
controlling the action of stressors in aquatic
systems as an exposure-effect problem follows the
approach taken in ecological risk assessment. Our
model given in terms of residence time,
processing capacity, and modifying factors
provides the means to determine effective
exposure and it can be of direct use to those
performing risk assessments related to the
development of TMDLs. If effective exposure
can be successfully related to observed effects by
accounting for the factors that alter this
relationship, we will have a robust method for
predicting the effects of pollutant loading in
aquatic systems.
When the evaluated and verified energy
systems models for the four classes of stressors
are simulated, the predicted temporal variations
and spatial patterns of risk and effect can be
examined. Energy systems models also help
organize thinking about a class of stressors and
~hey can serve as a checklist of the important
components and processes governing stressor
behavior. The simulation of emergy flows in a
detailed model of a stressor acting on an
ecosystem allows determination of the ecological
significance of a given change in the system,
which combined with the probability of that
change (risk) gives an accurate measure of
ecological importance, which is a unified and
comprehensive measure of the environmental
impacts of a stressor that can be used in decision-
making (Campbell 200 I b).
70

-------
Conceptual Models and Methods
The detailed models presented here strive to
capture the important processes and interactions
of the system components that control the action
of multiple reinforcing and agonistic stressors in
aquatic ecosystems of different kinds. To the
extent that these models are successful in
predicting ecosystem behavior under stress they
. should be useful in identifying and developing
specific indicators that could facilitate the
diagnosis of the causes of biological impairment
in cases where both single and multiple stressor
interactions are in play. .
71

-------
Conceptual Models and Methods
6.0 Glossary of Words and Concepts
aquatic ecosystems are material and energy
processing units organized by thermodynamic
. laws and principles that are contained in water
bodies of various kinds, which are arrayed over
the surface of the earth.
canonical an adjective used to describe the
standard form of an energy systems model (an
equation or system of equations), especially when
the model is simple.
emergy is the available energy of one kind
previously used up, directly and indirectly, to
make a product or service. It's unit is the emjoule.
energy and emergy signature Energy and material
distributions are highly variable in space and time,
and thus there are many combinations available to
support ecological organization. The energy and
material inflows available at a location, when
plotted in order of increasing transformity, make
up the energy signature of that place. If each
energy in the energy signature, is multiplied by its
transformity the resulting plot is the emergy
signature of the place. There is some indication
that qualitative and quantitative differences in the
emergy signature correspond to ecosystems of
different kinds that have different ecological
norms, i.e., expected values for empower, the
flow of emergy per unit time.
fundqmental watershed is a network of
ecosystems arrayed on the landscape and linked
by water flows that debouch into the open sea or
into one of the Great Lakes. These watersheds are
commonly the largest system within which
wetlands, stream segments, lakes and estuaries
must be managed to ensure that limits established
for pollutants will be effective: An exception to
this rule occurs when atmospheric or oceanic
inflows of a pollutant are large.
maximum power (empower) principle states that
ecosystems adapt and evolve to use available
energy and materials in a manner that maximizes
empower in the ecological network. Evolutionary
competition puts survival pressure on ecosystems
to adapt to use their inputs in a manner that
maximizes empower in their ecological network.
By definition a natural system is one that has had
the time to adjust its structure and function to
attain optimum efficiency for maximum power. P
and R are often but not always balanced in
natural systems that are adapted to their inflows.
The theory predicts that system indices will
approach a dynamic equilibrium that maximizes
power under a given emergy signature.
Paracelsus Axiom: the dose makes the poison.
The three pollutant stressors of concern to us, i.e.,
nutrients, toxic substances, and suspended and'
bedded sediments are hypothesized to be
materials that stimulate ecosystem production
when present in the quantities to which life has
adapted over evolutionary time scales. The
frequency of occurrence (relative rarity) of any
material in nature is different and the opportunity
for organisms to develop the capacity to process
any material is proportional to the probability of
encountering a given concentration of that
substance in nature. Thus, in general, organisms
have a greater ability to process more common
substances and less for rarer ones (Genoni 1997).
Of course there is considerable variation in the
concentrations of materials found in nature and
there is concomitant variation seen in the life.
forms that have developed different abilities to
process different concentrations of these
substances.
pollutant A pollutant is any substance that is
present within a system outside of its normally
expected range, usually a gaseous, chemical, or
organic waste.
state variable Any stored quantity within a
system which may be plotted over time. In an
energy systems model, there is one 15t order
differential equation for each state variable.
stress can be defined as a change in or
perturbation of the normal (long term or natural)
functioning of a system. Often stress causes
energy drains to development within the system.
Stress can also cause certain energy flows to
exceed there normal ranges. In general, stress is
72

-------
Conceptual Models and Methods
the result of a change in the emergy signature
of the place. The usual implication is that this
perturbation or change is the result of some
human activity. A change in the emergy signature
of a processing unit sets in motion the normal
adaptive mechanisms of an ecosystem that operate
to create a new network design that will maximize
power under the changed conditions, if the new
conditions become permanent, or return the
system to its normal state of operation, if the
perturbation is transitory. Ecosystems can adapt
to recurring natural perturbations, so that the
recurring pulse becomes necessary to the health
and preservation of the ecosystem, e.g., fire
climax in southeastern U.S. pine forests (Odum
1971). In this case the pulse or perturbation is a
stress to which the system has adapted. In cases
where ecosystems have adapted to recurring or
chronic stress they may be more resistant to the
effects of a similar but unnatural change in the
emergy signature.
stressor In the context of ecological systems the
word "stressor" may be defined as any force that
results in an injury to a system (Le., a decrease in
total system empower) often by over use or
exertion of some part of the system. Stressful
forces can be exerted by physical, chemical, or
biological entities and all three are encompassed
by the term "stressor".
transformity solar transformity is the solar
emergy required to make a joule of a service or
product. It is the solar emergy required for a
product divided by its heat content in joules. The
units of solar transformity are solar emjoules per
joule (sej/J).
unit model is an energy systems model or other
mathematical formulation applied in every grid
cell of a spatial analysis or simulation (some
components may be 0 in any particular cell).
73

-------
Conceptual Models and Methods
7.0 References
Abdelrhman MA. 2005. Simplified modeling of
flushing and residence times in 42 embayments in
New England, USA, with special attention to
Greenwich Bay, Rhode Island. Estuarine, Coastal
and Shelf Science 62:339-351.
Ashby J, Tennant RW. 1988. Chemical structure,
Salmonella mutagenicity and extent of carcino-
genicity as indicators of genotoxic carcinogenesis
among 222 chemicals tested in rodents by the U.
S. NCIINTP. Mutat. Res 204: 17-115.
Bailey, Robert G. 1998. Ecoregions, The
Ecosystem Geography of Oceans and Continents.
Springer, New York, NY, 176 pp.
Bridgman PW. 1928. The Logic of Modern
Physics. MacMillan Co., New York. 227 pp.
Burgess RM. 2000. Characterizing and
identifying toxicants in marine waters: a review of
marine Toxicity Identification Evaluations (TIEs).
International Journal of Environment and
Pollution. 13 :2-33.
Burgess RM, Ahrens MJ, Hickey CWo 2003.
Geochemistry ofPAHs in aquatic environments: a
synthesis of source, distribution and persistence.
In Douben PET. (ed.) PAHs: An Ecotoxicological
Perspective, John Wiley & Sons, Ltd., London
Brown MT, Ulgiati S. 1999. Emergy evaluation of
the biosphere and natural capital. Ambio 28:
486-493.
Campbell DE. 2000. Using energy systems theory
to define, measure, and interpret ecological
integrity and ecosystem health. Ecosystem Health
6(3): 181-204.
Campbell DE. 2001 a. An energy systems analysis
of constraints on economic development. In
Ulgiati S, Brown MT, Giampetro M, Herendeen
RA, Mayumi K.(eds.) 2nd International Workshop,
Advances in Energy Studies, Exploring Supplies,
Constraints, and Strategies. Servizi Grafici
Editoriali, Padova, Italy, pp. 175-187.
Campbell DE. 2001 b. Proposal for including
what is valuable to ecosystems in environmental
assessments. Environ. Sci. Technol.
35(14):2867-2873.
Campbell D.E. 2005. Evaluation and Emergy
Analysis of the Cobscook Bay Ecosystem..
Northeastern Naturalist 11 (Special Issue 2):
355-424.
Campbell DE, Wroblewski JS. 1986. Fundy tidal
power development and potential fish production
in the Gulf of Maine. Can. J Fish.. Aquat. Sci.
43( 1 ):78-89.
Carson R. (1962). Silent Spring. Houghton
Mifflin. Boston, MA.
Coakley J, Hodson PV, van Heiningen A, Cross
T. 2001. MFO induction in fish by filtrates from
chlorine dioxide bleaching of wood pulp. Water
Research 35:921-928.
Colborn T, Dumanoski D, Meyers JP. 1996. Our
Stolen Future: Are We Threatening Our Fertility,
Intelligence, and Survival?-A Scientific Detective
Story. Dutton/Penguin, New York, NX, 336 pp.
Cowardin, LM, Carter V, Golet FC, LaRoe ET.
1979. Classification of wetlands and deepwater
habitats of the United States. U.S. Fish &
Wildlife Service Pub. FWS/OBS-79/31,
Washington, D.C.
Daskalakis KD, O'Connor TP. 1995. Distribution
of chemical concentrations in US coastal and
estuarine sediment. Mar Environ Res. 40:
381-398.
Detenbeck NE, DeVore PW, Niemi GJ, Lima A.
1992. Recovery of temperate-stream fish
communities from disturbance: A review of case
studies and synthesis of theory. Environ.
Manage. 16(1):33-53.
74

-------
Conceptual Models and Methods
Doull J, Klaassen CD, Amdur MO. (1980).
Casarett and Doull's Toxicology: The Basic
Science of Poisons. Macmillan Publishing Co.
Inc. New York, New York.
Dow RL. 1977. Effects of climatic cycles on the
relative abundance and availability of commercial
marine and estuarine species. J. Cons. Int. Explor.
Mer. 37(3):274-280.
Gardner RH, Turner MG, Dale VH, O'Neill RV.
1992. A percolation model of ecological flows. In
Hansen AJ, diCastri F. (eds.) Landscape
boundaries: Consequences for biotic diversity
and ecologicalflows. Springer Verlag,
New York, NY.
Genoni GP. 1997. Toward a conceptual synthesis
in ecotoxicology. Gikos 80:96-106
Gentile JH, Harwell MA, Cropper W, Harwell
CC, DeAngelis D, Davis S, Ogden JC, Lirman D.
2001. Ecological conceptual models: a
framework.
General Accounting Office (GAO) 2002. Water
Quality, Inconsistent Approaches Complicate
Nation's Efforts to Identify Its Most Polluted
Waters. GAO-02-186. Report to Congressional
Requesters, GAO, Washington D.C. 40 pp.
Gibbs JP. 1993. Importance of small wetlands for
the persistence of local populations of wetland-
associated animals. Wetlands 13(1 ):25-31.
Grant WE, Pedersen EK, Marin SL. 1997.
Ecology and Natural Resource Management:
Systems Analysis and Simulation. John Wiley &
Sons, Inc., New York.
Hill, AB. 1965. The environment and disease:
Association or causation. Proceed. Royal Soc.
Medicine 58:595-300.
Ho KT, McKinney RA, Kuhn A, Pelletier MC,
Burgess RM. 1997. Identification of acute
toxicants in New Bedford Harbor sediments.
Environ. Toxico/. Chem. 16:551-558.
Ho KT, Burgess RM, Pelletier MC, Serbst JR,
Ryba SA, Cantwell MG, Kuhn A, Raczelowski P.
2002. An overview of toxicant identification in
sediments and dredged materials. Marine
Pollution Bulletin 44:286-293.
Holling CS. 1986. The Resilience of Terrestrial
Ecosystems: Local Surprise and Global Change.
In Clark WM, Munn RE. (eds.) Sustainable
Development in the Biosphere, Cambridge Univ.
Press, Cambridge, pp. 292-320.. .
Holmgren, D. 2002. Permaculture, Principles and
Pathways beyond Sustainability. Holmgren
Design Services. Hepburn, Victoria, Australia.
pp. 286.
Kagi JHR, Schaffer A. 1988. Biochemistry of
Metallothionein. Biochemistry 27: 8509-8515.
Kaufmann PR, Levine P, Robison EG, Seeliger
C, Peck and DV. 1999. Quantifying Physical
Habitat in Wadeable Streams, EPA/620/R-
99/003, U. S. Environmental Protection Agency,
Washington; D.C. 102 pp + Appendices.
Kuhn A, Munns WR, Poucher S, Champlin D,
Lussier S. 2000. Prediction of population-level
response from mysid toxicity test data using
population modeling techniques. Environ Toxicol
Chern 19:2364 - 2371.
Kuhn A, Munns WR, Serbst J, Edwards P,
Cantwell MG, Gleason T, Pelletier MC, Berry W.
2002. Evaluating the ecological significance of
laboratory response data to predict population-
level effects for the estruarine amphipod
Ampelisca abdita. Environ Toxicol Chem
21 :865-874.
LaFlamme RE and Hites RA. 1978. The global
distribution of polycyclic aromatic hydrocarbons
in recent sediments. Science 42:289-303.
Lijinsky W. (1989). A view of the relationship
between carcinogenesis and mutagenesis.
Environ. Molec. Mutagen 14:78-84.
Long ER. 1992. Ranges in chemical concentra-
tions in sediments associated with adverse
biological effects. Mar Pollut Bull 24:38-45.
Long ER, Robertson A, Wolfe DA, Hameedi J,
75

-------
Conceptual Models and Methods
Sloane GM. 1996. Estimates of the spatial extent
of sediment toxicity in major U.S. estuaries.
Environ Sci TechnoI30:3585-3592.
MacArthur RH, Wilson EO. 1963. An equilibrium
theory of insular zoogeography. Evolution
17:373-387.
Manahan SE. 1989. Toxicological Chemistry.
Lewis Publishers, Inc., Chelsea, MI, 317 pp.
Manley PN, Zielisnsky WJ, Stuart CM, Keane JJ,
Lind AJ, Brown e. Plymale BL, Napper CO.
2000. Monitoring ecosystems in the Sierra
Nevada: the conceptual model foundation.
Environmental Monitoring and Assessment 64:
139-152.
Muir D, Riget F, Cleemann M, Skaare J, Kleivane
L, Nakata H, Dietz R, Severinsen T, Tanabe S.
2000. Circumpolar trends of PCBs and
organochlorine pesticides in the Arctic marine
environment inferred from levels in Ringed Seals.
Environ Sci Technol 34:2431-2438.
Nacci D, Coiro L, Champlin D, Jayaraman S,
McKinney R, Gleason TR, Munns WR, Specker
JL, Copper KR. 1999. Adaptations of wild
populations of the estuarine fish Fundulus
heteroclitus to persistent environmental
contaminants. Mar Bioi 134:9-17.
Nacci DE, TR Gleason, Gutjahr-Gobell R, Huber
M, Munns WR. 2002. Effects of chronic stress on
wildlife populations: a population modeling
approach and case study. In Newman MC,
Roberts MH, Hale RC (eds.) Coastal and
Estuarine Risk Assessment, Lewis Publishers,
New York, NY, pp. 247-272.
Niemi GJ, DeVore P, Detenbeck N,Taylor D,
Lima A, Pastor J, Yount JD, Naiman RJ. 1990.
Overview of case studies on recovery of aquatic
systems from disturbance. Environ. Manage.
14(5):571-587.
Odum, E.P. 1971. Fundamentals of Ecology.
Saunders, Philadelphia, pp. 574.
Odum HT. 1968. Work circuits and system stress.
In Young H. (ed.) Mineral Cycling and
Productivity of Forests. University of Maine,
Orono, pp 81-146.
Odum HT. 1971. An energy circuit language for
ecological and social systems, its physical basis.
In Patten Be. (ed) Systems Analysis and
Simulation in Ecology, Vol. 2. Academic Press,
New York, pp. 139-211. .
Odum HT. 1976. Macroscopic Minimodels of
Man and Nature. In Patten B. (ed) Systems
Analysis and Simulation in Ecology, Vol. 4,
Academic Press, New York, pp. 249-280.
Odum, HT. 1983. Systems Ecology. John
Wiley&Sons, New York. 644 pp.
Odum HT. 1994. Ecological and General
Systems: An Introduction to Systems Ecology.
(Revised Edition of Systems Ecology) University
Press of Colorado, Niwot, 644 pp.
Odum HT. 1996. Environmental Accounting:
Emergy and Environmental Decision Making.
John Wiley&Sons, New York, 370 pp.
Odum HT, Odum EC. 2000. Modelingfor all
Scales. Academic Press, San Deigo, CA., 458 p.
Odum HT, Copeland BJ, MacMahan EA. (eds.)
1974. Coastal Ecological Systems of the United
States. Voll. The Conservation Foundation,
Wash. DC. 533 pp.
Odum HT, Kemp W, Sell M, Boynton, W,
Lehman M. 1977. Energy analysis and the
coupling of man and estuaries. Environmental
Management 1(4): 297-315..
Odum HT, WojcikW, Pritchard L, Ton S, Delfino
JJ, Wojcik M, Leszczynski S, Patel JD, Doherty
SJ, Stasik J. 2000. Heavy Metals in the
Environment, Using Wetlands for Their Removal.
CRC Press LLC, Lewis Publishers, Boca Raton,
FL. 325 pp.
O'Neill P. 1985. Environmental Chemistry.
George Allen & Unwin, Boston, MA, 232 pp.
76

-------
Conceptual Models and Methods
O'Neill RV, DeAngelis DL, Waide JB, Allen
THF. 1986. A Hierarchical Concept of
Ecosystems. Princeton University Press,
Princeton, New Jersey.
PoffNL, Ward JV. 1989. Implications of
streamflow variability and predictability for lotic
community structure: A regional analysis of
stream flow patterns. Can. J Fish. Aquat. Sci.
46:1805-1818.
Rand GM, Petrocelli SR. 1985. Fundamentals of
Aquatic Toxicology. Hemisphere Publishing
Corporation, New York, NY, 666 pp.
Redfield AC. 1958. The biological control of
chemical factors in the environment. American
Scientist 46:205-221.
Renne~ R. 2002. Newly deposited mercury may
be more bioavailable. Environ Sci Technol
36:226A-227 A.
Scienceman OM. 1987. Energy and Emergy. In
Pillet G, Murota T. (eds) Environmental
Economics B The Analysis of a Major Interface.
Roland Leimgruber, Geneva, pp. 259-276. .
Sklar FH, Browder JA.1998. Coastal
environmental impacts brought about by
alterations to freshwater flow in the Gulf of
Mexico. Environmental Management 22:547-562.
Stefan HG, Hondzo M, Eaton JG, McCormick JH.
1995. Validation of a fish habitat model for lakes.
Ecological Modelling 82:211-224.
Stefan HG, Hondzo M, Fang X, Eaton JG,
McCormick JH. 1996. Simulated long-term
temperature and dissolved oxygen characteristics
of lakes in the north-central United States and
associated fish habitat limits. Limnol. Oceanogr.
41: 1124-1135.
Swain WR. 1988. Human health consequences of
consumptiori of fish contaminated with organo-
chlorine compounds. Aquatic Toxicol.
11:357-377. .
Townsend OW, Cammen LM. 1988. Potential
importance of the spring phytoplankton bloom to
recruitment of juvenile demersal fishes.
Biological Oceanography. 5:215-229.
Townsend OW, Cammen LM, Holligan PM,
Campbell DE, Pettigrew NR. 1994. Causes and
consequences of variability in the timing of
spring phytoplankton blooms. Deep-Sea
Research 41 :747-765.
Turner MG, Romme WH, Gardner RH, O'Neill
RV, Kratz TK. 1993. A revised concept of
landscape equilibrium - disturbance and stability
on scaled landscapes. Landscape Ecology 8:
213-227.
United States Environmental Protection Agency
(USEPA) 1998. Condition of the Mid-Atlantic
Estuaries. EPA 600-R-98-147. U.S. Environmen-
tal Protection Agency, Office of Research and
Development, Washington, DC. 50 pp.
United States Environmental Protection Agency
(USEPA) 2000a. Stressor Identification
Guidance Document. EPA/822/B-00/025. U.S.
Environmental Protection Agency, Office of
Water, Office of Research and Development,
Washington, DC.
United States Environmental Protection Agency
(USEPA) 2000b. Mid-Atlantic Highlands
Streams Assessment. EPA /903/R-00/015. U.S.
Environmental Protection Agency, Region 3,
Philadelphia, P A. 64 pp.
United States Environmental Protection Agency
(USEPA) 2002a. The Twenty Needs Report: How
Research Can Improve the TMDL Program. EP A
841/B-02/002. U.S. Environmental Protection
Agency, Office of Water, Washington, DC. 43pp.
United States Environmental Protection Agency
(USEPA) 2002b. Aquatic Stressors, Framework
and Implementation Plan fqr Effects Research.
EPA 600/R-02/074. U.S. Environmental
Protection Agency, Office of Research and
Development, National Health and Environ-
mental Effects Research Laboratory, Washington,
DC. 197 pp.
United States Environmental Protection Agency
77

-------
Conceptual Models and Methods
(USEPA) 2002c. Consolidated Assessment and
Listing Methodology. Environmental Protection
Agency, Office of Wetlands, Oceans, and
Watersheds, Washington, DC. Downloaded from
http://www .eoa.gov/owow/monitoring/calm. html.
August 26, 2005.
United States Environmental Protection Agency
(USEPA) 2004. Classification Frameworkfor
Coastal Systems. EPA 600/R-04/061. U.S.
Environmental Protection Agency, Office of .
Research and Development, National Health and
Environmental Effects Research Laboratory,
Atlantic, Gulf, and Mid-Continent Ecology
Divisions. Washington, DC.
Valette-Silver NJ. 1993. The use of sediment
cores to reconstruct historical trends in
contamination of estuarine and coastal sediments.
Estuaries 16(3B):577-588
van der Valk AG. 1981. Succession in wetlands: a
Gleasonian approach. Ecology 62:688-696.
van Wezel AP, Opperhuizen A. 1995. Narcosis
due to environmental polIutants in aquatic
organisms: residue based toxicity, mechanisms,
and membrane burdens. Crit Rev Toxicol.
25:255-279.
West CW, Kosian PA, Mount DR, Makynen EA,
Pasha MS, Sibley PK. Ankley GT. 2001.
Amendment of sediments with a carbonaceous
resin reduces bioavailability of polycyclic
aromatic hydrocarbons. Environ. Toxicol. and
Chem. 20, 1104-1111.
Yount JL. 1956. Factors that control species
numbers in silver Springs, Florida. Limnology
and Oceanography 1 :286-295.
Ziemer RR. 2004. Scale considerations for linking
hillslopes to aquatic habitats. In Gucinski H,
Miner C, Bittner R.(eds.) Proceedings: Views
from the ridge: Considerations for planning at the
landscape scale. General Technical Report PNW-
GTR-596. Portland, OR: U.S. Dept. of Agricul-
.ture, Forest Service, Pacific Northwest Research
Station, pp. 22-32.
78

-------
Conceptual Models and Methods
ft EA~ United States
...~ Environmental Protection
,,.,. Agency
Office of Research and Development
National Health and Environmental
Effects Research Laboratory
Narragansett, RI 02882
Official Business
Penalty for Private Use
$300
EP A/600/R-06/024
May 2006
www.epa.gov

-------