Grantee Research Project Results
Final Report: Modeling and Multiobjective Risk Decision Tools for Assessment and Management of Great Lakes Ecosystems
EPA Grant Number: R825150Title: Modeling and Multiobjective Risk Decision Tools for Assessment and Management of Great Lakes Ecosystems
Investigators: Hobbs, Benjamin F. , Locci, Ana B. , Koonce, Joseph F.
Institution: The Johns Hopkins University , Case Western Reserve University
EPA Project Officer: Packard, Benjamin H
Project Period: October 1, 1996 through September 30, 1999
Project Amount: $620,259
RFA: Ecological Assessment (1996) RFA Text | Recipients Lists
Research Category: Ecological Indicators/Assessment/Restoration , Aquatic Ecosystems
Objective:
The 1994 State of the (Great) Lakes Ecosystem Conference recommended that an ecosystem approach be adopted for studying Great Lakes ecosystem problems and the stresses that cause them; that well defined ecosystem objectives be defined to measure success in restoring ecosystem integrity; and that roundtable and interdisciplinary approaches to decision making be taken that aim for consensus among stakeholders. The purpose of this project was to develop modeling and decision methods that could be used to respond to those recommendations.
In particular, the goal of the project was to develop and test an integrated ecological assessment and decision methodology for the Lake Erie ecosystem. The methodology was designed to assist managers and stakeholders involved in the Lake Erie Lakewide Management Plan (LaMP) and other Lake Erie management processes to define objectives and evaluate tradeoffs and risks associated with future uses. The research plan built on existing initiatives and addressed the following fishery and water quality management concerns:
? Interaction of invasions of exotic species, nutrient reductions, and fishery harvests;
? Influence of nearshore and tributary habitat on offshore community structure and productivity; and
? Sensitivity of emerging ecosystem objectives to climate change and stakeholder priorities.
The products of the research were to include:
? An expanded Lake Erie Ecosystem Model (LEEM). Modifications were to include habitat components, selection of the optimal level of spatial and species resolution, and development of risk analysis capabilities;
? Applications of LEEM to analysis of the effects of stresses on the Lake Erie ecology;
? Development and application of methodologies for decision making under
multiple objectives and risks; and
? Workshops in which Lake Erie managers
apply and evaluate LEEM and the multiobjective risk decision methods.
Summary/Accomplishments (Outputs/Outcomes):
The findings are grouped into three categories: enhancements of LEEM, analyses of ecological stresses, and multiobjective risk analyses of the management of Lake Erie. The references in square brackets [ ] refer to the publications and presentations listed later in this summary.
LEEM Improvements. The first category of products are enhancements to LEEM [J4,J6]. LEEM describes the population dynamics of 14 key members of the Lake Erie fish community, structured by life stage. Nutrient loadings are translated into lakewide rates of primary production. Lower trophic levels (e.g., phyto- and zooplankton, zoobenthos) are represented implicitly. Changes in fish abundance are driven by natural, predatory, and fishing mortality. Outputs include numbers of fish by species by life stage on an annual time step.
LEEM has been enhanced as follows. First, habitat limitations were included in three steps:
1. General characterization of habitat constraints upon fishery productivity in Lake Erie [P2];
2. Construction of habitat suitability index (HSI)-based models to include density-dependent and density-independent recruitment mortality mechanisms in LEEM [M4]; and
3. Inventorying of nearshore and tributary habitat crucial to fish recruitment. This work was undertaken in collaboration with the Great Lakes Fishery Commission and U.S. and Canadian fishery researchers. This last step culminated in attempts to set up demarcation criteria for aquatic biodiversity investment areas as part of the SOLEC process [R2,P37].
Inclusion of habitat limits permits analysis of tradeoffs involved in habitat improvement and preservation, which is the focus of follow-on work.
The second type of enhancement involved optimal aggregation of LEEM and
analysis of the effects of errors in model structure [S3]. It was our hypothesis
that there are temporal and spatial structures that act as natural filters of
variability and that control the propagation of error between interacting
phenomena at varying scales [M2]. It was found, for example, that aggregation
into annual year classes results in biases in estimates of parameters for growth
and survival [P6], and that use of a whole lake resolution would not allow
coexistence of all 14 species without parametric representations of spatial
heterogeneity (habitat overlap coefficients).
The third and final
enhancement has been inclusion of uncertainty propagation capabilities.
Uncertainties in LEEM parameters are propagated by Monte Carlo simulation using
Latin Hypercube sampling [P5,M3]. An importance metric was used to identify the
most sensitive parameters for inclusion in the sampling procedure. A novel
method for characterizing the implications of sample error for Bayesian Monte
Carlo analyses of the value of information was developed and demonstrated
[J5].
Application of LEEM to Questions Concerning Ecological Stresses. The availability of a comprehensive ecosystem model of Lake Erie makes it possible to address questions concerning the relative importance and interactions of various stresses upon the Lake Erie system. The questions addressed in this part of the research project included the following:
1. What are the potential stresses upon the Lake Erie ecosystem as a result of expansion of exotic invader species? [P40] Our conclusion was that recent instabilities of fish community structure in Lake Erie were more likely to be the result of historical fisheries management decisions and less likely to be the result of changes in nutrient loadings or the zebra mussel invasion. Another conclusion was that if the round goby (a predator of the zebra mussel) greatly expands its range further in Lake Erie, there would be no deleterious effects on populations of other fish species.
2. What are the stresses due to fishery exploitation? As just mentioned, the results of this analysis clearly show that historical variation in fishing exploitation is likely to have dominated variability due to zebra mussels and phosphorus loadings [P36,P39].
3. What are the joint effects of multiple stresses? Presently, fishery agencies make quota recommendations on a species-by-species basis with no formal consideration of their interactions. We have shown [M4,M5] how trophic interaction and nutrient loading affect the performance of various multispecies management procedures. Management uncertainty can be reduced by linking quota management through explicit consideration of ecosystem constraints.
4. What are the effects of degradation of nearshore and tributary habitat? Preliminary analyses indicate that habitat link may be a major limitation on fishery productivity [P36]. Lack of explicit inclusion of habitat and nutrient loading constraints in derivations of fishery quotas contributes substantial error to the calculation of allowable catches.
Multiobjective Risk Analyses of Lake Erie Management. Ecological restoration and integrity are fundamentally value-laden notions; operationalizing them requires explicit acknowledgment of uncertainty and subjective values [O1]. Four groups of ecosystem management questions were addressed in this research using multicriteria risk analysis methods that account for uncertainties and priorities:
1. Derivation of general ecosystem objectives for a large-scale Lake restoration planning effort (the LaMP);
2. Regulation of lake levels and their impacts;
3. Nutrient (phosphorus) regulation; and
4. Fisheries management and lower trophic level research project selection.
The project also supported the writing of general guidelines for analysis of
management alternatives by stakeholders under multiple criteria [B1,M1].
The
Lake Erie LaMP requires a set of ecosystem objectives to guide evaluation of
alternative measures to restore the ecological integrity of the lake. However,
because restoration to presettlement conditions is impossible, value judgments
must be made as to what directions restoration should take and how far it should
go. We provided assistance in designing a public consultation effort for the
LaMP Ecosystem Objectives Subcommittee. The analysis in which we collaborated
created a fuzzy cognitive map model [S2] that describes, in qualitative terms,
how different components of the Lake Erie terrestrial and aquatic ecosystem
relate to each other. The model was used to describe how different broad
management policies might affect the state of the ecosystem. The result was a
set of alternative ecosystem scenarios representing alternative visions for the
future of the Lake. Over the next year, these scenarios will be reviewed by Lake
Erie stakeholders, resulting in a set of recommendations concerning preferred
scenarios [R1].
In the lake levels analyses, the results of decision tree-based evaluations of alternatives for restoration of Metzger Marsh and for managing the levels of Lake Erie under climate change uncertainty were reported in [J2,J5,J7]. Different beliefs about the amount by which climate warming would decrease lake levels was shown to be potentially relevant to the decision to regulate lake levels, but not to the decision as to what restoration action is desirable for Metzger Marsh. In a workshop, the results of which were reported in [J3], Lake Erie managers compared several alternatives for including climate change risk in lake levels management; although sophisticated Bayesian analyses based on subjective probabilities were the theoretically most attractive methods, the managers found scenario analyses using interactive simulation models to be most useful. This also was the conclusion of fishery managers who analyzed the potential value of lower trophic level research to fisheries management [M3]. In the latter case, they preferred direct manipulation of an ecosystem model (LEEM) to a sophisticated Bayesian value of information analysis, in large measure because there was inadequate time during the workshop to use the Bayesian method in an informed manner.
The purpose of the phosphorus analysis was to revisit the issue of nutrient loadings in light of the apparent decreases in productivity of Lake Erie as a result of the zebra mussel invasion. There has been political interest in relaxing phosphorus standards. The application illustrated how decision analysis techniques for eliciting value judgments can be used to quantify the relative desirability of different levels of several ecological and social criteria [J1]. Results show that there are potential benefits to changing the historical policy of reducing phosphorus loads in Lake Erie, but that there also are large tradeoffs; the net benefits depend on the relative importance of the criteria, a subjective and political judgment. The decision analysis technique reveals what value judgments are crucial to this decision. In related research, a method was developed for explicitly characterizing the uncertainty associated with such value judgments, and the worth of additional efforts to refine those judgments [S1].
The last problem analyzed concerned the evaluation of research projects addressing the implications of phosphorus loadings and zebra mussels for lower trophic level energy and nutrient flows in Lake Erie. Large uncertainties concerning Lake Erie's lower trophic level make it difficult to predict the effect of alternative fishery management policies upon the lake ecosystem. Reduced uncertainty could improve management, as measured by 10 ecological and social criteria that U.S. and Canadian fishery managers identified in a workshop we held. These managers identified alternative hypotheses concerning the lower trophic level, defined research projects that could address those hypotheses, provided relative priorities for the decision criteria, and assessed subjective probabilities for the various hypotheses and for the outcomes of the research projects. These data were integrated in a decision analysis that quantified the value of information that would be provided by the research for fisheries management [M3]. It was shown that disregarding uncertainties can harm the expected performance of fisheries management, and that information from some of these projects could be very valuable to managers.
Journal Articles on this Report : 8 Displayed | Download in RIS Format
Other project views: | All 65 publications | 13 publications in selected types | All 8 journal articles |
---|
Type | Citation | ||
---|---|---|---|
|
Anderson RM, Hobbs BF, Koonce JF, Locci AB. Using decision analysis to choose phosphorus targets for Lake Erie. Environmental Management 2001;27(2):235-252. |
R825150 (Final) |
|
|
Anderson RM, Hobbs BF. Using a Bayesian approach to quantify scale compatibility bias. Management Science 2002;48(12):1555-1568. |
R825150 (Final) |
Exit |
|
Bloczynski JA, Bogart WT, Hobbs BF, Koonce JF. Irreversible investment in wetlands preservation: Optimal ecosystem restoration under uncertainty. Environmental Management 2000;26(2):175-193. |
R825150 (Final) |
Exit Exit |
|
Chao PT, Hobbs BF, Venkatesh BN. How climate uncertainty should be included in Great Lakes management: Modeling workshop results. Journal of the American Water Resources Association 1999;35(6):1485-1497. |
R825150 (1998) R825150 (1999) R825150 (Final) |
Exit |
|
Hobbs BF, Ludsin SA, Knight RL, Ryan PA, Biberhofer J, Ciborowski JJH. Fuzzy cognitive mapping as a tool to define management objectives for complex ecosystems. Ecological Applications 2002;12(5):1548-1565. |
R825150 (Final) |
Exit |
|
Kim JB, Hobbs BF, Koonce JF. Multicriteria Bayesian analysis of lower trophic level uncertainties and value of research in Lake Erie. Human and Ecological Risk Assessment 2003;9(4):1023-1057. |
R825150 (Final) |
Exit |
|
Linville CD, Hobbs BF, Venkatesh BN. Estimation of error and bias in Bayesian Monte Carlo decision analysis using the bootstrap. Risk Analysis 2001;21(1):63-74. |
R825150 (1998) R825150 (1999) R825150 (Final) |
|
|
Venkatesh BN, Hobbs BF. Analyzing investments for managing Lake Erie levels under climate change uncertainty. Water Resources Research 1999;35(5):1671-1683. |
R825150 (Final) |
Exit |
Supplemental Keywords:
multicriteria decision making, ecosystem health, fisheries management, Great Lakes Water Quality agreement, lake, risk, risk assessment, ecosystem, restoration, ecology, hydrology, limnology, Lake Erie., RFA, Scientific Discipline, Geographic Area, Ecosystem Protection/Environmental Exposure & Risk, Ecosystem/Assessment/Indicators, Ecosystem Protection, exploratory research environmental biology, Ecological Effects - Environmental Exposure & Risk, Environmental Monitoring, Ecological Risk Assessment, Great Lakes, risk assessment, interactive stressors, limnology, multiobjective risk decision tool, lake erie, assessment models, biodiversity, climate change impact, ecosystem assessment, wildlife, hydrological, modeling, ecological assessment, aquatic ecosystems, decision tool, water quality, fish , stakeholders, climate variabilityRelevant Websites:
http://129.22.156.18/leem.htm
http://129.22.156.152/ABIA/
http://www.ijc.org/boards/letf/letfreports.html
Progress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.