Grantee Research Project Results
Final Report: Social and Ecological Transferability of Integrated Ecological Assessment Models
EPA Grant Number: R825757Title: Social and Ecological Transferability of Integrated Ecological Assessment Models
Investigators: Deegan, Linda A. , Kremer, James , Webler, Thomas
Institution: Marine Biological Laboratory , University of Connecticut , Woods Hole Oceanographic Institution , Social and Environmental Research Institute
Current Institution: Marine Biological Laboratory , Social and Environmental Research Institute , University of Connecticut , Woods Hole Oceanographic Institution
EPA Project Officer: Packard, Benjamin H
Project Period: June 1, 1998 through May 31, 2001
Project Amount: $850,575
RFA: Water and Watersheds Research (1997) RFA Text | Recipients Lists
Research Category: Water , Watersheds
Objective:
The three objectives that drove this research were to: (1) create an empirically based numerical simulation model of broad generality that links land-use use patterns and nitrogen loading to ecologically important and socially relevant endpoints of water quality, eelgrass habitat, and fish diversity and abundance (Kremer); (2) measure estuarine fish habitat and community structure throughout a range of estuaries in southeastern New England, develop empirical relationships of fish abundance, diversity, and habitat quality (Deegan), and test the ecological transferability of an index of estuarine biotic integrity; and (3) investigate perceptions of ecological models and modeling science by town planners, with the objective of determining the best and most efficient way to encourage scientifically aware decisions at the crucial, local level of land-use debate (Webler).
Summary/Accomplishments (Outputs/Outcomes):
Development and Validation of CLUE—Changing Land Use and Estuaries Ecosystem Response Model
The intent was to capture the shifting dominance of major aquatic primary producers in response to changing nutrient loads delivered from watersheds undergoing land-use alterations. Nitrogen loading and freshwater inputs to the water body are calculated from the land-use mosaic using a process-based algorithm, including wet and dry, indirect and direct atmospheric deposition. Hydrography is based on a two-layer flushing model with direct and indirect freshwater inputs and net offshore exchanges of nutrients and phytoplankton. Hypsography, often ignored in simple box models, is included because it affects area- and volume-based calculations of planktonic and benthic metabolism. Stocks of phytoplankton and macroalgae are simulated mechanistically, although empirically based formulations were chosen over more theoretical alternatives. State variables include: phytoplankton and macroalgae biomass, DIN, DIP, pelagic, and benthic heterotrophy. Nutrient and oxygen stoichiometry are coupled to all appropriate rates, and an index of susceptibility to hypoxia is calculated. Submerged aquatic vegetation is not modeled dynamically. Rather, bathymetry and the simulated water clarity are used to predict what portion of the system offers a suitable habitat for eelgrass.
A primary goal is for this model to be general, and to use data as input that are mostly available to local managers without expensive technical site-specific studies. To evaluate the success of our model however, field sampling was conducted in a number of sites to be able to evaluate the model's generality. Extensive data were already available for the primary site for which the model was initially developed: the three sub-estuaries of Waquoit Bay, MA. In addition, a number of other sites were sampled during summers throughout the project: four in Buzzards Bay, MA; three coastal ponds on the Rhode Island south shore; and two in northeastern CT. The field sampling program provided a broad range of general indicators of the ecological state of each estuary or water body.
The model was developed and tested extensively on only the Waquoit Bay sites, and primarily the Childs River site. Thereafter, the unchanged model was supplied only the appropriate input coefficients for land use and basin bathymetry and run for the other sites. Although validation data are limited to summer data in most cases, the model results were acceptable for all the novel sites. These findings provide the strongest evidence that supports the general applicability of the model.
The model is generalized. A small set of site-specific inputs is required, but these relate to the geography and spatial scales of each site, and are likely to be readily accessible. None of the process-based parameters are site specific. The strategy was to use fixed relationships and parameters. Critical attention needs to be paid to how well this works. This approach needs to be tested further by application to additional sites.
The model is dynamic. Daily processes and material fluxes are modeled, so annual cycles of the various variables can be investigated in addition to time-average relationships. Changes in predicted stocks or descriptive indexes can be seen over the course of the year. For instance, probability of anoxia peaks during the summer months in the selected sites, but in other sites may be high year-round. This is an important advantage of this form of modeling. A range of stock and rate variables that the model produces can be compared to observations.
The model is predictive. As a result, after corroboration with observed data, a range of management scenarios can be investigated (flushing time may be altered through dredging/widening channels, nitrogen load could be reduced as a result of changes in specific land-uses or policies, etc.). This can be a valuable management and planning tool because the consequences of alternative scenarios can easily be explored. For example, changing a municipal by-law associated with reducing the number of septic beds could have essentially no impact if the vast majority of the loading comes from other land use such as fertilizers from farming. In another case, the construction of a new golf course may be enough to cause a significant increased chance of anoxia. The model can generate predictions of these possible outcomes. It may well be that the strength and reliability of these predictions will lie in the comparison of alternative scenarios that are rigorously internally consistent, rather than a less-than-perfect agreement with observed data.
The CLUE model is general, and appears to simulate reasonably well the overall patterns of ecologically important variables related to the eutrophication in a range of shallow coastal sites in MA, RI, and CT. The model must be applied to a greater range of geographically diverse sites to evaluate its generality, yet it already behaves well for multiple sites that are typical of many others in the locale. Although its formulations are empirical, they capture causal information about important underlying processes. Including a range of input compartments allows comparison with many different measures of rates and stocks, which increases the chance that the model is capturing the basic ecological interactions. This also means that the model is more likely to be valid when applied to scenarios different from the status quo.
Nitrogen loading rates have been estimated by a number of computational schemes. Despite the importance of this input rate to coastal waters, its value alone has proved to be of little practical use to planners and managers. The model extends the N-load estimate into the water body, providing information on how these aquatic ecosystems might change under various loading scenarios. Further, the model appears to have some generality with minimal site-specific requirements, and it may be useful to numerous localities in coastal New England. For example, citizens and town planners might explore with the model comparative scenarios of different zoning guidelines and development to buildout. Such results will be internally consistent; a point that may be more critical than absolute accuracy.
Changing Land Use and Effects on Estuarine Biotic Integrity
The effects of eelgrass habitat loss were quantified on fish abundance, biomass, species composition and species richness, life-history characteristics, and habitat use by examining the response of the fish community to eelgrass loss in Waquoit and Buttermilk Bays over an 11-year period (1988-1999), and in 14 other embayments of Buzzards Bay during 1993, 1996, and 1998. Overall, fish abundance, biomass, and diversity decreased significantly along the gradient of decreasing eelgrass habitat complexity. Loss of eelgrass was accompanied by significant declines in these measures of fish community structure.
Based on these data, an index of estuarine biotic integrity (EBI) was applied to 36 sites in 16 estuaries on Cape Cod and in Buzzards Bay, MA. The EBI is comprised of eight metrics that measure important characteristics of the fish assemblage: numerical abundance, biomass, total species, species dominance, and species composition by life history and activity zone, and individual health. The EBI tracked habitat degradation over time in Waquoit and Buttermilk Bays. Average EBI values in medium-quality habitats of Buzzards Bay estuaries during 1996 and 1998, were less than expected based on earlier EBI values from Waquoit and Buttermilk Bays, suggesting that many of these sites are in transition from medium to low quality. The results indicate that the EBI is sensitive to habitat quality change, and further suggest that low-quality habitats may approach a stable fish community structure that is well reflected by the EBI. The relationship of the EBI to an independent measure of water quality demonstrated inherent time lags between the degradation and improvement of water quality, fish habitat, and response of the fish community.
The transferability of the EBI beyond Cape Cod was tested by sampling estuarine sites in Rhode Island and Connecticut. As in estuaries of Buzzards Bay and Cape Cod, most fish caught were young-of-the-year specimens, and averaged about 1 g wet weight. The EBI was strongly correlated with habitat quality in the Rhode Island and Connecticut estuaries. The results suggest that the EBI is a useful indicator of habitat quality for these Rhode Island and Connecticut estuaries, and that the shallow-water fish community responds to habitat quality and habitat change in similar ways to the communities studied in Buzzards Bay and Cape Cod. The EBI shows that extant eelgrass habitats in Rhode Island and Connecticut estuaries, although having undergone serious declines in recent years, have retained high value to shallow-water fish communities. On the other hand, habitats that have lost eelgrass are typically enmeshed in an autocatalytic syndrome of eutrophication, high turbidity, and oxygen stress.
The transferability of the EBI, based on fish communities of Cape Cod, also was studied to sub-estuaries along the southwestern shore of Chesapeake Bay. The EBI and its metrics in Chesapeake Bay (28 sites in five sub-estuaries, sampled 1995) were compared to values in Cape Cod estuaries, where and when the EBI was developed and tested (nine sites in two estuaries, 1988 and 1989). Habitat quality at each site was assigned one of two classes based on eelgrass shoot density and wet biomass. Analysis of variance showed that in the Chesapeake, as in Cape Cod, metrics attained higher values in eelgrass-rich sites. However, mean metric values in either habitat quality class were generally greater in Cape Cod than in Chesapeake Bay. Adjustment of the critical or threshold values for metrics was required in the Chesapeake to accommodate differences between the ecoregions in fish community structure and use of habitat. After recalibration, the EBI achieved values in Chesapeake Bay that were similar to corresponding values by habitat quality in Cape Cod. The coherency and repeatability of the EBI as developed in Cape Cod and as modified for the Chesapeake sites shows that the fish community's qualitative response to eelgrass habitat quality is similar in both ecoregions.
Measures of biotic integrity—the ability of a habitat to support a diverse and productive community of abundant organisms—have been useful in assessing freshwater ecosystems. However, similar measures have been applied rarely in estuarine systems. Estuaries are the interfaces where the effects of upland development and changing land-use patterns meet coastal waters. The EBI index is a useful measure of ecosystem function in a wide variety of estuaries in southern New England, and with modification, is transferable to the Chesapeake Bay region. As coastal development continues apace, and efforts are made to restore lost habitats such as eelgrass meadows, the assessment of habitat function will be increasingly important. The EBI focuses on the fish community as integrator of stressors occurring at lower levels in the food web such as nutrient pollution (eutrophication), loss of eelgrass, depletion of oxygen, and altered diversity and abundance of primary producers and consumers. The EBI is presented as a tool to measure the effects of habitat change, and to assess the success of efforts to restore submerged aquatic vegetation habitats. This index provides a summation of fish community characteristics that is sensitive to habitat function and integrity, which may be used in future applications of the CLUE model.
Ecological Models and the Human Community
The social science research focused on three major components: (1) attitudes and beliefs about ecosystem models among local governmental officials in small towns in southern New England; (2) performance of a participatory ecosystem modeling approach as applied in Rhode Island by the cooperative extension service; and (3) characterization of views on ecosystems models and their application in local decisionmaking by modelers and outreach professionals in southern New England.
A survey instrument was developed to inquire of local governmental officials' perceptions about the usefulness and applicability of nitrogen-loading models as decisionmaking aids. More than 150 planning board, select board, and board of health members were surveyed in small towns in southeastern Massachusetts.
The cooperative extension service at the University of Rhode Island has developed a nitrogen-loading and risk assessment model called the MANAGE model. For the past 5 years, they have been taking the model into community settings, where they engage local planning board volunteers in running the model. These trainings also are focused around solving an immediate local environmental problem. Consequently, the MANAGE process allowed for the opportunity to pursue research objectives closely tied to our original proposal.
After identifying the MANAGE process as an exemplary instance of an ecosystem modeling tool applied in a participatory manner with local town board decisionmakers, negotiations took place with the URI team to allow us to study the way that local board members engaged the model as a decisionmaking aid. Data were gathered through observations and interviews.
Interviews were completed with 16 individuals who design computer models or who are professionally engaged in helping local town decisionmakers use or interpret the results of these models. These interview transcripts were analyzed using qualitative data analysis tools. Several salient themes were extracted:
- Among modelers and local governmental officials, there is no consensus on whether or not ecosystem and nitrogen-loading models intended to inform decisionmaking ought to include uncertainty information or not. Some argued that uncertainty information confuses the public, gives the "experts" someplace to "hide," and leads to the results being summarily dismissed without further consideration. Others argued that scientists have a professional obligation to convey uncertainty information, and the lay users can easily understand it. The research produced insight into this question that could be posed as hypotheses for future research. The research findings also contribute to a growing literature on perceptions of and responses to uncertainty in environmental decisionmaking.
- Most modelers and outreach professionals see ecosystem and nitrogen-loading models as important tools to help educate local governmental decisionmakers how to think in terms of ecological systems (e.g., they should realize that what happens in the watershed affects the fish in the bay). However, most local governmental decisionmakers are primarily interested in models telling the decisionmakers whether or not actions on the scale of the individual building lot will damage the ecology of the bay. Local boards want models that inform specific individual decisions. Modelers do not believe present models are capable of that level of resolution. Instead, they argue that models should only be used to inform local decisionmaking at much larger levels, such as the level of zoning districts or large subdivisions.
- Virtually no evidence was found to support the idea that local governmental board members would be interested in using ecosystem or nitrogen-loading models. There are many reasons for this. Chief among these reasons are that: (a) models are too complicated for lay users to use knowledgeably; and (b) the vast majority of lay users do not have the interest in spending the time necessary to learn how to use models, especially when the models cannot tell them what to do about immediate decisions at the scale of the individual building lot (e.g., allow someone to add a bedroom to their home).
- There are advantages to involving local citizens and local governmental decisionmakers in the development, calibration, validation, or operation of ecosystem and nitrogen-loading models intended to aid local governmental decisionmaking about land-use issues. The two key advantages are: (a) it is a cost-effective way to gain competent data about existing and likely future land uses; and (b) engaging local people in the model means they are more likely to pay attention to the model's results.
Models clearly have tremendous contributions to make to watershed planning. Accurately predicting ecological responses to land-use changes is surely one. There are good reasons to believe that models also can help watershed planning in other ways. These should be considered working hypotheses in need of verification: (1) they can promote public dialogue, (2) they can enhance people's understandings; and (3) they may lead people to begin to think more systematically about land-use development and environmental impacts. For all of these reasons, work toward developing better models should continue.
Realizing the opportunities to improve watershed management through bringing models into local decisionmaking requires a partnership between the modelers and the users. What do the community users of these models want and need to know? It was found that both modelers and outreach professionals made assumptions about what users want, and do not want. These assumptions are largely untested and based on personal experiences, rather than verifiable social science research.
Further difficulties arise in clarifying what users want from models because there has been so little communication between users and modelers. More attention needs to be given to communication between modelers and users so that questions of what can and what should be modeled are answered in ways that serve the needs, abilities, and interests of all involved.
Models also can be misused in numerous ways. Users may misinterpret model results, or use models strategically to support a preconceived agenda. These dangers can be mitigated by having outreach professionals—or science translators—run and interpret the model for the local decisionmakers. Modelers also can distort models by letting their personal values and policy judgments enter in. Instead, modelers should make certain that value judgments of any kind are left to legitimately democratic processes. Having local people involved in the design of policy scenarios is one important way to accomplish this. It would not only help ensure that the scenarios are realistic, but also draw people into the policy dialogue in a constructive way.
More attention should be given to the possibility that models can undermine democratic discussions and planning efforts. There is a real danger that the model may co-opt the voices of citizens. Regardless of the model's accuracy, how models like these can change the way democratic decisionmaking happens should be examined. In other words, improvements in environmental quality may come with certain costs. There should be an awareness of these costs, even as development of models continues.
Another obstacle to bringing models into policymaking has to do with uncertainty information. Some believe the best approach is to present uncertainty information and train people how to interpret it. They argued that lay people deal with uncertainty in their everyday lives—weather forecasting, for example—and it is just a matter of drawing the right analogies. Others think uncertainty information is the "kiss of death" in an adversarial political system in which people need plain simple answers.
A final obstacle to overcome will be the questions of social and ecological transferability. The keys to establishing that a model works are local data and experience. Although several modelers were skeptical about lay monitoring programs, it is obvious that the only affordable way to collect the volume of local data needed is to rely on citizen volunteers. Future research might focus on forming effective partnerships between modelers and lay monitoring programs.
Ecosystem and nitrogen-loading models will continue to play important roles in helping decisionmakers manage the environmental impacts of human activities. As the scientific complexity and credibility of models continue to evolve, how best to integrate these models into democratic discussions at the local level also must be learned. At cities and towns all along the coast, citizens elected to serve on local boards are making decisions about land use that will shape the environment for decades to come. The effective and appropriate employment of models in local democratic decisionmaking is best served by communication between users and those familiar with the models. However, connecting science with democratic policymaking demands a form of coordinated communication about which little is known. Future research should continue to investigate how modelers and users can talk with each other and coordinate their efforts toward watershed management goals.
Journal Articles on this Report : 12 Displayed | Download in RIS Format
Other project views: | All 67 publications | 20 publications in selected types | All 12 journal articles |
---|
Type | Citation | ||
---|---|---|---|
|
Brawley JW, Collins GN, Kremer JN, Sham CH , Valiela I. A time-dependent model of nitrogen loading to estuaries from coastal watersheds. Journal of Environmental Quality 2000;29(5):1448-1461. |
R825757 (1999) R825757 (Final) |
not available |
|
Collins G, Kremer JN, Valiela I. Assessing uncertainty in estimates of nitrogen loading to estuaries for research, planning, and risk assessment. Environmental Management 2000; 25(6):635-645. |
R825757 (1999) R825757 (Final) |
Exit |
|
Deegan LA, Wright A, Ayvazian SG, Finn JT, Golden H, Merson RR, Harrison J. Nitrogen loading alters seagrass ecosystem structure and support of higher trophic levels. Aquatic Conservation - Marine and Freshwater Ecosystems 2002;12(2):193-212. |
R825757 (Final) |
not available |
|
Dietz T, Tanguay J, Tuler S, Webler T. Making computer models useful: an exploration of expectations by modelers and local officials. Journal of Policy Analysis and Management. |
R825757 (Final) |
not available |
|
Graham S, Davis J, Deegan L, Cebrian J, Hughes J, Hauxwell J. Effect of eelgrass (Zostera marina) density on the feeding efficiency of mummichog (Fundulus heteroclitus). Biological Bulletin 1998;195(2):241-243. |
R825757 (Final) |
not available |
|
Hughes JE, Deegan LA, Weaver MJ, Costa JE. Regional application of an index of estuarine biotic integrity based on fish communities. Estuaries 2002;25(2):250-263. |
R825757 (Final) |
not available |
|
Hughes JE, Deegan LA, Wyda JC, Weaver MJ, Wright A. The effects of eelgrass habitat loss on estuarine fish communities of southern new England. Estuaries 2002, Volume: 25, Number: 2 (APR), Page: 235-249. |
R825757 (Final) |
not available |
|
Hughes JE, Deegan LA, Weaver MJ. Transferability of an index of estuarine biotic integrity based on fish communities. Transactions of the American Fisheries Society. |
R825757 (Final) |
not available |
|
Kremer JN, Reischauer A, D'avanzo C. Estuary-specific variation in the air-water gas exchange coefficient for oxygen. Estuaries 2003;26(4A):829-836. |
R825757 (Final) |
not available |
|
Kremer JN, Nixon SW, Buckley B, Roques P. Technical note: conditions for using the floating chamber method to estimate air-water gas exchange. Estuaries 2003;26(4A):985-990. |
R825757 (Final) |
not available |
|
Vaudrey JMP, Kremer JN. Ecosystem metabolism and nitrogen assimilation by dominant primary producers in three shallow temperate estuaries in relation to nitrogen availability. Estuaries. |
R825757 (Final) |
not available |
|
Wyda JC, Deegan LA, Hughes JE, Weaver MJ. The response of fishes to submerged aquatic vegetation complexity in two ecoregions of the mid-Atlantic bight: Buzzards Bay and Chesapeake Bay. Estuaries 2002;25(1):86-100. |
R825757 (Final) |
not available |
Supplemental Keywords:
atmosphere, water, watersheds, groundwater, land, soil, sediments, acid deposition, global climate, marine, estuary, precipitation, leachate, adsorption, chemical transport, effects, ecological effects, metabolism, animal, organism, population, stressor, acid rain, discharge, ecosystem, indicators, restoration, regionalization, scaling, terrestrial, aquatic, habitat, integrated assessment, alternatives, sustainable development, restoration, public policy, decisionmaking, community-based, cost benefit, conjoint analysis, observation, nonmarket valuation, contingent valuation, survey, psychological, preferences, public good, socioeconomic, conservation, environmental assets, sociological, environmental chemistry, biology, physics, social science, ecology, hydrology, geology, mathematics, zoology, EMAP, modeling, monitoring, analytical, surveys, measurement methods, general circulation models, climate models, satellite, landsat, remote sensing, northeast, Atlantic coast, Chesapeake Bay, states, Massachusetts, Connecticut, Rhode Island, Virginia, MA, CT, RI, VA, EPA Region 1., RFA, Scientific Discipline, Water, Ecosystem Protection/Environmental Exposure & Risk, Hydrology, Water & Watershed, Monitoring/Modeling, Ecology and Ecosystems, Watersheds, Social Science, social science research, model-based analysis, coastal watershed, fish, community-based research, nitrogen inputs, integrated assessment, aquatic ecosystems, data collection, ecological risk, nutrient monitoring , integrated ecological assessment model, water quality, ecology assessment models, public policy, social transferability, ecological models, nutrient transport modelRelevant Websites:
http://ecosystems.mbl.edu/Research/Clue/ Exit
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.