METHODS

Objective 1: Develop, improve, and evaluate watershed models and other approaches for TMDL development and implementation.

The ultimate goal of this objective is to improve the ability of TMDL development models to assess the impact of agricultural practices on in-stream water quality. At the first technical committee meeting, the specific needs of the models for TMDL development will be discussed. This discussion will also include establishing techniques for model evaluations. Needs for specific data, data parameters and criteria, and computer-compatible data formats will be mutually developed. As the result of this discussion, model evaluation and development and data collection responsibilities for the participating states/locations will be established. Results will be shared as the research progresses. An objective coordinator will provide overall leadership for the activities of Objective 1. These activities will be summarized to all project participants at the annual technical committee meetings. The objective coordinator will be selected by the participants of Objective 1 at the technical committee meetings and will likely serve a two-year term.

Task 1. Evaluate existing watershed assessment models (AGNPS, ANSWERS, HSPF, SWAT, GWLF, etc.) for their applicability for TMDL development in agricultural watersheds.

The ultimate goal of this task is to identify tools that assess the impact of agricultural practices on the hydrologic, chemical, biological, and economic response of a watershed and that are appropriate for TMDL development. If needed, existing models will be modified to improve their potential for TMDL development (Task 5). Virginia Tech, and the Universities of Florida and Georgia have either recently completed or are working on over a dozen TMDLs and TMDL like watershed assessment projects. Florida is just beginning the development of TMDLs for the Lake Okeechobee basin. Through these projects and others that participants have participated in, databases have been assembled that can be used to test different TMDL models. We will pick three or more impaired watersheds (in different physiographic regions) with comprehensive hydrologic, water quality, land use, pollutant discharge, etc. records and then use different TMDL models to develop TMDLs for the various impairments in each watersheds. A critical component of this task will be to identify watersheds that have both adequate data for running the models and that have the same impairments. This will be difficult, and additional watersheds may have to be identified for some impairments. Databases for these watersheds will be developed through Task 3. Once these TMDLs are developed for each watershed, the TMDLs developed with the different models will be compared and the strengths and limitations of each model for each impairment in each physiographic region will be assessed. It is also recognized that not all models are suitable for every impairment. For example, all of the likely models can simulate sediment loss, but they may not be able to simulate bacterial impairments. Consequently, not every model will be evaluated for every impairment. Watersheds for assessment and development of TMDL will be selected during and after the first regional project meeting. The following models will initially be evaluated: AGNPS (USDA-ARS Oxford and Tifton), ANSWERS (North Carolina State, Virginia), HSPF (Virginia), SWAT (USDA-ARS Tifton, Georgia, Virginia, Wisconsin), GWLF (Virginia).

Task 2: Develop new and improved systems to integrate existing data sources with models used for TMDL development.

Applications of models are frequently limited by input data requirements, interpretation of massive output data information, and the technical expertise of users. As models become more comprehensive, these issues are even more important. For example, ANSWERS and AGNPS have been underutilized because of the expertise, time and data collection expense necessary to operate them. During the previous project, S-273, GIS interfaces were developed for the ANSWERS (Questions Interface) and AGNPS models and the USEPA developed a similar interface, BASINS, for HSPF and the new release of BASINS 3.0 includes SWAT. Under Task 2, user interfaces will be further developed to enhance the use of watershed assessment and TMDL development models. Efforts will focus on using GIS-based and user-support systems.

A subcommittee will be formed to coordinate the activities of Task 2. The Task 2 coordinator will be selected by the subcommittee at the yearly technical committee meeting to provide overall leadership for completing the activities of Task 2. The subcommittee will supervise the collection of experimental and watershed information to be compatible with the GIS systems to be used in the regional project. Needs related to geographic regions will be considered by the subcommittee. The constraints and needs identified by the subcommittee will be shared with those participating states/locations evaluating and developing models (Task 1) to ensure that the outputs of their models are compatible with the interface software of Task 2. Likewise, the constraints and needs of the interface software for input data will be shared with the participants of Tasks 1 and 3. Compatibility of input and output data among model developers, interface developers and data collectors will be the responsibility of the Task 2 subcommittee. This subcommittee will meet at the annual committee meeting to monitor the previous year's progress in assembling and developing watershed information that is compatible with the GIS systems selected for use in the project. The subcommittee will identify changes and improvement to facilitate appropriate model development.

Geographic Information Systems (GIS)

Tools will be developed to combine disparate agro-ecological landscape data from two to three regions in the United States into a comprehensive GIS database for use in model evaluation and development. Model interfaces will also be developed to integrate the GIS database and analysis tools. The effectiveness of the tools and interface software will be evaluated using the databases and data collected as part of Objective 3. Alabama A&M University will investigate the application of GIS and GPS technology to develop environmentally sound water quality assessment methodologies for efficient water quality management. The objectives of this sub-task are three fold: (1) to demonstrate the application of GIS and GPS technology to evaluate the trend of water quality status in northern Alabama, (2) to build a GIS database for water quality analysis, and (3) to utilize information gathered in the past and rank watersheds and subwatersheds according to their potential for nonpoint source pollution of surface waters. Mississippi will coordinate efforts to develop GIS interfaces for the AGNPS model. Virginia will continue development of the QUESTIONS interface for ANSWERS.

User Support Systems

In general, interfaces that facilitate the use of models have been developed by researchers and consider research user's needs rather than the needs of the user community. The user's perspective (TMDL developers) will be carefully considered in the development of the GIS-interface software discussed in the previous section. Recent developments in computer visualization have permitted the creation of dynamic displays that incorporate spatial and temporal variability of complex processes. Computer visualization makes sophisticated interactive and effective decision-making capabilities available to those who are not experts in computer programming or modeling. For this project, visualization techniques, involving computer-animation and full-motion digital video data types, will be developed to permit dynamic displays of basic and derived data describing agroecosystem behavior and the landscape mosaic. The development of computer animation and visual displays of model input and output will be conducted by Georgia (Athens) and Iowa. All developed user support systems will also be designed to prevent or discourage use of inappropriate model parameters.

Task 3: Collect and assemble comprehensive databases to facilitate development and evaluation of models used for TMDL development.

Reliable, comprehensive, and complete data sets are necessary for the development, testing and validation of comprehensive watershed models. Computer development has driven the development of detailed distributed models that take into consideration variability in land use, geography and climate. Data to derive some input parameters required for comprehensive models used for TMDL development models are often limited. By utilizing existing data wherever possible, this project will provide data from plot, field, and watershed experiments for use in model development and testing activities under Task 1, with data put into the necessary framework for model interfaces developed under Task 2. We anticipate that three to four comprehensive databases will be assembled for model testing and evaluation. A subcommittee will be formed to oversee and coordinate the Task 3 efforts. The organizational structure is similar to that previously given for subcommittees under Tasks 1 and 2. This subcommittee will meet jointly with scientists involved in Tasks 1 and 2 to identify gaps in the current databases. Recommendations will be made to encourage technical committee members to collect the data necessary to fill these gaps. Common parameters being measured and common parameter estimation methods will be identified. The different methods being used will be compared, and those best suited for the goals of this project and TMDL development will be described. The Task 3 subcommittee will be responsible for ensuring commonality in the experimental techniques and data collection methods utilized in this project.

While the data-collection efforts outlined below may not at first appear to contribute toward a coordinated regional research thrust, the efforts are necessary to provide the input data required to complete successfully the activities of Tasks 1 and 2. Many of the data-collection efforts are already underway. The data collected, will be shared freely among project participants. The data will be used to test/evaluate the various watershed assessment and TMDL development tools to determine their performance and adaptability for various water quality impairments and under the variety of geological and land-use conditions represented in the project.

BMP assessment studies

Alabama will collect field data from a 6.7-acre watershed of cattle grazing pasture in north Alabama to assess the effectiveness pasture management BMPs. Five experiment plots will also be instrumented to measure sediment and nutrient losses to compare effects of alley cropping and conventional terraces. Thirty-nine 2 m by 6 m plots have been established at a Kentucky Agricultural Experiment station. Rainfall simulators will be used on these plots to collect data on the influence of grazing and manure application on chemical and biological quality of runoff. These plots have been used to assess the water quality effects of grass buffer strips, continuous versus rotational grazing, grazing duration and stocking density in reducing pollutants such as nutrients, solids, fecal coliform, heavy metals, hormones and antibiotics. Future plot studies will investigate the effectiveness of additional BMPs (Kentucky). Virginia Tech has been monitoring the Mossy Creek, Long Glade and Pole Cat Creek watersheds for the past three to six years and will continue monitoring each watershed for ten years to assess the water quality impacts of intensive agriculture and urbanization in agricultural watersheds. Extensive web-based water quality and land use databases are available for these projects. Florida will monitor the Lake Okeechobee basin in support of the Lake Okeechobee basin TMDL development effort and this data will be made available to project cooperators for model development and evaluation. These data, while collected at individual states and locations, are an important and contributing part of the overall effort of the regional project. The data will constitute much of the input necessary for the successful completion of Tasks 1 and 2.

Task 4: Develop better guidelines for calibration and estimation of model uncertainty.

A critical component of this task will be to determine what is the minimum amount of data needed to develop sound TMDLs. This may vary with the particular modeling approach used for TMDL development. This task is essential because the lack of water quality and flow data has been identified as one of the most severe problems facing the TMDL program (NRC, 2001). Although physically-based, distributed parameter models in theory have a direct interpretation to the physical reality, in practice it is impossible to collect sufficient measurements over a watershed to fully support these models. Instead, some model inputs have to be "effective parameters" representing physical properties of aggregate areas, such as hydrologic resource units, hillslopes, and subwatersheds. Most physically-based models may be regarded as conceptual models that require one or more effective parameters to be calibrated to the physically observable responses that are the object of prediction. Model calibration requires selection of values for model parameters such that the model closely simulates an observed response (e.g. streamflow hydrograph, sediment discharge, P loading). This approach has been criticized (Beven and Binley, 1992; Van Stratten and Keesman, 1991; Klepper et al., 1991) because calibrated parameter values may be 'tuned' to compensate for deficiencies in the model structure and data input. Significant advances in computational methods have recently emerged to objectively address these limitations. These methods use a combination of Monte Carlo analysis along with calibration on one or more objective functions (Gupta et al., 1998; Kuczera and Parent, 1998). Wisconsin and Virginia will investigate a multi-objective Monte Carlo analysis technique that can be used for model calibration. Performance of a model-parameter set combination are evaluated as a function (the objective function) of the deviation of the predictions compared to the observations of both stream flow, pollutant levels, or other observations. Some of the model realizations are rejected based on a criterion of acceptance (e.g. performance measured higher than a threshold value or a certain percentage of simulations with the highest performance measure score) using standard calibration techniques (Legates, 1999) and an interval of predictive uncertainty constructed from the retained models. A primary objective of the Task 4 subcommittee will be to assist the model evaluation subcommittee (Task 1) in evaluating the impacts of errors or uncertainty in land classification model predictions and resulting uncertainty in TMDL allocations.

Task 5: Extend capability of models for TMDL development.

If problems are identified with the existing models with respect to TMDL development, efforts will be made to improve the models if feasible (Task 5). For example, a pathogen transport submodel will be added to ANSWERS to facilitate its use in developing fecal coliform TMDLs (Virginia). Riparian buffers are also believed to play an important role in determining the water quality of watershed and are often suggested as a best management practice. North Carolina State University, USDA-ARS-Tifton, and Virginia Tech will evaluate the Riparian Ecosystem Management Model-REMM (Altier et al., 1993) and determine its utility for use in evaluating sediment and nutrient reductions in TMDLs. North Carolina will coordinate development of the buffer model VFSMOD (http://www.icia.es/vfsmod/) and investigate how it can be linked with watershed scale models to help assess the function and importance of conservation buffers and riparian buffers at the watershed scale. Some of the management options to be considered are the buffer widths, composition of plant species, and timing and extent of grazing and harvesting. North Carolina State University will coordinate these activities.

Considerable progress has been made in combining hydrologic and chemical models. The regional project will continue development of algorithms to predict the impact of agricultural practices on the physical and chemical constituents in water and we will seek to tie physical and chemical constituents to in-stream biological integrity (University of Georgia, Virginia).

To date much of the effort in the development of hydrologic and water quality models has been focused on either the upland erosion- and runoff-driven components or on subsurface movement of contaminants. Less attention has been given to how the models handle the delivery of sediment from the upland areas through the stream networks and out of the watershed. Though the modeling of the upland processes has become very sophisticated, process-based, and accounts for spatial and temporal variability in the model inputs, modeling of sediment and contaminant delivery often still relies on very crude delivery ratio estimates. This may be a significant weakness in using the models for TMDL implementation, as it is not the initial movement but rather the final delivery to the point of interest that is of primary concern. This project will examine the literature for a better understanding of the current approaches used for estimating sediment delivery in the watershed models, and will attempt to examine each of the approaches in light of the limited watershed delivery data (Tennessee).

Controlling water pollution can follow many courses. Economics has an important, if not vital, role to play in identifying policy strategies that can enhance water quality at least cost to landowners and taxpayers. An economic framework can coordinate policy formulation among different levels of government and help to unify policies across regions. Economics also helps determine the optimal level of water quality protection that balances public's desire for improved water quality and the public's willingness to pay for improved water quality. Society does not benefit from overly stringent or costly water quality goals. Measuring the benefits of water quality protection in economic terms is difficult, since many benefits occur outside the easily observable market conditions. Even where water quality impacts on markets are observed, it can be difficult to ascertain just how water pollution affects the ability of a resource to provide economic goods and services. Nevertheless, information on costs and benefits is essential to developing socially optimal water quality protection policies (Ribaudo et al., 1999). The ultimate goal of this objective is to convey societal costs (benefits, costs, and equity) associated with the current TMDL Program (development and implementation), and societal costs associated with recommended improvements to the overall TMDL Program. Current TMDL development models are used to assess the impact of agricultural practices on in-stream physical, chemical, and to a limited extent, biological water quality. However, data limitations are reducing the efficacy of such tools. In fact, data limitations are generating significant additional costs to water quality management efforts by misdirecting resources to ineffective management alternatives. Economic investigations of this project will closely parallel, and expand upon, the tasks identified in Objective 1; and support the efforts of Objective 3. As in the case of Objective 1, an Objective Coordinator will provide overall leadership for the activities of Objective 2. These activities will be summarized to all project participants at the annual technical committee meetings. The Objective Coordinator will be selected at the technical committee meetings by the participants contributing to Objective 2 and will likely serve a two-year term.

To accomplish the objective of Task 1, economists will work closely with modelers and state agencies to better quantify the economics of data collection, data accuracy, modeling assumptions, modeling application, and model improvement.

Data Collection Costs – Assessment data in state 305(b) reports are not complete because they do not represent all of a state's waters. In addition, there is substantial variation among states in virtually every aspect of water quality monitoring. A General Accounting Office Report (GAO/RCED-00-54 Water Quality) found variations in 1) the standards states use to assess water quality, 2) the way that states select their monitoring sites, 3) the kinds of monitoring tests that states perform and how they interpret results, and 4) the methods that states use to determine the causes and sources of pollution. For example, the 1996 National Water Quality Inventory revealed that only 19 percent of the nation's rivers and streams were assessed; one-half of which were assessed by means other than water quality monitoring. States have reported that they lack the data needed for two fundamental activities essential to the process of managing water quality: comprehensively assessing all state waters and compiling a list of waters that do not meet standards. Each state spends millions of dollars annually to assess the conditions and trends of their respective rivers and streams. Because states must develop TMDLs for waters on the list of impaired streams, there are tremendous economic ramifications associated with listing waters from limited data. Similarly, most water quality models require input parameters related to geospatially accurate land-use data. However, GIS products commonly used in modeling the land-water interface utilize land cover databases that require significant investments in labor, capital, and time. Economists working on this task will work closely with modelers to quantify the costs of data collection for the three watersheds selected under Objective 1. OSU, WVU, VT, and UT will lead a coordinated effort that will include collaborators from the institutions participating in Objective 1, Task 1.

Data Quality – As mentioned above the costs of data collection can represent a significant public investment. The natural resource management community would generally agree that additional investments are needed to acquire additional data to improve model applications. However, concerns over basic input data accuracy are increasing. For example, the methods that states use to select water quality monitoring sites have an impact on how the data can be used. EPA encourages states to use a random or statistical sample that will result in data, which is representative of the condition of all waters in a population. Some 39 states, however, use "evaluated" data to list streams on the list of impaired waters. Evaluated data can be old data, model output used in place of real data, qualitative information, and evaluations made by fish and game biologists. This type of information should be used as an indicator of potential water quality since data sources can vary in quality and reliability ((GAO-RCED-00-54-Water Quality). One of the other primary input data sets for most modeling activities is land use. Many institutions have accelerated efforts to geographically capture land patterns with the advent of GIS technology. In many instances land-cover data sets, derived from satellite imagery, are often inaccurately labeled as land use. There are significant distinctions between land-cover and land-use. Land-cover, most often derived from satellite imagery, identifies vegetative and development patterns to classify land in to general categories such as agriculture, forest, urban, etc. TMDL development and implementation efforts in Georgia have identified error rates in excess of 40 percent with land-cover data sets being utilized in TMDL modeling. Land use, most often derived from aerial photo-interpretation, identifies specific features on the landscape and makes inferences regarding the management of that land (agriculture - soil erosion rates, cattle accessing streams; forestry – adequacy of riparian buffers, road management; urban – erosion and sediment control, etc.). Therefore, land-use is a more descriptive representation of land influences on water quality. This component of Task 1 represents a critical element to the overall TMDL program nationally. In fact, it represents a critical element to all land-water interface simulations nationally as land-use data and water quality data are the foundational inputs upon which other algorithms are based. Uncertainty about the causes and consequences of environmental problems can be reduced through research. Environmental research generates information that is a pure public good, and so its cost is appropriately spread across the entire population. Economists addressing this task will work closely with state land and water monitoring agencies in the states for which the three watersheds are selected. The purpose of this interaction will be to assess the methods, accuracy, and costs of such data collection efforts. Accuracy will be quantified via a stratified two-stage area sample utilizing primary sample units and sample points. Costs will be compiled via interviews with, and documentation from, agencies that are responsible for generating such data sets. UGA (Athens) will lead this effort with collaboration from institutions in states where the three watersheds from Objective 1 are selected.

Model Applications/Improvements – In Objective 1, Task 1, a number of watershed assessment models will be evaluated for their usefulness as a TMDL tool. With limited federal, state, and local resources, economic efficiency, as opposed to a least-cost policy, becomes a primary goal. Economists addressing this issue will quantify the costs of model applications and improvements consistent with Objective 1, Tasks 1 and 5; and weigh those against larger societal benefits of improved ex ante conditions and TMDL program administration. Virginia will assume responsibility for leading this effort, with support from UGA (Athens).

Modeling Results - Modeling efforts can generate unrealistic, or uncertain, results for one of two primary reasons: data concerns (limited, inaccurate, etc.), or model assumptions (subjective bias, geographical bias, scale bias, etc.) When model results inaccurately, or inconsistently, simulate real world events; society incurs costs. These costs include the total public investment spectrum in natural resource management; from data collection and accuracy to model development and application to public policy and programs to individual land-user decision making. Estimating the significance of these costs for a given level of uncertainty will be the objective of this component of Task 1. WVU, UT, VT, and UM will lead this effort, with support from UGA (Athens).

To accomplish the Task 2, economists will work closely with Objective 1 modelers, state agencies, and watershed stakeholders to better quantify the economics of Social Feasibility, Social Equity, and Social Policy.

Social Feasibility - Modeling results can portray a number of "what if" implementation scenarios with a focus on the physical relationships between land management and water quality. The concept of efficiency, or even attainability, should be analyzed within the context, or constraints, of feasibility. In fact, TMDL program criteria from the Clean Water Act requires development of options that are feasible when considering measures to improve water quality. One foundational element of feasibility is information. Knowledge of economic and environmental impacts of alternative farming systems can assist decision makers in designing appropriate incentive mechanisms to encourage farming systems that alleviate agricultural externalities such as soil erosion and water pollution. Even though agricultural pollution can be studied at plot, field, and farm levels, the watershed is the most logical geographical unit for identifying holistic cause-and-effect water quality relationships, linking upstream uses to downstream effects, developing reasonable remediation efforts, targeting limited resources, and educating and involving the public. When agricultural nonpoint source pollution occurs, it does so at inefficiently high levels because producers, when making production decisions, have no incentive to consider the costs pollution imposes on others (Baumol and Oates, 1979). Economists refer to such costs as externalities because they are external to the production manager's decision framework. In a decentralized, competitive, economy, social welfare will not be maximized in the face of agricultural NPS pollution because of its external nature. An efficient solution is one that maximizes expected net economic benefits. Decisions must be based on the expectation of what damages will be since it is impossible to accurately predict damages due to the varying nature of pollutant runoff and transport. Consequently, the efficient solution is often referred to as the ex ante efficient solution, meaning that is the expected outcome as opposed to the actual or realized outcome (Ribaudo et al., 1999). Economists working on this component of Task 2 will develop alternative implementation plans with modelers in the three selected watersheds of Objective 1. Alternative implementation plans will include alternatives that identify the most economically efficient alternative, the most socially acceptable alternative, and the highest resource protection alternative as a minimum. UGA will lead this effort in collaboration with institutions of states in which the three watersheds from Objective 1 are selected.

Social Equity – Agricultural NPS pollution occurs at greater levels than are socially optimal because markets fail to accurately transfer the social costs of pollution to producers. At the same time producers often employ management strategies that provide public goods for which society does not pay. Economists and policy makers have suggested a variety of incentive-based instruments to control NPS pollution. However, no comparison of these instruments exists. As a result, social debates are intensifying in today's legal environment over which parties are financially responsible for environmental mitigation. Economists will assess current economic incentives to convey historical perspectives over resource equity in use and protection. Current economic incentive instruments will be evaluated and compared to identify those that are most efficient for specific watersheds and water quality impairments. These incentives include performance-based incentives, input- and technology-based incentives, designed-based incentives, compliance mechanisms, and market mechanisms. Virginia will lead this effort.

Social Policy – Since the passage of the Clean Water Act, private and public sectors of our economy have spent an estimated $541 billion on water pollution control. A vast majority of this money has been spent on point sources of pollution that are mainly municipal and industrial. While there have been improvements in water quality associated with point source pollution programs, NPS pollution programs have not been as effective. Designing comprehensive policies for controlling NPS pollution consists of defining appropriate policy goals, choosing appropriate instruments, and setting these instruments at levels that will achieve the goals at least cost. Difficulties with each of these steps derive from the complex physical nature of NPS pollution. An economically efficient outcome is generally unattainable because policy makers seldom have information about economic damages. Instead, a cost-effective approach to NPS control is typically preferred. A cost-effective outcome is an outcome in which policy is achieved at a least cost. A variety of policy goals exist; however, the physical nature of NPS pollution limits the way in which goals may be defined and also the economic properties of the goals. Apart from the economist's ideal outcome of economic efficiency, there are generally two types of goals: 1) physically-based goals (water quality, runoff, etc.), and 2) input- and technology-based goals (nitrogen application rates, etc.) (Ribaudo et al., 1999). Economists will compare physically-based goals to input- and technology based goals to identify the economically preferred method. Deterministic models will be developed to quantify costs and samples needed to achieve goals that will lead to water quality improvements.

To accomplish the objective of Task 3, economists will work closely with watershed stakeholders to better quantify the economics of Farm Attitudes and Preferences, and Environmental Tradeoffs.

Farm Attitudes and Preferences – Attitude is defined only as "a predisposition to act" and does not necessarily represent one's behavior. For example, some farmers may have strong conservation attitudes, but are influenced more by financial stresses. Also, lack of education or information may limit farmers in the implementation of conservation practices. A producer's disposition is usually a combination of attitudes on stewardship, profit, economic orientation, and government involvement. An attitude of stewardship, or belief that farmers have a moral obligation to conserve natural resources, has a positive effect on management activities that improve water quality. However, some studies have shown that economic returns are more influential. Another positive influence on the adoption of practices that improve or protect water quality is economic orientation, which is a farmer placing a high value on being his/her own boss. Nevertheless, not all attitudes have a positive effect. Many farmers have mixed opinions toward government involvement in agriculture. The majority of farmers do not support any type of regulatory pollution controls. They also feel that government is responsible for funding agricultural practices that provide public goods (i.e. clean water). In this section of the project, economists will capture producer attitudes and preferences associated with environmental protection and the TMDL development and implementation process. Investigators will collaborate closely with economists assessing social equity under Task 2.

Environmental Tradeoffs – At the farm level, an economically efficient solution is defined by three conditions: 1) For each input and each site, the marginal net private benefits from the use of the input on the site equal the expected marginal external damages from the use of the input; 2) A site should be brought into production as long as profits on this site are larger than the resulting expected increase in external damage; and 3) Technologies should be adopted on each site such that the incremental impact of each technology on expected social net benefits is greater than or equal to the incremental impact on expected damages (Ribaudo et al., 1999). These three efficiency conditions represent economic tradeoffs involving farm profitability and water quality. Economists will assess the economic impact of alternative TMDL implementation approaches and policy instruments (Objective 2, Task 2), on individual farms within each of the three selected watersheds. Virginia will lead this effort.

To accomplish the objective of Task 4, Economists will work closely with watershed stakeholders to better quantify the services provided by the local watershed. Valuing ecosystem services is controversial because of the potential importance such values may have in influencing public opinions and policy decisions (Costanza et al., 1998). Very little is known about how the public values ecological goods and services produced by mixed public and private land ownership at a watershed scale. However, a failure to quantify or value these ecosystem services could imply a zero value, whereas ecosystem services in most cases have value larger than zero. Of the 3.5 million miles of rivers in the United States, only 56 percent of them fully support a multiple use mandate. The multiple use mandate outlines that streams and rivers should supply drinking water, fish and wildlife habitat, recreations, and agriculture as well as flood prevention and erosion control. Economists will value ecosystem services in the three selected watersheds of Objective 1 using focus group sessions to place values on environmental "indicators". It is anticipated that valuing ecosystem services will also complement the assessment of the potential ecological benefits/implications of TMDL implementation at watershed level (Objective 3).

The Alabama watershed being modeled under Objective 1 (Task 1) has been the focus of previous research on the implementation of environmentally sound practices by landowners and the general public. In 1993, a telephone survey of residents was conducted in the Flint Creek Watershed of Alabama, which is made up of West Morgan, Northwest Cullman and East Lawrence Counties.  Most citizens believe that the current levels of environmental law enforcement are not strict enough, and 122 out of 202 respondents believe that we need more governmental regulation to protect our environment.  The majority also believed that protecting the environment should be given priority, even at the risk of slowing down economic growth. When respondents were asked why they were not more actively involved with water quality problems in their community, the majority (69% or 139 out of 201) said that they were not sure where to start. Sociologists at Auburn University will conduct a case study of changes and impacts of efforts to adopt environmentally sound practices through the TMDL program in landowner management decisions, community regulations and planning, as well as in the construction industry. Intensive interviews will be conducted with a network sample of informants knowledgeable about local conditions and circumstances. The study will extend and complement economic and engineering studies of TMDL issues, examining the institutional and landowner decision interface that shapes these implementation and maintenance of measures designed to improve environmental quality.

Most current field- or watershed-scale hydrologic models make estimates based on the assumption of relatively static systems. Though the models predict mass movement of sediment and contaminants through the system, they fail to take into account the capacitance (storage) inherent in such systems. Such effects are probably minimal in upland systems with negligible storage, but they can become very important in larger watersheds, and may mask the true effectiveness of BMP implementation. For example, sediment delivery from a severely-eroding watershed may be substantially reduced by deposition in floodplains, stream channels, and lakes. These may in turn act as source areas of sediment if the upland erosion is reduced, thereby buffering the resulting impact (Trimble and Crosson, 2000; Nearing et al., 2000). Tennessee and others will examine the literature for information related to such changes, and will attempt to link this to the more theoretical examination of how the watershed models estimate sediment delivery (Objective 1).

Increased sedimentation associated with common land use practices primarily impacts stream biota via a reduction in physical habitat quality (Bellamy et al. 1992, Wood and Armitage 1997). In fact, alteration of stream habitat quality likely has a stronger effect on the ecological integrity of aquatic systems than reductions in water quality (e.g., increased turbidity) (Waters 1995). Consequently, it is imperative that we quantify causative links between TMDL implementation, water quality and hydrologic characteristics, and instream habitat quality. The objective of this task will be to twofold. First we will obtain empirical data on the relationship between sediment loading rates and stream habitat quality across a range of stream types in West Virginia and other states. Study streams will vary on the basis of soil characteristic, topography, and the degree of agricultural activity. Second, we will use this empirical information along with models of water and sediment delivery to predict the impacts of TMDL implementation on stream habitat quality.

This task will involve establishing cause and effect relationships between ecosystem health (as measured by various biological measures, such as index of biotic integrity, etc.) and causative pollutant loadings. We will then attempt to use the water and habitat quality predictions of the predictive models, and link them to the impacts they are likely to have on stream ecological health. West Virginia will investigate watersheds impacted by agriculture and Acid Mine Drainage (AMD).  As with agricultural based TMDLs, there are a suite of watershed models that focus on landscape features, water quality, and economics, however, none of the models considers biological integrity.  West Virginia will attempt to add insect and fish community submodels to tools used for TMDL development.  The basic approach is a paired impact-control watershed approach.  In the control watershed, relationships between stream size, location, physical habitat, productivity, and fish communities are quantified.  This information is then used to assess the effects of human land use on fish communities at a watershed scale in the impacted watershed.  Combined, the information is used to predict how fish communities will respond to various TMDL implementation approaches.  The final objective is to produce a spatially explicit model that is used to direct TMDL implementation plans in ways that maximize ecological benefit under socioeconomic constraints. This task will be led by West Virginia and Virginia, which are in the process of developing TMDLs for AMD and benthically impaired streams.