Priorities for Coastal Ecosystem Science (1994)

In order to provide advice that can be applied easily within the Water Subcommittee's planning framework (that transcends freshwater resources and coastal marine environments), recommendations of high-priority science are organized under the five priority research areas being used by the Water Subcommittee: Integrated Monitoring, Water Availability and Flow, Water Quality and Aquatic Ecosystem Functions, Ecological Restoration and Rehabilitation, and Predictive Systems Management. This was difficult in many cases because many of the environmental problems discussed in Chapter 2 should be considered in more than one of the five priority areas. Scientific approaches such as observation and prediction systems, process-oriented ecosystem research, and modeling can also be relevant to more than one of the five areas.

For the purposes of this evaluation, the committee considered the contributions of science to include not only research activities but also monitoring, modeling, and assessment. The committee determined the scientific priorities on the basis of the following criteria:

• seriousness of the environmental problems to which the science is relevant;

• relevance to the Water Subcommittee's goals within the broader framework of the National Science and Technology Council (NSTC);

• opportunity for significant advancement in scientific understanding and/or application of scientific knowledge and information, and

• potential for linkage of science across the land-sea interface.

To maximize the results of the research recommended below, integration will need to occur across scientific disciplines, across government agencies at various levels, and between science and management.

Committee on Environment and Natural Resources Research (CENR) Objective : Develop and implement observation systems that focus on interactions among atmosphere, land, and water dynamics at time and space scales relevant to ecological, physical, and socioeconomic processes.

Monitoring of coastal environments is conducted for a range of purposes, including ensuring compliance with discharge and construction permits, long-term measurement of environmental status and trends, verification of predictive models, and determination of the effects of ecosystem restoration or rehabilitation (NRC, 1990a). There are widely held concerns, however, about the degree to which monitoring programs are effectively designed; relationships to research and modeling; whether the most appropriate ecosystem properties and processes are being monitored; the extent to which results are actually used in environmental management; and the degree of integration among local, regional, and national monitoring efforts (NRC, 1990a).

The need for effective monitoring of toxic chemicals in coastal waters has been documented extensively (NRC, 1990a). Prototype national programs such as the Environmental Protection Agency's (EPA) Mussel Watch Program led to the National Status and Trends (NS&T) program conducted by the National Oceanic and Atmospheric Administration (NOAA) (O'Connor and Ehler, 1991). This program has assessed a suite of chemical contaminants in bivalves, fish, and surface sediments at about 100 locations around the U.S. coast at approximately one-year intervals for the past six to seven years. There are numerous local and state programs that also measure chemical contaminants for a variety of reasons (NRC, 1990a,b). Some of these programs include assessments of the biological effects of toxicants on individual populations. Results have shown that several of the standard metal and organic contaminants monitored have declined in concentration in the tissues of organisms as a result of reductions in the sources of these contaminants (e.g., phasing out of lead additives in gasoline, discontinuation of the use of some organochloride pesticides, and improved waste treatment).

More recently, the EPA has developed and conducted pilot studies for the estuarine component of its Environmental Monitoring and Assessment Program (EMAP). EMAP-Estuaries includes measurement of contaminants in the environment and in coastal marine organisms, as well as activities that assess the

condition of benthic and fish communities and relate this to other environmental variables, including salinity, sediments, and dissolved oxygen. Because both the NS&T program and EMAP sample infrequently and depend primarily on relatively static properties, these national programs, and indeed most local and regional programs, are not able to conduct adequate monitoring of the trends and effects of eutrophication, which is a problem widely recognized by many coastal scientists as the most important marine pollution problem. Monitoring programs have demonstrated that contamination by at least some toxic compounds is declining, presumably as a result of regulatory activities, while contamination by other toxic chemicals remained the same over the period of 1986 to 1990 (O'Connor et al., 1994). Monitoring programs must be designed to answer specific questions and the use of the data must be determined before it is collected. Different monitoring programs may be needed, depending on the questions posed. Multiple programs should be coordinated, but not necessarily combined, if their purposes differ.

A high-priority science need is to establish a linked regional-national monitoring program to evaluate the present extent of eutrophication and to observe future trends. Because the environmental indicators and effects of eutrophication are variable over time, such a monitoring program poses significant challenges with regard to measurements of biological and functional indicators (which may consist of measurements of rates rather than absolute properties). The sections below describe important aspects of monitoring, such as measuring diffuse inputs, developing indicators of biological status and processes, and employing in situ and remote sensing. Monitoring of diffuse inputs and development of new indicators are necessary for studying both eutrophication and toxic effects.

There is a need for long-term and uniform monitoring of inputs, particularly of nutrients, to coastal waters from rivers and the atmosphere. This is particularly true in regions where coastal eutrophication may be a problem, such as sensitive areas on the continental shelf (such as the New York Bight), offshore the Mississippi Delta, and in selected estuaries. Although data on material fluxes into coastal waters are collected by federal, state, and local agencies, few (if any) data sets include total inputs of such critically important nutrients as nitrogen (see Box 3), phosphorus, and silicon. Usually, only inorganic nutrients are measured, requiring tenuous extrapolation to estimate total (inorganic plus organic) nutrient fluxes (Meybeck, 1982, 1988; Turner and Rabalais, 1991; Howarth et al., in press). The use of widely accepted standardized techniques (including quality control and quality assurance methods) should be promoted to enable comparisons among regions. Coastal measurement programs should be linked with surface and groundwater monitoring programs conducted

inland. Many data that are presently available may be adequate for discerning relative changes compared with other measurements made by the same program, but cannot be compared with data collected by other monitoring programs.

Trends in diffuse inputs (i.e., nonpoint source inputs) and ecosystem responses may also be inferred retrospectively from geochemical and paleonto-

logical analyses of sediment cores (Cooper and Brush, 1991; Turner and Rabalais, 1994; Eadie et al., in press). Such analyses should be expanded and used to complement the monitoring of contemporary inputs.

Indicators of biological effects and ecosystem functions are not well developed for use in monitoring programs, but are essential for understanding how ecosystems respond to a variety of factors (NRC, 1994d). In the past, benthic communities have been monitored because they represent stationary, relatively long-lived sentinels. However, benthic species integrate all the environmental stresses to which they are subject, so that the relationship of observed changes to a specific environmental stressor is not always clear (Howarth, 1991). EMAP-Estuaries relies heavily on indices of benthic community integrity, but it is difficult to apply a universal index over a broad array of environments. And, the effects of environmental stresses may not be evident until conditions become severely stressful to the indicator organisms. Clearly, more sensitive biological indicators are required to monitor the effects of environmental factors before they become extreme, to monitor water column as well as benthic communities, and to monitor incremental ecosystem recovery. Further development of indicators involving benthic macroalgae (Fujita, 1985; McGlathery, 1992) is also needed. The proper selection of indicators requires some understanding of material flow and species composition within an ecosystem and indicators may differ among different environmental stressors (e.g., between eutrophication and toxic contaminants). Thus, the selected indicator must be indicative of the ecosystem or process being studied.

Another important priority for improving monitoring of eutrophication effects is the development of spectrophotometric, molecular, and remotesensing techniques that reflect trends in phytoplankton communities. Rate measurements of important biological and biogeochemical processes show promise for use in monitoring because they may be sensitive to changes in the input of nutrients. Regular observations of the rates of biological and biogeochemical processes could also be useful for research and ecosystem water quality modeling purposes.

There is also a need for development of indicators that (1) detect short-term sublethal effects of toxic substances, as well as manifestations of longer-term chronic lethal effects, and (2) measure and predict biological availability of toxic chemicals in water and sediments. The focus should be on toxic substances that may alter reproductive processes at nanomolar concentrations; are of growing, rather than declining, prevalence in the environment; or are the target of hazardous waste remediation. Such measurements are technically feasible, although their cost would suit them for research purposes

and monthly or annual monitoring, rather than monitoring at more frequent time intervals.

Environmental responses to natural and human influences must be monitored at the appropriate space and time scales, developing monitoring instruments and systems based on indicators such as those recommended in the previous section. For eutrophication phenomena, relevant time scales range from hours to decades. Nutrient and oxygen concentrations and plankton biomass and metabolism can vary greatly over short time frames as a result of daily cycles, winds, and tides, whereas changes in community structure may only be evident over several decades. To improve understanding of many of the large-scale issues described in Chapter 2, the relevant spatial scales for monitoring, research, and modeling may extend over hundreds of kilometers. In order to monitor over short time scales and large areas, cost-effective means need to be developed for near-continuous in situ sensing and for remote sensing. This will require further development of reliable in situ autonomous and towed sensors of variables such as dissolved oxygen, temperature, salinity, nutrient and toxic substance concentrations, uptake rates, and measures of phytoplankton concentrations, activity, and general taxonomy, through the application of such meth-

ods as photometrics, acoustics, molecular biology, and fiber optics (NRC, 1993e). Improvements of platforms for coastal science will also be needed (see Box 4).

A major recommendation of the National Research Council's (NRC) (1990a) assessment of marine environmental monitoring is that effective coordination and linkages should be established among the national marine environmental monitoring programs (particularly NS&T and EMAP-Estuaries). These programs aim to provide a national assessment of environmental conditions and trends through regional monitoring, as well as location-specific assessments, for use in environmental protection, restoration, and resource management. Although there has been some progress in coordinating NS&T and EMAP, much remains to be accomplished to achieve the goal of fully integrated monitoring. EMAP-Estuaries has not been able to follow the timetable originally planned. Thus far, pilot studies have been conducted only in the Virginian (Mid-Atlantic) and Louisianan (Gulf of Mexico) provinces. Most monitoring programs of this type have been implemented at less than their full designed scope, due to funding limitations. Unfortunately, scaled-back programs may not meet the original objectives. Controversies have raged within the scientific and management communities about EMAP's design, measurements, interpretation, sustained effort, and relevance.

In 1992 Congress established the National Marine Monitoring Program under Title V of the NOAA Authorization Act, providing a statutory and institutional mechanism to stimulate regional marine monitoring where it would be useful and to provide incentives for these regional programs to contribute to a national marine monitoring program. However, funding to implement this program has not been appropriated. The Intergovernmental Task Force on Monitoring Water Quality (1993), which considered monitoring of inland as well as coastal waters, concluded that the lack of coordination in (1) monitoring program design, (2) selecting monitoring stations and parameters, (3) comparable sample collection and analytical methods, and (4) data-sharing formats and protocols have hindered sharing of data and the development of comprehensive information on water quality and environmental conditions. Efforts to improve comparability and utility of monitoring data among national monitoring programs and among national, regional, and local monitoring programs should be a high priority for the CENR Water Subcommittee. It is clear that, in order to be successful, the efforts will need to be based on a partnership with regional and state authorities and regulated industries and utilities. The committee is uncertain of the extent to which the federal agencies have explored the partnership issue with such entities. This is a major concern because the effort required from these partners will be substantial and cannot be assumed automatically.

The NRC (1990a) assessment of marine environmental monitoring noted that much, if not most, monitoring was conducted without a clear objective for use of the results. Data are frequently laboriously collected and stored but not synthesized or used in decision-making, often because the problem to be resolved has not been defined adequately. The NRC (1990a) recommended that monitoring should be considered as part of the environmental management system and designed based on rigorous criteria of sensitivity and utility. Federal and state agencies engaged in coastal science can improve the use of monitoring in management systems through support of (1) education and training in monitoring design, technologies, analysis, and implementation; (2) development of techniques to make monitoring results more understandable and useful to managers (e.g., computer visualization); (3) efforts to link research and predictive modeling with the design and interpretation of monitoring; and (4) feedback from management in an adaptive management strategy. The adequacy and usefulness of data from monitoring programs need to be assessed regularly.

A key to understanding many coastal environmental problems, including eutrophication and toxic contaminant pollution, is the ability to detect subtle inter-interannual changes in water and habitat quality and in ecosystem structure and function. Detection of long-term (decadal or more) trends in degradation of water and habitat quality requires long-term observations with frequent sampling (greater than twice per year) of key parameters of water and sediment chemistry, as well as ecological properties and processes. Furthermore, when process-oriented research is coupled with these long-term data collection programs, a deeper understanding of the fundamental mechanisms controlling ecosystem dynamics will emerge.

As described in the CENR Water Subcommittee strategy and implementation plan, this priority research area deals with freshwater resources in surface water and groundwater and the effects of modified hydrology within watersheds (CENR, 1994c). From the coastal perspective, not only are the movement and changes of water and materials within watersheds important, but the effects of water flow within coastal ecosystems also must be considered.

Quantification of the movement of water within the watershed and its influ-

ence on the flux of important materials (nutrients, toxicants, and sediment) to coastal ecosystems is a new and timely integrating factor that is bringing together the disparate scientific and management communities working on issues addressed by the Water Subcommittee. Meteorology, atmospheric chemistry, hydrology, freshwater biology and chemistry, agricultural sciences, and the marine sciences must be involved. The growing recognition of the effects of diffuse sources of pollutants from land on coastal ecosystems, particularly in causing eutrophication, provides another important reason to understand hydrology and materials flux for entire watersheds.

To achieve an understanding of the effects of water availability and flow on ecosystem integrity, research should focus on understanding (1) the coupling between watershed hydrology and materials flux and (2) physical forcing processes within the coastal environment. Such research is the foundation for understanding how coastal ecosystems function and for the development of atmosphere-watershed-coastal ecosystem models.

An important goal is to develop a quantitative understanding of the dynamics of the complex basins that drain into coastal waters and to use this improved understanding to manage land use to minimize adverse impacts on coastal ecosystems. The watershed is a logical geographic unit for research and land management, primarily because components of a given watershed are linked by the natural movement of water. The hydrology of both surface water and groundwater must be understood as a unit. Often, only about 5 percent of the annual precipitation actually moves downhill on the land surface and quickly reaches the nearest stream as a brief pulse. In contrast, groundwater that percolates slowly from the uplands of the watershed can enter coastal waters directly or through streams. Groundwater may be entrained in an aquifer for weeks to centuries before discharge, depending on the geology and size of the aquifer. Therefore, movement of both surface and groundwaters must be included in estimates of water and material fluxes.

It is important also to measure atmospheric deposition, both that which occurs as the washout of pollutants during precipitation events (wet deposition) and that which occurs as direct deposition of gases, aerosols, and particulates between storm events (dry deposition). The types of pollutants and the magnitude of the discharges per unit land area differ depending on land use. Thus, an unperturbed forest usually discharges relatively little nitrate, whereas an area of intensive agriculture usually discharges high amounts of nitrate (NRC, 1993d). Even in the case of forests, however, there is a pressing need to understand if and why forests become ''nitrogen saturated" as the result of large inputs of nitrate and ammonium from the atmosphere.

The output of a particular piece of land is a function of land characteristics

and use (e.g., farming practices), weather and climate conditions, urbanization, and soil characteristics. The delivery of contaminants will depend not only on land use but also on how overland storm flows and groundwater percolating from the fields interact with other habitats before discharge into the stream channel network. Thus, the positions of land-use patches on the watershed as well as their types and proportions are important. For example, riverside (riparian) forest buffers can remove over 80 percent of the nitrate in shallow groundwaters percolating through them from agricultural uplands to headwater stream channels (Lowrance et al., 1984; Cooper et al., 1987; Correll and Weller, 1989). In this case, the riparian forest acts as a sink for nitrate generated in upland agricultural areas. Obviously, such nitrate removal would not occur if the positions of the forest and agricultural areas were reversed. Finally, when a large watershed is composed of many streams, rivers, wetlands, and lakes, contaminants may be removed in each type of water body, and interactions among these hydrological elements must be known adequately to understand watershed-level processes.

Scientific priorities for coupling watershed hydrology with material fluxes to coastal waters include (1) high spatial resolution remote sensing; (2) improved geographic information systems, including detailed data on such factors as topography, geology, soils, land use, stream channels, aquifer recharge and discharge areas, wetlands, flow, residence, and land-use practices; (3) research on land-use management for reducing contaminant inputs to surface and groundwaters (NRC, 1993d); (4) research on the hydrological connections between surface and groundwaters; (5) investigation of material fluxes operating over scales ranging from small experimental plots to large watersheds of mixed land use; and (6) study of controls on export of nitrogen from forests, particularly as a function of atmospheric deposition and forest age.

Ecosystem modeling is considered in this section for coherence, but it has equal relevance to ecosystem function and habitat restoration and management, as discussed in subsequent sections. Calibrated and verified landscape and ecosystem models can be used to identify unknown ecosystem processes, synthesize research findings, and serve as management tools. These models must couple watersheds to both atmospheric and coastal ecosystem models and must integrate spatially explicit patterns with transfer, transformation, and response processes. It is especially important to develop models that can use, on an ongoing basis, data from all the important disciplines of coastal science. Present generation models often lack this capability, using data sets limited in time or in the type of variables they include. Coupled atmosphere-watershed-coastal models will need to include such a multidisciplinary data handling ability. Coupled models should not be developed as ends in themselves but as tools to organize

thoughts, identify important research questions, and quantitatively represent understanding of how complex systems work in a way that promotes more informed management decisions. Models should always be treated with some skepticism and be improved continuously and are often a critical and integral part of research and monitoring. Models must be improved to use data of many types, such as observations of the distributions of elemental and isotopic tracers and data from episodic events.

Watershed models need to be structured differently from simple hill slope or ecosystem models appropriate to one landscape patch. Appropriate models must quantitatively route water through the surface and subsurface of the watershed and accurately predict the transport and removal of pollutants along those hydrologic pathways. The models must account for the movement of water from one type of landscape patch to another along its trajectory. The models also need to include the effects of irrigation, in-stream processes, and the effects of lakes and reservoirs. Finally, the models must include the inputs to each watershed patch of contaminants from the atmosphere, point sources, and land-use practices.

At present, adequate estimates of the turnover times of many important aquifers are not available. More accurate measures of the volumes of water following various pathways through watersheds are needed, including direct and realistic measurement of evapotranspiration. It is important that the watershed models under development be calibrated with accurate and detailed data from each region. It is equally essential that these calibrated regional watershed models be verified with other data so that their transferability to other watersheds and other contaminants can be assessed.

Atmosphere-watershed models should be linked with hydrodynamic-ecosystem models of coastal receiving waters. A priority for such ecosystem modeling research is "to establish improved predictive capabilities for coastal ocean systems that link physical processes, biogeochemical cycles, and the interactions of living marine resources" (SUSCOS, 1993b). Many of the key frontiers in ecosystem modeling research are at the interfaces between existing modeling efforts. Coastal ecosystem modeling that focuses on the following topics would improve scientific understanding and management of the nation's coastal ecosystems:

1. Coupled models relating physical, biological, and biogeochemical processes in coastal environments need to be expanded and improved to include ecological feedback processes (e.g., coupling of benthic and water-column processes, nutrient recycling, top-down control from higher trophic levels), suspended sediment dynamics, and benthic biogeochemistry. Most existing models are primitive in their structures, being generally limited to: (a) simple nutrientalgal-oxygen interactions in conventional water quality models; and (b) simple phytoplankton-zooplankton-nutrient models commonly used to simulate basic processes in pelagic regions of the ocean. Special attention should be given to

the development of methods for coupling biological and physical models at time and space scales that are coherent and appropriate for simulation of fundamental biological and physical processes. It is also important for modelers to develop a clear understanding of the scale dependency of simulated ecosystem behaviors; that is, the extent to which models that use different time and space scales produce different simulations of the same processes and system scenarios.

2. Ecosystem process models should be coupled with population dynamics models, with special attention given to scale coherence. Traditionally, ecosystem process models have been used to address questions of how nutrient and other anthropogenic inputs to coastal systems affect water and habitat quality and overall ecosystem functions. Conventional population models of fish and other living resources have been used to analyze relations between fishery harvests and fish abundance. A growing recognition of key interactions between water and habitat quality and fish abundance, however, suggests that these two separate modeling approaches need to converge into an integrated framework. Moreover, the effect of physical transport processes on both recruitment into adult fish stocks and nutrient/trophic interactions reveals that these integrated process-population models should be driven by physical circulation models. Ecological feedback effects and material exchanges could be coupled between process and population models with simple seamless algorithms.

3. As a consequence of varying climate regime, geomorphology, hydrologic input, and hydrodynamic circulation, there is a great diversity of ecosystem types along the coasts of North America. Prudent cost-effective approaches to management of these ecosystems suggests the need for a systematic approach that recognizes similarities and differences among them. One component of this strategy would be to develop general conceptual models of coastal systems to facilitate cross-system comparative analyses of monitoring and ecological process data (see p. 47). These models could be modified systematically to represent different broad classes of coastal environments.

4. The development of ecosystem models capable of using data from remote and in situ observation systems is an important objective, particularly models that permit real-time comparisons between model predictions and observations. Techniques should be developed for using time-series data in model simulations as a diagnostic tool for model analysis and assessment to improve environmental forecasting (see pp. 54 and 55).

As water moves from the highest headwaters to the edge of the continental shelf, there is a continuity of processes and fluxes across coastal ecosystems. The task of understanding integrated terrestrial and coastal processes is daunt-

ing. The relevant time and space scales are relatively short compared with processes that operate on a global scale. The processes are three-dimensional and vary with the site. For example, major sources of nutrients are riverine in the Gulf of Mexico, whereas the source in the Gulf of Alaska is the deep ocean. Differences in the relative importance of wind forcing, freshwater fluxes, tides, topography, currents, air-sea interactions, storm events, and longer-term forcing, such as the El Niño-Southern Oscillation, are found in different regions.

The continuum of physical flow requires the development of an integrated understanding of the physical forcing upon coastal systems from the terrestrial, oceanic, lake, and atmospheric factors mentioned above. The fluxes of salt, heat, and buoyancy through the coastal ecosystem should be studied. These fluxes can vary on time scales from minutes to decades. After acquiring sufficient observations and research results, these fluxes should be modeled on a wide variety of temporal and spatial scales. Similar to actual processes, these models should be continuous across the coastal ecosystem.

Hydrological models have been developed to study variability in riverine inputs to estuaries (Howarth et al., 1991; Naiman et al., in press), but these have not been coupled with large-scale atmospheric models to address the effects of climate on coastal systems. Such coupled models should be developed. Because atmospheric models depend on sea surface conditions offshore, these are potentially important feedback mechanisms to terrestrial hydrology.

Coastal tides and alongshore currents, because they are relatively large and observable, have been the focus of successful research and modeling efforts. Coastal ocean areas can be relatively isolated from the open ocean, particularly bays, estuaries, and regions with broad continental shelves. In such areas the fluxes of materials off the shelf tend to be smaller than along the shelf. Cross-shelf fluxes are poorly understood and the quantification of such fluxes is a central objective of both the National Science Foundation (NSF) Coastal Ocean Processes (CoOP) initiative (Brink et al., 1992) and the Department of Energy (DOE) Ocean Margins Program (Jahnke et al., 1994). Cross-shelf exchange can be strongly driven by winds, producing coastal upwelling and associated influx of nutrients to surface waters and increased primary productivity.

Atmospheric effects on the coastal ocean and Great Lakes include previously mentioned effects on inland hydrologic cycles, as well as winds and storm events (see NRC, 1992e for a description of research priorities in coastal meteorology). The contrast in temperatures between land and water that occurs in coastal areas creates a variety of atmospheric conditions that are unique to the coast and that drive physical processes in coastal waters (Brink et al., 1992). Winds can enhance the flux of momentum and gases between coastal waters and the atmosphere. Tides and currents provide a means of structuring shallow-water benthic communities and affect the distribution of food, gametes, larvae, and planktonic organisms. Finally, storm events provide episodic physical forcing that can have extreme effects on coastal ecosystems by decreasing salinity

due to increased precipitation, increasing turbidity as bottom sediments are resuspended, and even destroying habitats and organisms.

No single federal agency is responsible for studies of physical forcing in coastal systems, and better cooperation is needed among the agencies, perhaps with CoOP as an integrating and coordinating mechanism, at least for continental shelf studies. Coordination should include research, observations, data exchanges, and modeling efforts. Also, investigations should relate physical forcing to important ecosystem processes, population dynamics, organismal health and survival, and production of living resources.

Inputs of nutrients, organic materials, toxic contaminants, and sediment can degrade water quality sufficiently to affect coastal ecosystem functions. These inputs may be direct waste discharges into the coastal waters or from nonpoint sources delivered from the watershed or the atmosphere. Additionally, these materials may be remobilized by human activities taking place within the coastal system itself (e.g., dredging).

Specific scientific priorities within this research area focus on (1) relating coastal nutrient fluxes to ecosystem dynamics, (2) conducting strategic scientific assessments of toxic effects, (3) understanding the role of sediments in coastal ecosystems, (4) relating resource use to ecosystem sustainability, (5) assessing the impact of multiple stressors, and (6) promoting comparative ecosystem science.

The importance of understanding the flux of nutrients within and between coastal ecosystems is not only due to trends toward greater eutrophication but also to the fact that nutrients are an important regulator of normal biological activity within these systems. Understanding the transformations of nutrients and organic matter and their relationship to ecosystem dynamics would help in studies of the interactions between adjacent coastal ecosystems and provide a basis for comparing the responses of coastal ecosystems in different geographic areas.

Physical models are necessary to describe the spatial and temporal variability of these nutrient fluxes (see previous section). The potentially important time

scales of physical processes include hourly (tidal forcing), daily, seasonal, annual, and interannual. The models should begin at the head of the estuary to include the stream and riverine influences on physical, chemical, and biological parameters of the estuary and consider the retention of materials within the estuary.

The exchange of material between nearshore regions and estuaries depends on the freshwater flow rates, coastal circulation, tides, waves, winds, bottom topography, and chemical and biological transformations of materials of interest. These materials are not passive, conservative, or all transported in the same manner. Understanding of the coupling between physical, chemical, and biological fluxes requires basic research on transformations and coupling processes. Biological productivity in coastal waters leading to eutrophication depends not only on the supply of nutrients but also on the ability of organisms to remain in the euphotic zone. This ability is dependent on the vertical density structure (stratification) of the water column, which is related to the temperature and salinity structure, which, in turn are related to the fluxes of heat and salt (i.e., hydrology). Therefore, coupled models of physical, chemical, and biological processes are necessary to understand coastal nutrient dynamics and eutrophication phenomena (see p. 39). A strategy needs to be developed whereby generic models can be made specifically applicable to the wide range of coastal ecosystems across the nation.

Aspects of ecosystem dynamics that deserve attention as they relate to ecosystem nutrient fluxes include (1) the regulation of primary productivity and the algal and plant species responsible for it, (2) the relationship of nutrient enrichment to harmful algal blooms, (3) the flux of organic matter through the pycnocline and to the seabed, (4) rates of nutrient remineralization, (5) the effects of top-down control of algal and plant species by filter feeders and grazers, (6) food chain structure, and (7) the consequences of nutrient enrichment on harvested species.

There are a number of existing and planned programs that could individually make contributions to those goals, but they could make even greater contributions if they were better integrated and coordinated. NSF's Land-Margin Ecosystem Research Program (LMER; see LMER Coordinating Committee, 1992) supports investigator-initiated, multidisciplinary studies of estuaries and coastal embayments, all of which include investigations of nutrient transformations and biological responses. The CoOP program (Brink et al., 1992), which is funded primarily by NSF, has developed a plan to address the issues of cross-shelf transport of contaminants and nutrients; early projects deal with nearshore processes on a topographically simple shelf and with wind-driven currents. NOAA's Coastal Ocean Program (NOAA, 1991), through its Coastal Ecosystems Health theme, has addressed aspects of coastal eutrophication (e.g., in its Nutrient Enhanced Coastal Ocean Productivity study off the Mississippi Delta and the Atmospheric Nutrient Inputs to Coastal Areas projects). Also, DOE is

launching a comprehensive study of nutrient dynamics and offshelf transport of organic material off Cape Hatteras (Jahnke et al., 1994).

Scientific research and monitoring of the coastal ocean and Great Lakes and use of this information for policy and management decisions have reduced the impacts of toxic chemicals on human health and life. In the late 1950s and early 1960s the marine scientific community undertook studies and provided assessments that informed policy-makers and managers (and the general public) about the dangers associated with indiscriminate release of artificial radionuclides to coastal waters. In the late 1960s and early 1970s further losses of birds and other organisms were avoided once the widespread nature and severity of the effects of dichlorodiphenyltrichloroethane (DDT) and other chlorinated pesticides were identified.

These successes in preventing acute toxicity should not be cause for complacency. Concerns have shifted from chemical substances of immediate toxicity (within hours to days) to substances with longer-term (years to decades), more subtle, but serious, effects such as mutagenicity, carcinogenicity, and impairment of reproductive activity (see Box 2). Many of these effects are manifested through antagonistic and/or synergistic interactions among substances. Although there is clear evidence of the decrease in concentrations of some toxic synthetic chemicals and human mobilized chemicals [e.g., DDT, polychlorinated biphenyls (PCBs), lead] in organisms and sediments of many coastal areas, inputs to coastal waters of chemical contaminants of known or suspected adverse effects continue (e.g., polycyclic aromatic hydrocarbons), and there are many places where significant reservoirs of chemical contaminants (e.g., PCBs) exist in sediment. These reservoirs of past inputs and the continuing inputs of toxic chemicals pose threats to coastal ecosystems and to human health by transfer through aquatic food chains (Dawe, 1991; IOM, 1991; Hamelink et al., 1994). The fate, transformation, and bioavailability of various contaminants vary widely as a result of differences in their reactivities and uptake and incorporation into organisms (e.g., Fisher et al., 1991; Fisher and Reinfelder, 1991).

Scientific assessments should focus on the most important, strategic, components of toxic effects questions, particularly on compounds of continuing or emerging concern, realistic exposure conditions, chronic rather than acute exposures, and experimental determination of sublethal and ecosystem-level effects. Poorly quantified estimates of atmospheric and groundwater inputs of toxic chemicals should be improved. Research should be directed to improve understanding of the synergistic and antagonistic effects of toxic chemicals such as carcinogenic and mutagenic effects on aquatic organisms (Dawe, 1991; IOM, 1991; Colburn et al., 1993). Further development, application, and long-term testing of models that link physical transport processes, biogeochemical trans-

formations, biological uptake, and effects on organisms are needed for multiple toxic chemicals entering, or already present in, coastal ecosystems.

Coastal ecosystems differ from those in the open ocean in a number of important respects. Perhaps the most profound difference is the functional importance of the seabed and suspended sediments in coastal ecosystem processes. Not only is there a continuous supply of sediment from rivers and shoreline erosion, but coastal ecosystems are characterized by the frequent resuspension of deposited sediment and the prevalence of biologically created or modified particulates in the form of organic-mineral aggregates, calcareous or siliceous skeletons, and fecal pellets. Coastal sediments, both deposited and suspended, play a number of important roles in coastal ecosystem function. They (1) affect light availability; (2) are sites of intense activity by associated microbes; (3) greatly influence the transport and fate of chemical constituents, including organic and inorganic contaminants; (4) accumulate the ecosystem's waste products; (5) are important in the cycling of biologically important materials, including carbon, nutrients, and sulfur; and (6) determine living conditions for benthic organisms.

Scientific investigations of the processes that affect the transport, deposition, and resuspension of sediments are seldom included in studies of coastal ecosystems; such "geological" studies are generally pursued separately. Basic studies of sediment dynamics, particularly of fine sediments, in different coastal environments should be expanded and integrated with studies of ecosystem function (ASLO/ERF/SAML, 1990; USGS, 1994). To a certain degree, programs such as LMER (in which the Columbia River study focuses on the role of the turbidity maximum zone in the estuary), CoOP (Brink et al., 1992), the DOE Ocean Margins Program (Jahnke et al., 1994), and certain elements of the U.S. Geological Survey's National Marine and Coastal Geology Program (USGS, 1994) attempt to integrate sediment processes into ecosystem studies.

More concerted research is needed on such subjects as (1) particle and biogeochemical dynamics of the turbidity maximum zones, (2) soil-building processes in coastal wetlands, (3) how sediment dynamics influence biogeochemical fluxes across the sediment-water interface, and (4) sediment accumulation and mixing rates.

Important questions for relating the human exploitation of living resources to the integrity of coastal ecosystems that sustain those resources concern the effects of harvesting activities on habitats and nontarget species and the effects of selective removal of the target species, which are often mid- to top-level predators in their ecosystems. For example, there is ample evidence from fresh-

water ecosystems that modification of populations of intermediate or top predators will cause changes in the population structure of the prey organisms. There are also examples of marine ecosystem modifications that result from such top-down effects (see Box 5). The top components of marine and Great Lakes food webs have been severely modified in recent history by fishing. Examples (described in Chapter 2) include groundfish species on Georges Bank, oysters in the Chesapeake Bay, and sea otters in the Pacific Ocean. What effects have the severe depletions or effective removal of these consumers had on the structure and function of the food webs that supported them? In some cases the niches occupied by these organisms have, to some degree, been reoccupied by other species. This might suggest that there should be no significant effects at lower trophic levels. In other cases, however, there has not been an obvious replacement (e.g., for menhaden), and lower trophic levels might be modified as a consequence.

Coastal waters exhibit tremendous spatial and temporal variability, with the additional complexity that different scales are important for different processes or groups of organisms. If the effects of overexploitation on the structure and composition of coastal ecosystems are to be determined, studies will need to be planned in the context of this variability.

Research should be undertaken to determine the effects of exploitation of living resources on ecosystem structure and function. This research should strive to (1) increase understanding of ecosystem controls at the population level (see p. 40 for a discussion of the integration of population dynamics and ecosystem function) and (2) increase understanding of causes of variability.

Attempts to identify the ranges and important scales of variability have

often been made through routine monitoring. In some cases these efforts have yielded correlations that provide significant insights. For example, fisheries recruitment is often correlated with some physical or meteorological parameter, suggesting mechanisms by which recruitment may be controlled. These correlative approaches do not, however, provide understanding of the underlying causes of interannual variability of populations, which should be investigated through process-oriented research.

Our approaches to understanding and managing the effects of water quality on coastal ecosystem function are too often highly reductionist (e.g., determining the effects of a single contaminant on a specific target organism under laboratory conditions that hold other variables constant) or too vaguely holistic (e.g., monitoring ecosystem health through abiotic and biotic condition indicators as is done by EMAP). To improve predictions for use in environmental management, the effects of multiple stressors must be assessed in ways that differentiate the individual and collective effects of stressors. The consequences of these effects must be applied over a range of the spatial and temporal scales. This notion is at the heart of understanding what are often called cumulative impacts.

Mitigation of multiple-stressor effects requires research to determine the most important factors, how they interact, and over what scales their effects act. The distribution, concentrations, bioavailability, and fate of toxic organic materials are related to the degree of eutrophication present in a coastal system. Integrated monitoring provides a means to study toxic contaminants, eutrophication, and other coastal environmental problems as they relate to one another, instead of maintaining the artificial boundaries that now exist between these problems.

Federal sponsorship for research on multiple stressors and scales—admittedly a daunting challenge—has been meager and should be increased. This high-risk, but essential, research must involve the interaction of diverse disciplines ranging from toxicology to ecology to modeling and must link observational, experimental, and theoretical studies.

Coastal ecosystems vary greatly in their sensitivity to nutrient enrichment, habitat modification, resource exploitation, and toxic chemical stress. Yet we lack a sound theoretical and practical basis for extending results from one location to another or even a framework for comparing ecosystem responses in a way that would enhance environmental protection and management. The increasing emphasis on place-based ecosystem management, rather than on uniform, single medium regulation (Gore, 1993) requires, ironically, more comparative ap-

proaches to coastal ecosystem science rather than fewer. We simply cannot afford to study each place with the same level of intensity and must rely on results extended from the broader base of knowledge (Cole et al., 1991).

Unfortunately, the way coastal science is supported and conducted has not been conducive to the development of comparative coastal science. Much of the support for science has been applied to regionally focused efforts. For reasons related to sponsorship as much as to logistics, coastal scientists tend to work near their home institutions. There has been relatively little support or encouragement from federal agencies for the advancement of comparative coastal science. For example, NSF has only just begun to treat the individual LMER projects in a way that should stimulate the rich opportunities, both for science and its applications, that would come from an implicitly comparative approach. The NSF Long-Term Ecosystem Research (LTER) sites have attempted integration, achieving mixed success. The committee believes that the national coastal science community has recognized this major shortcoming and is now prepared to work across parochial boundaries.

The federal agencies should promote the development of comparative ecosystem science by supporting comparative studies and synthesis and by encouraging cross-fertilization through interregional research collaboration among coastal scientists. Such efforts must be conducted over an extended period of time. Extending the results from one region to other regions requires that the principles determining ecosystem structure and function be understood. Simple characterization of a system is insufficient to provide understanding that is transportable to another system. Long-term projects such as those supported by NSF's LTER and LMER programs should be expanded—with greater emphasis on ecosystem comparisons—because they seek understanding of ecosystem principles (Likens, 1989).

In the past, most scientific efforts directed to the protection of coastal ecosystem integrity have focused on demonstrating threats or determining the precautions needed to protect these ecosystems. However, because of widespread degradation of coastal ecosystems and the extensive modification of coastal habitats, it is now necessary or desirable in many cases to rehabilitate coastal ecosystems through active intervention. Protection against future impacts is simply not sufficient to ensure ecosystem integrity. The growing emphasis on restoration poses significant challenges to the scientific community to provide guidance for

effective restoration, as well as to environmental management, which is generally oriented toward regulatory protection rather than active restoration.

It is easier to describe what caused an ecosystem to degrade than to prescribe what should be done to restore it to a former or otherwise improved condition. Thus, recreating the original conditions may result in a different outcome than the original ecosystem. Although understanding the causes of ecosystem degradation provides some information useful for restoration, much additional knowledge is required to determine how the environment can be manipulated to accomplish the desired restoration goals.

The committee organized its assessment of science priorities for coastal ecosystem restoration and rehabilitation within three themes: (1) determining the effects of habitat loss and degradation on biodiversity and productivity, (2) advancing scientific approaches to restoration, and (3) guiding the remediation of toxic contamination.

Extensive efforts over the past decade have documented the extent and rate of loss of certain important coastal habitats, such as wetlands, seagrass beds, and coral reefs. Other physical changes to habitats, such as siltation of bottom habitats and alterations to salinity and flow patterns, are less well documented. To guide restoration and rehabilitation, it is not only necessary to chart habitat loss and degradation but also to understand the effects of this degradation on living resources and biodiversity. Research agendas have been offered on this subject for biologically generated habitats such as wetlands (NRC, 1992c), seagrass beds, and reefs (NRC, 1994b). Habitat changes due to altered hydrodynamics and the introduction of nonindigenous species also must be considered. For example, diversion of fresh water increases the salinity of coastal marine ecosystems and can diminish the supply of sediment and nutrients to coastal systems (NRC, 1993c; Boesch et al., 1994)

Estuarine and coastal habitat restoration will be influenced by present and future modifications in freshwater flow into the coastal zone. In particular, salinization and sediment input have a marked influence on the type and rate of plant establishment in wetland restoration projects. Nutrient and contaminant loading levels affect subtidal habitat and submerged vegetation by increasing turbidity or toxicity. To date, most restoration efforts have been short term and are planned assuming steady-state conditions. Substantial effort is needed to understand how to restore habitats in the coastal zone under various water management scenarios through seeking improved knowledge about the interactions of hydrological changes with establishment, succession, and function of restored coastal ecosystems. Research should address restoration rates and variation in wetland functions under modified hydrological regimes such as wet-dry cycles,

altered tidal regimes, and different rates of sediment and nutrient loading. Greater emphasis is needed on how water management affects coastal habitats and on how to integrate anthropogenic changes into long-term management of coastal habitats. Research should be supported for studies of the role of water management in controlling physical, chemical, and biological processes within the coastal zone and the examination of means by which large-scale water management can be incorporated into coastal restoration.

Coupled with research that relates habitat condition to water flow, research should be conducted on species and community requirements under conditions of managed hydrology. Floral and faunal responses to managed hydrological regimes need to be understood, especially if restoration is to succeed in coastal habitats affected by highly managed watershed tributary systems. Research may also suggest appropriate management to promote habitat rehabilitation.

Restoration of coastal ecosystems may also be influenced by the presence of nonindigenous species, which can be aggressive invaders. Unfortunately, restoration projects present opportunities for the establishment of nonindigenous species, which are often adapted for colonizing disturbed habitats, and eventual displacement of the anticipated natural community. Given the long time frame over which restored communities must establish themselves, they are subject to invasion for decades after their initial creation. Some invading species alter the community structure so seriously that establishment of a native community similar to nearby mature communities may be impossible.

Restoration often involves maintaining some degree of management control. For example, subsided coastal wetlands restored to vegetated marshes are contained within levee systems or impoundments in which water levels are managed or coastal wetlands are restored within urbanized areas containing sources for nonindigenous species introduction. Restoration efforts need to focus not only on initial control but also on long-term management to reduce deleterious effects of nonindigenous species in the future. Along the east coast, control of the marsh plant Phragmites is a prime example of the necessity of understanding altered hydrological regimes on competition between invasive and natural plant communities.

Research is needed on the causes of vulnerability to invasion by nonindigenous species, competitive interactions with native species, human-caused changes in native communities that result in invasion, and mechanisms by which these species persist. In addition, environmentally acceptable methods to prevent and control invasive species need to be developed. Chemical and mechanical means to control nonindigenous species have often proven to be expensive or environmentally unacceptable. The use of ''integrated pest management" in coastal habitats should be examined. Such an approach has been successful in freshwater environments for the control of aquatic weeds and may find applicability in coastal environments.

Federal science agencies should encourage rapid advancement of the science and engineering of ecosystem restoration and rehabilitation as called for by two recent NRC assessments (1992c, 1994b). Mitigation, if properly designed, can provide a significant mechanism for reducing loss of habitat, whereas rehabilitation, enhancement, and restoration can actually increase our depleted natural resource base. The success of restoration and rehabilitation depends on scientifically sound design and performance criteria, effective technical implementation, and monitoring of performance. Science priorities include research to provide a firmer basis for the design of restoration projects, various modeling approaches to yield understanding from quantitative measures of design and performance, and monitoring the effectiveness of the restoration and rehabilitation.

Research, including small-scale experiments and larger pilot programs, is needed to eliminate shortcomings in our understanding of coastal habitat needs, functions, and processes. In particular, research should be directed to providing a capability for predicting the effects of hydrodynamic and other physical processes on coastal habitats, with important factors for habitats ranging from wetlands to reefs. In addition, research is needed on the following subjects:

• the relationships between habitat structure and function;

• sediment properties, transport, and accretion as they influence the physical and biological performance of habitat rehabilitation, restoration, and creation projects (see p. 45);

• the relationship between nutrient dynamics and establishment of wetlands and submerged aquatic vegetation;

• recruitment and population and community development in restored ecosystems;

• processes that regulate and control interannual variability in populations of coastal biota;

• techniques, including the use of dredged material, for coastal habitat restoration; and

• methodologies for economic evaluation of alteration and restoration of coastal habitats.

Improved physical and biological models would greatly help advance the engineering of ecosystem restoration. In environments in which wetlands or seagrasses are restored, the boundary conditions in shallow water and tidal channels present unique problems to physical and chemical modelers. Sedimentation rates and biogeochemical cycles also could be studied in conjunction with improved modeling. Biological responses such as reproduction and propagation should be modeled for the species that structure the habitat (e.g., vascular plants

or oysters) and models linked with environmental quality factors and with production of living resources. With improved information on interactions between vegetation, soil development, and organisms, models can be useful tools in predicting and designing wetlands restoration and rehabilitation.

Coastal wetlands restoration has often been labeled unsuccessful or ineffective in replacing natural ecosystems. However, few projects have been monitored sufficiently long or with enough detail to understand the rates at which restoration proceeds. Restoration sites may have similar vegetative structure and habitat values as natural wetlands in less than a decade, but the establishment of equivalent nutrient retention and transformation and biodiversity may take decades and is highly scale dependent. From a practical standpoint, information about relevant time and space scales can be used to establish appropriate monitoring programs and in setting regulatory standards for wetland use permits.

Successful project performance is the most productive, conclusive, and reliable means of demonstrating the viability of restoration technology and building public and professional confidence in its application. Structural and functional monitoring before, during, and after project implementation is crucial for determining the effectiveness of the engineering methods, technologies, and practices used and their relation to natural functions. Many coastal habitat restorations do not establish well-defined criteria to assess their performance or monitoring programs to establish or document performance relative to scientific and engineering parameters. Refinement of individual project design during implementation to meet project-specific conditions is not often included in project plans, but it is especially important when innovative approaches and emerging technologies are used.

Many toxic contaminants are widespread in the environment and occur at low concentrations. Removing these contaminants from the coastal environment is often expensive and impractical. For example, DDT is found in California coastal sediments (off Los Angeles) due to past discharges from a chemical manufacturing plant. The DDT-contaminated sediments are found at depths exceeding 300 m, and removal would be costly if not infeasible. Many harbors and estuaries also have elevated levels of contaminants in sediments. The presence of contaminated sediments and the possible remobilization of toxic contaminants during dredging and disposal or during attempts to remove these sediments from the environment raise concerns about effects on the ecosystem, contamination of coastal fishery stocks, and impacts on recreational uses (NRC, 1989).

New technologies are being developed to remove contaminants from soils on site through the use of biological agents or chemical deactivation. Although some effort has focused on microbial degradation of oil and plastics in aquatic

environments, little research has been completed on the wide array of contaminants in coastal sediments and water. In addition to processing of sediments in situ, procedures and technologies to use contaminated soils in offsite habitat restoration activities are needed. Low-level contaminated soils might be buried in basins that are then restored with marsh vegetation that can act as an erosion control method to retain the sediments in place.

Scientific activities needed to guide the rehabilitation of contaminated sediments include studies of the fate and transport of toxic materials in sediments, the water column, and organisms; development and demonstration of in situ technology to eliminate or reduce contamination in sediments; demonstration of successful habitat restoration technologies that remediate low-level contamination within the coastal environment; application of risk management; and risk communication to include the concerns of the citizens of nearby communities.

Despite the long list of research needs identified earlier in this report, it must be noted that scientists have accumulated a great deal of knowledge about how coastal ecosystems work and how human activities affect them. Often, the lack of application of existing scientific knowledge to management decisions is as much a barrier to good decisions as is lack of sufficient knowledge. Good management decisions demand quantitative scientific information that can be used as a means of predicting possible and likely outcomes of these decisions. Likewise, good management decisions also require accurate and effective communication of information among scientists, policy-makers, and managers. Scientists and managers should strive to make environmental science more predictive by implementing observation and prediction systems that technological advances now allow and by employing ecosystem models as management tools. However, nature can and often does defy prediction and scientists and managers are not totally objective observers of nature or users of knowledge. Thus, given the limitations associated with using ecosystem models, coastal scientists must be prepared to contribute effectively to adaptive management as a means of dealing with uncertainty. There must be sufficient cross-training of scientists and managers to allow the communication of relevant knowledge and predictions effectively. To advance predictive systems management, federal agencies should work together to implement observation and prediction systems, employ ecosystem models as management tools, advance adaptive eco-

system management, and stimulate interactions between science and management at the science-management interface.

Rapidly developing technologies for in situ environmental sensing, remote sensing (see Box 6), and data management and communication are increasing our ability to observe environmental phenomena in near real time. This opens opportunities for the application of such information for a variety of purposes, including providing coastal storm warnings and marine advisories, weather forecasting, tracking oil spills, and fishing reports, as well as for use in scientific research and environmental management. Converging with these technological developments is a growing quantitative understanding of environmental processes and predictive modeling capabilities that, especially when coupled with near-real-time data, allow accurate "now-casting" and forecasting of environmental conditions.

Federal agencies should work to implement multipurpose observation and prediction systems in selected areas of the U.S. coastal ocean and Great Lakes (see also NRC, 1994d). NOAA (1993a) has developed a strategic plan for a coastal forecast system. Such a system should serve broad needs, such as those related to coastal hazards, maritime safety, and weather forecasting, in addition to serving the needs of environmental research and management. For research purposes these systems should include the kinds of observations identified by the CoOP program, including observations along "corridors"

extending from estuaries to the shelf break (Vincent et al., 1993). The requirements of the Global Ocean Observing System should also be criteria for the design of a national ocean observing system (NRC, 1994d). In addition, these observation and prediction systems should be linked closely with regional monitoring and should employ innovative in situ and remote sensing, as discussed above in the Integrated Monitoring section. Finally, observation and prediction systems should be integrated with both the predictive and adaptive aspects of ecosystem management, as discussed in the next two sections.

A basic premise of this report is that the effectiveness of coastal ecosystem and resource management can be improved by increasing the use of high-quality science in coastal environmental decision-making. Scientific understanding of the complex environmental problems that confront coastal regions worldwide requires an integrated multidisciplinary approach. Ecosystem modeling is one tool that allows integration of diverse scientific information and dynamic simulation of interactions embodied in such information. In addition, ecosystem models can serve to focus scientific understanding toward resource management questions. Therefore, ecosystem models offer tools for scientific integration and for applying integrated science to improve resource management. Coastal circulation models are the basis for ecosystem models. For example, such a model is being used to predict water quality in the Massachusetts Bay-Cape Cod Bay system (Box 7).

Modeling strategies should strive to capture the key mechanisms embodied in ecosystem processes and population dynamics (see p. 40), and equations should be calibrated with the highest-quality data available. To achieve maximum credibility, uncertainties underlying model coefficients (and model formulations) need to be stated in quantitative terms and sensitivity analyses should be used to reveal the consequences of these uncertainties on model performance. Models used as research tools and for coastal resource management must also be calibrated rigorously and validated with independent data sets. Modeling activities should strive to achieve real-time comparisons between model behavior and observations from key coastal environments. Techniques need to be developed to assimilate these data into model simulations as a diagnostic tool for model analysis and in the future as means for improved forecasting.

Coastal research programs that address regional and global-scale questions tend to use models as integrative (and occasionally predictive) tools. For example, several of the LMER programs have successfully used mass balance and numerical simulation models to relate watershed and atmospheric inputs of nutrients to biological production, transport, and accumulation of organic matter in coastal environments. Models have also been used in LMER and the Joint

Global Ocean Flux Study to extrapolate from regional observations in estuaries and shelf areas to global estimates of oceanic carbon balance.

An example of the kind of linked atmosphere-watershed-coastal ecosystem model discussed on page 38 is one being used to guide efforts to reduce nitrogen and phosphorous inputs to the Chesapeake Bay (see Figure 4). It includes a large-scale atmospheric deposition model; a watershed model that uses inputs on meteorology, land use, soil, and geophysical characteristics; and a hydrodynamic water quality model for the mainstem of the bay. The three-dimensional mainstem model is highly sophisticated, is relatively accurate in predicting water quality (e.g., dissolved oxygen), and is being refined to predict effects on living resources. The watershed model is, however, less well developed and based on surface water flow, yet it is critical to making land-use and management practice decisions to reach environmental quality goals in the bay most efficiently. The Green Bay Mass Balance Model is another example of a coupled coastal ecosystem model. It was developed by the EPA Great Lakes National Program Office with cooperation from NOAA, USGS, the State of Wisconsin, and numerous academic institutions. This model of the fates of toxic contaminants integrates watershed, tributary, atmospheric deposition, and ecological processes. A few coastal research initiatives (LMER and the NOAA Coastal Ocean Program Ecosystem Health Theme) are attempting to develop generic simulation models that can be used for comparative analyses of ecosystem processes across a wide spectrum of coastal environments. In principle, such generic models could be-

FIGURE 4 Chesapeake Bay modeling strategy. (pers. comm., Lewis Linker, EPA Chesapeake Bay Program Office, Annapolis, Md.)

come useful tools for adjusting management policies that have been used in one coastal region to similar situations in other systems although achieving compatibility of such models among a range of systems may require relatively simple models that can be tailored to individual systems.

Adaptive environmental management (Walters, 1986) provides an approach to develop and use science in environmental management in the face of high uncertainty surrounding both scientific predictions and the outcomes of management decisions. It assumes that management is essentially an experimental process through which scientists can learn and managers readily accept, even encourage, alternative views of how ecosystems may respond to management actions. Adaptive management allows the exploration of a broader range of options than are typically pursued by environmental management. Science provides a "compass," but management is also steered by a "gyroscope" of bounded conflict among the stakeholders (Lee, 1993). In adaptive environmental management, scientists are significantly involved in the process of management because management depends on regular reassessment of ecosystem performance

and changes in science-based management strategies over time. Adaptive management is a dynamic process, different from most of today's environmental regulations that remain relatively static for years or decades at a time. Attitudes and tactics appropriate to adaptive management differ from those conventionally adopted (Walters, 1986). Generally, adaptive approaches require embracing alternative explanations and approaches rather than seeking narrow consensus and precise predictions.

A lack of understanding of coastal systems will impede attempts to manage them. Therefore, fundamental science must proceed in cooperation with adaptive management. New observations, research, and modeling studies will be needed as new areas of uncertainty and management failure are identified. Management strategies can evolve as new information and techniques become available. This will demand different approaches to research, more rapid reporting of results, and different modes of group interaction and communication among and between scientists and managers. Federal science agencies should provide leadership and support in involving scientists in adaptive ecosystem management.

Improvements in the communication (both the transmission and reception) of scientific information and policy requirements is sorely needed to aid policy formulation and management decision-making that influence coastal ecosystem integrity. The development and use of science-based management models and other innovative ways to transmit meaningful information and the intense interactions between scientists and managers involved in adaptive ecosystem management will help. To make real progress, however, federal agencies involved in coastal science and/or environmental management should make a more concerted effort to improve the understanding of management issues by scientists and of science by managers. This would promote cross-sectoral literacy and a better understanding and appreciation of the different cultural perspectives and requirements of the two sectors.

Presently, interactions between scientists and managers generally take place over brief, intense periods, such as during workshops and conferences, which have inherent limitations in penetrating the science-management interface. Federal agencies could help bridge this gap by (1) supporting the exchange of scientists to management organizations and managers to research institutions, (2) supporting traineeships in science and technology transfer for scientists and managers, and (3) encouraging the education of specialists in the translation of scientific information for environmental management.