REVISED 12/17/96

Watershed analysis-the evaluation of processes that affect ecosystems and resources in a watershed-is now being carried out by Federal land-management and regulatory agencies on Federal lands of the Pacific Northwest. Methods used differ from those of other implementations of watershed analysis because objectives and opportunities differ. In particular, Federal management must take into account the full range of uses and values present, so interagency watershed analysis includes wildlife, social, and economic components in addition to the physical and aquatic evaluations shared by other watershed analysis efforts.

Interagency watershed analysis is intended to describe process interactions and ecosystem functions that shape the environment in 50- to 500-km² watersheds and to describe the causes of environmental change in those areas. Analysis is intended to be based primarily on existing information. It is not an exercise in data gathering or inventory, and neither does it provide prescriptions for management. The first wave of analyses has shown the major operational difficulties to be integrating information from multiple disciplines into an interdisciplinary analysis, lack of participation of multiple agencies, and discomfort with the broad-scale approaches required to evaluate a large area in a short time.

Watershed analysis describes processes and interactions that influence ecosystems and resources in a watershed. Watershed analysis has become a major preoccupation of Federal land management agencies in the Pacific Northwest. Even in an era of budget cuts and personnel reductions, adoption of the Northwest Forest Plan requires Federal agencies to take on new responsibilities and to adopt a new approach to land management and interagency cooperation, and watershed analyses is a cornerstone of the new management approach. Considerable analysis work needs to be done over the next several years, but few understand the type of work required or its intended scope. Confusion about watershed analysis is widespread both within Federal agencies that are responsible for the analyses and among State agencies and individuals who may become involved in the analysis process.

Confusion arises in part because several types of watershed analysis are now being carried out in the region. One type of watershed analysis has been adopted by the Washington Forest Practices Board (1993) as the basis for evaluating the potential cumulative impacts of land management activities. Other versions of watershed analysis are being considered for similar applications by the National Council of the Paper Industry for Air and Stream Improvement (NCASI 1992) and by the State of Idaho (Idaho Department of Lands 1995). The purpose and scope of the Federal approach to watershed analysis (referred to here as "interagency watershed analysis") differs from each. This paper outlines the original objectives for watershed analysis on Federal lands, describes the intended approach, and discusses the types of problems that appeared as the first analyses were carried out. The interpretations described here are by no means universally shared among those carrying out or managing interagency watershed analyses, and some of these interpretations undoubtedly will change as the analysis process evolves.

Watershed analysis became a required step in land-management planning on Federal lands in the Pacific Northwest because of controversy over survival of the northern spotted owl (Strix occidentalis caurina). The Scientific Panel on Late-Successional Forest Ecosystems (a.k.a. the "Gang of Four") was convened by the U.S. House of Representatives to devise management alternatives for the spotted owl (Johnson et al. 1991). Logging effectively stopped on Federal lands within the range of the northern spotted owl when the 9th Circuit Court of Appeals decided (Seattle Audubon Society v. US Forest Service 1991) that Forest Service management plans were insufficient for maintaining the owl's viability, as required by the Endangered Species Act, the National Forest Management Act (NFMA), and the National Environmental Policy Act. The Court also required the agency to prepare an environmental impact statement on management of the owl that also addressed viability of a variety of old-growth-dependent animals, including several species of Pacific salmon. Meanwhile, plans for logging on Bureau of Land Management lands were also challenged, and in 1992 further logging was enjoined until the BLM evaluated the likely effects of their timber management on the northern spotted owl.

The Forest Service submitted the required environmental impact statement (EIS) in 1992, and it was rejected by the Court after the Seattle Audubon Society argued that the EIS did not address the viability of old-growth-dependent species to the extent required by the NFMA. Later in 1992, the Forest Service convened the Scientific Analysis Team to assess the required viabilities. As part of this effort, a committee composed primarily of Forest Service Research scientists was given the task of designing a management strategy for Pacific salmon. The committee outlined objectives for aquatic habitat conservation, suggested widths of riparian reserves necessary to maintain in-stream habitat, identified particular "key" watersheds that provide especially good habitat, listed standards and guidelines for management activities, and required that processes influencing aquatic habitat conditions in a particular watershed be evaluated before the recommendations were modified for that watershed (Thomas et al. 1993). This evaluation of processes that affect aquatic and riparian habitat was referred to as "watershed analysis", after a similar method being used to evaluate cumulative watershed effects in Washington State (Washington Forest Practices Board 1993).

The Scientific Analysis Team report (Thomas et al. 1993) was published just as President Clinton convened the Forest Conference in Portland, Oregon, in early April 1993 to review the controversy surrounding forest-related economic and environmental issues. As a result of the Forest Conference, the President directed the Federal agencies to work together to devise a uniform approach to Federal land management in the Pacific Northwest that would provide for both ecosystem and economic sustainability and satisfy the requirements for lifting the injunction on logging on lands managed by the Forest Service and BLM. The interagency Forest Ecosystem Management Assessment Team (FEMAT) was created to develop this approach. The resulting report (FEMAT 1993) adopted many of the recommendations of the Scientific Analysis Team. However, FEMAT expanded the objectives of watershed analysis to include evaluation of terrestrial ecosystems and to incorporate explicitly a socio-economic component.

The FEMAT findings were formally adopted as Federal policy with the publication of the Record of Decision and The Standards and Guidelines for Management of Habitat for Late-Successional and Old-Growth Forest Related Species Within the Range of the Northern Spotted Owl (USDA and USDI 1994; the combined Record of Decision and the Standards and Guidelines are here referred to as the "ROD", and the strategy is known as the Northwest Forest Plan). The ROD describes the new land-management strategy, and it presents watershed analysis as an important component of this strategy. Federal lands in western Washington, western Oregon, and northwestern California will be managed according to the environmental needs and opportunities of each area, and watershed analysis is the primary tool for identifying those environmental needs and opportunities.

There is no widely shared image of what watershed analysis is, so people have different expectations of what an analysis should include. Indeed, "watershed analysis" is as general a term as "cumulative effects analysis": no specific procedure is implied by the term, so the goals and scope of an analysis approach are defined by those who mandated that particular implementation.

The intent for interagency watershed analysis is so simple that it is usually overlooked, and people consistently try to attribute more to the idea than was originally intended. The agencies have a history of implementing well-designed, well-intentioned, and irrelevant projects (Ziemer 1994). Expensive restoration projects are carried out at sites where they will have little effect (Frissell and Nawa 1992); massive resource inventories must be redone using a different method or at a different scale; projects designed to benefit one component of the ecosystem damage others (Fuller and Lind 1992); and guidelines are followed to the letter in places they do not apply. In each case, troubles result from an institutionalized approach to project planning that considers only the particular site or issue of concern and not its areal, temporal, and interdisciplinary context. It became evident that each project has to be examined within a much larger context, and that a single broad-context analysis could provide much of the background information needed for a variety of projects in an area. Watershed analysis was intended to provide the generalized understanding of the processes and interactions that shape the physical, biological, and social landscapes in an area, and it was intended to provide an overtly interdisciplinary view of the ecoscape.

Analyses are to be carried out in 50- to 500-km² watersheds, but even these are not large enough to encompass all of the issues and interactions that are relevant to project planning in an area. Watershed analysis was thus designed as a second analysis step; analyses are to be preceded by "river-basin assessments" that provide an even broader view. A basin assessment identifies the types of issues and interactions in the major river basin that contains the watershed to be analyzed (ROD p. B-22), so it provides the context necessary both to ensure that the watershed analysis provides information relevant to broader concerns, and to interpret the analysis results as they relate to the broader concerns.

Interagency watershed analysis is intended to describe patterns of process interactions and ecosystem function (ROD p. E-20, par. 4) rather than to provide specific information necessary for designing particular projects. One objective of analysis is to identify the types of information needed "for subsequent analyses, planning, or decisions" (ROD p. E-21, par. 1); analysis identifies what we don't know and determines how badly we need to know it. Watershed analysis itself "will be an information-gathering and analysis process, but will not be a comprehensive inventory process" (ROD p. E-20, par. 6). In general, it will "organize, collate, and describe existing information", although if crucial information for the analysis is lacking, completion of the analysis may be postponed until it is available (ROD p. E-21, par. 1). Extensive data sets already exist for many areas, but resource specialists rarely have had time to examine the broader implications of these data or to consider their interdisciplinary significance. Interagency watershed analysis is intended to provide a framework for making these connections.

Many worry that watershed analysis results will not be precise enough to be useful, and that analyses will need to be redone to design particular land-use projects. For example, watershed analysis will not produce a landslide map, so landslides will eventually need to be identified at each project site as the project is planned. However, since any landslide map is out of date as soon as a new slide occurs, this would be necessary even if each landslide in the watershed had been mapped during watershed analysis. Such site-level detail is not the intent of interagency watershed analysis. Instead, analysis will show the broader patterns of cause and effect so that future project analyses can focus on the site-specific issues. Watershed analysis thus might describe the types of sites susceptible to landsliding in an area and describe the land-use activities with which slides are associated. Project analysis would then use this information to aid in evaluating stability conditions at the project site. The results of watershed analyses are likely to demonstrate that routine mapping of landslides is not useful in many watersheds because landslides are so frequent, so rare, or so predictable in their distribution patterns.

Others suggest that a broad overview such as a watershed analysis can only disclose the information we already know to be true, such as the observation that environmental conditions improve when sediment loads decrease. However, many of these "facts" are found in retrospect either not to be true or to apply only to particular sites. At one time, for example, woody debris was consistently cleared from streams to improve fish habitat (Reeves et al. 1991), and a high proportion of the educated public still believes that logging decreases base flows despite 50 years of research literature to the contrary (Bosch and Hewlett 1982, Stednick 1996). Each assumption must be reexamined in the context of particular watersheds and multiple disciplines. In addition, there are many places where major environmental problems arise from obscure interactions that would not be recognized without an interdisciplinary analysis. To restore a stream channel, for example, we must understand what caused it to destabilize in the first place (Hartman et al. 1996, Frissell and Nawa 1992), and this usually requires a sophisticated understanding of the interactions between riparian vegetation, sediment transport, and watershed hydrology. The breadth of expertise required for evaluating such a problem can rarely be convened for project-level work, but it is the essential ingredient of watershed analysis.

The first phase of implementation for interagency watershed analysis was the selection of 15 watersheds for "pilot" analysis during 1994 to develop and test analysis techniques. The best of the pilot analyses are being used as models for future efforts, but it will require several years for analytical procedures to be developed fully (ROD p. 55, par. 4). An interagency team has prepared a manual to guide future analyses (Regional Ecosystem Office 1995).

The lands to be analyzed

The ROD divides Federal lands within the range of the northern spotted owl into key watersheds and non-key watersheds (ROD p. 10). A variety of reserve types, including Riparian Reserves, are located within each of these land categories. "Matrix lands" comprise the parts of non-key watersheds that are not included in reserves, and management can proceed in these areas by following the standards and guidelines outlined by the ROD. In all other areas, analysis is required before significant management activities can occur. Watershed analysis is required in all key watersheds, and it is also necessary in non-key watersheds if modifications to Riparian Reserves are desired. These analyses are to be carried out on 50- to 500-km² watersheds by interdisciplinary, interagency teams. Prioritization of watersheds for analysis was originally intended to be carried out using input from multiple agencies, but currently each agency is unilaterally prioritizing watersheds on the lands it manages.

Interagency watershed analyses must evaluate all lands within a watershed, and some of this land is usually privately owned. Information about private lands provides a context for judging the influence of Federal land management and is essential for understanding the variety of influences that contribute to downstream environmental changes. Because only representative sites need to be visited, lack of access to private lands usually will not hinder analysis. Much of the information necessary for analysis is available on aerial photographs or can be inferred using data from adjacent Federal lands. In other cases, on-the-ground information may be needed, or the landowner may own data sets relevant to the analysis. The Record of Decision states that public input is important for watershed analysis (ROD p. B-21), and strong efforts should be made to encourage cooperation with private landowners. It will usually be in the landowners' interest to participate, because the resulting data base and analysis will be useful for their own management planning and for completing their cumulative effects evaluations.

Watersheds as analysis units

FEMAT adopted watersheds as the geographic unit for reporting the habitat needs of many organisms, such as the spotted owl, that are not directly affected by watershed processes. Wildlife biologists understandably ask the question, "why should our analysis be constrained by watershed boundaries when our organisms are not?" Similarly, social and economic issues have little relation to watersheds; many physical processes are better described according to geologic and climatic types that cut across watershed boundaries; and information from multiple river basins is needed to understand the condition of anadromous fish stocks. Not only does each issue have a different scale that is relevant to it, but each issue must be examined at several different scales (Lewis et al. 1996, Rabeni and Sowa 1996).

At the same time, it is essential that interdisciplinary evaluation be brought to bear on the suite of issues important in a particular area if those issues are to be understood (Ludwig et al. 1993). For example, to evaluate the history, distribution, and future of physical impacts in an area, it is necessary to understand the history of land use there, the economic setting, and the biological changes that have occurred (Meyer and Helfman 1993). Since no single area is appropriate for all issues, it does not really matter what size of area is selected or how it is delineated. In essence, the fundamental understanding that has been derived by each discipline using the scales relevant to that discipline is applied to the area in question. Fifty- to 500-km² areas were selected as being small enough to analyze at a useful level of precision, while being large enough to exhibit many of the interactions important to environmental issues. Watersheds were selected as the analysis unit because they can be defined at the appropriate scale, they are identifiable on maps and on the ground, and they do not change much through time.

Watersheds also are a geographic unit that holds relevance for off-site impacts that influence biological, socio-economic, and physical resources and values. Environmental and economic concerns often focus on channel changes triggered by upstream land use, and it is necessary to understand conditions in the channel's watershed to understand these changes. Since downstream changes (which generally correspond to off-site cumulative watershed effects) are an important consideration everywhere and would need to be evaluated in each analysis area, it makes sense that the analysis unit would be selected to make this type of analysis possible. Downstream changes strongly influence floodplain land use, aquatic and riparian biological communities, and terrestrial ecosystems that are affected by riparian and aquatic communities.

Many people are uncomfortable with the large size of the watersheds to be evaluated, but the size range was selected specifically to keep the analyses relevant to the types of broad-scale problems they are intended to address. Analysis watersheds need to be large if the connection between land-use activities and impacts is to be explained. Many past cumulative impact analyses have failed because they focused on too small an area for important processes to be recognized (Reid 1993). For some issues, such as the survival of anadromous salmonids, even the watershed scale is too small and a river basin or province scale must be used.

Because of the strategy used for interagency watershed analysis (Regional Ecosystem Office 1995), the size of the area to be evaluated does not hinder evaluation. Large areas can be divided into smaller areas of uniform character, and representative sites in a sub-area can then be observed to characterize that sub-area. Large areas can be evaluated using this "landscape stratification" approach almost as quickly as small ones (Reid and Dunne 1996). In addition, analyses are intended to show distribution patterns and qualitative categories, not specific locations or detailed measurements. Analysis focuses on which processes are active and how they generate impacts, not on how rapidly they operate at particular sites. Patterns often are easier to recognize and understand in large areas than in small ones.

The issue focus

Watershed analysis is to be carried out for large areas over short periods of time. Analysis is greatly simplified if it concentrates on only the most important issues in an area. However, land-use planning has a history of overlooking-and often being crippled by-some very important issues. A first step of watershed analysis is thus to use public outreach to identify the issues of concern in and around the analysis watershed. Issues of concern at the scale of the river basin are identified during a preliminary river-basin assessment, and the watersheds in which these issues will need to be considered are also identified at this stage. An additional set of issues are those already well-known to the various disciplines carrying out the analysis: water quality, biodiversity, threatened species, and so on. A final set of issues that might be important in the future can be identified according to the trends in socio-economic, biologic, and physical conditions in the watershed. The analysis will then be planned so that it will be capable of speaking to the identified issues. In most cases, the list of issues can be prioritized to focus the analysis more closely; issues that become important in the future would be evaluated during future iterations of the watershed analysis. Without this type of prioritization, it is very easy for analysts to be diverted by the details they recognize in their own area of expertise. It should be noted that analysis is not intended to be guided solely by the concerns of the public, but that public concerns are identified early so that the relevant ones can be addressed.

Related to the need for focus is the need to avoid gratuitous detail. In many cases, only qualitative information or order-of-magnitude measurements are needed to address the problems. This is a difficult concept for most experts (Holling 1993): our training has primarily been in observing the fine details of our problems, in the acquisition of precise and accurate measurements, and in the need for large data sets. Interagency watershed analysis demands that analysts step back from the detail and define general patterns and relative importance. The appropriate level of effort or detail must be judged by its relevance to the problem at hand, rather than by the attainable precision or by the possibility that a piece of information might be useful in the future.

Time required for a watershed analysis

The specialists' desire for detail and precision argues for a lengthy time commitment for completion of a watershed analysis. In contrast, land managers see watershed analysis as a hurdle to be leaped before any activities are carried out, so there is strong pressure for analyses to be completed quickly. A watershed analysis could take decades to finish, or it could be completed in a day. These products would differ primarily in their level of detail; both would be useful for particular applications, and neither would be complete. Ultimately, the usefulness of an analysis will be judged by how well it meets planning needs, and these needs include both timeliness and detail. Designers of the interagency analysis approach originally selected two months as a period over which a useful level of detail could be achieved for identifying appropriate future projects, but which would allow the majority of analyses to be completed during the next ten years.

In a few cases, the expected two-month duration for analysis may not be appropriate. Some areas may need only a brief examination of a few key issues, while others may involve such complex problems that more time is required. Different types of Federal lands may be better suited by different levels of analysis detail. For example, problems in national parks are often concentrated in a few watersheds, and park personnel may be able to devote long periods to their detailed analysis.

The earliest analyses are expected to take longer than later ones as methods and approaches are developed, and as information relevant to large areas is catalogued. Abundant data exist for many areas, but it takes a long time to find and catalog the information so that it can be used. A pre-analysis scoping of issues could prioritize data types so that most effort could be devoted to the most useful data, and data compilation could then begin long before the actual analysis. This preliminary scoping might be a part of a river-basin assessment. Some types of information are likely to be important for all analyses, and data compilation for any future analysis could begin once these are identified.

Several early analyses have suffered from inappropriate budgeting of efforts: parts of the analysis that were expected to be accomplished in several days have taken months. It may be useful to establish a time-line that indicates the level of effort expected for different parts of the analysis. An early draft of the Pilot Interagency Watershed Analysis Manual (Reid and McCammon 1993) suggested the following time budget:

Stage 1: Preliminary work: accumulate and catalog available information (occurs simultaneously for several watersheds over several months)

Stage 2. Identify issues and concerns (approximately 1 week)

Stage 3. Identify mechanisms through which environmental change could occur and use this information to plan analysis strategies (approximately 1 week)

Stage 4. Stratify the watershed according to important issues (less than a week)

Stage 5. Describe existing conditions (4 weeks, with stages 6 and 7)

Stage 6. Describe the mechanisms of environmental change (see stage 5)

Stage 7. Describe likely future environmental changes (see stage 5)

Stage 8. Prepare report (approximately 1 week)

Later versions of the manual have modified the analysis stages, but the relative effort to be put into each phase is clear. A lot of time will be devoted to the preliminary identification of available data (stage 1), which can then be used to identify and plan the work that needs to be done (stages 2-4). Interdisciplinary analysis (stages 5-7) then accounts for half the time budget. The preliminary data compilation can occur at one time for many different watersheds, and it does not require much oversight by resource specialists.

One function of interagency watershed analysis is to guide future monitoring and inventory by disclosing data gaps, describing large-scale and interdisciplinary relationships, and identifying the information needed to better understand the watershed and ecosystem. New information will thus continue to become available after a watershed analysis is "completed", and subsequent project-level analyses will use both information from the watershed analysis and the supplementary data. Project-level analyses will also add site-specific data to the watershed database. In some cases, the importance of new data or changing issues may make it useful to revisit parts of the original analysis or even to update the entire document. A watershed analysis should thus be considered as never "done", but as an open file that is continually modified as new information becomes available. Watershed analysis is described by the ROD as "an ongoing, iterative process" (p. E-20, par. 6).

Consistency and standardization

Watershed analysis is complicated because every watershed has a unique set of characteristics, conditions, processes, and issues. It is thus futile to collect the same types of data, perform the same analyses, and use the same methods in each watershed. However, some consistency is required so that results can be compared for watersheds throughout a river basin or region to understand the causes of environmental change at a larger scale. Standardization falls into three categories: data standards, method standards, and product standards.

Because interagency watershed analysis is not itself a data-gathering exercise, data standards are not of direct concern to analysis. However, watershed analysis uses existing data and would thus be easier if inventories and monitoring adhered to uniform standards. Data are most easily compared if they share a uniform precision, accuracy, scale, units of measure, and format, and if they are collected using standard methods. On the other hand, issues present in one area may demand more detailed information than in others, so no single data standard is appropriate for all issues in all areas. Data collection usually is guided by the discipline most concerned with a particular type of data, and often the information is collected in a form that makes it useless to other disciplines. Interdisciplinary communication during watershed analysis will allow future data collection efforts to be designed to meet the needs of multiple disciplines. Future analyses would be facilitated if data standards and standardized measurement techniques were established for commonly inventoried attributes. However, even if uniform data standards were adopted for Federal lands, analyses would still be faced with multiple standards and data gaps because different types of information are available for non-Federal lands.

Many watershed analyses will use geographic information systems (GIS) for data compilation and analysis, and GIS introduces additional demands for standardization. Many types of data are not readily managed using GIS, and it is important to preserve the original integrity of the data rather than translating them to the uniform scales that GIS often demands. GIS is a useful tool for data compilation and analysis, but it is not a prerequisite for watershed analysis, and its requirements and limitations cannot be allowed to control the analysis. Spatially explicit information is rarely necessary until particular projects are proposed, and the site-specific information required for project design cannot be attained at the scale of watershed analysis.

Neither can watershed analysis methods be standardized. Because of the variety of issues and settings that analyses will address, no method will be valid for every site. However, some consistency of analysis is necessary to communicate and compare results. Consistency will come about because analyses are to be conducted by competent experts in the appropriate disciplines, and analysts will use methods consistent with the standards in their fields. Within this broad guideline, analysts must have the freedom to use methods appropriate for the setting. In particular, each issue will require examination at the scale relevant to it (Holling 1993), and different analyses will provide different levels of detail for particular problems. Quality control is introduced because methods must be carefully described and must stand up to peer review. As experience grows, certain approaches will be found to be particularly useful, and eventually a suite of methods may become widely applied. Standardization may thus eventually occur through peer review. The lack of a "cookbook" is expected to benefit the analyses: cookbook methods tend to be applied blindly without judging their appropriateness, and they prevent people from seeing the unique analysis opportunities that each area possesses. In addition, premature selection of a standard method may institutionalize an invalid procedure and prevent its weaknesses from being assessed. The most useful "manual" in the long run may simply be examples of successful analyses that exhibit a variety of useful approaches and methods.

Even standards for an analysis product are not particularly useful, since areas rich in data will allow detailed analyses while data-poor watersheds may permit only a generalized analysis. What is essential, however, is a standardized vision of how the analysis product will be used. Standardization is again primarily useful for communicating and comparing results, and this is particularly important for allowing aggregation of watershed information at the basin level. A product of basin assessment could be a description of the types of information that need to be standardized during watershed analysis to address issues at a basin scale. The most effective route to a useful level of standardization may well be to arrange meetings of a variety of analysis groups to allow critiques of methods, sharing of useful techniques, and comparison of analysis products. The most useful approaches and methods will be converged on-standardized-through repeated comparisons.

Interagency participation

The Forest Service and the Bureau of Land Management were initially given large budgets to complete watershed analysis, but regulatory agencies such as the Environmental Protection Agency, Fish and Wildlife Service, and National Marine Fisheries Service have not. Many key agencies are understaffed even for their existing workload, and they have little inclination to expand the workload to beyond what they already do not have time to do. Federal agencies will participate in a watershed analysis if they see the analysis as a way to achieve their agency's goals or if they are told to by their agency's leaders. Through FEMAT, Federal agencies have been directed to accept watershed analysis as a priority, but the degree and manner of each agency's participation remains unclear. State agencies have no mandate to participate and are likely to be involved only if it satisfies their needs. Lures for State participation might include the general utility of a better information base, and the extent to which watershed analysis can contribute to State requirements for land-use planning or regulation.

Many of the regulatory agencies that are unlikely to participate in analyzing a watershed retain an important role because they participate on the Regional Interagency Executive Committee that reviews proposed actions to determine whether they are consistent with the objectives of the ROD (ROD p. 35, par. 4). To avoid confrontations at later stages, mechanisms therefore must be found to bring all interested agencies together to agree on an analysis approach in a watershed, and the agencies must be kept informed as analysis progresses. Agencies are likely to participate more intensively only if they manage or have regulatory responsibility for land within the watershed. An effective prioritization of watersheds for analysis may contribute to more interagency participation by taking into account the particular concerns and capabilities of the interested agencies.

Other types of watershed analysis are being applied to forest lands of the western United States or will be in the near future. None of these were found to be appropriate for addressing the challenges posed by the ROD.

The most thoroughly implemented version of watershed analysis is that being used in Washington State to evaluate cumulative effects in watersheds with multiple land owners. Interdisciplinary representatives of the forest-related interests in each 40- to 200-km² watershed work together to identify causes of environmental change and to define area-specific standards and guidelines that would avoid detrimental changes. Evaluation focuses on the effects of timber management on fisheries and water quality; wildlife issues are not yet addressed. Adherence to the analysis recommendations is voluntary, with the incentive that those following the recommendations will need to do no further environmental assessment work. No restrictions are placed on what land uses are allowed at what times at particular sites. Instead, guidelines describe how activities should be carried out in different parts of the watershed. Analysis is carried out for all lands, public and private, in the designated watersheds. A manual describes analysis modules for evaluating certain environmental characteristics, including aquatic habitat condition and several erosion processes (Washington Forest Practices Board 1993). Methods were selected that compilers believe to be effective and practical in Washington, and the manual is updated as methods are modified. A week of training introduces analysts to the procedures in the manual, but analysts are expected to have adequate training in their own fields. In practice, most analyses have been carried out by experienced specialists, and methods have not been limited to those outlined in the manual.

The Washington approach is not directly applicable to the needs of Federal land management for several reasons. First, it does not address the full range of issues that must be considered on Federal lands, where wildlife and social issues are accorded as much attention as aquatic issues. Second, the approach assumes that types of land-use activities will not be proscribed anywhere in the watershed, although the way that they are carried out can be controlled by prescribing on-site Best Management Practices. On Federal lands, however, a much broader range of management options is possible. The distribution and timing of activities over large areas can be scheduled on Federal lands, and some activities can be proscribed completely. Planning for scheduling or proscription requires some information that is not necessary for the design of on-site Best Management Practices. Third, the Washington manual compiles methods that were selected to suit Washington conditions, and many of these methods are not flexible enough to apply to areas of as diverse character as occur throughout the range of the northern spotted owl or have not been tested adequately (Reid, in press). Fourth, the Washington approach results in a prescription of Best Management Practices instead of simply describing the functioning of the ecosystem and watershed, and such land-use prescriptions go beyond the intent for analysis of Federal lands described by the ROD.

Despite the differences in intended application, elements of the Washington approach were used in the design of the interagency approach to watershed analysis. In particular, its recognition of the importance of including multiple interest groups, its use of interdisciplinary teams, and its focus on understanding general patterns of process interactions throughout a watershed are shared by the interagency analysis procedure.

Everyone has a different vision of what analysis should be, so everyone can be assured that analysis results will fall short of their expectations. Every interest group has tried to assign to watershed analysis the tasks that would most benefit the goals of that group. Planners would like watershed analysis to produce land-use plans, managers want it to delineate the boundaries of Riparian Reserves, the US Fish and Wildlife Service would like it to provide enough information for consultations concerning threatened and endangered species, and resource specialists want the analyses to provide the level of detail needed for project-level planning. The Record of Decision indicates that all of these expectations are unfounded: watershed analysis will provide information on general ecosystem and physical interactions, but will not contribute comprehensive inventory information (ROD p. E-20, par. 6) or take the place of project-level data collection (ROD p. E-21, par. 1) or identify Riparian Reserve boundaries (ROD p. E-20, par. 4). Even so, the information provided by watershed analysis is intended to be useful for planning Riparian Reserves, restoration programs, and monitoring, and to contribute to the understanding of process interactions, ecosystem function, and causes of environmental change that will be useful for general land-use planning.

Watershed analysis and Riparian Reserves

There has been widespread confusion over what tasks are to be done at the watershed-analysis level and what at the project level. For example, early documents were unclear about whether Riparian Reserve boundaries were to be established during watershed analysis or during project planning. The ROD has clarified the issue: boundaries are to be established during project planning, but watershed analyses will provide information useful for deciding how to draw the lines (ROD p. E-20, par. 4). For example, the analysis will describe the ecological and physical role of riparian zones in different parts of the watershed. This division of applications is necessary for three reasons: first, it is not possible to map the reserves to the detail required without visiting each site; second, the design of the reserve will reflect the nature of the project, and this cannot be known throughout the watershed at the time of watershed analysis; and third, any decisions or recommendations about land allocation would require compliance with procedural provisions of the National Environmental Policy Act (NEPA).

However, some information about the potential extent of Riparian Reserves is needed for a watershed analysis. For example, evaluation of potential wildlife habitat would need to consider the extent and character of potential Riparian Reserves. To obtain this, analysts would need a rough idea of the extent of the stream network, the amount and distribution of potentially unstable land, and the logging history in different parts of the watershed. For watershed analysis, a description of the likely size and characteristics of reserves in different parts of the watershed will usually be sufficient, and this information generally can be obtained by observations at representative sites. These types of information are also needed for establishing Riparian Reserve boundaries during project design; but more detailed, site-specific information is required at that later stage.

Understanding ecosystems

Watershed analysis will identify what information exists about aquatic and terrestrial ecosystems, what is needed, and what the likely fish and wildlife issues are in different parts of the watershed. Analysis may identify and prioritize the types of inventories and monitoring information needed, identify areas of particular importance, and indicate the type of information needed for project-level planning. Results of a watershed analysis might be used to justify a watershed-scale survey to determine the distribution of particular species.

A watershed analysis can be amended as additional information becomes available from future inventories, monitoring, and project-level analysis. However, there will never be enough on-the-ground data available for a "complete" analysis, so it is important for analysts to identify the information they do have, describe the important data gaps, and restrict inferences to those that can be supported by existing data. Habitat models are useful only if they have been well-tested and if sufficient data exist for their valid application in an area. Most watershed analyses will incorporate little fieldwork, and much of the field time needs to be devoted to developing cross-disciplinary understanding rather than to addressing problems peculiar to a single discipline. The agencies are well-practiced at data collection, but now is the time to tackle the more difficult task of integrating and interpreting the information they already have.

Cumulative effects analysis

The Forest Service is required to evaluate the potential for a proposed action to contribute to existing or potential future environmental impacts by way of a cumulative effects analysis. Cumulative effects analyses evaluate the effects of particular projects, so these analyses cannot be completed during interagency watershed analysis because future projects are not known at the time of watershed analysis. In addition, management strategies are changing rapidly on Federal lands, so the potential for environmental change will need to be reassessed for each future project according to the conditions at that time. However, watershed analysis results will be useful during future cumulative effects analyses anywhere in the watershed. Watershed analysis will have identified existing data, described the nature and cause of existing cumulative effects in the watershed, and discussed the interactions that could cause future impacts. New cumulative effects procedures are likely to be developed that will take full advantage of this information. In the meanwhile, existing cumulative effects methods can make use of watershed analysis results. Watershed analysis will provide much of the watershed-scale information necessary for cumulative effects evaluations of Timber Harvest Plans on private lands in California, for example, and this will decrease the work necessary for completion of such plans.

The Washington State watershed analysis approach has been adopted as a cumulative effects analysis procedure (Washington Forest Practices Board 1993) by consensus of the disparate interest groups in the state. In this case, analysis identifies Best Management Practices that are expected to prevent future cumulative effects, but it does not specify what types of land use are permissible at specific sites. This approach is possible on private lands because cumulative effects analyses are guided by State, rather than Federal, requirements.

Watershed analysis is new. Federal agencies have no history of such analyses, but they are now required to produce hundreds of analyses over the next several years. A humbling array of analytical, cultural, and procedural barriers will need to be overcome before watershed analysis becomes recognized as a routine phase of management planning.

The major analytical hurdles center on the need for an interdisciplinary approach and on the size of the area to be evaluated. Because of our culture's mono-disciplinary approach to education and the requirements of specialization, most resource specialists have little experience with true interdisciplinary problem solving. Past interdisciplinary efforts have often led instead to multidisciplinary work, where representatives of different disciplines all simultaneously and independently examine the same area. We still have a tendency to believe that we are capable of solving our part of the problem, not realizing that we do not have a part of the problem because the problem does not have discrete parts. Resource specialists need to look at a problem and see a web of interdisciplinary relationships, and they need to work hard to see beyond their own disciplinary boundaries and communicate with other team members. Meanwhile, terrestrial biologists fail to see the relevance of hydrologic and geomorphic change to the issues they are concerned with; aquatic biologists do not consider the controlling influences of terrestrial ecosystems; hydrologists are oblivious to evidence of past vegetation changes; and few understand the pervasive role of the social setting and cultural history in determining past, present, and future ecosystem characteristics.

Many of us have picked up the subconscious belief that our discipline is the one that is central to solving the world's problems. We have a tendency to view other disciplines as they relate to ours, rather than as they relate to the world. This disciplino-centrism is embodied in the hammer analogy: "If the only tool you have is a hammer, the whole world looks like a nail." Watershed analysis in many cases requires that the hammer be used as a grappling hook or fishing weight-still useful, but not in the way we were trained to use it. Each expert has a responsibility to devise creative ways for their hammers to be used to solve problems in others' fields. This requires that each expert on the watershed analysis team understand and have an appreciation for the other disciplines represented on the team.

The scale problem is also inherited from the western philosophy of science: we are more comfortable with looking carefully at small things than with getting a broad overview of a large thing. We are taught to value precision and accuracy in detailed measurements, and these are then assembled as pieces of a conceptual jig-saw puzzle to eventually make the "big picture" understandable. Watershed analysis takes the opposite approach. Here, the emphasis is on recognizing the broader patterns of interactions to determine which are most critical to understand at what level of precision. It is difficult for us to grasp the realization that we may need only qualitative or order-of-magnitude information for many applications, or that we may not need to know the precise location of a feature to understand its role in the overall system. We tend to approach the world through inventory and mapping, while watershed analysis primarily uses pattern recognition and sub-sampling to characterize the elements of the patterns. Watershed analysis proceeds simultaneously at multiple scales, and this approach is an unfamiliar skill that must be acquired through practice.

Some of the agencies that are intended to participate in watershed analysis have not accorded it the importance that the primary land management agencies have. Even within the BLM and Forest Service, there are inconsistencies between administrative units in the staffing provided and their mandate. In some Forest Service districts, analysis is being seen as simply a variation of the types of work done in the past. Managers expect that their resource specialists can do their watershed analyses in isolation while keeping up with their other assigned duties. At the other extreme, some administrative units perceive watershed analysis as a massive, one-time "fire-storm" that can be dealt with by devoting large teams and budgets to getting it done so that normal business can be resumed. Both approaches miss the point: watershed analysis is now normal business, and it is new normal business. Future analyses will become easier when agencies develop mechanisms for staffing and funding analyses at appropriate levels as part of their normal operating procedure.

Watershed analysis requires an uncustomary level of communication and cooperation between agencies and between agencies and the public. The investment of time that such efforts require has persuaded many of the pilot analysis groups to avoid the efforts. Socio-economic analyses have not been included in many of the reports because people with training in such fields are not yet available within the agencies. Agency personnel also have a widespread and erroneous perception that the Federal Advisory Committee Act does not allow public participation in watershed analysis except in the customary form of public meetings. As a result, some efforts at issue identification have been carried out with no public input at all.

Within agencies, there are strong institutional barriers to sharing resources or personnel; agency employees and equipment are expected to work on their own agency's lands. Faced with such bureaucratic roadblocks, many analysis teams have found it easiest simply to avoid soliciting participation of the other agencies.

In all cases, the final arbiter of what is appropriate, sufficient, and useful in an analysis is how well that analysis achieves its goals. It is likely that the attitudes currently hindering analysis will change only when it becomes apparent what the most useful analyses look like.

Bosch, J.M., and J.D. Hewlett. 1982. A review of catchment experiments to determine the effect of vegetation changes on water yield and evapotranspiration. Journal of Hydrology 55:3-23.

FEMAT. 1993. Forest ecosystem management: an ecological, economic, and social assessment. Report of the Forest Ecosystem Management Assessment Team, July 1993. U.S. Government Printing Office: 1993-793-071.

Frissell, C.A., and R.K. Nawa. 1992. Incidence and causes of physical failure of artificial fish habitat structures in streams of western Oregon and Washington. North American Journal of Fisheries Management 12:182-197.

Fuller, D.D., and A.J. Lind. 1992. Implications of fish habitat improvement structures for other stream vertebrates. Pages 96-104 in: R.R. Harris, D.C. Erman (technical coordinators) and H.M. Kerner (editor): Proceedings of the Symposium on Biodiversity of Northwestern California, 28-30 October 1991, Santa Rosa, California. Wildland Resources Center, University of California, Report 29.

Hartman, G.F, J.C. Scrivener, and M.J. Miles. 1996. Impacts of logging in Carnation Creek, a high-energy coastal stream in British Columbia, and their implication for restoring fish habitat. Canadian Journal of Fisheries and Aquatic Sciences 53 (Supplement 1): 237-251.

Holling, C.S. 1993. Investing in research for sustainability. Ecological Applications 3(4):552-555.

Idaho Department of Lands. 1995. Forest practices cumulative watershed effects process for Idaho. Idaho Department of Lands.

Johnson, K.N., J.F. Franklin, J.W. Thomas, and J. Gordon. 1991. Alternatives for management of late-successional forests of the Pacific Northwest. A report to the Agriculture Committee and the Merchant Marine Committee of the U.S. House of Representatives. 59 p.

Lewis, C.A., N.P. Lester, A.D. Bradshaw, J.E. Fitzgibbon, K. Fuller, L. Hakanson, and C. Richards. 1996. Considerations of scale in habitat conservation and restoration. Canadian Journal of Fisheries and Aquatic Sciences 53 (Supplement 1): 440-445.

Ludwig, D., R. Hilborn, and C. Walters. 1993. Uncertainty, resource exploitation, and conservation: lessons from history. Science 260:35-36.

Meyer, J.L., and G.S. Helfman. 1993. The ecological basis of sustainability. Ecological Applications 3(4):569-571.

NCASI. 1992. Status of the NCASI cumulative watershed effects program and methodology. National Council of the Paper Industry for Air and Stream Improvement, Technical Bulletin 634. 31 p.

Rabeni, C.F., and S.P. Sowa. 1996. Integrating biological realism into habitat restoration and conservation strategies for small streams. Canadian Journal of Fisheries and Aquatic Sciences 53 (Supplement 1): 252-259.

Reeves, G.E, J.D. Hall, T.D. Roelofs, T.L. Hickman, and C.O. Baker. 1991. Rehabilitating and modifying stream habitats. Pages 519-558 in: W.R. Meehan (ed.). Influences of forest and rangeland management on salmonid fishes and their habitats. American Fisheries Society Special Publication 19. Bethesda, Maryland. 751 p.

Regional Ecosystem Office. 1995. Ecosystem analysis at the watershed scale, version 2.2. Regional Ecosystem Office, Portland, OR. US Government Printing Office: 1995 - 689-120/21215 Region no. 10. 26 p.

Reid, L.M. 1993. Research and cumulative watershed effects. USDA Forest Service General Technical Report PSW-GTR-141. 118 p.

Reid, L.M. In press. Cumulative watershed effects and watershed analysis. Chapter 18 in: R.J. Naiman and R.E. Bilby, The Ecology and Management of Streams and Rivers of the Pacific Coastal Ecoregion. Springer Verlag, New York, NY.

Reid, L.M., and T. Dunne. 1996. Rapid evaluation of sediment budgets. Catena Verlag GMBH, Reiskirchen, Germany. 164 p.

Reid, L.M, and B.P. McCammon. 1993. A procedure for watershed analysis. Review draft prepared for the Forest Ecosystem Management Assessment Team, 14 July 1993. USDA Forest Service Region 6, Portland, OR.

Stednick, J.D. 1996. Monitoring the effects of timber harvest on annual water yield. Journal of Hydrology 176:79-95.

Thomas, J.W., M.G. Raphael, R.G. Anthony, E.D. Forsman, A.G. Gunderson, R.S. Holthausen, B.G. Marcot, G.H. Reeves, J.R. Sedell, and D.M. Sollis. 1993. Viability assessments and management considerations for species associated with late-successional and old-growth forests of the Pacific Northwest. Report of the Scientific Analysis Team. USDA Forest Service. Portland, Oregon.

United States Department of Agriculture and United States Department of the Interior. 1994. Record of Decision for amendments to Forest Service and Bureau of Land Management planning documents within the range of the northern spotted owl; Standards and Guidelines for management of habitat for late-successional and old-growth forest related species within the range of the northern spotted owl. U.S. Government Printing Office 1994 - 589-111/00001 Region no. 10.

Washington Forest Practices Board. 1993. Standard methodology for conducting watershed analysis under Chapter 222-22 WAC. Version 2.0. Department of Natural Resources Forest Practices Division, Olympia, WA.

Ziemer, Robert R. 1994. 25. Cumulative effects assessment impact thresholds: myths and realities. Pages 319­326 in: Kennedy, A.J., ed. Cumulative Effects Assessments in Canada: From Concept to Practice. Alberta Association of Professional Biologists. Edmonton, Alberta, Canada. 333 p.