Bernard T. Bormann
USDA Forest Service
Pacific Northwest Research Station
Forestry Sciences Laboratory
3200 SW Jefferson Way
Corvallis, OR 97331

Ecosystem sustainability is the stated goal for Federal land managers; adaptive management is proposed as the principal means by which sustainability can be achieved. Ecosystem sustainability, adaptive management, and other terms now have multiple meanings, and any debate about their use must begin with careful definitions.

For the purposes of this discussion, ecosystem sustainability is defined as the condition of overlap between what people want for themselves and for future generations and what is ecologically possible in the long run (fig. 1). This goal framework is complex and depends on considerable knowledge of societal values and ecological capacity, and their interactions and changes through time.

The concept of adaptive management is often presented as a simple learning cycle

(fig. 2). How people interpret this simple concept is at the heart of questions that frequently arise, including:

How are goals set and changed?

Does management on the ground have to change?

Does adaptive management just mean adding monitoring or increasing public participation?

Does research fit in?

What is the role of information management?

Figure 2. The Northwest Forest Plan (ROD 1995) view of adaptive management as a process of feedback (monitoring and evaluation) and adjustment (plans and actions).

The term "adaptive management" was first proposed for use in natural resource management by Holling (1978) and Walters (1986). Various perspectives on adaptive management are rooted in parallel concepts in business (total quality management), in experimental science (hypothesis testing), and in systems theory (feedback control), to mention a few (Bormann et al. 1994b). All of these views of adaptive management have a common thread: feedback and adjustment. Feedback is knowledge or data on the effects or results of an action, purposefully collected and used to improve future actions. Adjustment is using knowledge and data produced by feedback to redirect subsequent action.

Changes in goals drive changes in policies (defined here as plans for regions, Forests, and Districts) that guide prescriptions (defined here as management plans for specific places) and practices (defined as activities used in prescriptions). Information gained from policies, prescriptions, and practices can help people evaluate and change goals. The three major approaches to adaptive management--reactive, passive, and active (Hilborn 1992)--differ mainly in the nature of their feedback and adjustment steps, and the quality of information they produce to influence future goals (fig. 3).

In the reactive approach, change is driven by stimuli external to the management system, including Congress, lawsuits, public reactions, and research findings. Certainly, this approach is a type of adaptive management because feedback occurs and adjustments are made. Problems arise when different stimuli conflict, and the rate of change outstrips the rate of learning. Crisis management tends to emerge, and creating and maintaining a long-term strategy becomes extremely difficult. Scientists and citizens supply criticisms that may or may not be constructive and may or may not be strongly considered by managers.

The passive approach recognizes that more can be learned from a management action if attention is paid to what actually happened. Learning is advanced when the questions and anticipated outcomes are clearly defined and monitoring plans are written before management takes place. When a commitment to monitoring wanes, the strategy reverts to a reactive one. This approach requires patience to allow sufficient time for learning. Scientists and citizens continue to offer criticism from outside; they may also help to frame questions, anticipate outcomes, and help design and implement a monitoring plan.

An active approach seeks to learn more rapidly than under reactive or passive strategies by designing suites of policies that can be directly compared in "management experiments" that become the focus of monitoring and evaluation. Because an active approach compares different policies simultaneously, learning is more rapid than a passive approach that compares different policies sequentially through time.

Global trends in climate, air and water pollution, energy supply and demand, and population growth will rapidly change societal demands into the next century. Time for society to learn to manage sustainably is limited.

Land management is becoming much more complex

A focus on a single resource or even groups of resources is being supplemented with concerns over biodiversity, long-term site productivity, global change, and public participation. No longer do single disciplines represent each resource in developing independent "best" practices.

Research is difficult at the scale of management

Research agencies and universities have always had difficulty providing needed information at the scales of geography, time, and complexity now called for. Managers and management systems must actively produce information along with other resources to inform future decisions.

Incentive system for continued monitoring

Clear and informative comparisons in management experiments create incentive to continue monitoring to help avoid slipping back to a reactive mode.

Learning can extend beyond Federal lands

Federal lands occupy a small fraction of forests in the United States. The actions of private and industrial landowners will determine in large part whether sustainability can be reached. Having direct comparisons of different policies that people can observe for themselves is more likely to change policy than arguments over existing scientific knowledge and best practices.

Concepts of active adaptive management are being developed and tested in a pilot study on the Oregon Coast Province, one of 12 Provinces created in the Northwest Forest Plan (ROD 1994). The Province, about 35% Federal land, contains most of the Siuslaw National Forest and parts of the Salem and Eugene Districts of the Bureau of Land Management. Scientists and Province managers are seeking to develop possible active adaptive management alternatives and to explore possible new roles for citizens as well.

Multiple policies and multiple prescriptions would be compared in management experiments (fig. 4). Because policy and prescription experiments will take many years to produce needed information, retrospective studies would seek to learn from past events. Because management experiments will be very complex and difficult to interpret, more controlled research experiments would seek to tease apart causes and effects. Ultimately, the role for research is to anticipate future problems and to build new theory on which new prescriptions and policies can be based. Management experiments, with or without replication and randomization, would be carried out, primarily by managers, at the scale and complexity of management.

Figure 4. An active adaptive management system to link management strategies with societal values and scientific theory through collaborative design of management experiments and associated research. Managers would focus on simultaneously implementing sets of different policies (B1 to Bn) and prescriptions (p1 to pn).

Citizens could play a new role in Federal land management by helping to design policy and prescription experiments and plans to monitor them. We hypothesize that, the more individual treatments in a policy experiment reflect the views of large societal groups, the greater likelihood that more people will relate to and support management because their ideas will be implemented somewhere on the landscape. Past approaches with a single policy either reflected one group's view, or worse yet, sought to "average" societal views, producing a "consensus" or a least-common denominator with little support from anyone. Understanding why single policies have been fluctuating widely and have not been strongly supported is not difficult. In effect, implementing multiple policies rather than a single one endorses, as a starting point, multiple world views. Perhaps, even in these largely cynical times, many inquisitive citizens and political leaders could support an open evaluation of multiple world views.

This prototype plan for the Oregon Coast Province is far from mature. A prescription experiment being implemented and a proposed policy experiment illustrate a range of possibilities for active adaptive management on Federal forest lands.

A prescription experiment is a management activity--not to be confused with research--that compares different ways of achieving a policy objective on a specific forest area (fig. 5). Implementing a prescription experiment may or may not include a timber sale, an environmental assessment, or an environmental impact statement. A prescription experiment is more complex than a "practice experiment", in that each prescription is a series of practices. These experiments may sometimes be designed to simultaneously test scientific theories; then researchers would monitor additional variables. Prescription experiments differ from many past administrative studies because they explore a wider range of treatments and will be based on closer collaboration with scientists and citizens.

Most prescription experiments will require random allocation of treatments to areas that are initially similar. Increasing the size of the treatment areas may reduce the variability in initial conditions by, for example, including more types of past management or topographic position in every treated area.

First, a basis for creating a range in policies is needed. We started by postulating that citizens tend to hold one of two fundamentally different philosophies about managing forests. One world view, widely held, is that the "hand of man" is likely to have complex, unpredictable, and usually adverse effects on other species and ecosystems. The more hands-on the management, the more likely are unpredictable negative effects, under this view. Post-settlement people are not thought of as integral parts of nature, and nature is thought to be its own best restorer. Often, pre-settlement conditions are thought to be more desirable than are modern-day conditions. An alternative world view is that people are considered more of an integral part of nature, and people can restore or even improve nature when management is careful and thoughtful. Past positive and negative effects of human actions are recognized as a starting point for shaping the future. Both natural and management activities are thought to help restore ecosystems and build sustainability.

The Northwest Forest Plan focuses on restoring old-growth and riparian conditions and at the same time producing commodities to support local communities. In essence, the Plan seeks to blend hands-on and hands-off philosophies by allocating some land for hands-on management and others for hands-off management. A Province policy experiment can be designed that does not violate the Plan or substantially change resource production goals. Rather, the experiment would seek to redistribute management across the landscape to learn from management more efficiently.

Treatments in this experiment could be applied to all of the 21 large watersheds that have more than 50% Federal land in the Province. Watersheds with less than 50% Federal land were excluded under the assumption that interpreting responses from these watersheds would be difficult. Three treatments are proposed that would be assigned randomly to watersheds dominated by Federal land (fig. 8). Prescription experiments would be concentrated on the large-river watersheds (of the Nestucca, Alsea, and Siuslaw rivers).

Figure 6. Proposed large-watershed policy experiment comparing policies that are based on hands-off or hands-on philosophies for restoring old-growth and riparian ecosystems and producing some commodities for local communities. The mixed hands-on, hands-off treatment is an assemblage of smaller policy treatments (see fig. 7).

Hands-off treatment

In seven of the large watersheds selected at random, apply the standards and guides from the ROD without local modification. This treatment is not entirely hands off because it includes harvesting on the ridgetops where the matrix lands are. Light commercial thinnings will also occur in some areas outside the pre-watershed-analysis riparian buffers.

Hands-on treatment

In another seven of the large watersheds selected at random, the ROD would be applied by distributing management throughout the landscape as much as possible by using local knowledge and judgment gained through watershed analysis, Provincial assessments, and experience. Active restoration, including wide spacings in intermittent riparian buffers, will begin after watershed analysis.

Mixed hands-off and hands-on policies on small watersheds

Although the comparison of hands-on and hands-off policies will be interesting and provide valuable management, a wider range of policies is certainly possible, but will not fit on the limited number of large watersheds. For this reason, four other policies are proposed for the smaller watersheds within the remaining seven large watersheds, selected at random. These treatments would broaden the range of policies studied from more strictly hands-off to hands-on management under the stewardship concept (fig. 7).

Proposed small-watershed treatments:

Minimal management, restricting human influence and access, to let pre-European conditions develop with minimal human intervention. This treatment could include removing or closing roads and recreation sites. Fire suppression might be limited to where fires are likely to move into other watersheds. Underburning may be considered on some sites.

No management, exploring a possible policy of minimal funding for Forest Service management and research. No management may mean reducing fire protection, closing roads, and abandoning recreation sites.

Hands-off management, reserved for future treatments yet to be conceived. New ideas from prescription experiments and other sources are likely to emerge. These areas would be set aside to be added into the experimental mix later.

Hands-on management under the stewardship concept, where private citizens or corporations would compete for contracts to manage small watersheds to achieve hands-on management objectives.

The idea of random allocation of treatments worries many managers, who have spent careers seeking to apply a single consensus view by finding the best mix of practices, designed for each place. Managers are also concerned about the effect of these designs that might not result in a policy that fits well with the dominant philosophy of a nearby community. As a result, managers may decide to back away from a strictly random design. If nearby watersheds can be shown to be quite similar, then a non-random allocation of treatments would provide important information, but perhaps not of the quality of a random design. A comparison of approaches on the quality and quantity of information produced will help managers make the tradeoffs between production of information and other resources.

Adaptive management works with any given goal. In addition to improving policies over time, adaptive management focuses on providing information to change future goals by comparing alternative policies, prescriptions, and practices geared to achieve currently stated goals.

Why does management on the ground need to change?

Most past management appears to fall in the reactive strategy; therefore, changes are needed to implement either the active or passive models. The magnitude of these changes, however, may not be as large as they first appear. The range of management actions may be somewhat similar, but they may be distributed differently to maximize learning.

Why is adding monitoring not sufficient?

Monitoring by itself is not sufficient to implement an active adaptive management strategy because too many variables are possible to monitor. Monitoring focused on management experiments will be more efficient than monitoring for passive adaptive management--by helping to rule out some random factors, like changing climates and shifts in ocean currents, and by comparing policies simultaneously, rather than sequentially through time. Coordinating inventory and monitoring at multiple geographic scales and across issues and disciplines is required for either a passive or an active strategy.

Why is public involvement and participation essential?

To build trust in management, managers need to find ways to get input that is meaningful to citizen participants, and a variety of social values need to be better reflected in on-the-ground management. Management experiments may provide a means to reflect those values in side-by-side management and a focus for participation through design, monitoring, and analysis of these experiments.

Why does planning have to change?

Added steps are needed in planning to focus on increasing learning from management by designing management to produce information simultaneously with other resources. An active strategy explicitly recognizes information as a resource to be managed and produced like other resources. For example, important information is contained in existing stands and landscapes, in the form of existing studies, side-by-side ownership or disturbance comparisons, fence-line comparisons, old plantations, species trials, tree rings, and sediment deposits, to name a few. Management plans must seek to manage, grow, and harvest information resources. Perhaps planners, collaborating with scientists, can identify and manage these resources, including identifying priority retrospective studies to extract knowledge products. Ultimately, knowledge production should be identified as a "purpose and need" in NEPA documents.

Why will research need to become part of the management system?

A key to active adaptive management is better coordination and collaboration between managers and scientists (fig. 8). A new look at roles and responsibilities for managers and scientists is sorely needed. Greater interaction should not mean, for example, that scientists will be doing less science. A new role for researchers will be to assist in planning and designing management to produce knowledge. The most valuable knowledge that scientists can give managers is the knowledge of how they can learn more efficiently by themselves. If management experiments are designed well, they can help to test scientific theory as well as management strategy; thus, monitoring data could provide opportunities for research publications. Traditional roles for scientists in synthesizing existing knowledge, producing new knowledge and options, identifying problems in advance of crises, and managing scientific theory remain.

Why is a greater emphasis on information management required?

Complexity and knowledge requirements increase exponentially as more objectives are identified under sustained-ecosystem management. Expanded monitoring will add to the information burden. Active strategies will require even more record keeping and wider access to information. New information technologies appear likely to be able to handle most of the new requirements, but investment in an adequate hardware and software infrastructure is essential. An often neglected area of information management is synthesis and translation of information between users with vastly different backgrounds. Especially important are people who can translate and communicate complex science and management strategies to the public effectively.

How do we integrate management activities into an adaptive management system?

Since the Northwest Forest Plan was introduced, an improved adaptive management system has been proposed (fig. 5) that identifies a sequence of activities to blend management and research and facilitate effective learning (Bormann et al. 1994b). Information synthesis is especially important, given the large and complex information requirements. Feedback must include a wide array of monitoring, public input, and research that challenges the system. Implementing linked management and research projects is central to generating information that can feed future information synthesis and adjustment actions. Current planning and decisions must jump-start the cycles by identifying priority linked actions, feedback, and information synthesis

Can we afford to learn from every application of practice, prescription, and policy? Not likely. Can we learn fast enough to realistically approach sustainability? I hope so.

Bormann, B. T.; M. H. Brookes; E. D. Ford; A. R. Kiester; C. D. Oliver; J. F. Weigand. 1994a. A framework for sustainable-ecosystem management. Gen. Tech. Rep. PNW-GTR-331. Portland, OR: U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station. 61 p.

Bormann, B. T.; P. G. Cunningham; M. H. Brookes; V. W. Manning; and M. W. Collopy. 1994b. Adaptive ecosystem management in the Pacific Northwest. Gen. Tech. Rep. PNW-GTR-341. Portland, OR: U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station. 22 p.

Hilborn, R. 1992. Can fisheries agencies learn from experience? Fisheries. 17(4):6-14.

Holling, C.S., ed. 1978. Adaptive environmental assessment and management. New York: John Wiley and Sons, Inc. 377 p.

ROD. 1994. Record of decision for amendments for Forest Service and Bureau of Land Management within the range of the northern spotted owl. USDA Forest Service, USDI Bureau of Land Management. Variable pagination.

Walters, C.J. 1986. Adaptive management of renewable resources. New York: McGraw Hill. 374 p.