An approach to management of natural resources that emphasizes how little is known about the dynamics of ecosystems and that as more is learned management will evolve and improve. Natural systems are very complex and dynamic, and human observations about natural processes are fragmentary and inaccurate. As a result, the best way to use the available resources in a sustainable manner remains to be determined. Furthermore, much of the variability that affects natural populations is unpredictable and beyond human control. This combination of ignorance and unpredictability means that the ways in which ecosystems respond to human interventions are unknown and can be described only in probabilistic terms. Nonetheless, management decisions need to be made. Adaptive management proceeds despite this uncertainty by treating human interventions in natural systems as large-scale experiments from which more may be learned, leading to improved management in the future.
A key first step in the development of an adaptive management program is the assessment of the problem. During this stage, existing knowledge and interdisciplinary experience is synthesized and formally integrated by developing a dynamic model of the system. This modeling exercise helps to identify key information gaps and to postulate hypotheses about possible system responses to human intervention consistent with available information. Different management policies have to be screened in order to narrow down the alternatives to a few plausible candidates.
The second stage involves the formal design of a management and monitoring program. To the extent that new information can result in improved future management, adaptive management programs may include large-scale experiments deliberately designed to accelerate learning. Some management actions may be more effective than others at filling the relevant information gaps. In cases where spatial replication is possible (such as small lakes, patches of forest, and reefs), policies that provide contrasts between different management units will be much more informative about the system dynamics than those that apply the same rule everywhere. There are other barriers to the implementation of large-scale management experiments. Experiments usually have associated costs; thus, in order to be worthwhile, benefits derived from learning must overcompensate short-term sacrifices. Choices may be also restricted by social concerns or biological constraints, or they may have unacceptably high associated risks.
Once a plan for action has been chosen, the next stage is to implement the program in the field. This is one of the most difficult steps, because it involves a concerted and sustained effort from all sectors involved in the use, assessment, and management of the natural resources. Beyond the implementation of specific initial actions, putting in place an adaptive management program involves a long-term commitment to monitoring the compliance of the plan, evaluating the effects of management interventions, and adjusting management accordingly.
No matter how thorough and complete the initial assessment and design may have been, systems may always respond in manners that could not be foreseen at the planning stage. Ecosystems exhibit long-term, persistent changes at the scale of decades and centuries; thus, recent experience is not necessarily a good basis for predicting future behavior. The effects of global climatic change on the dynamics of ecosystems, which are to a large extent unpredictable, will pose many such management challenges. Adaptive management programs have to include a stage of evaluation and adjustment. Outcomes of past management decisions must be compared with initial forecasts, models have to be refined to reflect new understanding, and management programs have to be revised accordingly. New information may suggest new uncertainties and innovative management approaches, leading to another cycle of assessment, design, implementation, and evaluation.