Restoration of riverine ecosystems is often stated as a management objective for regulated rivers, and floods are one of the most effective tools for accomplishing restoration. The National Re- search Council (NRC 1992) argued that ecological restoration means re- turning "an ecosystem to a close approximation of its condition prior to disturbance" and that "restoring altered, damaged, O f destroyed lakes, rivers, and wetlands is a high-priority task." Effective restoration must be based on a clear definition of the value of riverine resources to society; on scientific studies that document ecosystem status and provide an understanding of ecosystem processes and resource interactions; on scientific studies that predict, mea- sure, and monitor the effectiveness of restoration techniques; and on engineering and economic studies that evaluate societal costs and benefits of restoration.

In the case of some large rivers, restoration is not a self-evident goal. Indeed, restoration may be impossible; a more feasible goal may be rehabilitation of some ecosystem components and processes in parts of the river (Gore and Shields 1995, Kondolfand Wilcock 1996, Stanford et al. 1996). In other cases, the appropriate decision may be to do nothing. The decision to manipulate ecosystem processes and components involves not only a scientific judgment that a restored or rehabilitated condition is achievable, but also a value judgment that this condition is more desirable than the status quo. These judgments involve prioritizing different river resources, and they should be based on extensive and continuing public debate.

In this article, we examine the appropriate role of science in determining whether or not to restore or rehabilitate the Colorado River in the Grand Canyon by summarizing studies carried out by numerous agencies, universities, and consulting firms since 1983. This reach of the Colorado extends 425 km between Glen Canyon Dam and Lake Mead reservoir (Figure 1). Efforts to manipulate ecosystem processes and components in the Grand Canyon have received widespread public attention, such as the 1996 controlled flood released from Glen Canyon Dam and the proposal to drain Lake Powell reservoir.

With a focus on resources of the Colorado River ecosystem below Glen Canyon Dam, the Glen Canyon Dam Adaptive Management Program has included a variety of experimental policy tests, ranging from manipulation of water releases from the dam to removal of non-native fish within Grand Canyon National Park. None of these field-scale experiments has yet produced unambiguous results in terms of management prescriptions. But there has been adaptive learning, mostly from unanticipated or surprising resource responses relative to predictions from ecosystem modeling. Surprise learning opportunities may often be viewed with dismay by some stakeholders who might not be clear about the purpose of science and modeling in adaptive management. However, the experimental results from the Glen Canyon Dam program actually represent scientific successes in terms of revealing new opportunities for developing better river management policies. A new long-term experimental management planning process for Glen Canyon Dam operations, started in 2011 by the U.S. Department of the Interior, provides an opportunity to refocus management objectives, identify and evaluate key uncertainties about the influence of dam releases, and refine monitoring for learning over the next several decades. Adaptive learning since 1995 is critical input to this long-term planning effort. Embracing uncertainty and surprise outcomes revealed by monitoring and ecosystem modeling will likely continue the advancement of resource objectives below the dam, and may also promote efficient learning in other complex programs.

An Adaptive Environmental Assessment and Management workshop process was used to assist Grand Canyon scientists and managers in developing conceptual and simulation models for the Colorado ecosystem affected by Glen Canyon Dam. This model examines ecosystem variables and processes at multiple scales in space and time, ranging from feet and hours for benthic algal response to diurnal flow changes, to reaches and decades for sediment storage and dynamics of long-lived native fish species. Its aim is to help screen policy options ranging from changes in hourly variation in flow allowed from Glen Canyon Dam, to major structural changes for restoration of more natural temperature regimes. It appears that we can make fairly accurate predictions about some components of ecosystem response to policy change (e.g., autochthonous primary production, insect communities, riparian vegetation, rainbow trout population), but we are moderately or grossly uncertain about others (e.g., long-term sediment storage, response of native and non-native fishes to physical habitat restoration). Further, we do not believe that existing monitoring programs are adequate to detect responses of native fishes or vegetation to anything short of gross habitat changes. Some experimental manipulations (such as controlled floods for beach/habitat- building) should proceed, but most should await development of better monitoring programs and sound temporal baseline information from those programs.

Many case studies in adaptive−management planning for riparian ecosystems have failed to produce useful models for policy comparison or good experimental management plans for resolving key uncertainties. Modeling efforts have been plagued by difficulties in representation of cross−scale effects (from rapid hydrologic change to long−term ecological response), lack of data on key processes that are difficult to study, and confounding of factor effects in validation data. Experimental policies have been seen as too costly or risky, particularly in relation to monitoring costs and risk to sensitive species. Research and management stakeholders have shown deplorable self−interest, seeing adaptive−policy development as a threat to existing research programs and management regimes, rather than as an opportunity for improvement. Proposals for experimental management regimes have exposed and highlighted some really fundamental conflicts in ecological values, particularly in cases in which endangered species have prospered under historical management and would be threatened by ecosystem restoration efforts. There is much potential for adaptive management in the future, if we can find ways around these barriers.

Even unmanaged ecosystems are characterized by combinations of stability and instability and by unexpected shifts in behavior from both internal and external causes. That is even more true of ecosystems managed for the production of food or fiber. Data are sparse, knowledge of processes limited, and the act of management changes the system being managed. Surprise and change is inevitable. Here we review methods to develop, screen, and evaluate alternatives in a process where management itself becomes partner with the science by designing probes that produce updated understanding as well as eco- nomic product.