At this point, it is useful to note that the decision analysis framework that we are promoting is not mutually exclusive with many of the concepts articulated in the conventional approach. Rather, the decision analysis framework works hand-in-hand with the different types of monitoring and sampling designs to complete the informational loop among goals, decisions, and actions.

Figure 5.1 illustrates this informational loop and the essential role played by decision analysis. The process originates with general goals and direction at multiple scales. These goals and direction may be articulated in the purpose and need of regional directives such as the President's Northwest Forest Plan or individual Forest Plans, or from more local concerns. The first step in making goals and direction operational is a decision analysis, which can be subdivided into more discrete steps.

The first steps in decision analysis are to identify the relevant ecological processes at appropriate scales, and identify the range of management options that are available. These two component feed into the more formal analysis of options, which included use of the influence diagrams as discussed in previous sections. The formal analysis explicitly includes a sensitivity analysis to identify key uncertainties and dependencies, suggesting important indicators for monitoring. A potential list of indicators can be further screened for technical or administrative constraints (i.e., do we know how and can we afford to measure each indicator). A rough estimate of the feasibility and cost of monitoring a particular set of actions can help in choosing among alternatives.

The formal analysis should indicate the relative preferableness of different options, given the overall goals. This leads directly to implementation decisions. These decisions initiate an entire set of consequences not shown on the diagram, that is, the actual implementation of management activities. Making an implementation decision also leads back to the formal analysis, but for a different purpose. Now the analysis is conditioned on the decisions that have been made, and the monitoring design begins in earnest. Using the influence diagrams and screening indicators for feasibility and affordability, a key set of management and ecological indicators are identified. This set of indicators is passed to the sampling design phase wherein the previous discussion about model-based versus sample-based designs comes to the fore.

Once a suitable sampling design is chosen, data collection and analysis soon follows. The results of this analysis can lead back to the formal analysis step or to an evaluation phase. Depending on the results of the evaluation of progress, feedback to the implementation decision step may occur.

The data collection and analysis step, and its feedback to the formal analysis of options step encompasses all of the monitoring types identified in the conceptual model (Figure 5.2). Implementation monitoring, for example, ensures that decisions were implemented as planned. Baseline monitoring is necessary to set up initial conditions in the network and parameterize some of the conditional dependencies. Validation monitoring provides the information necessary to examine the overall diagram structure for its agreement with observed data, and updates conditional probabilities matrices based on new observations. Finally, the overall linkage between decisions and final outcomes provides the linkage necessary for effectiveness monitoring, in the sense of measuring progress towards an objective in a manner that the true effect of management actions is distinguishable.

One of the insights that are sure to arise from the formal analysis of options is that there are certain management options that lead to outcomes with potentially high value, but also high uncertainty. That is, the most likely outcome given the option may be highly favorable, but the chance of unfavorable outcomes is also high. In some cases, this uncertainly is an inherent property of the natural system and cannot be avoided. In such instances, the management option is inherently risky, and there is no apparent way to avoid the risk.

Many times, however, the reason an option has high uncertainty is because of ignorance about how the system operates. It may be that the particular option has never been tried under a given set of circumstances, so one cannot say with certainty what the result might be. If the potential gain form the action is high, it might seem rational to try the option on a limited scope and carefully observe the outcome. This type of active probing of the system to increase knowledge of system behavior is central to the idea of adaptive management. Such management experiments must be carefully designed such that the results can be accurately attributed to the correct causal factors. Such designs cannot be achieved without something akin to the influence diagram approach.

Next Section | Table of Contents | References