used in statistics? How can a model be used to set priorities for supplemental 

 data gathering for calibration purposes? 



5. Model validation . A definitive theory of this critical phase in 

 modeling procedures has not yet been written. There are tacit philosophical 

 and logical considerations which underlie present validation practice. These 

 need to be drawn together in a formal validation theory. Probably no topic 

 under "Technical Basis of Modeling" is more important than establishing exact 

 rationales, procedures, and criteria for validation, verification, corroboration, 

 etc. 



6. Output to User . Just as a model is a conceptual and mathematical 

 analogy of the real world, so its output for useful purposes should be subject 

 to feedback considerations which relate it to the original problem definition. 

 The problem of simplifying model output into user-relevant information should 



be amenable to the same kinds of formal considerations as model conceptualization 

 itself. 



The surest way to increase confidence in models and modeling is to have 

 them firmly grounded in sound theory. Ecological modeling now is an art; it 

 needs to become a formalized exact science. 



Recommendations 



1. The theory of modeling in all areas should be considered high priority 

 research. There is a dearth of such theory in ecological modeling, and improved 

 theory is certain to increase reliability and usability of models. 



2. The number of possible calibration solutions to a model can be reduced 

 if the model is required to represent adequately process-rate and state-variable 

 measurements. This will not eliminate problems of nonuniqueness . However, we 

 may be forced to use such models; and calibration against both types of data 

 substantially reduces the degrees of freedom. 



3. Studies of applicability of stochastic modeling procedures are 

 encouraged. Model uncertainty or confidence levels should accompany any model 

 projection made. 



III. INTERACTION AMONG MANAGERS, MODELERS, AND FIELD SCIENTISTS 



The success of any modeling effort for resource management depends heavily 

 upon clear, continuous communication between the modelers and the managers who 

 will use the model results. This fact is generally accepted, but too often 

 lines of communication are severed before the management decision is made. We 

 feel the dialogue should continue up to and beyond the decision and lead to 

 refinements in the model and in the data collecting scheme. By extending 

 communication, both the short-term decision-oriented management goals and the 

 long-term ecosystem-oriented scientific goals will be realized. 



Several studies could be cited to illustrate these points. Moreover, 

 formats for the interaction of managers, modelers, and field scientists are 

 available (e.g., Adaptive Environmental Assessment Technique). It is beyond 

 the scope of this report to treat the subject exhaustively. Instead, we suggest 



249 



