The Panel thus stressed the need to explore each of these criteria when 

 contemplating a modeling effort. The Panel further stressed the point that 

 these criteria should be examined very early in the planning stages of the 

 modeling effort. Model development should not progress beyond the conceptuali- 

 zation stage until the above criteria have been considered in detail and met. 



Another viewpoint expressed by the Panel concerns the modeling effort. 

 This viewpoint focused on the importance of ecosystem modeling as a research 

 tool and framework for formulating problems, e.g., in evaluating logic, 

 information flow, and structuring the development of important questions which 

 would not otherwise evolve. Planning and development and/or application of an 

 ecosystem model will always require that a screening or preliminary assessment 

 by conducted by both "expert modelers" and "decisionmakers." 



The Panel expressed concern that there was a serious lack of information 

 and data for many ecosystem modeling applications, particularly when these 

 applications were oriented to specific societal problems. The Panel feels that 

 this serious deficiency stems from conceptual gaps in biology and socio- 

 economics; if the utility of the modeling process is to be improved, these 

 conceptual gaps will need to be identified and rectified. The Panel further 

 noted that unsuccessful ecosystem modeling attempts sometimes resulted not from 

 the model itself, but from the quality and applicability of of the data base 

 used to conceptualize and quantify the model. Thus, if a model is structured 

 to be highly dependent on the quality and applicability of the input information, 

 the modeling results will be no better than the data base. Inadequate input 

 information will always yield less-than-satis factory modeling results under 

 these circumstances. On the other hand, the Panel recognized the need for some 

 models to be based upon inadequate information/data bases so that these models 

 could be useful in specifying data needs and design of sampling programs and 

 studies to elucidate unknown or poorly understood ecological mechanisms and 

 processes. A careful consideration reveals this to be another difference 

 between the types of models and their intended purposes (as suggested above). 

 If predictive models are to be useful, they must be designed for application to 

 real world, resource management decisions and must be based on good, 

 scientifically defensible data. Research models, on the other hand, may be 

 designed on inadequate data and theoretical constructs since one of their key 

 values is to delineate the information needs that must be addressed in the next 

 round of data generation and investigation. 



In concluding its deliberations, the Panel felt that there is a strong 

 potential for ecosystem modeling applications within the context of a socio- 

 economic framework. Modeling could provide administrators and decisionmakers 

 with a prediction of the outcomes of their decisions, albeit this would have 

 more practical value if presented as probabilities of occurrence. As such, 

 these models would improve the decisionmaking process. However, the value that 

 the Panel saw in the ecosystem modeling approach was not without qualification. 

 In a practical sense, models need to be more relevant to the needs of 

 decisionmakers. This will occur, not through simply better communication 

 between modelers and decisionmakers, but through explicit plans for modelers 

 and decisionmakers to interact throughout the course of the modeling exercise. 

 The objective is for the decisionmaker and his or her staff to obtain a general 

 understanding of the operation and critical elements of the model, its strengths 

 and weaknesses in a given application, and its accuracy and precision. Of 

 articular importance in this regard is the need for modelers to develop models 



267 



