these historical and scientifically valuable data sets are scattered among 

 individuals, universities, research centers, industries, and local, state, and 

 Federal government agencies. For all practical purposes, these data sets are 

 inaccessible to those who have to complete a task or project within the realistic 

 constraints of time, money, and personnel. As a result, many ecological issues 

 are probably addressed incompletely, and environmental decisions are made based 

 on only a small fraction of the total data which is directly pertinent to the 

 issue. 



As a first step towards rectifying this situation, an annotated list of 

 existing data sets could be compiled and regularly updated. Although this 

 listing would never be 100% complete, it would be a valuable starting point for 

 initially identifying relevant data sets and their residence. As a start, many 

 of the larger and more extensive data records residing with government agencies, 

 research centers, and universities could be catalogued. 



Eventually the mechanisms should be developed to incorporate some of these 

 larger data sets into one data bank that performs the statistical and associative 

 functions which data-base management systems commonly perform. This system 

 should be "user friendly," accessible, and have a user-feedback mechanism so 

 that the data sets can be evaluated and updated. 



VII. HIGH RETURN, HIGH PRIORITY RESEARCH ACTIVITIES 

 -Highest Priority- 



1. Model validation tests 



Conduct comparisons of several (minimum of three to five) ecosystem 

 models to one or more Marine Ecosystem Research Center experiments . 1 MERC is 

 also a model in the sense that it is similar to macrocosms of Narragansett Bay. 

 Three or more ecosystem modeling projects of approximately $50,000 each would 

 enable the testing of modeling methodology to see if the models can predict the 

 macrocosm outcome and how well they do it. 



Another approach to model validation is to compare model output and 

 environmental data for several before and after case studies for areas where 

 good data sets exist. Examples might be power plant sitings, sewage outfalls, 

 etc. Results will be directed to questions of how well ecosystem models predict 

 either mean ecosystem state and/or how much variability occurs in the ecosystem. 



2. Environmental quality criteria and cost-benefit analysis 



Further development of methodology, which should include: 



a. identification of simulatable ecosystem properties and products 

 that can serve as useful environmental quality criteria. 



1 The Marine Ecosystem Research Center of the University of Rhode Island 

 is offered here as an example; it is the program which the panelists are familiar 

 with. Similar programs at other institutions would also be suitable for 

 verification, calibration, and testing of model structure and formulation. 



254 



