90 



evaluated, but other related tools, such as fit to a lognormal 

 distribution, should be looked at as well. It may be that H* diversity 

 is simply not the appropriate index to illuminate the finer nuances of a 

 heterogeneous system like Casco Bay. 



Classification Analysis 



Classification or cluster analysis is a useful way of objectively 

 examining patterns in complex data sets which cannot easily be uncovered 

 by other techniques. It is a hypothesis generating technique which 

 can suggest relationships between biological and physical factors that 

 may be causal to observed community distributions. Although the method 

 is numerical, and therefore objective, the interpretation is subjective. 

 One way to minimize subjectivity is to produce a large number of 

 site-groups and species groups and then combine them until the most 

 meaningful pattern is produced using a minimum of groups. The goal is 

 to produce the most comprehensive but simple explanation for the 

 observed phenomena. 



We are at an intermediate step in this process. We have produced 

 station and species dendrograms based on faunal data. We have defined a 

 moderate number of site-groups and species-groups and have initiated a 

 comparison of the groups with the extrinsic factors now on hand. In 

 order to conserve resources we intend to wait until all of the physical 

 data are available before finalizing our interpretation. We have, 

 however, included our analysis to date because it does provide useful 

 information about Casco Bay and illustrates the strength of the analysis 

 that will be available soon. 



