Community structure can only be understood fully when many kinds of 

 data are available. Some of these data can be obtained from quanti- 

 tative sampling but others must be based on detailed long terra 

 observations. Under most conditions the presence of the observer 

 is a disturbance that lessens the chances of making meaningful 

 unbiased observations. The use of rebreather units, rather than 

 conventional scuba, makes possible longer periods of observation 

 with much less disturbance to the fishes. Future studies should 

 include concentrated work on seasonal changes, feeding habits, and 

 social interactions as well as expanding the observations described 

 here. 



VI-269 



