August 1981 at Giants Neck and resulted in a similarity dendrogram where 

 there is only a 3.2% difference between the replicates on the upper 

 level of the rack and those on the lower level (Fig. 7). Therefore, the 

 vertical bias in the distribution and abundance of species can affect 

 replicated data in a variable way depending on which species, stations, 

 and exposure periods are considered. In extreme cases true replication 

 has been lost for a series of six panels. 



Natural variability . Variability in the data caused from seasonal 

 effects and year to year trends has been reported since 1968. Since 

 1979, the assessment of quantitative data from quarterly collection 

 periods has helped to define seasonal effects. Four well colonized 

 samples, whose consecutive exposure periods are over lapped by three 

 months, have established seasonal trends in community composition and 

 abundance over the last three years (Fig. 8). These trends are based on 

 the seasonal reproduction periods of the various fouling and woodboring 

 organisms in the Millstone Point area. Three years of sampling have 

 provided data concerning how these seasonal trends in species composition 

 and abundance can vary from year to year. 



Adequacy of six replicates 



For this evaluation, the adequacy of six replicate panels for 

 sampling fouling and woodboring organisms will be assessed based on the 

 effectiveness with which species are collected and the percision with 

 which their abundances can be estimated. Species area curves will be 

 used to illustrate how many replicate panels are necessary to adequately 

 sample the community. The 95% confidence limits surrounding the mean 

 will be used to describe the precision with which abundances can be 

 estimated . 



18 



