Assume that triplicate determinations of a given larval species at five locations 

 along the cross-section of an estuary have been made at three depth levels. Since 

 each station location has three replicate samples at each depth, the classifications 

 "depths" and "locations" are completely crossed, while the triplicate determinations 

 provide three replications for each combination of depth and location. In a 

 tabulation of the data there would be three columns corresponding to depths, five 

 rows corresponding to the five locations, and 1 5 cells each containing three replicate 

 observations. 



In the general case there are x^-.q^ observations, i = l,...,p rows, j - l,...,q 

 columns, a = 1,...,t? replicates. The model for the analysis is assumed to be: 



V = ^ + ^ + ^j + ^ij'^^iF- 



In the above model, At is the mean, ^ represents row effects (location 

 variability), and m represents column effects (depth variability). The 

 interaction effects Xj; represent any variations which may be peculiar to a 

 particular combination of station and depth, and the effects Sj^ are normally 

 distributed random components with average value zero for each ij. 



The general analysis of variance table (Table 22-1) for this particular model 

 is shown. From this table the partitioning of the degrees of freedom can be 

 determined as well as the appropriate tests of significance determined. Two 

 points can be made concerning this model. They are: 1) the variation between 

 sample locations can be estimated, and 2) the systematic variations between 

 sample locations can be eUminated from the study of other effects, i.e. depth. 



SINGLE STATION EXPERIMENT BACKGROUND 



A single sample site was chosen in the West Passage of Narragansett Bay, Rhode 

 Island to examine within station variability at a water depth of approximately 50 

 feet. At the time of this report, all of the detailed analyses of the data have not been 

 completed. However, some indication of the nature of the data and the approach to 

 be utilized are presented herein. 



Sampling was conducted on four days, hourly for 24 hour periods. During each 

 hour p »ired 60 cm Bongo net hauls (505 m mesh) were made. One tow was taken in 

 the direction of the tide and the other against it. A data set consisting of 275 samples 

 taken on 28 June, 1 July and 7 July 1 976 are being used for detailed analysis. Oblique 

 tows of standardized length and duration were made in all cases. 



The data collected were separated into two categories: 1) Physical data; and 2) 

 species counts. The species counts were further separated into three classes: 

 ichthyoplankton eggs, ichthyoplankton larvae, and zooplankton. 



340 



