GRAHAM: RETENTION OF LARVAL HERRING 



Table 1. — Statistical procedures for a 2^ factorial design using larval catches per 100 m^ of water strained from 

 an experiment during Nov. 8, 1967. I) Factors are coded, II) Catches are assigned to the factorial order, III) 

 the sums of squares are determined, IV) they are applied to an analysis of variance, and V) a conclusion is de- 

 termined from the analysis. 



Factors and their codes (A). 



Landward location 

 2.64(8) 5.54(8) 



86.09(6) 



48.60(6) 



Landward location 

 33.28(7) 24.73(7) 



Ebb 



Deep 



0.44(1) 



4.73(1) 



26.75(5) 



4.58(5) 



II. Determination of the sums of squares lor the 2^ factorial design using logarithms of the catches; 1 has been added to each characteristic. 



rates are used because they are more familiar 

 than logarithms as measures of concentra- 

 tions of larval fishes in their environment. 

 The results of the statistical analysis are given 

 by Graham and Davis (1971), Significant dif- 

 ferences between catch rates with depth and lo- 

 cation and the interaction of these two factors 

 with tide occurred primarily in the autumn and 

 spring. Of eight experiments conducted from 

 late October to early December (1965-67) five 

 had significant differences between larval catch 

 rates. Of 10 experiments completed between 



mid-March and mid-May (1966-68) six had sig- 

 nificant differences between catch rates. The 

 distributions of larvae from experiments that 

 had significant differences in larval catch rates 

 will be described, and an explanation for the 

 absence of significance in the other experiments 

 will be discussed. 



RESULTS AND DISCUSSION 



Concentrations of larval herring were evident 

 within the channel when the distributions of 

 catch rates that differed statistically within the 



301 



