ZOOPLANKTON ABUNDANCE IN THE CENTRAL PACIFIC 



135 



Table 13. — Cruise 8: Analysis of variance of volumes of 

 paired samples taken by day and by night in about the 

 same latitude 



[Two criteria of classification] 



Least significant difference for latitudinal means = . 01 r.!> 



•"Indicates a highly significant (P<0.01) mean square value. 



Thus, despite the important variation resulting 

 from diurnal migration of the zooplankton, there 

 still remain in our data significant differences 

 among latitudes. Zooplankton populations oc- 

 curring near the Equator are significantly greater 

 in abundance than those of adjoining areas. 



Variations between cruises, latitudes, and longitudes 



To examine differences between cruises and be- 

 tween longitudes and to inspect further the differ- 

 ences among latitudes, an analysis of variance with 

 multiple classification as outlined by Snedecor 

 (1946, pp. 304-309) was utilized, employing as 

 many of the cruise-5 and cruise-8 data as were 

 available for similar latitudes and longitudes 

 (table 14). The analysis is of the same general 

 type as that used by Winsor and Clarke (1940) 

 in their study of variation in the catch of plank- 

 ton nets. 



Table 14. — Analysis of variance of volumes (cc./m.*) of 

 zooplankton samples collected on cruises 5 and 8 



[Three criteria of classification] 



Least significant difference for latitudinal means-=.0218 



• Indicates a significant (P<0.0'>) mean square value. 

 •• Indicates a highly significant (P<0.01) mean square value. 



The analysis was first carried out in full to 

 determine the significance of the first-order inter- 

 actions. As these proved to be nonsignificant, 

 they were pooled (following Kendall L948, p. 201) 

 with the second-order interaction to give a new 

 sum of squares with 64 degrees of freedom and 

 a new mean square which was then used as the 

 error term for testing the main effects. The fol- 

 lowing observations may be made from the tests 

 of significance : 



1. No significant differences (P>0.05) are demon- 

 strated between the means for series (longitudes) 

 (0.0284 cc./m. 3 for 172° W. longitude, and 0.0326 cc./m.' 

 for 158° W. longitude). 



2. Significant differences (P<0.01) are demon- 

 strated between the means for cruises (0.0366 cc./m.' 

 for cruise 5 and 0.0244 cc./m.' for cruise 8). 



3. Significant differences (P<0.02) are demonstrated 

 among the means for latitudes (these vary from 0.0550 



