This period^ from Augiist tc November, marked the end of the 



1951 season for the live-bait skipjack fishery of Hawaii o While our 

 data show that the zooplanktoa abuadancs decreased considerably during 

 these same months (fige 12) „ we have no evidence that this reduction 

 in the amount of zooplankton was a limiting factorg but rathsr the 

 decline in both faional elements may have had a common cause in some 

 other factor of the environment o 



The skipjack catch for the Territory of Hawaii for the summer 

 of 1951 was considerably above average, whereas the catch for the 1952 

 season was somewhat below average and seemed particularly poor after 

 the excellent season of the previous year»^'/ Unfortxinatelyj, there were 

 no plankton cruises during midsummer of 1952 to provide data compar- 

 able to those of 1951o Smith cruise 17, howeverj, occurred in Septeirber 



1952 during the decline of the skipjack season and provided an adjusi;- 

 ed zooplankton mean (26,4 cCo/lOOO mo-^) not differing greatly from 

 that of Smith cruise 10 (25 o6 cCo/lOOO mo^) of July 1951 and somewhat 

 larger than the mean (20o3 cCo/lOOO mo^) of Smit h cruise 12 of October- 

 November 1951 „ From this small amount of evidence v^e can conclude at 

 least that the marked difference in skipjack catch between the 1951 

 and 1952 seasons was not a reflection of a corresponding reduction in 

 zooplankton abiondancee 



^/ According to Snedecor (1946) the analysis of variance is a valid 

 test of individual and population differences if the groups of 

 samples are randomly drawn from a normally distributed population 

 and have similar varianceso He states^ however^ (po 221) "ooobut 

 it has been found that little bias is introduced into the test of 

 significance by moderately skewed distributions o" Much of our 

 plankton volume data possesses a moderate skewness which is nct^ 

 in many instances j corrected by the usual logarithmic transforma^ 

 tiono Since the distribution of the data included in tab is 9 was 

 improved by a logarithmic transformation., and since the F value for 

 visits was quite near the Oo05 level of probability^ it iwas consider- 

 ed advisable to recalculate the analysis using the logarithms of 

 the volumes, The new F values were Ool3 for stations and 2ol9 for 

 visitsj which are very similar to those obtained previously and do 

 not change our conclusions <> The hypothesis that the groups have 

 similar variances was tested by Bartlstt's Test (Snedecor 1943^ 

 Po 250) o The chi-square value obtained was well belOT\'- the Oo05 

 level of probability, permitting the conclusion that the separate 

 variances of these 6 groups do not differ sufficiently to disturb 

 the validity of the F tests o We believe that the data in table 9 

 are representative of those included in the other statistical tests 

 appearing in the reports 



_/ From records supplied by the Territory of Hawaii, Division of 

 Fish and Game the average skipjack catch for the season liay to 

 September for the 3 years 1943, 1949, and 1950 was 6,576,000 

 pounds o The 1951 catch for these 5 months was 11^235,000 pounds, 

 while the 1952 catch for this same period dropped to 5^795,000 

 pounds o 



34 



