mean sizes were plotted against the gonad index 

 in lieu of the 95th centile of the total ova-size 

 frequency distribution. This difference should 

 not affect greatly the comparison of my results 

 with those of Schaefer and Orange. 



Figure 5 is a scatter diagram showing my 

 results, with the regression line obtained by 

 Schaefer and Orange for their Area II data 

 superimposed. Schaefer and Orange concluded 

 that the gonad index was a reasonably reliable 

 measurement of the degree of ova development; 

 however, the largest gonad index they found was 

 about 36, as compared to my 96. My data fit 

 Schaefer and Orange's regression line fairly well 

 up to a gonad index of 36. Above this value the 

 relation seems to break down. There is a big 

 overlap in gonad index between developing and 

 advanced ovaries; therefore, it was not possible to 

 make any inferences about the stages of develop- 

 ment of skipjack in this area from the gonad index. 



STAGE OF SEXUAL DEVELOPMENT AND 

 SCHOOLING BEHAVIOR 



As mentioned earlier, the sampling of skipjack 

 by pole-and-line fishing was such that each sample 

 represented fish captured from a single school. 

 This circumstance made it possible to examine 



the within-scliool distribution of developmental 

 stages. 



A gross examination of the data (table 2) indi- 

 cates that skipjack in the different stages of 

 development were not distributed randomly within 

 schools. Although fish in all stages of sexual 

 development were found simultaneously in certain 

 months, there was no instance in which all three 

 stages of development were found within a school. 

 Most of the schools were represented by skipjack 

 in a single stage of development. A contingency 

 test of homogeneity (Hoel, 1954, pp. 172-175) was 

 applied to the data, testing the hypothesis that, 

 with respect to stage of sexual development, 

 skipjack are distributed randomly within schools. 

 The probability (X 2 =510, df = 144) of obtaining 

 the observed distributions by chance alone was 

 less than 1 in 100. I conclude that there was 

 indeed a relation between schools and stage of 

 sexual development. The fact that fish of similar 

 stages of sexual development tended to occur 

 together, as well as the fact that those in several 

 stages of development were found in a number of 

 months, should be considered in the design of any 

 study of spawning. Obviously, examination of a 

 large number from only a few schools could lead 

 to erroneous conclusions. 



50 



I 



b 



I" 



G 



UJ' — 



Jjtfi 



■a 



-0.6 



w 



p- 



- 



< 

 W 



s 



0.7 MM. 





• .• • 





* * „ * 



» SCHAEFER 3 ORANGE (1956J 



FIG. 10 (AREA II) 



ADVANCED 

 DEVELOPING 

 EARLY DEVELOPING 



20 



30 



40 50 60 



GONAD INDEX 



70 



80 



90 



100 



Figure 5. — Gonad index and mean diameter of largest group of ova. 

 SKIPJACK SPAWNING IN MARQUESAS AND TTJAMOTTJ AREAS 



485 



