SKIPJACK IN HAWAII FISHERY 



295 



F^GUKE 11. — Oahu region catch statistics, 1952 and 1953, biweekly periods. 



The results of these tendencies in catch and 

 effort curves on the catch per imit-of-effort curve 

 are as follows: (1) During midyear when effort 

 tends to be constant, the catch per unit of effort 

 will closely follow the fluctuations in catch. (2) 

 During the onset and decline of the season, effort is 

 changing more rapidly than catch and the catch 

 per unit of effort will change at an intermediate 

 rate. Random fluctuations in the catch per unit- 

 of-effort at this time may be somewhat at variance 

 with those in the catch, as occurred in the interval 

 from period 24, 1952, to period 3, 1953. 



Correlation analysis between raw catch and 

 catch per unit-of-effort data yields a coefficient of 

 + 0.92. It was necessary to use rank correlation 

 methods (Snedecor, 1956, p. 190) because the dis- 

 tribution of catches is skewed toward small 

 catches. In order to determine the amount of 

 agreement between the random fluctuations in the 

 two indices, the first differences were correlated 

 and yielded a coefficient of +0.79. Conventional 

 methods were used here because the first differ- 

 ences are distributed more normally tiian the 

 original series. The probability that correlation 

 coeflScients this large would occur by chance is less 

 than 0.01. 



The variate-difference technique (Kendall, vol. 

 II, p. 387-390) was used to obtain an estimate of 

 the variance in the random component of each 

 index compared to the variance of the original 



series. The values obtained for the random com- 

 ponents were 24 percent for the raw-catch series 

 and 18 percent for the catch per unit-of-effort 

 series. Since it exhibits a smaller random com- 

 ponent, the catch per unit-of-effort series appears 

 to be somewhat more reliable than the catch as an 

 index of slapjack availability, but only slightly so. 

 In summary, the raw catch is almost as accurate 

 as the catch per unit of effort in indicating the 

 seasonal variation in skipjack abundance in the 

 Hawaiian fishery. During the middle of the year 

 the raw catch is in good agreement with the ran- 

 dom fluctuations in catch per unit of effort, but 

 during the off-season of the fishery, when effort 

 and catch are either declining rapidly or are at a 

 low level, random fluctuations in catch per unit of 

 effort may not vary in agreement with fluctuations 

 in catch. For most puqjoses, total catch would 

 appear to be as useful an indicator of availability 

 as the catch per unit of effort, especially in prob- 

 lems where the seasonal trend in the fisheiy is 

 apparent. 



USE OF CATCH RECORDS TO DETERMINE 

 POPULATION COMPOSITION 



The catch statistics of the Hawaiian fishery are 

 the only continuous source of information which 

 provides a means of assessing the nature of the 

 skipjack population which supports the fishery. 

 Inferences about this population must be made 



