FISHERY BULLETIN: VOL. 74, NO. 3 



where b = 1 --— ^ = a parameter depicting the 



°° theoretical proportion of 



potential growth in length 



that occurs after hatching; 



or by substituting t = t + /M and be''"' = 1 - =^ 



into Equation (2a) "" 



A+ A, =1,6-"^' + L^a-e"'^') (2b) 



where Af = time increment between points 

 of measured length. 



GROWTH OF STRIPED MARLIN 



Results— Analysis of Cohorts 



Age-groups were successfully separated by sex 

 and by quarter, within years, using the computer 

 program ENORMSEP (Table 1). In general, the 

 mean length estimates for females were slightly 

 larger than those for males of the same age- 

 groups. Quantitatively, the goodness of fit of the 

 separated age-groups to the observed frequency 

 distribution can be assessed with the chi-square 

 values in Table 1. The largeness of the chi-square 

 values indicated poor fit, but it was found that the 

 tails of the distribution, having frequencies too 

 small for the separation of age-groups, contribut- 

 ed disproportionately to the total chi-square value. 

 Qualitatively, the goodness of fit was deemed 

 adequate for the following reasons. In all years 

 there was close agreement between sexes in the 

 number of age-groups separated within quarters: 

 approximately 2, 3, 4, and 3 age-groups for the 

 first through fourth quarters, respectively. There 

 was also close agreement among years and 

 between sexes in the mean lengths and length 

 composition of age-groups within quarters. 



The progressions of mean lengths were set up as 

 depicted by the connected open circles in Figure 1. 

 In the third quarter of every year for both sexes, 

 there was an age-group with mean length of about 

 167 cm that did not fit into the progression of 

 age-groups. By assigning the same assumed age to 

 this age-group as to a similar size group in the first 

 quarter, and allowing the same time between 

 spawning and the attainment of this age, this 

 age-group could have resulted from a spawning in 

 January. On the basis of gonad indices, Kume and 

 Joseph (1969) believed that striped marlin from 

 the eastern South Pacific spawn from November to 



December, and Royce (1957) indicated that striped 

 marlin with ripe gonads have been collected in 

 February in the equatorial region of the central 

 Pacific. Hence, we concluded that this age-group 

 belonged to a different spawning stock and should 

 not be used in the calculation of the von Ber- 

 talanffy growth curves for the central North 

 Pacific stock. Also, for females there were two 

 age-groups having mean lengths of 267.5 and 200.0 

 cm in the third quarter 1964 and 1966, and for 

 males there were four age-groups having mean 

 lengths of 204.6 and 272.0 cm in the third quarter 

 of 1968, 271.8 cm in the fourth quarter 1968, and 

 266.0 cm in the second quarter 1969. These could 

 not be assigned with certainty to any cohort. 



Several qualitative aspects of the observed 

 growth of the cohorts were noteworthy (Figure 1). 

 First, there seemed to be a cyclical pattern in the 

 mean size at recruitment but no upward or down- 

 ward trend. Second, the progression of age-groups 

 during the first year in the fishery was fairly 

 smooth and consistent between cohorts. Third, 

 after about a year and a half in the fishery, there 

 seemed to be a regression or slowing down in the 

 apparent growth that persisted for two or three 

 quarters. Fourth, the mean length of the last 

 age-group in each cohort varied considerably. 



There were a sufficient number of age-groups 

 for the 1959-65 cohorts to fit a von Bertalanffy 

 growth curve but not for the 1957-58 and 1966-68 

 cohorts. The calculated growth curves were shown 

 as smooth curves in Figure 1. As expected from the 

 variation shown in the progression of the observed 

 mean sizes in Figure 1, the standard errors of 

 estimates were moderate, and there was variation 

 in parameter estimates between cohorts (Table 2). 



To investigate the variation in estimates of 

 mean length for age-groups shown in Table 1 and 

 in growth parameters shown in Table 2, a series of 

 nonparametric Friedman two-way analyses of 

 variance was performed. No difference in the 

 number of age-groups by sex, quarter, or sample 

 year could be demonstrated (all 5 test statistics 

 were insignificant with probability P>0.05, Hol- 

 lander and Wolfe 1973:139). In testing the effect of 

 sex on the mean length of age-groups within 

 quarters, three significant effects (S = 5.44, 

 P50.05; S = 8.99, PsQ.oi; and S = 5.44, Ps0.05) 

 were found out of 16 comparisons. However, using 

 a sign test for trends in mean length between 

 sexes showed that females tended to be larger 

 than males (test statistic P = 0.00005, or 

 probability P^O.Ol, for data from all years). In 



556 



