148 



Fishery Bulletin 94(1), 1996 



ered to represent different recruitments, and only 

 the one that could be followed based on biological 

 grounds was used in growth parameter estimation 

 (Schnute and Fournier, 1980). Recruitment was usu- 

 ally first detected in May or June for the spring co- 

 hort and a minor recruitment in September or Octo- 

 ber in some years (Tables 1 and 2). The birthday of 

 the spring cohort, based on the spawning study of 

 Mohammed et al. (1981) was considered to be Janu- 

 ary 1. The age of the 1986 spring cohort on 18 May 

 which was the survey date was then calculated to be 

 4.58 months or 0.38 yr. The ages of the cohort in the 

 following months were calculated in the same way, 

 and the weighted mean carapace length for the co- 

 hort at each month or age was then calculated by 

 using the defined central range (Table 3). 



equality of the covariance matrices between sexes or 

 cohorts was rejected in all cases except for the com- 

 parison of 1986 and 1987 females (Tables 5 and 6). 

 Therefore, the differences between vectors of the growth 

 parameters were tested by using Equation 2 for all the 

 pairs except for the 1986 and 1987 females which were 

 tested with Hotelling's T 2 . Significant differences in 

 growth between male and female shrimp were detected 

 (P<0.005) for all four cohorts (Table 5). The differences 

 in growth of female shrimp between cohorts were not 

 significant except for the 1986 and 1988 cohorts 

 (P<0.05). The vectors of growth parameters for male 

 shrimp differed significantly between 1986, 1987, and 

 1988 cohorts, but the growth of the 1989 cohort was 

 not found to be significantly different from the others. 



Estimation of the seasonal growth 

 parameters 



t , were esti- 



The growth parameters, L m , K, t , C 



mated for males and females of the 1986, 1987, 1988 



and 1989 spring cohorts (Table 4). The hypothesis of 



Discussion 



Accuracy of the proposed method 



The bootstrap technique (Efron, 1982) was applied 

 to examine the effects of variations in length fre- 



