Xu 



and Mohammed An alternative approach to estimating growth parameters 



151 



data in Table 1 by using five scenarios, i.e. modes, 

 1-2 mm, 2-3 mm, 3-4 mm, and as wide as possible 

 to spread from the mode to each side. Scenario 1 is 

 similar to the modal progression analysis (George 

 and Banerji, 1964) and scenario 3 is the selection 

 used in this study. The growth estimates for the five 

 scenarios were very similar, and the growth curve 

 estimated with scenario 3 was located in the middle 

 of the five curves (Fig. 2). The insensitivity of the 

 proposed method to the subjective selection of the 

 spread width, as mentioned in the Material and 

 Methods section, may be due to the fairly symmetri- 



cal length distribution around the distinct mode of 

 the cohort. The spread from the mode can only go up 

 and down for 2-3 mm for the old age classes (e.g. 



Asymptotic CL K 



Growth Parameters 

 Figure 1 



Boxplot of the bias in the estimates of sea- 

 sonal growth parameters, based on 100 

 bootstrap simulations compared with true 

 values. 



Table 6 



Test of equality of covariance matrices of growth parameters between cohorts using the statistic MC~ l and test of differences in 

 growth between cohorts of green tiger prawns, Penaeus semisuleatus, in Kuwait waters using Hotelling's P 2 for the pair of 1986 

 and 1987 cohorts of female and Ti for the rest. 



Cohorts 



1987 



1988 



198!) 



II v 



H o :A=0, 



