284 



Fishery Bulletin 103(2) 



values among months (F=1.63, df=7, P=0.129), likely 

 because of the large variation in increment by months. 

 Under the statistical criteria established in our study, 

 all growth models fitted the data well (Table 3). For 

 males and females, models were highly significant 

 (P<0.001) and exhibited high coefficients of determina- 

 tion (r 2 a0.88). Residual mean square error (MSE) was 

 lowest for the logistic models. Notably, MSE was much 



Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 



Figure 3 



Mean marginal increment analysis (MIR) by month for combi 

 of spinner sharks (Carcharhinus brevipinna). Vertical bars 

 standard deviation of the mean and numbers above each m 

 resent the sample size. 



higher for the von Bertalanffy model males than for 

 any other model. Plots of the residuals against pre- 

 dicted sizes indicated no pattern in the residuals for 

 any model. The standard deviation of the residuals was 

 lowest for the logistic models (Table 3). 



Estimates of the asymptotic size varied depending 

 on sex and model (Table 3; Figs. 4 and 5). For males, 

 the highest asymptotic length was produced by the von 

 Bertalanffy model (L 5 . = 421 cm FL), inter- 

 mediate lengths came from the von Berta- 

 lanffy model with a size-at-birth intercept 

 (L x =279 cm FL) and the Gompertz model 

 (L,=200, G = 1.38), and lowest length was 

 produced by the logistic model (W x = 60.2 

 kg, -161 cm FL). For females, asymptotic 

 sizes were highest and similar with the 

 von Bertalanffy, von Bertalanffy model 

 with a size-at-birth, and the Gompertz 

 models (226, 202, and 220 cm FL, respec- 

 tively) and lowest with the logistic model 

 (62.6 kg or -162 cm FL). 



Among models with comparable growth 

 coefficients, the von Bertalanffy model 

 produced the lowest growth coefficient for 

 both males and females (& = 0.03 and 0.08/ 

 yr, respectively). Growth coefficients were 

 higher and fairly similar for the other two 

 length models. The growth coefficient from 

 the logistic weight model was 0.44 and 

 0.37 for males and females, respectively. 



ned sexes 

 are ± the 

 onth rep- 



