854 



Fishery Bulletin 101(4) 



by the three models with the statistically suitable absolute 

 error structure because it provided both residual random- 

 ness and homoscedasticity. The computed value of 0.15 for 

 the winter point of the seasonalized von Bertalanffy growth 

 model indicated that the lowest growth rate occurred at 

 about two months after the birth date, meaning that the 

 growth rate was reduced in winter. 



No significant difference was found between the von 

 Bertalanffy growth models with a likelihood ratio test 

 (F=0.15<Fqq5 2 1560^' ^^^ ^ significant difference was 

 found between the von Bertalanffy growth models and the 

 Schnute growth model with the same test (■f>-f'o 05 2 1560^' 

 therefore, the specialized von Bertalanffy growth model 



10 II 12 1 2 3 4 5 6 7 8 9 

 I 2000 I 2001 I 



Month 



Figure 2 



Mean monthly marginal increment from otoliths with one and 

 two, three, and more than three annuli, representing fast, mod- 

 erate, and slow-growing individuals of P. incisus off the Canary 

 Islands. Standard errors are identified by the bars. 



was chosen because it has fewer parameters making it sta- 

 tistically more robust, its parameters are commonly used in 

 mortality estimates and per recruit modelling, and because 

 it allows for comparison between growth studies conducted 

 on other species (Booth, 1997). No significant differences 

 were found between mean lengths-at-age of males and fe- 

 males with a Student's t-test {t=0.52<tQQ^.jg^=l.65) or be- 

 tween the von Bertalanffy growth curves for separate sexes 

 with a HoteUing's T- test (T2=4.73<To2(, 05 3 78^=7.89). The 

 data were pooled as a single growth model. In fitting the 

 growth curve to the age-length data, age was disregarded 

 because this age group was not well represented. The von 

 Bertalanffy growth curve derived from the observed weight 

 at age for males and females was not significantly dif- 

 ferent (HoteUing's T^-test, T^=6.91<Tf'^g^^^^g^=1.89; 

 Fig. 4). 



Fish total length and otolith radius were closely 

 correlated (Fig. 5). The value of the derived constant 

 {u=0.851) was different from one (Student ^-test; 

 ^=14.61 > ;gQggjQ = 1.65). Backcalculatcd size at time of 

 annulus formation was used to provide length-at-age 

 data unbiased by differences in sampling date and 

 to estimate the von Bertalanffy equation (Table 3). 

 Backcalculated lengths were similar to those pre- 

 dicted by the growth models. Growth parameters 

 estimated from the backcalculated sizes at age were 

 L„=315.23 mm; A=0.217/year; and tg=-1.733 year. The 

 data was pooled as a single growth model because no 

 significant differences in the growth parameters were 

 found between males and females (HoteUing's T'-'-test, 

 7^=48.3>ro2Q.o5,3,784=7-89). 



Discussion 



Age estimation in fishes is complicated by the phe- 

 nomenon of "stacking" of growth zones towards the 

 otolith margin, particularly in older fish (Buxton and 

 Clarke, 1991). In many cases age determination is 

 difficult because whole otoliths are so thick that light 

 does not pass through (Buxton and Clarke 1991); how- 

 ever, in the bastard grunt off the Canary Islands the 

 translucency of the otoliths allows aging with relative 

 ease. The values of the lAPE and the CV suggested 

 that the precision levels obtained are according to the 

 reference point values indicated by Campana (2001). 

 The oldest age estimate obtained in the present study 

 was seven years and the phenomenon of stacking was 

 not evident. 



The otoliths of the bastard grunt have a ring pat- 

 tern common to teleost fishes. Marginal increment 

 analysis demonstrated that one annulus, consisting of 

 one opaque zone and one hyaline zone, is formed an- 

 nually. These rings are believed to be deposited during 

 periods of fast and slow growth, respectively (Williams 

 and Bedford, 1974). Seasonal growth cycles might be 

 related to physiological changes produced by the influ- 

 ence of temperature, feeding regime, and reproductive 

 cycle (Morales-Nin and Ralston, 1990). The seasonal- 

 ized von Bertalanffy growth model reveals the reduc- 



