462 



Fishery Bulletin 97(3), 1999 



core or confluence with the sulcus acousticus. 

 Opaque bands were initially counted as an- 

 nuli until they could be properly validated. 

 Annuli were counted without reference to fish 

 length or date of capture. Where counts dis- 

 agreed, otolith sections were re-examined 

 jointly, and most disagreements were re- 

 solved. Unresolved counts and illegible 

 otoliths were excluded from the analysis. 

 Structural aberrations in otoliths judged 

 unsuitable for age estimation included poorly 

 defined annuli, unusual calcification, and 

 erosion of the ventral lobe. Terminology for 

 otolith readings followed definitions of Wil- 

 son et al. (1987). 



We determined the periodicity of annulus 

 formation and validated our ageing technique 

 by marginal-increment analysis. As recom- 

 mended by Beamish and MacFarlane ( 1983 ), 

 all age classes were included in the analysis. 

 Measurements for marginal-increment 

 analysis were made in the ventral lobe of the 

 magnified (30x) section by using a digital 

 imaging system. Distances were measured 

 ventrally from the sulcus along an axis pass- 

 ing through the center of the lobe and ex- 

 tending from the otolith's core to the outer 

 margin of the section. The distance from the 

 proximal edge of the ultimate annulus to the 

 otolith's margin (marginal increment) was ex- 

 pressed as a percentage of the distance be- 

 tween the proximal edge of the last two an- 

 nuli formed on the otolith. This procedure 

 was adapted for age 1 fish by expressing the 

 marginal increment as a percentage of the distance 

 from the edge of the first annulus to a hypothetical 

 second annulus (Crabtree et al., 19961. Mean per- 

 cent marginal increments were plotted for all age 

 groups and collection years combined by month of 

 capture. 



The von Bertalanffy (1957) theoretical growth 

 equation, FL^ = Ljl-exp [-K(t-tQ)]), was fitted to 

 observed age-length data with the nonlinear regres- 

 sion procedure of Statgraphics (1994). Likelihood- 

 ratio tests (Kimura, 1980; Cerrato, 1990) and ap- 

 proximate randomization tests (Helser, 1996) were 

 used to compare growth parameter estimates for 

 males and females. Sexed YOY were included in the 

 growth models. 



Observed ages at lengths for all years combined 

 were used to derive an age-length key for each sex 

 (Ricker, 19751. Aged fish (n=565) were assigned to 

 50-mm length intervals, and age distribution (as 

 percent) was then calculated for each size interval. 

 Age-length keys were used to convert length frequen- 



cies to age frequencies by assigning ages to unaged 

 fish >838 mm FL from which a catch curve (Ricker, 

 1975) was constructed for 1987-92. We estimated 

 instantaneous total mortality (Z) by catch curve 

 analysis (Beverton and Holt, 1957; Everhart and 

 Youngs. 1981) based on fully recruited fish. 



Results 



We examined 1005 cobia that ranged from 335 to 

 1651 mm FL, 33 of which were YOY (age 0) and 

 ranged from 335 to 510 mm FL. External sexual di- 

 morphism was not evident in R. canadum. Males 

 («=275) ranged from 345 to 1450 mm FL (mean=952 

 mm) and from 0.3-29.0 kg (mean=10.5 kg); females 

 (« =7301 ranged from 335 to 1651 mm FL (mean=1050 

 mml and from 0.3 to 62.2 kg (mean=16.6 kg). The 

 length-frequency distributions of males and females 

 (Fig. 3) were significantly different (Kolmogorov- 

 Smirnov two-sample test, df =0.432, P<0.05). Females 



