814 



Fishery Bulletin 97(4), 1999 



Figure 1 



Structures used to estimate the age of kingfish from New South Wales; (A) otolith 

 i860 mm FL, age 3i; (B) scale (618 mm FL, age 3); and (C) vertebra (720 mm FL, age 3). All 

 structures viewed with reflected light against a black background. Size of scale bars are 

 indicated on each figure. Growth zones are indicated by squares; abbreviations; A = ante- 

 rior, P = posterior. D = dorsal, V = ventral. 



o^ are estimates of the standard deviation for the 

 observation a at age 1 and A respectively. The third 

 parameter, a, determines nonhnearity of aib), the 

 standard deviation of the observation a. The fourth 

 parameter, /3, controls the extent that the classifica- 

 tion matrix may be dominated by its diagonal en- 

 tries (Richards et al., 1992). Two possible represen- 

 tations of the classification matrix were used, namely 

 a normal and an exponential representation isensu 

 Richards et al., 1992, p. 1803) and the a parameter 

 was constrained to during one fit of each of the 



normal and exponential representations, allowing 

 four different cases of the model to be fitted for each 

 aging structure. The appropriate model structure, 

 or best fit model, for the classification matrix was 

 then selected by using the Akaike information crite- 

 rion (AIC) where a model with a low AIC value in 

 relation to other models was considered to provide a 

 good fit (Richards et al.. 1992). 



Initially, data from each aging structure (otoliths, 

 scales, and vertebrae) were analyzed separately and 

 an age assigned to each fish from each structure. The 



