Tracey and Lyle: Age validation, growth modeling, and mortality estimates for Latns lineata 



175 



tion of annual periodicity (Fig. 4). However, inferences 

 about population age structure cannot be drawn from 

 the age-frequency histograms because some sample siz- 

 es were low and there was discriminatory sampling (by 

 gear type) over the period. For instance up to 1996-97 

 most of the aged samples were from inshore gillnet 

 catches, whereas subsequent samples were derived pri- 

 marily from hook catches. 



Precision of repeated age estimation was high. Second 

 readings by the primary reader were 79% in agreement 

 with first readings, yielding an APE of 0.93%. Eighteen 

 percent of second readings gave rise to a one-year dif- 

 ference and 3% of second readings differed by 2 years, 

 and no significant tendency to overestimate or underes- 

 timate age was evident. An age bias plot did not differ 

 significantly from 1:1 for the primary reader (Table 2). 



