FISHERY BULLETIN: VOL. 85, NO. 4 



80- 



2 



u 



X 



(- 



o 



u 



60- 



40- 



20- 







I <^- 



1979 



I I I I I I I I I I I I — r 



1 5 10 



1980 



I I I I I I I r I — I I I I 



15 10 



AGE (YRS) 



Figure 3. — Comparison of mean length at age estimated from scales ( — A — ), converted 

 dorsal fin-ray ages {^), and length-frequency analysis ( — # — ) for 1979 and 1980. Vertical 

 line indicates the 95% confidence interval from the 30 simulation runs. 



-^ 60 



g40H 



q: 

 O 



Q_ 



O 



I 20 







1979 



^ ^' ♦^  ♦♦^ 



10 

 AGE 



(YRS) 



1980 



10 



Figure 4. — Comparison of age composition estimated from scales ( — A — ), converted dor- 

 sal fin-ray ages (o<), and length-frequency analysis ( — 9 — ). Vertical line indicates 95% 

 confidence interval of the 30 simulation runs. 



length-frequency analysis. In 1980, the propor- 

 tions estimated by using the scale method were 

 somewhat higher than for the other two methods 

 at ages 2 and 3 (Fig. 4), but were lower at ages 4 

 and 5. Since the classification probabilities and 

 scale age-length keys used in this study were 

 derived independently from different years, the 

 method of age conversion appears to be relatively 



insensitive to interannual differences in the clas- 

 sification probabilities. 



DISCUSSION 



Although Westrheim and Shaw (1982) vali- 

 dated the interpretation of the annuli on scales 

 for age groups 1 and 2, this validation was not 



720 



