FISHERY BULLETIN: VOL. 85. NO. 4 



since the dorsal fin-ray method provides more ac- 

 curate ages. 



Age-length keys, derived from scales collected 

 in 1979 and 1980 NWAFC demersal trawl sur- 

 veys in the eastern Bering Sea, and length- 

 frequency distributions obtained from these sur- 

 veys were used as basic data to be converted. The 

 classification probabilities for fin rays vs. scales 

 were constructed from 966 age readings on dorsal 

 fin rays and scales from the same fish collected in 

 1983-84 (Table 2). Each fish in the 1979 and 1980 

 scale age-length key was assigned a "pseudo" dor- 

 sal fin-ray age, which was a random variate gen- 

 erated from the subprogram GGDA in IMSL 

 (International Mathematical and Statistical 

 Library). The GGDA subroutine used the prior 

 probability density in Table 2 corresponding to 

 the scale age and length group to which the fish 

 belongs. This was done 30 times to create 30 con- 

 verted age-length keys each for 1979 and 1980. 

 Then, mean lengths at age were determined from 

 each converted age-length key, and age composi- 



tions were estimated by the method described in 

 Lai (1987). These simulated results were com- 

 pared with those from length-frequency analysis 

 (Lai 1985). 



RESULTS 



Table 1 shows that the age readings from dorsal 

 fin rays had the highest percent agreement (68%), 

 followed by coracoids (63%), scales (62%), otoliths 

 (57%), and pectoral fin rays (55%). Furthermore, 

 the percent agreement for the major age classes 

 in the fishery (ages 4-7) were much higher than 

 for the other methods. To confirm this result, a 

 log-linear model was fitted to the data in Table 1. 



Table 3 shows that the best log-linear model 

 (Appendix A) was 



e,.^=log(m,,^) = |x + Xf + Xf +\f 



ik jk y 



(2) 



Table 2. — Observed classification probability for age readings from scales and dorsal fin-rays. 



Scale Length 

 age (cm) 



Dorsal fin-ray age 



8 



10 



11 



12 



13 



14 Total 



0.029 



0.111 



0.090 



0.167 



20 

 6 



8 

 8 



2 



3 

 32 

 23 



15 



53 



122 



137 



23 



14 



92 



189 



39 



2 



18 

 60 

 32 

 17 

 4 



9 

 11 

 16 



5 

 6 



716 



