BOEHLERT: AGE DETERMINATION IN FISHES 



samples of 50 male and 50 female S. pinniger were 

 taken from the remaining 1980 samples not used 

 in the calibration subsample. These subsamples 

 were representative of the length and age ranges 

 in the overall collection. Ranges of whole otolith 

 age, NMFS age, and otolith section age in these 

 subsamples were 4-26, 4-25, and 4-29 for females 

 and 7-35, 7-32, and 8-45 for males, respectively. 



Whole otolith age was estimated from the ap- 

 propriate whole otolith age model for males and 

 females. Values of model estimated age, whole 

 otolith age, and NMFS age as a function of length 

 are plotted in Figure 8. Female S. pinniger ages 

 are similar for all three age determination 

 methods. For males, model-estimated age is simi- 

 lar to the whole otolith age but both are less than 

 the NMFS age (Fig. 8). Deviations from the whole 

 otolith age by the otolith reader whose ages were 

 used to calibrate the model are shown in Figure 9. 



26 

 24 

 22 



20 

 18 

 16 

 14 

 12 

 10 



8 



6- 



D— Q 



MALES 



V) 



< 

 UJ 



-6 

 -8 



-lOf- 



"I 1 1 1 1 r 



MALES 



9 ° 



> 



Q -2 

 -3 

 -4 

 -5 



6 8 10 12 



1 1 1 1 r 



FEMALES 



16 18 20 22 24 



1 1 ' r 



ntxv. 



■A ^/ 



J \ L 



_L 



_!_ 



J_ 



6 8 10 12 14 16 18 20 22 24 



WHOLE OTOLITH AGE (YEARS) 



Figure 9. — Mean deviations of whole otolith ages from the 

 confirmation subsample oi Sebastes pinniger. Triangles repre- 

 sent model-induced variability, circles within-agency variabil- 

 ity, and squares between-agency variability. 





 22 



J I I I I I 



J I L 



36 38 40 42 44 46 48 50 52 54 56 



in 

 < 



UJ 



> 



UJ 



< 



z 

 < 



UJ 



S 



oL 



T r 



25 30 35 40 45 50 



FORK LENGTH (cm) 



55 



J 



60 



Figure 8. — Comparisons of mean whole otolith ages at length 

 from the confirmation subsample of Sehastes pinniger. Trian- 

 gles represent age from reader B, circles the age estimated by the 

 model, squares the age determined by another laboratory. 



The explanation of these deviations is the same as 

 for S. diploproa with the exception that the within- 

 agency variability is a between-reader rather than a 

 within-reader variability. One-way ANOVA within 

 these deviations shows significant differences among 

 the groups for both females (Table 13) and males 

 (Table 14). Multiple range testing (least significant 

 difference, a = 0.05) demonstrates that for females, 

 mean between-agency variability and model-induced 

 variability are equal but are both less than within- 

 agency variability (for S. pinniger this was based 

 upon two different readers). For males, between- 

 agency variability is less than model-induced vari- 

 ability which is less than within-agency variability. 

 For the purposes of this comparison, however, the 

 model-induced variability is significantly closer to 

 zero than either of the other sources of variability 

 (Table 14). 



In the confirmation subsample, section ages 

 estimated from the multiple regression model are 



113 



