FISHERY BULLETIN: VOL. 83, NO. 2 



Table 13. — Results of one-wa.v analysis of variance and multi- 

 ple range tests comparing deviations of age from the standard 

 age in Sebastes pinnlger females. Group 1 = between-agency 

 variability; group 2 = model-induced variability; group 

 3 = within-agency, between reader variability. 



Source 



df 



Sum of 

 squares 



Mean 

 squares 



Analysis of vanance 



Between groups 2 88.69 



Within groups 147 751.84 



Total 149 840 53 



Group n Mean SD 



44.34 

 5.11 



8.67 



0,001 



50 

 50 

 50 



-0.320 



-0.021 



1.44 



2.817 

 1.516 

 2 260 



Multiple range test (least significant difference, a = 0.05) 

 Group 1 = Group 2 Group 3 



Table 14. — Results of one-way analysis of variance and multi- 

 ple range tests comparing deviations of age from the standard 

 age in Sebastes pinniger males. Group 1 = between-agency var- 

 iability; group 2 = model-induced variability; group 

 3 = within-agency, between reader variability. 



Source 



df 



Sum of 

 squares 



Mean 

 squares F 



67.43 -0.001 



Multiple range test (least significant difference, a = 0.05) 

 Group 1 Group 2 • Group 3 



compared with conventional section ages in Fig- 

 ure 10. The two ages are similar and as a whole are 

 not significantly different for females but are sig- 

 nificantly different for males (paired ^-test, 

 a = 0.05). This is presumably a result of the con- 

 sistently overestimated otolith section age for S. 

 pinniger males. The ages estimated from the hy- 

 brid model (Fig. 11) are not significantly different 

 from those determined by the appropriate conven- 

 tional age (paired /-test, a = 0.05). 



DISCUSSION 



The results of this research demonstrate the 

 potential for using objective criteria and mul- 

 tivariate models to determine age in fast- and 

 slow-growing members of the genus Sebastes . Past 

 studies have used weight of the eye lens for esti- 

 mates of age in fishes, amphibians, and certain 

 mammals (Crivelli 1980; Malcolm and Brooks 

 1981). In fishes, however, this technique is only 



114 



good for fast-growing species and provides poor 

 estimates of age after several years when length at 

 age becomes highly variable (Crivelli 1980); the 

 same problems exist in estimating age from modal 

 lengths. Growi;h of most body parts, including the 

 eye lens, is allometric with length rather than age. 

 Growth of the otolith, however, as described above, 

 is a complex function of age as well as length. After 

 a certain size is reached, the fish otolith does not 

 increase in length or width, but continues to in- 

 crease in thickness, and therefore weight, with 

 age (Fig. 3). The increasing thickness is a function 

 of addition of aragonite crystals only on the inter- 

 nal surface of the otolith (Fig. 1). 



Similar patterns of otolith growth in length, 

 width, thickness, and weight have been observed 



38 40 42 44 46 48 50 52 54 56 



30 



35 40 45 



FORK LENGTH (cm) 



50 



55 



60 



Figure lO. — Comparisons of mean otol ith section ages at length 

 from the confirmation subsample of Sebastes pinniger. Trian- 

 gles represent otolith section age and circles the model esti- 

 mated section age. 



