646 



Fishery Bulletin 104(4) 



Results 



Stained cross sections improved the precision of Green- 

 land halibut age estimates for the larger, presumably 

 older, specimens in trial 2 but did not improve precision 

 of estimates for specimens in trials 1 and 3. Percent 

 CVs were 11.33, 16.31, and 8.11 for surface ages and 

 19.68, 9.64, and 9.96 for cross-section ages from trials 

 1, 2, and 3, respectively (Table 1). A similar pattern 

 occurred in the symmetry of age estimates. Bowker's 

 test of symmetry indicated that surface age estimates 

 in trial 2 were significantly biased between age readers 

 (P<0.0342) and that cross-section estimates were not 

 {P<0.2159), whereas in trials 1 and 3 significant bias 

 occurred in the cross-section estimates (P<0.0001 and 

 P<0.0012, respectively) (Table 1). 



These equivocal results were primarily caused by 

 difficulty interpreting the second annuli on cross sec- 

 tions. Reader 1 tended to count a small diameter mark 

 close to the nucleus as the second year whereas reader 

 2 considered it a check. A post hoc correction of this 

 bias in trial 1 (i.e., adding 1 year to each of reader- 

 2's cross-section estimates) yielded better precision 

 (CV=7.68) and no significant bias (P<0.2440) (Table 1). 

 This problem in interpretation occurred in all trials but 

 the resulting bias was most noticeable in trials 1 and 3 

 where fish age estimates were younger. 



Definitive cross-section ages were significantly 

 greater (older) than definitive surface ages for trial 2 



Trial 2 



D surface 



D cross-section 



n n n ri nl _ n n_ 



6 7 8 9 t) n e B u 5 -B 17 B B 20 21 22 23 34 25 26 27 28 29 30 31 32 33 34 35 36 



B 



Trial 3 



LteL 



D surface 



D cross-section 



rfi n 



Age estimate (yr) 



Figure 3 



Age frequencies for age estimates from sagittal surfaces and from 

 sagittal cross sections of Greenland halibut (Reinhardtius hippo- 

 glossoides) in trial 2 (A) and trial 3 (B) of this study. 



(^=17.32, df = 225, P<0.0001). Mean cross-section age 

 was 17.1 years and had a range from 9 to 36 years, 

 whereas mean surface age was 12.4 years and had 

 a range from 7 to 28 years (Fig. 3A). Differences be- 

 tween definitive cross-section ages and definitive sur- 

 face ages in trial 3 were not significant (/=1.74, df=74, 

 P<0.0858). Mean stained age was 4.29 years and had 

 a range from 1 to 7 years, and mean surface age was 

 4.15 years and had a range from 1 to 8 years (Fig. 3B). 

 Von Bertalanffy growth parameters calculated from the 

 definitive surface ages (trial 2 and 3 combined) were 

 L, =103.7, iC=0.104, and t„=-0.333. Parameters from 

 definitive cross-section ages were L_^ = 86.2, /<'=0.125, 

 and /|| = -0.233. The models varied significantly from 

 each other (P=40.58, P<0.0001) (Fig. 4). 



Discussion 



In larger Greenland halibut (i.e., in trial 2), the precision 

 of age estimates can be improved by aging stained cross 

 sections of sagittae rather than aging sagittal surfaces 

 (Table 1). The sagittae of larger, older Greenland halibut 

 are very difficult to interpret from the surface. Marginal 

 growth increments are very small and are interrupted 

 by the fingerlike projections on the otoliths. Age readers 

 in our study were more confident in the age estimates 

 they made from stained cross sections. The clearest 

 annuli were encountered in the perisulcular region of 

 left sagittae. This region appears to grow 

 more consistently than other areas of the 

 otolith. Staining allowed resolution of very 

 narrow increments in this region that were 

 not visible on the surface of sagittae. 



Precision did not increase in trials 1 and 

 3 (Table 1) because these trials contained 

 many smaller specimens. The benefits of 

 cross-sectioning and staining are not as 

 great in otoliths that are still growing rap- 

 idly in sagittal diameter. The magnitude of 

 the difference in age estimates from whole 

 surfaces and stained cross sections did not 

 exceed 1 year in fish less than 46 cm and did 

 not exceed 2 years in fish less than 57 cm. 

 A second confounding factor was interpreta- 

 tion of the second annuli in cross sections. 

 This consistent one-year bias between read- 

 ers outweighed any improvements that may 

 have resulted from cross-sectioning otoliths 

 in smaller specimens. We feel that more in- 

 terreader calibration and validation of cross- 

 sectioned annuli by the Peterson method 

 (Ricker, 1975) can resolve this problem. 



The increase in precision in trial 2 was 

 accompanied by age estimates that were 

 significantly greater (older) (Fig. 3A). In 24 

 cases, the cross-section estimate was 10 or 

 more years greater than the surface age es- 

 timate, and in two cases the cross-section 

 estimate was 22 years greater. The mean 



-JLa 



