NOTE Gregg et al : Otolith preparation methods adapted for fragile sagittae 



645 



Figure 2 



Stained cross section of the left sagitta from an 84-cm Greenland halibut tReinhardtiiis 

 hippoglossoides). Arrows point to presumed annual marks in the perisulcular tuberosity 

 where annuli were most readily resolved. Fish age was estimated at 13 years. 



tae. In trial 1, a training trial, sagittae of 93 Greenland 

 halibut were examined to compare precision between the 

 two aging methods and to calibrate age readers with 

 respect to the first few annuli on the stained cross sec- 

 tions. This sample included fish that ranged from 12 to 

 84 cm TL, but it was dominated by smaller fish (mean 

 TL=40 cm, median=31 cm). Two readers independently 

 aged sagittal surfaces and stained cross sections. Sur- 

 face aging necessarily preceded cutting and staining, but 

 no consultation occurred between readers until the end 

 of the trial. Readers were aware of fish length during 

 aging. At the end of trial 1 the independently determined 

 age estimates were compared and readers re-examined 

 otoliths that had resulted in age discrepancies. 



In trial 2, 226 otolith pairs were examined. This 

 sample contained sagittae from many larger specimens 

 (mean TL=75 cm, median=72 cm, and range=57 to 98 

 cm). Ages were determined in the same manner as in 

 trial 1. However, after surface aging, readers re-exam- 

 ined discrepancies together and assigned, by mutual 

 agreement, a definitive surface age to each sagitta prior 

 to cutting and staining. Similarly, the cross sections 

 were aged independently and then assigned a definitive 

 cross-section age by mutual agreement. This process is 

 similar to that used for production aging of other species 

 at the AFSC (Kimura and Lyons, 1991) and allowed not 

 only a comparison of precision between methods but also 

 a comparison of the final age estimates that resulted 

 from the two methods. In trial 3, sagittae were exam- 



ined from 76 Greenland halibut with a size range of 12 

 to 63 cm TL (mean=37 cm, median=39 cm). This trial 

 was conducted in the same manner as trial 2, with the 

 exception that fish length was not provided to the read- 

 ers. We felt that criticism could arise if length data were 

 known because of the potential for reader bias when ag- 

 ing small fish that fall into distinct size classes. 



Two age readers performed each trial. Reader 1 was 

 relatively inexperienced with six months of experience 

 aging larval otoliths and one month of experience ag- 

 ing adult otoliths. Reader 2 had 14 years of experience 

 aging several species, including other Bering Sea flat- 

 fishes. Neither reader had previously aged Greenland 

 halibut. 



Between-reader agreement and coefficient of varia- 

 tion (CV) were calculated for each aging method from 

 each trial. CV was used as the measure of preci- 

 sion (Chang, 1982). Percent agreement is not a good 

 measure of precision because it is highly dependent 

 on the age structure of the sample. Bowker's test of 

 symmetry (Hoenig et al., 1995) was used to assess be- 

 tween-reader bias. Definitive ages from trials 2 and 3 

 were compared by using a two-tailed matched pairs 

 t-test (Snedecor and Cochran, 1967). Von Bertalanffy 

 growth parameters were estimated from surface and 

 cross-section ages combined from trials 2 and 3. An 

 f -test based on the residuals of nonlinear least-squares 

 fit was used to test for difference between the resulting 

 models (Quinn and Deriso, 1999). 



