130 



Fishery Bulletin 102(1) 



: v. ; - ■*■-■ 



count from here 



Figure 3 



Example of an image of a baked sablefish otolith which has been annotated with a 

 mark. The image is an example of one of the images provided to three researchers in 

 order to obtain cross-reading comparisons. 



Following standard age determination procedures (Chil- 

 ton and Beamish. 1982), if a hyaline zone was not visible 

 on the edge between January and March, then the edge 

 was counted. If a mark was not visible on the edge between 

 April and May and there was a wide opaque zone, then the 

 edge was counted as a mark. If a mark was visible on the 

 edge and the month was after May, the edge was not count- 

 ed. This procedure is used to properly assign the fish to an 

 annual cohort. Because the reader was not given the month 

 of recapture, the ages were adjusted based on the count of 

 hyaline zones, the month of recapture, and whether the 

 edge had been counted. This adjustment provided a cor- 

 rected reader count of annual marks. The corrected count 

 was compared to the number of annual marks that would 

 have been present if marks were laid down annually. 



Previous experience suggested that there are differ- 

 ent patterns of sablefish otolith growth. We attempted to 

 classify and characterize these different types of growth 

 patterns based on morphology of the otoliths as seen in 

 cross section. After the otoliths had been examined, we 

 developed a standard classification scheme of morphologi- 

 cal classes and types which could be used to classify the 

 most commonly observed morphological types. The otoliths 

 were re-examined and reclassified to see if difficulties and 

 discrepancies in aging were associated with morphological 

 type. It was hoped that this process could be used to refine 

 the aging criteria and improve precision. 



Because sample size was small, we used a Fisher exact 

 test (Agresti, 1990) to test for independence of morphological 

 type versus tendency to over-estimate, correctly estimate, or 

 under-estimate the number of annual marks. The columns 

 in the test indicated whether the fish had been over-aged. 



correctly aged, or under-aged. The rows in the test were the 

 four morphological types identified in this study. 



Examination of the otoliths by the age readers 



To determine how age readers would count the marks on 

 the otoliths, we selected a subsample of 25 otoliths to be 

 aged at four West Coast fisheries laboratories. The otolith 

 selection was based on having good quality images and 

 otoliths. The images of the baked otoliths (not the compos- 

 ite images ) were annotated with a mark ( Fig. 3 ). The mark 

 was placed in a location which could be readily located on 

 the actual otolith by the readers — on the zone just inside of 

 the OTC mark. Readers were given the following: a set of 

 printed images, an electronic file of the images for viewing 

 on a computer screen, the embedded otolith, the month of 

 capture, the size and sex of the fish from which the otolith 

 came, and a set of instructions for examining the otoliths. 

 Readers were not told where the mark on the image was 

 placed in relation to where the OTC mark was in order to 

 reduce bias from readers who may have known when the 

 fish were injected and recaptured. Readers were asked to 

 provide the following: the number of annual marks vis- 

 ible outside the mark on the image, whether the edge was 

 counted, how confident they were of their readings, and any 

 comments they might have. 



Three readers participated in this analysis, two of whom 

 had extensive, long-term experience in aging sablefish. 

 The readings and age determination criteria (including 

 edge count criteria) were compared to each other and to 

 the time known to have passed between OTC marking and 

 recapture. 



