Pearson and Shaw: Age determination errors for Anoplopoma fimbria 



135 



after the OTC mark are shown in Tables 6, 7, and 8. In 

 these tables, it is assumed that the readers should not have 

 counted the zone in which the OTC mark occurred because 

 that mark is presumed to have formed in the summer of 

 1991. Readers 1 and 2 tended to overestimate, whereas 

 reader 3 (the least experienced age reader) had generally 

 good agreement. Reader 1 agreed with the expected count 

 24% of the time, reader 2 agreed with the expected count 

 4% of the time, and reader 3 agreed with the expected count 

 44% of the time. The result for reader 3 is deceptive, how- 

 ever, because that reader did not follow accepted methods 

 of when to count the edge. 



Reader 1 and reader 2 agreed on whether to count the 

 edge of the otolith in 24 of 25 otoliths (Table 9). Reader 3 

 agreed with reader 1 on whether to count the edge in 16 of 

 25 otoliths and 17 of 25 otoliths with reader 2. Had reader 

 3 followed accepted practice, agreement with the expected 

 count would have been much less. 



Efforts to determine what factors (depth of capture, loca- 

 tion of capture, sex, size of the fish, and otolith morphologi- 

 cal type) resulted in a miscount of the true number of an- 

 nual marks were inconclusive. We first corrected the count 

 for the fact that all readers counted the mark in which 

 the OTC mark had occurred by subtracting one from their 



counts, and we then eliminated the readings from reader 

 3 because of his lack of experience and anomalous age de- 

 termination criteria. Then we examined the relationship 

 of how many otoliths had been over-aged, correctly aged, 

 and under-aged to the above factors. Depth of capture was 

 divided into two groups: less than 600 m and 600 or more 

 m. Location was divided into two groups: north and south 

 of latitude 39 north. Sizes were divided into two groups: 

 <55 cm FL and ;>55 cm FL. And finally, we tested each of 

 the four otolith morphological types. 



We used Fisher exact tests to determine the probability 

 that differences were due to chance alone. There were no 

 detectable differences from the null hypothesis for depth, 

 sex, or location of capture (Table 10); however, there was 

 some evidence that fish length and otolith morphological 

 type might be related to miscounting. Small fish showed a 

 slightly greater tendency to be over counted (more rings 

 than should have been present) than larger fish (P=0.150). 

 Otolith morphological type showed some departure from 

 randomness: thick types appeared to be more likely to be 

 undercounted (fewer rings than should have been pres- 

 ent) and wide types were more likely to be over counted 

 (P=0.066). 



Discussion 



Position of mark 



There was no visible mark on 22 of the 191 otoliths ( 11.5%). 

 Beamish et al. ( 1983 ) reported that 14 of 129 OTC-injected 

 fish ( 10.9%) had no detectable mark. They attributed this to 

 improper handling of the fish after recapture. The similar- 

 ity in the number of otoliths failing to show the OTC mark 

 between their study and our study suggests that some 

 portion of the population may not absorb sufficient OTC to 

 produce a visible mark. 



The finding that most of the OTC marks were in a 

 hyaline zone is important. This indicates that many of 

 the sablefish in our study laid down a prominent hyaline 

 zone in the summer. Age readers who conventionally as- 

 sume that an annual mark is laid down only in the winter 



