Heifetz et aL: Age validation and analysis of ageing error for Anoplopoma fimbria 



257 



difficult species to age, and age determination is highly 

 subject to an age reader's interpretation. 



For this paper, sablefish tagged and released as 

 young juveniles (Rutecki and Varosi, 1997a) provided 

 a relatively large sample (n=49) of known-age fish, 

 2-9 years old when recaptured. This sample provided 

 a unique opportunity to evaluate the accuracy of pro- 

 duction ageing methods for young sablefish. "Produc- 

 tion ageing" is the routine ageing of large samples of 

 fish, usually to obtain age and growth information 

 or age composition. We conducted a statistical analy- 

 sis of ageing errors and compared ageing errors based 

 on among-reader agreement with ageing errors based 

 on reader and known-age agreement. 



Materials and methods 



Known-age fish 



From 1985 to 1991 about 23,000 age 0-2 sablefish 

 were tagged with individually numbered Floy anchor 

 tags (Rutecki and Varosi, 1997a, 1997b). By 31 De- 

 cember 1993, 1160 of these fish had been recaptured, 

 mostly from commercial fishing operations; otoliths 

 were taken from 49. In Rutecki and Varosi's studies, 

 cohorts of young (aged 0-2) sablefish were periodi- 

 cally sampled in interior bays of southeastern Alaska 

 from spring through early fall. Depending on the time 

 of year, usually only one or two age classes were 

 present at the sampling locations, and these age 

 classes could be distinguished easily from one an- 

 other on the basis of nonoverlapping length frequen- 

 cies (see Figs. 5 and 7 in Rutecki and Varosi [1997b] ». 

 Rutecki and Varosi ( 1997b) confirmed assignment of 

 ages by ageing otoliths from a subsample of the fish 

 measured for length. Of the 49 known-age sablefish, 

 one was tagged at age 0, 41 at age 1, and seven at 

 age 2. Known age of a fish was calculated by sub- 

 tracting the year of release from the year of recap- 

 ture and then adding the age at release. For example, 

 the known age of a fish tagged in 1986 at age 1 and 

 recaptured in 1992 would be 7 years. 



Ageing methods 



Extracted otoliths were stored in a SO'/r ethyl alco- 

 hol solution. A sample of otoliths from 140 fish, in- 

 cluding the 49 known-age fish, were aged by two ex- 

 perienced readers in the Age and Growth Labora- 

 tory at the Alaska Fisheries Science Center (AFSC) 

 in Seattle, Washington. Using normal production 

 ageing procedures, the readers were told the fork 

 length of each fish and the month and day of sam- 

 pling. The year of sampling was withheld. Age read- 



ers were aware that known-age otoliths were in the 

 sample but did not know what percentage of the 

 sample were known-age specimens nor the age range 

 of the known-age fish. 



The primary age reader (hereafter referred to as 

 "primary reader") prepared all the otoliths and was 

 first to examine the sample. The ageing methods 

 generally followed Beamish and Chilton ( 1982). The 

 ageing routine began with placing each otolith in a 

 water-filled petri dish with a black background and 

 examining it through a light microscope with fiber- 

 optic light. This gave the primary reader a general 

 idea of an age range to expect before the otolith was 

 broken and burned. Next, the otolith was broken dor- 

 soventrally along the focus, and the broken surface 

 was passed over a flame. The burnt half was then 

 mounted into clay for support, and the burnt surface 

 was coated with cedar oil for clarification. The burnt 

 surface consisting of light and dark zones was then 

 examined by using a light microscope and fiber-op- 

 tic light. The light areas are known as opaque zones 

 and the dark areas as translucent zones. Notes kept 

 by the reader included the reasons for ageing deci- 

 sions when uncertainties occurred. Typically in pro- 

 duction ageing, an age reader may opt not to assign 

 an age to an otolith if the otolith pattern or condi- 

 tion is considered so poor that the reader's confidence 

 at reproducing roughly the same age again is ques- 

 tionable; this policy was followed in our study. 



In production ageing at the AFSC, after the pri- 

 mary reader completes the age determinations, an- 

 other reader (hereafter referred to as the "tester") 

 typically rereads a 20% random sample of the otoliths 

 without knowledge of the primary reader's ages. This 

 is done mainly to control quality (Kimura and Lyons, 

 1991). Because the known-age specimens were not 

 known to the readers, the randomly chosen 20% 

 sample included only 10 of the known-age otoliths. 

 To compare agreement on all known-age otoliths, a 

 second subsample was drawn, which included all the 

 remaining known-age otoliths interspersed with 

 some unknown-age otoliths. The tester did not know 

 the percent composition of known- to unknown-age 

 otoliths in either subsample. 



In addition to the ageing methods of Beamish and 

 Chilton (1982), in production ageing at the AFSC, 

 readers use the marginal increment component for 

 ageing, i.e. they decide whether the most recent year's 

 annulus was formed along the margin of an otolith. 

 Many experienced sablefish age readers have con- 

 cluded that the annulus usually forms in spring, but 

 some annuli may form in summer. However, an age 

 reader must decide whether an annulus was depos- 

 ited before the date of otolith collection by consider- 

 ing the date of the otolith collection and the amount 



