BOEHLERT: AGE DETERMINATION IN FISHES 



in other species offish, but the information has not 

 been applied to the estimation of age, with the 

 exception of preliminary tests using discriminant 

 techniques by Brander (1974). Templeman and 

 Squire (1956), however, noted the importance of 

 this information: "In many fishes, in which accu- 

 rate age reading is doubtful, otolith weights, 

 which are more factual, may offer a better separa- 

 tion of fish populations than growth rates which 

 are dependent on the judgement of the scale- or 

 otolith-reader." Weight and otolith measurements 

 are valid criteria for age determination based 

 upon the models (Tables 3-5, 10-12) and provide 

 good estimates of age compared with other reading 

 methods (Tables 6, 7, 13, 14; Figs. 4-6, 8-11). Based 

 upon published patterns of otolith growth, these 

 techniques should work for other species of 

 Sebastes (Sandeman 1961; Beamish 1979b), Pacific 

 hake (Beamish 1979a), haddock (Templeman and 

 Squire 1956), plaice, sole, turbot, and horse 

 mackerel (Blacker 1974a), and cod (Trout 1954; 

 Blacker 1974a), among others. This technique 

 may therefore be amenable to a wide variety of 

 species of fishes. 



Ages determined by scale or otolith readers are 

 generally based on subjective decisions by the age 

 reader, who reads annuli but must distinguish 

 from "false checks", "metamorphic checks", and 

 "spawning checks" (Trout 1961; Bailey et al. 1977). 



FORK LENGTH (cm) 



Figure ll. — Comparison of ages determined from otoliths and 

 those predicted by the hybrid regression model for Sebastes pin- 

 niger males. Otolith ages were based upon whole otolith ages if 

 the difference between section and whole otolith ages were s5; 

 otherwise, otolith sections were used. Triangles represent 

 whole otolith or section age and squares the model estimated age. 



With experience comes reduced individual vari- 

 ability, but aging variability among different 

 otolith readers and especially among different 

 agencies is great; such variability can have impor- 

 tant effects upon the estimates of growth 

 parameters important for fisheries management 

 (Sandeman 1961; Brander 1974; Hirschhorn 1974; 

 Kimura et al. 1979). While otolith or scale ex- 

 changes are occasionally made between agencies 

 for calibration purposes, this represents addi- 

 tional time spent for gaining greater consistency 

 in ages (Westrheim and Harling 1973; Blacker 

 1974b), and difficulties may remain if disagree- 

 ment in aging techniques cannot be resolved. 

 Blacker (1974a) noted that "Recent progress in the 

 use of otoliths for age determination has been lim- 

 ited mainly to the development of new techniques 

 for preparing otoliths for reading and for photog- 

 raphy so that aging methods can be readily com- 

 pared." The techniques described in the present 

 study represent a new approach to the systematic 

 and repeated age determination in species for 

 which continued age determination is necessary; 

 once calibrated and implemented, the models 

 would reduce between-reader and between-agency 

 variability in age determination. Further re- 

 search, however, should be conducted on varia- 

 tions in thg models over seasons, regions, and dif- 

 ferent years to determine to what extent repeated 

 calibration is necessary. 



Ancillary benefits of the proposed methodology 

 include its simplicity. Reliable, repeatable esti- 

 mates of age require a great deal of experience on 

 the part of an otolith or scale reader using conven- 

 tional aging methodology (Blacker 1974a). It is 

 often difficult to maintain a staff of trained otolith 

 readers and retraining may require a large time 

 commitment. The techniques described here re- 

 quire no special training, since the criteria (otolith 

 length, otolith width, and otolith dry weight) are 

 objective and can be measured with simple dial 

 calipers and balance. Time expended for age de- 

 termination by different methods is as follows: An 

 experienced otolith reader averages about 17 

 ages/h on whole otoliths, but only 6 to 8 ages/h 

 when otolith sections are used due to the addi- 

 tional preparation necessary. An untrained tech- 

 nician, however, can determine the measurements 

 necessary for the model-based age estimates at a 

 rate of about 40 otoliths/h on a long-term basis. 

 Since the criteria for age are measurable, the 

 techniques will be amenable to automation. Sev- 

 eral attempts have been made in the past to auto- 

 mate or semiautomate age analysis using imaging 



115 



