USING OBJECTIVE CRITERIA AND MULTIPLE REGRESSION MODELS 



FOR AGE DETERMINATION IN FISHES 



George W. Boehlert^ 



ABSTRACT 



Analysis of the age structure of exploited fish papulations is necessary for models upon which manage- 

 ment decisions are made, but existing aging methodology for many species is hindered by subjective 

 criteria used in age determination. A new technique is described in which age is estimated using 

 multiple regression models based upon the measurable parameters otolith weight, otolith length, and 

 otolith width in the splitnose rockfish, Sebastes diploproa , and the canary rockfish, S. pinniger. Models 

 were calibrated using ages determined by interpretation of both whole otoliths and otolith sections 

 which differ within these species, particularly at greater lengths. The models typically explained from 

 70 to 92% of the variability in age depending upon species, sex, and method of age analysis. In another 

 sample used to verify the precision of the models, variability associated with model-estimated ages was 

 generally less than that induced by variability in ages between different agencies. Based upon the 

 pattern of otolith growth in length, width, and weight in these and other species, it is suggested that 

 these methods would be applicable to a wide variety of fishes. Implementation of this type of age 

 determination methodology could result in savings in time and cost for fisheries management agencies 

 while decreasing variability among age estimates between different laboratories. 



Virtually all methods of age determination in 

 fishes involve a certain degree of subjectivity. De- 

 ciding whether a mark on an otolith or scale con- 

 stitutes 1 year's growth is difficult; precision in fish 

 aging improves only with experience. Even so, var- 

 iability between experienced readers may be 

 great. Sandeman (1969), for example, observed 

 only 9% agreement between readers for a wide age 

 range of otoliths of Sebastes marinus and S. men- 

 tella, and noted greater variability with increas- 

 ing age of the fish. Kimura et al. (1979) suggested 

 that bias between readers within a given agency is 

 likely to be much less than among different agen- 

 cies. In a situation such as exists on the Pacific 

 coast, where several management agencies may 

 routinely determine ages for the same species, 

 interagency calibrations are necessary but are 

 rarely achieved. Williams and Bedford (1974) 

 suggested ". . . that otolith reading remains, for the 

 present at least, as much an art as a science, and 

 that proficiency cannot easily be achieved without 

 examination of very large numbers of otoliths." 

 Clearly, objective, repeatable age determination 

 methodology which will minimize variability is 

 desirable. 



Traditional methodology for age determination 



^Oregon State University, College of Oceanography, Marine 

 Science Center, Newport, Oreg.; present address: Southwest 

 Fisheries Center Honolulu Laboratory, National Marine 

 Fisheries Service, NOAA, PO. Box 3830, Honolulu, HI 96812. 



in fishes generally involves some calcified struc- 

 ture; in Sebastes, Six and Horton (1977) tested 25 

 different structures. By far the most commonly 

 used structures, however, are the otolith and 

 scales. Scales are often best for short-lived, fast- 

 growing species because annuli become indistinct 

 near the margin in long-lived, slower growing 

 species (Power 1978; Maraldo and MacCrimmon 

 1979). When this is the case, the otolith becomes 

 the superior structure for age determination; even 

 in the otolith, however, annuli may become indis- 

 tinct on the margin as otoliths thicken and become 

 opaque with age. For this reason several inves- 

 tigators have used broken or sectioned otoliths to 

 determine age from internal banding patterns. 

 While some studies using otolith sections have 

 provided clear continuation of growth patterns 

 obvious on whole otoliths from younger specimens, 

 others have suggested maximum ages which are 

 double or triple those estimated from whole 

 otoliths. Power (1978), for example, suggested ages 

 of >50 yr in Salvelinus namaycush and Coregonus 

 clupeaformis and provided confirming evidence 

 based upon population structure. In the redfish, 

 Sebastes marinus, Sandeman (1961) suggested 

 that specimens exceeding 50 yr of age were pres- 

 ent in the population; ages up to 80 yr have since 

 been estimated (Sandeman^). Similarly, Beamish 



Manuscript accepted April 1984 



FISHERY BULLETIN; VOL. 83, NO. 2, 1985. 



^E. J. Sandeman, Biological Station, St. John's, Newfound- 

 land, Canada, pers. commun. July 1978. 



103 



