Lowerre-Barbieri et al.: A comparison of otolith and scale ageing methods for Cynosaon regalis 559 



To evaluate the applicability of scales and otoliths 

 for back-calculation, it was necessary to first ana- 

 lyze separately their total length to hardpart rela- 

 tionships. Seasonal effects were assessed by compar- 

 ing hardpart size of one age class taken from differ- 

 ent seasons to that predicted by the linear regres- 

 sion of total length on hardpart size for all fish. Only 

 one age class ( age 3 ) was used to remove any confound- 

 ing effects of age. This age class was chosen because it 

 was well-represented throughout the seasons. 



Back-calculation relationships for both scales and 

 otoliths were based on the "body proportional" hy- 

 pothesis (Francis, 1990) proposed by Whitney and 

 Carlander(1956): 



L=\g(S i )lg(S c )]L c< 



where g is the total length on hardpart radius func- 

 tion, L ; is back-calculated TL at age i, S is the mea- 

 sured hardpart size at annulus i, and S . and L. are 

 the respective hardpart size and total length at cap- 

 ture. Only fish collected in April and May — the be- 

 ginning of the somatic growth season — were used, 

 to remove seasonal effects from the back-calculation 

 equations (Ricker, 1992). Because body-proportional 

 back-calculation is based not just on the relationship 

 of hardpart size to total length but also on the rela- 

 tionship of hardpart size to consecutive annuli, mean 

 annual growth increments were also calculated and 

 compared between scales and otoliths. 



The tendency for older fish to produce smaller back- 

 calculated lengths at younger ages than observed, 

 known as Lee's phenomenon (Smith, 1983), was 

 evaluated by calculating mean SAR and mean OAR 

 for each age at capture. In this way it was possible 

 to determine if older fish demonstrated slower 

 hardpart growth at younger ages, i.e. true Lee's phe- 

 nomenon (Smale and Taylor, 1987). 



Data were analyzed by using % 2 tests and regres- 

 sion methods available through the Statistical Analy- 

 sis System (SAS 1988). Rejection of the null hypoth- 

 esis in statistical tests was based on a=0.05. Assump- 

 tions of linear models were checked by residual plots 

 as described in Draper and Smith ( 1981). 



Results 



Preliminary comparison of hardparts 



All four hardparts showed concentric marks that 

 were interpreted as annuli (Fig. 2). However, marks 

 on the dorsal spines and pectoral rays were incon- 

 sistent, often blurred or impossible to follow around 

 most of the section and therefore difficult to inter- 



pret. Presumed annuli on scales were distinctly 

 clearer and more regular than those on dorsal spines 

 and pectoral rays, but they still required some sub- 

 jective interpretation. Presumed annuli on otoliths were 

 exceptionally clear, consistent, and easy to interpret. 



Typical otolith sections showed an opaque nucleus 

 surrounded by a translucent zone followed by a pat- 

 tern of thin, opaque zones alternating with wide, 

 translucent zones along the sulcal groove (Fig. 2A). 

 In some sections the translucent zone between the 

 nucleus and the first opaque zone was relatively 

 small and made more opaque by a number of fine, 

 circular, opaque bands. However, in all sections the 

 first opaque zone beyond the nucleus was easily iden- 

 tified and considered to be the first annulus. 



Presumed annuli on scales were harder to iden- 

 tify than those on otoliths but were usually identifi- 

 able as a clear zone in the anterior field, where circuli 

 are either absent or more widely spaced, and by cut- 

 ting over in the lateral fields (Fig. 2D). Checks were 

 most apparent in the anterior field. A clear zone in 

 the anterior field was considered a check if it was 

 not accompanied by distinct cutting over in the lat- 

 eral fields. The first annulus was the hardest to iden- 

 tify. It rarely showed a clear band in the radii zone, 

 although cutting over was sometimes apparent. Its 

 position was based predominantly on the first point 

 at which a large number of secondary radii originated. 



Presumed annual marks on dorsal spines were 

 fairly clear in some sections but incomplete or blurred 

 in others (Fig. 2C), whereas pectoral-fin ray sections 

 were consistently hard to interpret (Fig. 2B). Pre- 

 sumed annual marks on both these hardparts ap- 

 peared as wide, opaque, semicircular bands alternat- 

 ing with narrow translucent zones. 



Otoliths showed the greatest precision, with 100% 

 average agreement within and between readers. 

 Scales also had high average agreement: 89% within 

 readers and 80% between readers. Dorsal and pectoral 

 fin sections showed the lowest agreement (Table 1 ) and 

 little confidence was attached to their age assignments. 



