136 



Fishery Bulletin 94| 



1996 



alcohol after fixing in 10% formalin for two weeks 

 (15%); frozen (12%); or processed fresh (2%). Indi- 

 vidual centra were soaked in a 5.25% solution of so- 

 dium hypochlorate for 10-40 minutes and the con- 

 nective tissue was then removed with forceps. 



Ring-enhancing methods 



Five methods to enhance visibility of growth rings 

 were tested. X-radiography 1 and silver nitrate stain- 

 ing 2 failed to clarify the rings and were discarded. 

 The crystal violet staining technique (CV method) of 

 Schwartz (1983) was modified in that the corpus 

 calcareum was stained with a 0.01% solution of crys- 

 tal violet for 2-10 minutes and then destained with 

 50% isopropyl alcohol until the desired color inten- 

 sity was obtained. In the second (UT) method the 

 entire cleaned, unstained vertebra was viewed 

 through a dissecting microscope with transmitted 

 light and a dark field. In the "bow tie" (BT) method, 

 as described by Branstetter and McEachran ( 1986), 

 a sagittal section was cut from the centrum and suc- 

 cessively polished with 200-, 400-, and 600-grit sand- 

 paper to a thickness of 0.5 cm. The section was 

 mounted on a glass microscope slide with clear ep- 

 oxy resin for viewing. 



Ring counts 



A growth ring was defined as a band pair, composed 

 of one calcified, opaque band and of one less-calci- 

 fied, translucent band. A change in angle on the cen- 

 trum face, a result of the difference between fast in- 

 tra-uterine and slower post-natal growth (Walter and 

 Ebert, 1991), was regarded as the birth mark. Birth 

 was defined as age zero and two embryos which were 

 close to term, a 42 cm male and a 41 cm female, were 

 included in establishing the von Bertalanffy growth 

 curves (VBGC). 



Three nonconsecutive ring counts for each method 

 were made without knowledge of the animal's length 

 and previous counts. Count reproducibility was de- 

 termined by using the average percentage error 

 (APE) and the APE index as described by Beamish 

 and Fournier ( 1981 ). An upper limit in the APE was 

 arbitrarily set at 20% for each vertebra. Samples 

 were discarded if, after a single recount, they were 

 still above this limit. For each method, a mean of the 

 three counts was used as an age estimate. In the 

 fourth (ALL) method, a mode of all nine readings was 



taken as an age estimate. This method was chosen 

 to compensate for possible over- or under-counting 

 tendencies of the different methods. 



The von Bertalanffy growth parameters (VBGP) 

 were computed with Marquardt's algorithm (Draper 

 and Smith, 1981 ). The von Bertalanffy growth equa- 

 tion was fitted by using the nonlinear regression pro- 

 cedure of STATGRAPHICS 5 '. The equation is 



L (1-e 



-KU-t ) 



J) 



1 Standard x-radiography. mammography, and a technique pro- 

 ducing zerograms. 



2 A modification of the method of Stevens ( 1975). 



where L t = length at age t in years, L = maximum 

 theoretical length, K = the rate at which L m is 

 reached, and t = the theoretical age at length zero. 

 The VBGC were then examined for their goodness 

 of fit. A poor goodness of fit in all four methods may 

 be due to a wide variation in the growth of individu- 

 als, whereas a poor fit in a single method would sug- 

 gest inaccurate age estimates. 



Centrum analyses 



Confirmation of the annual periodicity of growth 

 rings (Cailliet et al., 1983, a and b) was attempted 

 with centrum analyses. The band immediately after 

 the change in angle and the outermost band were 

 classified as translucent or opaque. The nature of 

 the last deposited band was related to the month of 

 capture, and the observed and expected ratios of 

 translucent to opaque last bands were compared. 



Distance from the focus to the outer edge of the 

 opaque band of each growth ring, marginal incre- 

 ments, dorsal diameter, and dorsal "birth diameter" 

 were measured in a transverse plane along a straight 

 line through the focus. A relationship between cen- 

 trum diameter and animal length was investigated 

 and the Dahl-Lea method of back calculation 

 (Carlander, 1969) was used, in which 



PCL t = CD t x PCLJCD C 



where PCL t = length at age /, CD t = centrum diam- 

 eter at age t, PCL c = length at capture, and CD c - 

 centrum diameter at capture. 



Results 



The APE indices were 8.1% for the CV method (n=87), 

 8.9% for the UT method (rc=86), and 9.4% for the BT 

 method (n=87). These results indicated that all three 

 methods were of similar reproducibility. The APE 

 index for the ALL method was 13.0% (n=80). 



