160 



Fishery Bulletin 97(1), 1999 



assumed to occur in summer or in winter. The exclu- 

 sion of borderline cases, however, resulted in a sig- 

 nificant difference between observed and the ex- 

 pected ratios (x' test, P<0.001, n-39), assuming 

 opaque band deposition in summer, but no signifi- 

 cant difference (X" test, P>0.05, n=39), assuming a 

 translucent band deposition in summer. 



The vertebra of shark BT433 was difficult to read 

 (Fig. IB), when compared with other sharks of simi- 

 lar length and mass (Fig. lA). For this reason, we 

 also examined the vertebra with transmitted light 

 (Wintner and Cliff 1996), and our interpretation of 

 GRs is based mainly on this method. The OTC 

 marker was visible in the opaque band (injected 30 

 October 1994). The deposition of OTC when injected 

 intramuscularly occurs after a couple of weeks 

 (Holden and Vince, 1973) or 21-35 days (Brown and 

 Gruber, 1988). The opaque band would therefore have 

 been deposited in summer (November or December). 

 The last band in the vertebra was an opaque band. 

 Because the shark was recaptured on 27 May 1997 

 (the beginning of winter), the translucent band might 

 have been in the process of being formed. 



MIR analysis of the entire sample (Fig. 5), how- 

 ever, did not show a distinct time of GR formation 

 because mean and minimum ratios did not get close 

 to zero. The results of a Kruskal-Wallis analysis in- 

 dicated that there is no relation between MIR and 



month (P=0.39). In view of the results from the above 

 recaptured shark and if the peak in July was re- 

 garded as a single peak, we assumed that one GR 

 was formed annually, combined with the minimum 

 MIR trend. With the latter, GR formation may occur 

 during December or January. Because of the ambi- 

 guity of the results of the centrum analyses, how- 

 ever, annual periodicity of GR could not be confirmed. 



Age and growth 



The von Bertalanffy growth function was the most 

 appropriate model for the data sets obtained by both 

 methods SC-A and XR-A. The length-GR data set 

 obtained by method SC-B was not investigated fur- 

 ther because two growth models. Putter no. 2 and 

 Gompertz (Punt and Hughes, 1989), provided a bet- 

 ter fit than did the von Bertalanffy growth function. 

 In addition, the L^ values obtained by the two mod- 

 els were too low to be realistic according to observed 

 sizes of white sharks. Sexes could not be compared 

 because there were no females larger than 300 cm; 

 consequently no meaningful von Bertalanffy growth 

 function could be fitted. Although the VBGP for the 

 two methods SC-A and XR-A differed, there was little 

 difference in the calculated GR over the range of 

 lengths sampled (Fig. 6). Method SC-A had lower 

 relative standard errors and a more realistic L than 



