Hampton: Growth parameters of Thunnus maccoyii from tagging 



data 



587 



previous studies did not restrict the data 

 to those observations where release length 

 was at least 50 cm and time at liberty at 

 least 250 days (except Kirkwood 1983 in 

 the case of the latter restriction). 



The estimates of L m and K based on 

 length-frequency data only (Shingu 1970, 

 Hynd and Lucas 1974, Kirkwood 1983) 

 and those based on scale (Yukinawa 1970) 

 and otolith readings (Thorogood 1986) 

 must be treated with caution because 

 they were estimated from samples which 

 consisted of no, or very few, larger fish 

 (in the case of length-frequency data, 

 modes are rarely visible beyond 100 cm). 

 As discussed by Knight (1968), this al- 

 most inevitably leads to biased estimates 

 of the growth parameters. 



The analyses of both real and simulated 

 data indicate that biased parameter esti- 

 mates may result if model error is ig- 

 nored. In addition, the incorporation of 



