Hampton: Growth parameters of Thunnus maccoyu from tagging data 



583 



Time at liberty (yr) 



Figure 1 



Plot of standardized residuals against time-at-liberty follow- 

 ing an initial fit of model 1 to the primary data set (n 1800). 

 The parameter estimates used were L„= 195.614cm, K = 

 0.131914/year, and o e 2 = 24. 8193cm 2 . The observations 

 labeled A and B are those identified as outliers. 



6- 



4" 



2 





 -2 

 -4" 



ji iiiiiiii- 



ijli 



B 



30 40 50 60 70 80 90 100 110 

 Length at release (cm) 



Figure 2 



Plot of standardized residuals against release length follow- 

 ing an initial fit of model 1 to the primary data set (n 1800). 

 The parameter estimates used were L„ = 195.614cm, K = 

 0.131914/year, and o e 2 = 24.8193 cm 2 . The observations 

 labeled A and B are those identified as outliers. 



residuals were recalculated and checked for outliers. 

 Models 2-7 incorporate individual variability in growth 

 parameters, necessitating the use of a quite severe 

 criterion in the definition of an outlier. An observation 

 was classified as an outlier only if the absolute value 

 of its standardized residual was greater than 4.4172. 

 Under the assumptions of model 1, only one in 2000 

 observations would exceed this value due to chance 

 alone. Two observations were classified as outliers on 

 this basis and rejected from further analyses. These 



observations are indicated in Figures 1 and 2. The final 

 data set to which all models were fitted consisted of 

 1,736 observations. 



Parameter estimates 



Parameter estimates for the seven models fitted to the 

 final data set are given in Table 1 . The estimates differ 

 substantially among some models, which is not surpris- 

 ing since they are based on rather different assump- 



