FISHERY BULLETIN: VOL. 81, NO. 4 



TABLE 5. — Possible combinations of stabilizing catches (in tonnes) in the four component 

 fisheries when fishing effort index is specified for Western Australia (WA), South Australia (S A), 

 and New South Wales (NSW) and calculated for Japaa 



TABLE 6. — Results of ordinary sensitivity analysis 

 of the CB estimate of 30,012 t derived by using 

 method I. 



'All values in this table represent relative changes( expressed in 

 percentages! in CB. 



'All values (i.e.. for / = 2-20) are simultaneously perturbed by 

 the same percentage indiated in the first row. 



3 The f ; values tot i = 2-6 are simultaneously perturbed by the 

 percentage indicated in the first row, and all reamining f, values 

 (i.e., for / = 7- 1 0) are changed by a percentage such that 

 20 



£ f y is equal to one. 

 i=2 



Table 7. — Results of ordinary sensitivity analysis 

 of the CB estimate of 29,013 t derived by using 

 method II. 



All values in this table represent relative changes (expressed in 

 percentages) in CB 



2 AII values (i.e., for/ = 2-20) are simultaneously perturbed by 

 the same percentage indiated in the first row. 



TABLE 8.— Effects of perturbations in M,'s(i = 2 20) and conse- 

 quent perturbations in No, and PS upon the CB estimates of 30,012 

 and 29,013 t derived by using methods I and II, respectively. 



1 AII values in this table represent relative changes (expressed in percentages) in 

 CB. 



paper. These examples are presented to illustrate 

 the use of the sensitivity analysis technique in the 

 case of the CB estimation rather than to make a final 

 judgment of the validity of the CB estimates present- 

 ed. The knowledge, lacking in the case of southern 

 bluefin tuna, of probability distributions of all input 

 parameters would allow the application of stochastic 

 sensitivity analysis (Majkowski and Waiwood 1981; 

 Majkowski et aL 1981a; Majkowski 1982, 1983; 

 Majkowski and Hampton 1983, in press) and assist in 

 making such a judgment. 



CONCLUDING REMARKS 



The methods presented are mathematically very 

 simple. The system of equations associated with 

 method I is linear with respect to the CB and No, 

 variables. Therefore, the estimate of CB related to a 

 specified set of f, coefficients can be obtained an- 



730 



