FISHERY BULLETIN: VOL. 73, NO. 1 



Table 9. — Estimated parameters for the five replicated stochastic catch histories using the 

 equihbrium approximation and transition prediction approaches for fixed estimates of m. 



'Program PRODFIT; /( = 4, weighted estimates option. 

 ^Program GENPROD; KK = 3, DEL = 5, weighted estimates. 



Table 10. — Summary of Ymax estimates by alternative strategies with the equilibrium 

 approximation and transition prediction approaches for five replicated stochastic catch 

 histories. Empirical value of y^gx is 5.60. 



'Program PRODFIT. 

 ^Program GENPROD. 



set unless either there was no fishing prior to the 

 first record of the set or some information is avail- 

 able on the approximate level of catch and effort. 

 One should check carefully to ensure that critical 

 points (those being the only points at high, low, or 

 intermediate levels of fishing) are not excluded or 

 that the fitted model does not deviate greatly from 

 where they might reasonably be expected to lie. If 

 fishing effort was reasonably constant or negligi- 

 ble prior to the first record, dummy data of length 

 ^ - 1 can be employed to allow use of the first few 

 data points. Also, since the average fishable dura- 

 tion, T, is less than the number of significant 

 fishable year classes, k , the unweighted averaging 

 method will result in fewer data being excluded 



In any case, the sensitivity of the parameter esti- 

 mates to alternative averaging times should be 

 explored. 



No data points are excluded with the transition 

 prediction approach, a positive factor which 

 should be considered even if one is satisfied with 

 the parameter estimates obtained with the 

 equilibrium approximation approach. On the 

 other hand, the transition prediction approach 

 utilizes five parameters while the equilibrium ap- 

 proximation approach utilizes only three, so that 

 vdth few significant year classes in the fishable 

 population there is little difference between the 

 required number of data points. For example, the 

 transition prediction approach statistically 



34 



