652 



Fishery Bulletin 88(4), 1990 



Particularly in the three- and five-species cases, it 

 appeared that the optimal multispecies strategy was 

 to maintain stable catches for the species accounting 

 for the majority of total yield, such as shortbeily or 

 widow rockfish. Catches of less-abundant species (e.g., 

 bocaccio or splitnose rockfish) varied depending on the 

 abundance of the dominant species. For example, in 

 the five-species case, the annual catch of splitnose 

 rockfish was in 24-30 of 100 years for the logh 

 policies. 



Errors in biomass estimates 



Constant F policies were essentially identical when 

 different CVs for biomass estimates were used. This 

 indicates that the current management approach for 

 Pacific Coast groundfish is robust to random, non- 

 autocorrelated errors in estimating biomass. One ob- 

 vious impact in the simulated fishery of introducing 

 errors in estimating biomass was that the variance for 



total catch increased substantially (Tables 3-.5, P"'ig. 4). 

 This increased variability was due to differences be- 

 tween the intended (F) and actual (F') fishing mortal- 

 ity rate. Such errors are nonlinearly related to esti- 

 mated stock size and can be substantial (Rivard 1981 ). 

 Rivard (1981) argued for higher precision in estimates 

 of stock size in order to reduce the biological risk of 

 applying F'»F or the economic loss if F'«F. In this 

 study, the variation in F due to errors in estimating 

 biomass was much greater than that caused by using 

 the variable F policies. For example, when CV = 0%, 

 Fs for chilipepper in the five-species case ranged from 

 0.00 to 0.24 when using the multispecies logh model. 

 When CV = 25%, the actual Fs ranged from 0.06 to 0.61 

 when the intended policy was a constant F of 0.22. 

 Actual Fs ranged from 0.00 to 0.63 when CV = 25% and 

 a multispecies logh policy was used, which suggests 

 that multispecies policies do not introduce substantial 

 additional risk, relative to the risk introduced by esti- 

 mating biomass with error. 



