Hightower: Rockfish harvesting policies in Washington-Oregon-California trawl fisheries 



649 



ings of splitnose and shortbelly rockfish have been 

 negligible to date; consequently, parameter estimates 

 for those species should be viewed as highly prelim- 

 inary. 



Results and discussion 



Objective functions 



Results were similar for constant F policies using dif- 

 ferent objective functions. Optimal constant Fs gen- 

 erally were comparable with M, although the F for 

 chilipepper (M = 0.20) was higher than that for bocac- 

 cio (M = 0.25) (Table 2). That apparently was due to the 

 higher age at first recruitment for chilipepper. (For 

 bocaccio and chilipepper, the slight differences in con- 

 stant Fs in the two- and three-species cases were due 

 to random error.) Optimal Fs were lower for the five- 

 species case than for the two- and three-species cases 

 because harvests in years 1-74 were excluded from the 

 five-species objective function. Policies obtained assum- 

 ing a transient period may be unacceptable on socio- 



economic gi-ounds but can probably be viewed as a prac- 

 tical lower bound for F. 



For each of the models, the differences in F, total 

 yield, and variance for total yield showed a consistent 

 trend: highest for maxh policies, intermediate for k)gh 

 pohcies, and lowest for negx policies (Tables 2-5). 

 Relative to the variance for maxh policies, logh con- 

 stant F policies resulted in about a 2% reduction in total 

 variance, whereas variances for the negx policies were 

 3 (2-species case) to 10% (5-species case) lower. These 

 results illustrate that moderately lower variances can 

 be achieved by using slightly lower Fs and accepting 

 slightly lower total yields. 



Constant vs. variable F policies 



The variable F policies (5)-(6) for maximizing harvest 

 did not increase total yield significantly compared to 

 the constant F policies (Tables 3-5). Results from 

 earlier studies suggest that the degree of increase due 

 to a variable F policy may depend on the assumed life- 

 history characteristics. For example, a 17% increase 



