744 



Fishery Bulletin 90(4). 1992 



82 83 84 

 Year 



Figure 6 



Medians, 2.5th percentiles, and 97.5th percentiles of the out- 

 put distributions from the Monte Carlo simulations, (a) 

 Distribution of estimates of the population size of swordfish 

 Xiphius gladius aged 5 and above; (b) distribution of estimates 

 of fishing mortality for swordfish aged 5 and above. 



6b and illustrates how very easily the Monte Carlo 

 approach lends itself to hypothesis testing. 



Of course, the goals of an assessment are not re- 

 stricted to estimating population sizes and mortality 

 rates. Interest is often centered on catch projections 

 and quotas, effort regulations, and risk analyses. For 

 swordfish assessments, it is useful to contrast the 

 estimated current level of fishing mortality against 

 reference points such as Fq.i and Fmax- The uncertain- 

 ty in such comparisons (e.g., the ratio of current F to 

 Fo i) can easily be quantified using the Monte Carlo 

 procedure. For each simulation run, we computed the 

 multiplier that would be necessary to bring the esti- 

 mated vector of age-specific fishing mortalities in the 

 terminal year to the Fq.i and F^^^x levels (Fig. 8). For 

 the computations, we used the run-specific natural 



400- 



200 



0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5 

 Relative Change in Current F to Reach: 



I Fmax j 



I F0.1 



Figure 8 



Multipliers necessary to bring the vector of age-specific fishing 

 mortalities in the terminal year to the F^ ^ and F„„ levels, 

 for 1000 simulated data sets for swordfish Xiphias gladius. 



mortality rate and the weight-at-age relationships used 

 by ICCAT in the 1989 assessment. No uncertainty was 

 specified for weight relationships although this could 

 easily be added if appropriate information were avail- 

 able. From Figure 8 it is evident that, to achieve the 

 Fo.i goal, fishing mortality must be cut to ~25% of its 

 current value. With respect to F^^^, it appears that 

 fishing mortality must be cut by ~50% (Fig. 8). Note, 

 however, that this conclusion is considerably less 



