250 



Fishery Bulletin 97(2), 1999 



shaded. The diagonally shaded 5° cells indicate where 

 the billfish proportion of the catch within the 5° area 

 exceeded the 4.36'?^ threshold required to obtain a 

 50*^ reduction in the 1986-91 billfish bycatch indi- 

 cated in Table 2. The horizontally shaded 2" cells in- 

 dicate those cells where the billfish proportion ex- 

 ceeded the threshold for the 2" spatial resolution, and 

 the filled 1° cells indicate those cells where the bill- 

 fish proportion exceeded the 1° threshold. Inspection 

 of these plots indicates a temporal pattern in the dis- 

 tribution of fishing effort — from southern waters in 

 the winter to northern and eastern waters during 

 the summer and fall. Also the minimum number of 

 cells that would have to have been eliminated from 

 the 1986-91 fishery to reduce the billfish bycatch by 

 50*^ occur during the months of September through 

 November There is also a tendency for the affected cells 

 to move north fi-om the southernmost regions fished in 

 November and December to the entire U.S. coastline 

 by July and August. 



Similar plots are presented for marlins in Figure 4, 

 with the respective threshold percentages of marlin 

 in the catch. As for billfishes combined, the number 

 of cells above the threshold percentages is at a mini- 

 mum in October but increases thereafter to a peak 

 in April. The affected cells move north during the 

 spring and summer from primarily southern lati- 

 tudes in the winter. However, some cells in the Car- 

 ibbean exceed the 507f thresholds of Table 3 in al- 

 most every month. 



To test the temporal and spatial stability of the 

 areas identified in Figures 3 and 4, I calculated the 

 percent reductions in billfish and marlin bycatch and 

 target species that would have occurred if those ar- 



eas had been closed in each year from 1992 through 

 1995 for each level of spatial resolution considered 

 (Tables 4 and 5). These estimates assumed that the 

 time-area strata identified for closure would not be 

 redirected elsewhere but did include strata fished in 

 1992-1995 that were not present in the 1986-91 base 

 period. For billfish combined, billfish would have 

 been selectively protected by closing the time-area 

 cells identified from the 1986-91 data evaluation 

 each year from 1992 through 1995 with all levels of 

 spatial resolution examined. The agreement between 

 predicted and 1992-95 observed values increased 

 with increasing grid size (Table 4). Mean billfish re- 

 ductions increased ft-om 28.6% to 41.6% fi-om 1° to 5° 

 cells compared with the predicted levels of 50%. The 

 mean reductions for targeted species were slightly 

 less than the predicted values. The observed mean 

 values were 12.8, 11.9, and 16.7%, compared with 

 the predicted values of 13.9, 17.6, and 19.2% for the 

 1', 2° and 5° cells, respectively. In general, the larger 

 5° grid produced results that were closest to the val- 

 ues predicted from the 1986-91 data set and showed 

 the least year-to-year variability in the percent re- 

 duction of billfish bycatches. 



For the analyses considering marlin species only, 

 marlins would also have been selectively protected 

 by closing the time-area cells identified from the 

 1986-91 data evaluation each year from 1992 

 through 1995 with all levels of spatial resolution ex- 

 amined. The agreement between predicted and 1992- 

 95 observed values also increased with increasing 

 grid size (Table 5) but were closer to the predicted 

 values than for the evaluation of billfish combined. 

 Mean marlin reductions increased from 35.7%^ to 



