332 



Fishery Bulletin 94(2). 1996 



A 



Dolphin sets 

 All areas 



10 20 30 40 50 



I) 



Log sets 

 Area 1 



10 20 30 40 50 



Log sets 

 Area 3 



V 



10 20 30 40 50 



10 20 30 40 50 



Observed discard (t) 



Figure 2 



(A-El Observed and fitted per-set tuna discards for school, dolphin, 

 and log sets for the U.S. tuna purse-seine fleet fishing in the eastern 

 tropical Pacific Ocean, 1989-92. Geographic areas are defined in Fed- 

 eral Register ( 1989) (Fig. 1 ). Bars indicate observed frequencies, lines 

 indicate fitted frequencies from the negative binomial with added ze- 

 ros model. Top portion (above horizontal division) of the fitted frequency 

 for the zero bin of dolphin and school set data represents "true zeros," 

 bottom portion represents "rounded" zeros from negative binomial (NB) 

 component. Vertical axes have been cut and scaled to make NB compo- 

 nents directly comparable. Not shown: (C) 3 observation >50 tons, 1 1)> 

 16 observations >50 tons, (E) 12 observations >. r i() tons 



of discard or in the absence of a plausible model for 

 the measurement errors, we treated the discard 

 weight estimates as exact measurements. 



Discard weight was recorded for 59% (2,110 of 

 3,590, Table 1) of observed dolphin sets, 76% (960 of 

 1,266) of observed school sets, and 75% (998 of 1,328) 

 of observed log sets. These sets generated 134, 1,098, 

 and 9,819 tons of reported discard, respectively. The 



relatively small discard totals for school and dolphin 

 sets were due to the large numbers of those sets with 

 zero discard reported. Positive amounts of tuna dis- 

 card were reported in 65% (650 of 998, Table 1 ) of log 

 sets for which discard was recorded, but in only 8 r /c 

 (80 of 960) of school sets and only 0.9% ( 19 of 2,110) 

 of dolphin sets for which discard was recorded. We 

 ignored log and school fishing in area 2 (see next sec- 



