544 



Fishery Bulletin 97(3), 1999 



1986-90 and 1992-93 marine mammal surveys 

 (Table 1). These seven surveys each occurred between 

 28 July and 10 December, referred to here as the 

 "study period." We used sightings only from search 

 effort within the NE offshore stock boundaries, for 

 which observers were actively searching ("on-effort") 

 and for which Beaufort sea state was 5 or below. We 

 excluded sightings farther than 5.5 km from the 

 trackline because school size estimates were gener- 

 ally unreliable for such distant schools. A total of 499 

 sightings were included in this study. 



The median estimated school size was 106 dol- 

 phins; 48% of the estimates were less than 100, and 

 only 17c greater than or equal to 1000 ( Fig. 2A ). These 

 quantities are not corrected for size selection bias 

 (see "Statistical model" section). The true distribu- 

 tion of school sizes was estimated to include a larger 

 percentage of small schools. 



Tuna vessel set data 



Since the late 1970s, the Inter-American Tropical 

 Tuna Commission (lATTC) and NMFS have placed 

 trained observers aboard a significant percentage of 

 ETP tuna purse-seiner vessels larger than 400 tons 

 capacity. A single observer is placed aboard a vessel 

 when a trip is selected for observation. These observ- 

 ers collect data on the dolphins with which commer- 

 cial species of tuna in the ETP commonly associate, 

 and monitor dolphin mortality due to purse-seine fish- 

 ing operations. For a more detailed description of the 

 observation programs, see lATTC (1989), lATTC,' or 

 Jackson (1993). 



Detailed observer data were available from tuna 

 sets made by U.S. vessels. From these data, we used 



observer estimates of school size and species compo- 

 sition from spotted dolphin schools set on by tuna 

 vessels within the NE offshore stock boundaries and 

 during the study period for the years 1986-90. We 

 used the observer's final "best" estimates, made af- 

 ter the set, which included counts of any animals that 

 evaded or were cut out of net encirclement'^ (lATTC, 

 1991; NMFS, 1992). The tuna vessel data used here 

 included only schools that were actually set on, and 

 not observations of schools that were sighted but not 

 set on. We assumed that there was no "observer ef- 

 fect" on a captain's choice of schools to be set on. A 

 total of 3454 set observations were included in this 

 study (Table 2). 



The tuna vessel set data represent about 77'^ of 

 the search effort during the study period for the U.S. 

 purse-seine fleet (29 to 40 vessels during the years 

 1986-90). To increase search efficiency, tuna vessels 

 often travel at 15 knots, and crews use helicopters 

 and communicate in "code groups" (Orbach, 1977) in 

 addition to searching with binoculars from the ship. 

 Not all schools that are sighted are set upon, and 

 thus the observed sets represent a much larger effec- 

 tive sample size in terms of sightings, and the "miss- 

 ing" schools tend to be small. The median estimated 

 size of schools that were set upon was 560 dolphins. 

 Only 4% of the estimates were less than 100, whereas 

 26% were greater than or equal to 1.000 (Fig. 2B). 



^ Using the pre-encirclement school size means that although we 

 are estimating the rate at which schools of a given size were set 

 on, it is likely that some dolphins involved in a given set were 

 chased by the tuna vessel and its speedboats but not actually 

 encircled by the purse-seine net. For the purposes of estimat- 

 ing numbers of sets as a measure of adverse impact, we do not 

 distinguish chase from capture. 



