FISHERY BULLETIN: VOL. 84, NO. 3 



location, and estimated tuna catch of each set, and 

 other information describing the fishing operation. 

 For sets involving dolphins (dolphin sets), the ob- 

 servers collected additional data, including the 

 number of dolphins killed by species. 



In 1973, all sampled trips were arranged with 

 vessel captains under a voluntary sampling pro- 

 gram. Beginning in 1974, trips to be sampled were 

 determined from a randomly ordered list of vessels. 

 Which trips were actually sampled during 1974 and 

 1975 depended on several factors, including the 

 cooperation of the captains. Because of uncertain- 

 ties about the cooperation of the captains and the 

 number of fishing trips that would be made in a year, 

 observers were placed on vessels as soon as possi- 

 ble. Thus, before 1976, the planned number of sam- 

 pled trips was frequently obtained in the first half 

 of the year. The sampling process became more ran- 

 dom starting in 1976, when participation in the 

 sampling program became mandatory for captains 

 making sets on dolphins. 



I extracted independent information for the 

 population of all fishing trips by U.S. registered tuna 

 purse seiners in the ETP from the Inter- American 

 Tropical Tuna Commission (IATTC) logbook data 

 base. This data base contains abstracts of vessels' 

 logbooks obtained by IATTC personnel. An in- 

 dividual entry in the logbook data base provides 

 information about one or more sets, including num- 

 ber of sets, set type, date and location, estimated 

 tuna catch by species, but not numbers of dolphins 

 killed. 



The logbook data are incomplete in that number 

 of sets may be missing, set type may not be re- 

 corded, and information for some sets of a trip and 

 for all sets of some trips may be omitted (Punsly 

 1983). To compensate for these omissions from the 

 logbook data, Punsly (1983) estimated the total 

 number of dolphin sets made by all U.S. and 

 non-U. S. seiners in the ETP. He then modified this 

 procedure to estimate the total number of dolphin 

 sets made by U.S. seiners only (Table 1). 



METHODS 



I stratified the data to allow for potential differ- 

 ences in dolphin kills. If kills do indeed differ among 

 strata, then a stratified estimator may be superior 

 to an unstratified estimator in two respects. First, 

 a stratified estimator will have a smaller standard 

 error and thus be more precise. Second, if the sam- 

 ple data are unrepresentative of the population with 

 respect to these strata, a stratified estimator will 

 be less biased. 



Therefore, estimates of incidental dolphin mortal- 

 ity by species or species grouping were computed 

 using a stratified kill-per-set ratio estimator, follow- 

 ing the general approach described by Lo et al. 

 (1982). I excluded trips which made no dolphin sets 

 and experimental-gear trips from the sample and 

 the population. However, I added dolphin kills in- 

 cidental to the experimental-gear trips as constants 

 to the mortality estimates. 



I stratified the dolphin set data by four factors 

 used in previously published estimates: 1) year of 

 the set, 2) fish-carrying capacity of the vessel, or 

 simply vessel capacity, 3) period within year of the 

 set, and 4) geographic location of the set. Vessel 

 capacity was divided into two categories— small and 

 large. The breakpoint between categories was deter- 

 mined by examining the cumulative distribution of 

 sampled trips by capacity. Periods were defined to 

 be quarters of the year, considering the results of 

 Wahlen and Smith (1985). The ETP was divided in- 

 to three geographic areas— North Inside, North Out- 

 side, and South (Fig. 1)— because mean kill (per set) 

 after 1978 has been shown to differ among these 

 areas. 3 



In previous estimates, the amount of tuna caught 

 in the set was included as a stratification factor. 

 However, Hammond and Tsai (1983) found that 

 stratification by this factor made very little differ- 

 ence in their estimates. For that reason and to avoid 

 possible overstratifi cation, I omitted amount of tuna 

 caught as a stratification factor. 



I pooled dolphin set data over strata when it was 

 determined that between- strata differences in mean 

 kill were not statistically significant or that sample 

 sizes were otherwise too small. I prorated the esti- 

 mated numbers of dolphin sets made by U.S. seiners 

 (Table 1) among the pooled strata according to pro- 



3 K. -T. Tsai, Inter-American Tropical Tuna Commission, c/o 

 Scripps Institute of Oceanography, La Jolla, CA 92093, pers. com- 

 mun. December 1983. 



Table 1.— Estimated number of dolphin sets 

 made by U.S. purse seiners fishing in the 

 eastern tropical Pacific, by year. 1 



Year 



Number of dolphin sets 



1973 

 1974 

 1975 

 1976 

 1977 

 1978 



8,341 

 7,475 

 7,902 

 7,126 

 7,239 

 4,214 



'Peterson, C. L. (editor). 1984. The quarterly 

 report October-December 1983 of the Inter-American 

 Tropical Tuna Commission. Inter-Am. Trop. Tuna 

 Comm., c/o Scripps Inst. Oceanogr., La Jolla, CA 

 92093. 



560 



