Wade: Estimates of incidental kill of dolphins by the purse-seine tuna fishery 



353 



versus my estimate of 0.46. My CV's from 1966 to 

 1972 (from 0.13 to 0.20) are in the same range as 

 bootstrap CVs reported for 1979-87 (IATTC, 1989) 

 for which there were larger sample sizes of observed 

 sets. These data indicate that there was less varia- 

 tion in MPS between trips in the data set used here 

 than was usual in succeeding years. This may have 

 been an artifact of a relatively small sample size. 

 Alternatively, as MPS declined after 1972, the vari- 

 ance in MPS may have actually increased. One pos- 

 sible explanation for increased variance may be that 

 MPS declined for sets that went as planned but did 

 not decline for high-mortality sets. This could lead 

 to a decrease in average MPS and an increase in the 

 variance. 



As expected, CVs of the kill for each stock were 

 higher than those for the total kill because they in- 

 cluded additional uncertainty in species proportions. 

 CVs for northeastern spotted dolphins were on av- 

 erage 0.05 more than CVs for total dolphin mortal- 

 ity. CVs were even greater for eastern spinner dol- 

 phins, averaging 0.10 more than for total dolphin 

 mortality. 



Variance estimates presented here are estimates 

 of the precision given the assumptions of the analy- 

 sis, such as constant MPS rates in 1959-72. They 

 cannot provide a complete measure of uncertainty of 

 how many dolphins were killed during this period, 

 because some untestable assumptions were made 

 which could lead to bias. However, this is always the 

 case for statistical variance estimates, unless one can 

 be assured that all assumptions have been met or 

 unless one accepts subjective beliefs and incorporates 

 them into a Bayesian statistical analysis (Press, 

 1989). For example, the assumed linear increase in 

 the use of the backdown procedure between 1959 and 

 1964 (Lo and Smith, 1986) is probably open to ques- 

 tion, but it is unlikely that there are data available 

 to test these kinds of assumptions. 



Although my estimates of the numbers of dolphins 

 killed in the early years of the fishery come from a 

 limited set of data on MPS and seem quite large, 

 these estimates are consistent with statements from 

 scientists involved in the fishery then (e.g. "in the 

 early years of the fishery these incidental mortali- 

 ties were very high," Joseph, 1994 ). Additionally, my 

 estimates are reasonable when compared with esti- 

 mates from later years that had substantially more 

 data on MPS. Even after expected declines in MPS 

 due to the passage of the MMPA in 1972 and the 

 increased use of the Medina panel, Wahlen (1986) 

 estimated a kill of 197,000 in 1973, a reduction of 

 about one-third from the average kill of 308,072 per 

 year in the two previous years (Table 4). As recently 

 as 1986, the mortality of dolphin was estimated to 



be 133,174 (IATTC, 1989). Of those, 52,000 were es- 

 timated to be northeastern spotted dolphins, repre- 

 senting 7.1% of the populations estimated abundance 

 in 1986-90 (Wade and Gerrodette, 1993), a level of 

 mortality that is probably not sustainable. 



Events in recent years have shown how quickly 

 MPS can change. A variety of efforts to decrease MPS 

 in the fishery (Joseph, 1994) have caused the total 

 kill to decline dramatically to levels of 27,292 in 1991 

 (Hall and Lennert, 1993), 15,539 in 1992 (Hall and 

 Lennert, 1994), and 3,601 in 1993 (Hall and Lennert, 

 in press). The 1993 mortality of northeastern spot- 

 ted dolphins was 1,139 or 0.16% of the 1986-90 abun- 

 dance estimate, and all other stocks had even lower 

 percent mortality levels. The low 1993 levels of mor- 

 tality (if continued I should allow future growth and 

 recovery of these populations, even if they are cur- 

 rently depleted. 



Acknowledgments 



I would like to thank the observers on the tuna ves- 

 sels who collected the mortality-per-set data and 

 Randall C. Rasmussen and Ken Wallace for helping 

 to guide me through the NMFS tuna vessel observer 

 databases. I would especially like to thank Michael 

 G. Hinton for so efficiently providing the set effort 

 data. This manuscript was improved considerably by 

 comments from Robert Brownell, Tim Gerrodette, 

 Michael G. Hinton, Keith Mullin, Joyce E. Sisson, 

 and Steve Reilly. I thank them for their contribution 

 and note that the author is solely responsible for the 

 contents. 



Literature cited 



Au, D. W. 



1991. Polyspecific nature of tuna schools: shark, dolphin, 

 and seabird associates. Fish. Bull. 89:343-354. 

 Barham, E. G., W. K. Taguchi, and S. B. Reilly. 



1977. Porpoise rescue methods in the yellowfin purse seine 

 fishery and the importance of Medina panel mesh 

 size. Mar. Fish. Rev. 39(5):1-10. 

 Dizon. A. E., W. F. Perrin, and P. A. Akin. 



1994. Stocks of dolphins (Stenella spp. and Delphinus 

 delphis) in the eastern tropical Pacific: a phylogeographic 

 classification. U.S. Dep. Commer., NOAA Tech. Rep. 

 NMFS 119, 20 p. 

 Edwards, E. F. 



1989. Using tuna-vessel observer data to detect trends in 

 abundance of dolphin populations: history and research to 

 date (1988). U.S. Dep. Commer.. NOAA Tech. Memo. 

 NMFS-SWFC-122, 123 p. 

 Efron, B. 



1982. The jackknife. the bootstrap, and other resampling 

 methods. SIAM, monograph 38, CBMS-NSF, 92 p. 



