WAHLEN: INCIDENTAL DOLPHIN MORTALITY 



as the square of the product of (a) the number of 

 dolphin sets 4 (Table 10) and (b) the corresponding 

 estimated standard error of the total kill-per-set 

 ratio (Table 11). I computed the estimated stratum 

 variance for each species or species grouping, sub- 

 stituting values for the species or species grouping 

 for the total in (b) above. I estimated the standard 

 errors (Table 11) using mean number of dolphin sets 

 per trip calculated from the sample rather than from 

 the population (Lo et al. 1982). 



My annual estimates of the total number of 

 dolphins killed incidentally in the U.S. purse seine 

 fishery of the ETP ranged from a maximum of 

 125,000 dolphins in 1973 to a minimum of 14,000 

 dolphins in 1978, with coefficients of variation (CV) 

 no greater than 10% (Table 13). The estimated mor- 

 talities of the two species most often exploited, 

 spotted and spinner dolphins, together accounted 

 for about 80-90% of each annual total. 



DISCUSSION AND CONCLUSIONS 



The kill-per-set ratio, or mean kill, declined from 

 15 dolphins/set during the 1973-76 period to 3 

 dolphins/set during the 1977-78 period (Table 11, 

 pooled over quarter and area). Many changes affect- 

 ing dolphin kill were made during these periods, 

 including improvements in fishing gear and dolphin- 

 release procedures and introduction of federal 

 regulations. The change in mean kill between 

 1973-76 and 1977-78 coincided with the first NMFS 

 notice in late 1976 prohibiting fishing on dolphins 

 for the remainder of the year. This one example of 

 a correspondence between a change in the number 

 or distribution of dolphins killed during purse seine 

 sets and an identifiable legal or management action 

 is not necessarily indicative of a cause and effect 

 relationship. There are, however, two other ex- 

 amples of such a temporal correspondence present 

 in the data from 1964 through 1982. 



In the second example, the data prior to 1973, 

 while sparse, suggest that the mean kill was substan- 

 tially higher than during the 1973-76 period. Lo and 

 Smith (1986) reported a mean kill of 46 dolphins/set 

 based on 1964 through 1972 data, pooled over vessel 

 capacity and catch of tuna. They found no consis- 

 tent differences in annual mean kill during that 

 period. The decline in the mean from 46 dolphins/ 

 set during the 1964-72 period to 15 dolphins/set dur- 



4 Numbers of dolphin sets were treated as constants since no 

 variances were provided for these estimated quantities. Therefore, 

 my variances of estimated mortality are underestimated to an 

 unknown, though likely small, degree. 



ing the 1973-76 period coincided with the passage 

 of the Marine Mammal Protection Act in late 1972. 



In the third example, Wahlen and Smith (1985) 

 demonstrated a difference between the two periods 

 from 1979 through March 1981 and from April 1981 

 through 1982 in the frequency distributions of 

 number of dolphins killed during purse seine sets. 

 While the difference in mean kill during these two 

 periods was not significant, the percent of dolphin 

 sets in which no dolphins were killed (zero-kill sets) 

 decreased significantly. This decrease coincided with 

 a court order in March 1981 which prohibited using 

 data collected by NMFS observers to monitor com- 

 pliance of vessel captains with dolphin-release 

 procedures. 



These three examples suggest that significant 

 legal or management actions can affect kill rates, 

 measured by the kill-per-set ratio or by the percent 

 of zero-kill sets. Furthermore, such effects on the 

 kill-per-set rate are reflected in the series of esti- 

 mates of total numbers of dolphins killed presented 

 here. For example, between 1976 and 1977 the num- 

 ber of dolphin sets increased slightly (Table 10), yet 

 total estimated mortality decreased by nearly 80% 

 (Table 13), due primarily to the significantly lower 

 kill rate after 1976. The further decline in estimated 

 mortality to 14,000 dolphins in 1978 reflects both 

 the lower kill rate and a decline in the number of 

 dolphin sets. Thus, the decrease in the kill rate 

 following the first enforcement of dolphin kill limits 

 in 1976 is reflected in the decrease in the estimates 

 of total mortality after 1976. 



My estimates of total annual dolphin mortality in 

 U.S. purse seine fishing from 1973 through 1978 

 (Table 13) are lower, except for 1976, and more 

 precise than those in the Status of Porpoise Stocks 

 Workshop Report (SOPS) (Table 14). However, for 

 each year except 1973, my estimated total is con- 

 tained inside an approximate 95% confidence inter- 

 val around the estimated total (t) in SOPS, where 

 the confidence interval is computed as T ± 2 • CV • T. 

 Thus, the differences between my point estimates 

 and those in SOPS are small when the imprecision 

 (large CV) of the SOPS estimates is considered. 



The lower precision of the SOPS estimates may 

 be due to overstratification because of a concern that 

 the sample might not be representative of the 

 population, particularly during the 1973-75 period. 

 Thus, in order to minimize bias, a large number of 

 strata (32 per year) were defined. Tests for between- 

 strata differences in mean kill were not made, but 

 strata were pooled to the degree that each pooled 

 stratum contained some sample data. However, even 

 after pooling, some strata during the years 1973-76 



567 



