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



349 



In this study, I duplicated the method of Smith 

 (1983), using the observed proportions of dolphins 

 killed in each stock from the 1971-72 MPS data, with 

 the same assumption about whitebelly spinner kill. 

 However, to calculate the variance of the estimate of 

 the number of dolphins killed in each stock in each 

 year, I used the bootstrap method (Efron, 1982): each 

 fishing trip was a resampling unit and there were 

 1,000 iterations. Thus, on each bootstrap iteration, 

 20 fishing trips were resampled with replacement 

 from the 1959-72 pooled observer data, and the MPS 

 rates were recalculated and multiplied by the strati- 

 fied set totals to estimate the total kill, which was 

 then prorated to stock by the recalculated species 

 proportions. The variance of the kill of each stock in 

 each year was then estimated as the variance of the 

 1,000 bootstrap estimates of that stock. This method 

 automatically incorporates into the variance uncer- 

 tainty due to the observed proportions of each stock 

 killed in the 1971-72 MPS observer data. Estimates 

 of number of dolphin sets were precise, having coef- 

 ficients of variation less than 17c in all years except 

 1959 and 1960 (Punsley, 1983) and were therefore 

 treated as constants. 



For comparative purposes, I made two additional 

 calculations. First, I investigated the effect of the 

 1964, 1965, and 1968 trips by recalculating the kill 

 of dolphins using only the MPS data collected dur- 

 ing the observer program in 1971 and 1972. Second, 

 I calculated stratified MPS rates from the 1973 ob- 

 server data to compare to MPS rates used here from 

 1964 to 1972. 



Results 



Average MPS for each of the 20 observed trips be- 

 tween 1964 and 1972 showed that most of the trips 

 had similar kill rates (Table 2). Calculated estimates 

 of the MPS for each category of year, vessel size, and 

 set type (Table 3) were, as expected, equivalent to 

 the values reported by Lo and Smith ( 1986). The six 

 R tljk values used in Equation 4 are also shown in 

 Table 2. 



Estimating total dolphin mortality with my MPS 

 rates (Tables 2 and 3) and the number of sets from 

 Lo and Smith ( 1986, their Table 3 ) resulted in slightly 

 different estimates of the total number of dolphins 

 killed than those they reported in their Table 4. How- 

 ever, when I used the six MPS values reported in 

 their Tables 1 and 2, I obtained their estimates. Of 

 six MPS values in their Tables 1 and 2, two were not 

 the same as the MPS values required by their Equa- 

 tion 4. Their Table 1 reports MPS for both large and 

 small vessels by category of successful set, but pooled 



over backdown or no backdown (R. n . and #. 21 ., re- 

 spectively), whereas Equation 4 requires the MPS 

 for both large and small vessels by category of suc- 

 cessful set and by backdown status (R. lu and -R. 211 , 

 respectively). I duplicated the results of Lo and Smith 

 (1986) exactly using R, n . and -R. 21 . and conclude 

 that they inadvertently used these values, versus 

 R. 1U andi?. 2n as they intended, because their equa- 

 tion is correct. 7 The estimates reported here (Table 

 4) were calculated by using Equation 4 and R. m and 

 R. 2U of 29.9 and 65.8, respectively. 



Lo and Smith ( 1986) did not prorate their estimates 

 of total dolphin mortality to stocks. Using reported 

 proportions from the 1971-72 MPS observer data, 

 Smith (1983) prorated 0.70, 0.23, 0.03, and 0.04 of 

 the total mortality to offshore spotted, eastern spin- 

 ner, whitebelly spinner, and other dolphins, respec- 

 tively. My results assign 0.694, 0.241, 0.034, and 

 0.029 of the mortality to the same categories. 8 



Annual estimates of incidental mortality of dol- 

 phins ranged from a low of 23,485 in 1959 to a high 

 of 558,572 in 1961; CVs ranged from 0.13 to 0.48 

 (Table 4). Nearly 5 million dolphins were estimated 

 to have been killed by the purse-seine fishery over 

 the fourteen-year period considered here, an aver- 

 age of 347,082 per year. More northeastern spotted 

 dolphins were killed than from any other stock; a 

 total estimate for the period was 3.0 million (2 11,612 

 per year). Totals for other stocks were 0.4 million 

 (29,361 per year) western/southern spotted dolphins, 

 1.3 million (91,739 per year) eastern spinner dol- 

 phins, 0.05 million whitebelly spinner dolphins, and 

 0.15 million dolphins from other species. CVs of the 

 annual kill estimates ranged from 0.17 to 0.53 for 

 the northeastern spotted, 0.19 to 0.54 for the west- 

 ern/southern spotted, 0.32 to 0.47 for the eastern 

 spinner, 0.45 to 0.50 for the whitebelly spinner, and 

 from 0.84 to 0.89 for other dolphins. 



When I recalculated dolphin mortality using only 

 MPS data from the 17 trips for 1971-72, total dol- 

 phin kill for 1959-72 was estimated to be 5.1 mil- 

 lion. In all years except 1972 the estimates were 

 slightly higher than those in Table 4 that were cal- 

 culated by using all 20 1964-72 trips. The higher 

 value in 1972 was due to a slightly higher value for 

 the ratio of MPS without backdown to MPS with 

 backdown (C in Eq. 3). Owing to the decrease in the 



' Original records were not available to confirm which values were 

 used (Lo, N. C. H., Southwest Fisheries Science Center, NMFS, 

 La Jolla, CA, 92038, and T. D. Smith, Northeast Fisheries Sci- 

 ence Center, NMFS, Woods Hole, MA, 02543. Pers. commun., 

 1993). 



8 Differences are likely due to round-off error. Additionally, they 

 may be also due to revisions made to the NMFS observer data- 

 base (Rasmussen, R. C, Southwest Fisheries Science Center, 

 NMFS, La Jolla. 92038. Pers. commun., 1993). 



