LO and SMITH: INCIDENTIAL MORTALITY OF DOLPHINS 



ratios vary without a consistent trend over time 

 (Table 2). 



In unsuccessful sets the use of the backdown pro- 

 cedure was more variable because the conditions of 

 the set are more diverse For example, only a few 

 or no dolphins may be captured, and the net may not 

 be retrieved in the usual manner. Because of this 

 diversity and because so few observations are avail- 

 able, we consider one kill rate for all unsuccessful 

 sets. 



Reexpressing Equation (2) to account for a 

 constant ratio of kill rates for successful sets when 

 backdown was used and when it was not used, and 

 for no difference in kill rates for unsuccessful sets, 

 yields 



Mil 



i=i y_i k-1 



■K'ijk-X-tijk 



= 2. l^nll V^till + CX H12 ) + R. i2 .X tl2 .) (3) 



1=1 



where C = R.. l2 IR,. n and the subscript . is used 

 when that stratifying variable is not considered. For 

 example, R.^ is the kill-per-set not stratified by 

 year t, and X H2 . is the total number of sets not 

 stratified by use of backdown. 



Estimates of the total number of sets involving 

 dolphins from 1959 to 1972. with approximate 

 variances, are given by Punsly (1983). He also gives 

 partial estimates of the numbers of successful and 

 unsuccessful sets, but does not provide estimates of 

 the numbers of sets by vessel size Punsly's data did 



not indicate the use of the backdown procedure 



The coefficients of variation (CV) of Punsly's 

 estimates are <1% in all years except 1959 and 1960, 

 when it was 8%. The percentage of unidentified sets 

 in 1959-61 was higher than subsequent years because 

 set type was not recorded systematically 

 (Hammond 4 ). We assume these estimates are in fact 

 constants, because in most years, and in the absence 

 of additional information in 1959-61, the CVs are 

 small compared with the CVs of the kill rates 

 (0.13-1.0, Table 1). 



By applying the proportions of successful and un- 

 successful dolphin sets from Punsly's partial 

 estimates to his totals, we obtained numbers of suc- 

 cessful and unsuccessful dolphin sets. We further 

 prorate these estimated numbers of successful and 

 unsuccessful sets to large and small vessels by 

 multiplying by the estimates of proportions from 

 NMFS (Anonymous 1976 5 ) of sets made by vessels 

 of each size class (Table 3). The slight differences 

 between the totals for each year given by Punsly are 

 due to rounding. 



The number of sets during which backdown was 

 used can be estimated from the estimated total 

 number of sets involving dolphins (Table 3) and the 

 observed proportion of successful sets in which back- 

 down was used (Table 2). The observed proportions 

 increase from 0.79 in 1964-65 to almost unity (0.96) 

 by 1972. The backdown procedure was reportedly 



4 P. S. Hammond, Sea Mammal Research Unit, British Antarctic 

 Survey, Cambridge, England, pers. commun. 1983. 



6 Anonymous. 1976. Report of the workshop on stock assess- 

 ment of porpoises involved in the eastern Pacific yellowfin tuna 

 fishery (La Jolla, July 27-31, 1976). Southwest Fish. Cent., Ad- 

 min. Rep. LJ-76-29, 54 p. + app. 



Table 2. — Mean number of dolphins killed (R) during purse seine sets in the eastern tropical 

 Pacific Ocean when the backdown dolphin-release procedure was and was not used. Also 

 given are the ratio of numbers killed with and without backdown (C), the proportion of suc- 

 cessful sets where backdown was used (P), the number of sets (A/), number of trips, and stan- 

 dard error in parentheses. 



1 From Smith and Lo (1983). 



2 Kill rates tor 1966 omitted because incomplete data were collected. 



3NMFS records. 



<SE(£) = 6 [CV2(rt.. 12 ) + C\z2(tf. #11 ) - 2cor(A.. 12 , 4.. 11 )] 1 ' 2 ; where C = A.. 12 ff). a11 . 



29 



