OBSERVER EFFECT ON INCIDENTAL DOLPHIN MORTALITY IN 

 THE EASTERN TROPICAL PACIFIC TUNA FISHERY 



Bruce E. Wahlen' and Tim D. Smith^ 



ABSTRACT 



Scientific observers placed aboard a sample of purse seine vessels collect data that are used to estimate 

 the total number of dolphins killed incidentally in the eastern tropical Pacific tuna fishery. If the presence 

 of these observers, who are not crew members, affects incidental kill levels, then the kill estimates will 

 be biased, lb test for the existence of such an observer effect, we compared dolphin kill data that had 

 been recorded by observers who differed in levels of obtrusiveness according to their purposes for data 

 collection. Some observers were placed on board primarily to collect data for estimating the total number 

 of dolphins killed annually. Other observers collected data both for that purpose and for monitoring com- 

 pliance with dolphin-release regulations. Our results confirm that the presence of an observer does affect 

 dolphin kill. The primary effect is an increase in the proportion of sets with no dolphins killed, and a 

 decrease in the proportion of sets with one to nine dolphins killed. While the magnitude of the effect 

 of observers cannot be estimated from our data, estimates of total dolphin mortality based on data col- 

 lected by the scientific observers are biased downward. 



Schools of dolphins of several species, primarily 

 Stenella attenuata and 5. longirostris, have been 

 used since the late 1950s by purse seine fishermen 

 in the eastern tropical Pacific Ocean (ETP) to locate 

 and catch yellowfin tuna, Thunnus albacares. Per- 

 rin (1969) described the process of deploying, or set- 

 ting, the net around the tuna and dolphins, and then 

 releasing the dolphins while retaining the tuna. 

 Significant numbers of dolphins have been killed in- 

 cidentally in this fishery by becoming entangled in 

 the purse seines (Smith 1983). 



The National Marine Fisheries Service (NMFS) 

 and the Inter-American Tropical Tbna Commission 

 (lATTC) place scientific observers who are not crew 

 members aboard a sample of tuna purse seine vessels 

 to collect data related to dolphin kill. Both the NMFS 

 and lATTC have used the data collected by these 

 scientific observers to estimate the total number of 

 dolphins killed annually by the entire tuna purse 

 seine fleet (Lo et al. 1982; Hammond and Tbai 1983). 



Additionally, the NMFS uses these data to monitor 

 dolphin kills relative to annual kill limits establish- 

 ed for the U.S. registered fleet (Lo et al. 1982). 

 Periodic estimates of the cumulative numbers of 

 dolphins killed are compared with the annual limit. 

 If the limit is exceeded, U.S. vessels must stop fishing 



on the affected populations for the remainder of the 

 year. 



Data collected by the NMFS observers have also 

 been used to monitor compliance of vessel operators 

 with dolphin-release regulations, including the 

 release of all live dolphins from the net (Federal 

 Register 1977, 1980). Until recently data collected 

 by an NMFS observer could be used as evidence to 

 prosecute vessel operators for violations of these 

 regulations. 



Observer effects have been defined in a general 

 context as measurement procedures which influence 

 and thereby change the behavior of the subject 

 (Johnson and Bolstad 1973, p. 38). Researchers have 

 encountered such effects in a variety of empirical 

 sciences, including psychology (Johnson and Bolstad 

 1973), social science (Webb et al. 1966, p. 18), and 

 biology (Ricker 1975, p. 87). 



We defined an observer effect on the number of 

 dolphins killed as a differential in levels of dolphin 

 kill between trips made with and without a scien- 

 tific observer. The existence of such a differential 

 would introduce a bias into estimates of the total 

 number of dolphins killed (Smith 1983; Powers^). 

 Large numbers of sets involving dolphins (dolphin 

 sets) are made each year (Punsly 1983), so even a 

 moderate observer effect could result in a substan- 



'Southwest Fisheries Center La Jolla Laboratory, National 

 Marine Fisheries Service, NOAA, P.O. Box 271, La Jolla, CA 92038. 



^Southwest Fisheries Center La Jolla Laboratory, National 

 Marine Fisheries Service, NOAA, P.O. Box 271, La Jolla, CA; pres- 

 ent address: Northeast Fisheries Center Woods Laboratory, Na- 

 tional Marine Fisheries Service, NOAA, Woods Hole, MA 02543. 



Manuscript accepted November 1984. 

 FISHERY BULLETIN: VOL. 83, NO. 4, 1985. 



^Powers, J. E. 1979. A discussion of incidental mortality by 

 unobserved United States purse seiners. Unpubl. manuscr., 7 

 p. Southwest Fisheries Center La Jolla Laboratory, National 

 Marine Fisheries Service, NOAA, P.O. Box 271, La Jolla, CA 92038. 



521 



