REACTION OF DOLPHINS TO A SURVEY VESSEL: 

 EFFECTS ON CENSUS DATA 



Roger P HEvmT^ 



ABSTRACT 



A field experiment is described in which a helicopter was used to observe the efficiency of shipboard 

 line-transect sampling of dolphin populations in the eastern tropical Pacific Ocean. Nineteen dolphin 

 schools were tracked; 13 of these were detected by observers aboard the ship and 5 of these reacted to the 

 approach of the ship by altering the direction and/or the speed of their movement; however, only 1 school 

 reacted prior to shipboard detection. The results suggest that dolphin schools only occasionally react to 

 the approach of a survey vessel prior to their detection by shipboard observers and that the use of a 

 monotonically decreasing detection function is adequate to minimize bias. Aerial and shipboard 

 estimates of school size and species composition for six schools compared favorably. 



The Marine Mammal Protection Act of 1972 pro- 

 hibits the harvest of marine mammals and 

 specifies that the Federal Government may issue 

 permits for their take only under special cir- 

 cumstances. One such circumstance involves the 

 incidental kill of dolphins associated with the yel- 

 lowfin tuna fishery in the eastern tropical Pacific 

 Ocean. Before issuing the permits, the govern- 

 ment must first certify the viability of the affected 

 dolphin populations. To meet this requirement, 

 scientists at the Southwest Fisheries Center define 

 stocks and monitor their population demography, 

 reproductive output, and abundance. 



The vital statistics are derived primarily from 

 specimens obtained from the tuna fishery. How- 

 ever, to estimate abundance, surveys are con- 

 ducted using ships and aircraft independently of 

 the fishery. The surveys, using line-transect 

 methods (Burnham et al. 1980), have yielded esti- 

 mates of the density of dolphins in the eastern 

 tropical Pacific Ocean (Holt and Powers 1982). A 

 critical assumption in the application of the 

 method is that the animals do not move, in reac- 

 tion to the observer, prior to their detection. In 

 practice, a detection function, which is relatively 

 insensitive to nonrandom movement, is used to 

 describe the probability of observing a school of 

 dolphins given its position relative to the ob- 

 server's transect. A field experiment. was designed 

 with the following objective: 



^Southwest Fisheries Center La Jolla Laboratory, National 

 Marine Fisheries Service, NOAA, EO. Box 271, La Jolla, CA 

 92038. 



Msinuscript accepted April 1984. 



FISHERY BULLETIN: VOL. 83. NO. 2, 1985. 



1) Test the assumption that the animals do not 

 alter their movement in reaction to the ap- 

 proach of a survey vessel prior to shipboard 

 detection. 



During a survey the unit of observation is a 

 school of dolphins. In addition, species composition 

 and the number of individual animals in a school 

 (school size) are estimated. Surveys routinely col- 

 lect information to determine the precision of 

 these estimates by recording independent obser- 

 vations of several observers; however, determining 

 their accuracy is more difficult and attempted less 

 often (Holt and Powers 1982). Six schools Were 

 closely approached and observed from both an air- 

 craft and a ship with the following objective: 



2) Compare shipboard and aerial estimates of 

 school size and species composition. 



Although not an absolute determination of accu- 

 racy, the comparison yielded estimates from two 

 very different viewpoints (high-altitude plan view 

 versus low-altitude profile view). 



A similar experiment was conducted using the 

 NOAA Ship Surveyor and a ship-supported 

 helicopter in 1977 (Au and Perryman 1982). They 

 observed the reaction of eight dolphin schools to 

 the approach of a ship; all eight schools swam 

 away from the projected trackline of the ship. Au 

 and Perryman also suggested that, in some cases, 

 avoidance began beyond the visual range of ship- 

 board observers. The present study was intended 

 to collect additional data under a wider variety of 

 conditions. 



187 



