Barlow: Abundance of cetaceans in California waters, ship surveys 



I I 



Probability of detecting trackline groups 



The probability of detecting a trackline group of ani- 

 mals,^), varied between 0.74 and 1.0 (Table 2 ). The 

 data clearly indicated that small groups are much 

 more likely to be missed than are large groups. This 

 is intuitively obvious and justifies stratifying by 

 group size when estimating ^(0) values. The fraction 

 of trackline harbor porpoise seen in calm seas has 

 been estimated previously to be 0.78 (with five ob- 

 servers on a similar platform in California, Barlow 

 [1988] and 0.70 (with six observers in the Gulf of 

 Maine, Palka [1993]). The higher value ofg(0) esti- 

 mated here for "cryptic species" with only three ob- 

 servers (0.81) may be due to the inclusion of Dall's 

 porpoise which may be easier to see or may simply 

 be an artifact of small sample size. 



These estimates of the fraction of animals seen 

 include only animals that were available to be seen. 

 Availability bias is likely to be large for species such 

 as beaked whales, which have extremely long dive 

 times, and harbor porpoise and Dall's porpoise, which 

 have shorter dive times but seldom are seen more 

 than 0.5 km from the ship and may therefore remain 

 submerged during the entire time they are within 

 visual range. Correcting for availability bias is more 



difficult than for perception bias. Attempts 

 that have been made so far have involved 

 detailed modeling of the surfacing behavior 

 of the animal and the searching behavior of 

 the researchers (Doi, 1971, 1974; Barlow et 

 al., 1988; Stern, 1992; Kasamatsu and 

 Joyce 7 ). In addition, there are still problems 

 with estimating perception bias because the 

 methods used here assume that all animals 

 are equally available to be seen if they sur- 

 face. Heterogeneity in sightability (e.g. ani- 

 mals that splash vs. animals that do not) gen- 

 erally will result in an underestimate of the 

 fraction missed. Additional work is needed 

 to obtain complete estimates of the fraction of 

 trackline animals seen for all species. 



Previous studies of Dall's porpoise have 

 shown that attraction to the vessel is a 

 greater problem for estimating the abun- 

 dance of this species than are missing 

 trackline animals (Turnock and Boucher 8 ). 

 Turnock and Quinn (1991) estimated a cor- 

 rection factor of 0.2378 (CV=0.3391) to ad- 

 just Dall's porpoise abundance estimates for 

 ship surveys (effectively then, g =4.2). That 

 study was based, however, on a design that 

 used only one observer who searched with 

 7x binoculars and unaided eyes. In the 

 present study, very few Dall's porpoise ap- 

 peared to be attracted to the vessel; of those 

 sighted in calm conditions and used for abundance 

 estimation, only 10% (9 of 88) of the Dall's porpoise 

 groups approached the vessel to "ride the bow wave," 

 and 89% (78 of 88) were exhibiting a "slow roll" sur- 

 facing behavior at the time they were first sighted. 

 Because attraction to the vessel was less than in 

 other studies and because most Dall's porpoise were 

 sighted before showing any apparent reaction to the 

 vessel (perhaps because 25x binoculars were used), 

 the magnitude of bias is probably less than that es- 

 timated by Turnock and Quinn ( 1991). 



Statistical precision 



An attempt was made to account for most sources of 

 sampling error in the bootstrap estimates of confi- 

 dence intervals and coefficients of variation. How- 

 ever, several sources of variation could not be easily 

 included. The process of selecting a stratification 



7 Kasamatsu, F., and G. G. Joyce. 1991. Abundance of beaked 

 whales in the Antarctic. Int. Whaling Comm. working paper 

 SC/43/012. 



8 Turnock, B. J., and G. C. Boucher. 1990. Population abundance 

 of Dall's porpoise, Phocoenoides dalli, in the western North 

 Pacific Ocean. Int. Whaling Comm. working paper SC/42/SM10. 



