24 



Fishery Bulletin 93(1). 1995 



expected to be smallest for species which tend to oc- 

 cur in large groups, such as common dolphins, and 

 largest for species which spend relatively little time 

 at the surface, such as porpoise, beaked whales, and 

 sperm whales. 



Dive studies (Barlow et al., 1988) may provide 

 information on the magnitude of availability bias, 

 but each species requires a separate assessment of 

 the average proportion of time it spends at the sur- 

 face (and hence is 'available'), and adequate estimates 

 are not currently available for most species in Cali- 

 fornia waters. Rough estimates can be made for Dall's 

 porpoise and humpback whales based on prior stud- 

 ies. Dall's porpoise have similar sighting character- 

 istics to those of harbor porpoise (both have a small 

 body size and generally are found in small groups); 

 thus, assuming that dive patterns are similar and 

 applying the correction factor of 3.1 (CV=0.17) for 

 harbor porpoise, 6 one would obtain a corrected esti- 

 mate of approximately 26,200 Dall's porpoise. Based 

 on a very small sample, a correction factor of 2.7 has 

 been estimated for humpback whales. 7 This would 

 yield a corrected abundance estimate of 861 humpback 

 whales. Clearly, given the magnitude of these correc- 

 tion factors, availability bias can be substantial. 



Potential upward bias in line-transect analysis can 

 result if factors other than distance to the trackline 

 affect the probability of seeing a school. School size 

 has been shown to affect the probability of detection 

 (Drummer, 1985; Holt and Sexton, 1989), and this 

 can lead to an upward bias in the abundance esti- 

 mate (Quinn, 1985; Drummer and McDonald, 1987; 

 Buckland et al., 1993a). To counteract this effect, we 

 have stratified small cetacean sightings by group size 

 and estimated abundances separately for small and 

 large groups of the same species. This is an artificial 

 separation, but it reduces potential biases that are 

 due to large variation in group size within a single 

 species, such as common dolphins or Pacific white- 

 sided dolphins. Within each stratum, correlations of 

 perpendicular sighting distance with group size are 

 weak and not significant at oc=0.05 (r=0.195 for small 

 cetaceans in groups of 1—10 animals; r=0.169 for 

 small cetaceans in groups of greater than 10 animals; 

 and r=0.183 for whales in groups of all sizes). 



6 Calambokidis, J., J. R. Evenson, J. C. Cubbage, P. J. Gearin, 

 and S. D. Osmek. 1993. Development of a correction factor for 

 aerial surveys of harbor porpoise. Draft Final Contract Report 

 to the National Marine Mammal Laboratory, NMFS, NOAA, 

 7600 Sand Point Way NE, BIN C-15700, Seattle, WA 98115. 

 36 p. 



7 Calambokidis, J., G. H. Steiger, J. C. Cubbage, K. C. Balcomb, 

 and P. Bloedel. 1989. Biology of humpback whales in the Gulf 

 of the Farallones. Final report for Contract CX-8000-6-0003 to 

 Gulf of the Farallones National Marine Sanctuary, NOAA, Fort 

 Mason Center, Bldg. 201, San Francisco, CA 94123, 93 p. 



In summary, we have attempted to correct for per- 

 ception bias by estimating the fraction of animals 

 missed during these surveys and have minimized 

 potential upward bias with a poststratification by 

 school-size range. However, species-specific availabil- 

 ity bias cannot currently be estimated, and overall our 

 abundance estimates are likely to be biased downward. 



Precision 



Estimation of variance for line-transect abundance 

 calculations can be difficult. We have attempted to 

 include most of the sources of sampling error in the 

 bootstrap procedure, which reestimates n, s, and/10) 

 (in Eq. 1) for each replicate. Our analysis revealed 

 that the choice of segment length used for the boot- 

 strap did not affect the resulting estimates of preci- 

 sion within the range of appropriate segment lengths 

 for this study (5-100 km; longer segments would not 

 be appropriate because surveys extended only 100-150 

 km offshore). However, potential heterogeneity due to 

 the pooling of different species and group sizes for esti- 

 mation of /(0) andg(0) was not accounted for in preci- 

 sion estimates. Furthermore, we did not include the 

 variance ing(0) or in the estimation of group size for 

 each school encountered (however, the variance in the 

 estimated mean group size for the survey was included 

 in the bootstrap procedure). Thus, the coefficients of 

 variation for the abundance estimates (Table 3) are 

 likely to be underestimated and the confidence inter- 

 vals are likely to be too narrow. 



Considerations for future aerial surveys 



Two species of common dolphins, short-beaked and 

 long-beaked, are recognized in California waters 

 (Rosel, 1992; Dizon et al., 1994; Heyning and Perrin, 

 1994). Although clear differences in color pattern, 

 size, and beak length exist between these two forms, 

 it is not currently possible to differentiate them dur- 

 ing aerial surveys; therefore the abundance estimate 

 here is a combined estimate. Unless reliable means 

 of identifying the two species from the air are devel- 

 oped, aerial surveys will not be adequate for future 

 assessments requiring separate estimates of short- 

 beaked and long-beaked common dolphins. 



Similarly, it was difficult to distinguish between 

 the smaller species of beaked whales during our 

 aerial surveys. The estimates presented for the 

 beaked whales as a group are therefore a combined 

 estimate for Ziphius cavirostris and Mesoplodon spp. 

 All unidentified beaked whale sightings could be 

 narrowed down to these two genera. The only other 

 beaked whale species known to occur in this region, 

 Berardius bairdii, can be readily distinguished based 



