BARLOW: SHIP SURVEYS OF HARBOR PORPOISE 



be more accustomed to vessel traffic there than 

 along the majority of the coast. It is likely that some 

 harbor porpoise are missed because they do not sur- 

 face near the vessel; however, it is not possible to 

 quantify this source of bias without additional study. 



Trackline animals may be missed even if they do 

 not avoid the ship and do surface within visual range 

 of the observers if their surfacing is not detected. 

 In another study comparing ship surveys to aerial 

 and shore surveys, Kraus et al. (1983) found that 

 observers on ships saw only about 50% of the har- 

 bor porpoise in an area. In that study, however, ship 

 observers stood only 2.5 m above the sea surface 

 (versus 10 m in this study), and the estimate of 50% 

 was based on all groups, not just on trackline 

 animals. Based on the experiment using monitor 

 observers in the present study, an estimated 22% 

 of harbor porpoise that surface on the trackline are 

 missed by the usual team of 5 observers. If this is 

 underestimated by some percentage, population size 

 would be underestimated by the same percentage. 



The third critical assumption is that group size is 

 estimated without error. In the case of harbor por- 

 poise, group size is small and estimates are typical- 

 ly based on actual counts. For tropical dolphins, 

 which school in groups of several hundreds, the 

 problem of group size estimation is more acute (Holt 

 and Powers 1982; Hammond and Laake 1983). Only 

 in two instances did harbor porpoise group size 

 exceed 20: in Monterey Bay and near Point Arena, 

 both in California. Excluding these two sightings, 

 mean group sizes are 2.05, 2.33, 2.03, and 1.59 for 

 surveys 1, 2, 3, and 4 (respectively); including the 

 two sightings, means are 2.30 and 2.26 for surveys 

 1 and 3. These values are comparable to other esti- 

 mates of mean group size for coastal populations of 

 harbor porpoise: 2.2 based on aerial surveys in 

 California (Dohl et al. fn. 4), 2.6 based on ship 

 surveys in the Gulf of the Farallons (Szczepaniak 

 and Webber fn. 12), 2.3 based on shore surveys in 

 northern Oregon (see footnote 5), and 2.75-3.23 

 based on aerial surveys along California, Oregon, 

 and Washington (Barlow et al. 1988). The consis- 

 tency of all these estimates from different platforms 

 indicates that group size estimation from ships is 

 not likely to be a major source of bias in abundance 

 estimation. 



Variance Estimation 



Although the estimates of standard error for 

 abundance and density are very high, these may still 

 be underestimates because the choice of a trunca- 

 tion criterion was based on minimizing variance and 



because all possible sources of sampling errors were 

 not considered. The model upon which relative abun- 

 dance in the various depth strata was based is too 

 crude to allow reasonable estimates of its variabil- 

 ity. Estimates based on alternate models of depth 

 distribution indicate that abundance estimation is 

 relatively sensitive to the choice of models. Addi- 

 tional field work may help refine this model and 

 allow estimation of variance for the parameters 4 

 in Equation (4). 



ACKNOWLEDGMENTS 



Surveys of this magnitude could not be executed 

 without the help of many people. I thank the survey 

 crews for their many hours of labor: S. Beavers, 

 P. Boveng, S. Bragg, S. Chivers, S. Diamond, 

 V. DoUarhide, J. Flanders, B. Goetz, S. Hawes, 

 S. Heimhch-Boran, A. Hohn, S. Kruse, S. Mizroch, 



F. Mann, M. Newcomer, R. Rasmussen, A. Read, 



A. Robles, M. Scott, K. Sechiguchi, S. Sexton, 



G. Silber, I. Szczepaniak, B. Taylor, B. Troutman, 

 M. Webber, J. Wexler, and K. Wynne. The heli- 

 copter crew included R. Holt, C. Oliver, and 



B. Taylor. The initial survey was planned in col- 

 laboration with D. DeMaster and T. Jackson. I thank 

 S. Buckland, R. Holt, and S. Sexton for aid in 

 analysis and interpretation of line transect data. 

 B. Taylor provided unpublished data from shore 

 observations of harbor porpoise in northern Oregon. 

 Geographic areas were computed by K. Forney. 

 R. Allen prepared the figures presented here. This 

 manuscript was improved by the critical reviews of 

 P. Boveng, R. Brownell, D. Chapman, D. DeMaster, 

 S. Diamond, D. Goodman, D. Hanan, A. Hohn, 

 R. Holt, J. Lecky, S. Sexton, G. Smith, and 

 B. Taylor. 



LITERATURE CITED 



Adams, L. 



1951. Confidence limits for the Petersen or Lincoln index 

 used in animal population studies. J. Wildl. Manage. 15: 

 13-19. 

 Barlow, J., C. Oliver, T. D. Jackson, and B. L. Taylor. 

 1988. Harbor porpoise, Phocoena phocoena, abundance esti- 

 mation for California, Oregon, and Washington: II. Aerial 

 surveys. Fish. Bull., U.S. 86:433-444. 

 Buckland, S. T. 



1985. Perpendicular distance models for line transect sam- 

 pling. Biometrics 41:177-195. 



BURNHAM, K. P., AND D. R. ANDERSON. 



1976. Mathematical models for nonparametric inferences 

 from line transect data. Biometrics 32:325-336. 

 BuRNHAM, K. P., D. R. Anderson, and J. L. Laake. 



1980. Estimation of density from line transect sampling of 

 biological populations. Wildl. Monogr. No. 72, 202 p. 



431 



