BARLOW ET AL.: AERIAL SURVEYS OF HARBOR PORPOISE 



ble to compare 1985 transects at 1.85 km with 1984 

 transects at 0.61 km in geographic regions 1 and 

 3 (Table 6). Sample sizes are very small, but den- 

 sities are roughly comparable (0.070 vs. 0.057, 

 respectively for 1.85 and 0.61 km from shore). Sam- 

 ple sizes in 1985 can be increased if Beaufort 2 is 

 considered instead of Beaufort & 1 (still with <25% 

 cloud cover). Based on geographic regions 1 through 

 5, porpoise densities are virtually identical at 1.85 

 and 0.61 km from shore (0.188 vs. 0.192, respective- 

 ly, Table 7). 



In a preliminary model of harbor porpoise depth 

 distribution. Barlow (1988) hypothesized that har- 

 bor porpoise density is constant from the shore to 

 the 75 m isobath. Ship data did not, however, in- 

 clude any transects inshore of 18.3 m depth. In this 

 shallow area, the model was not based on any data. 

 Although sparse, data from aerial surveys show that 

 densities are similar in areas that were surveyed by 

 ship (1.85 km from shore) and in areas that were 

 too shallow to be surveyed by ship (0.61 km from 

 shore). These data are consistent with the model pro- 

 posed by Barlow. 



Comparison of Ship 

 and Aerial Density Estimates 



Estimates of harbor porpoise density from 1984 

 to 1985 ship surveys (Barlow 1988) can be compared 

 with adjusted and unadjusted estimates from the 

 aerial surveys (Table 8). The overall density for all 

 regions is higher for the ship surveys than for the 

 unadjusted aerial. Adjusted estimates from the 

 aerial surveys are very close to the overall estimates 

 from ship surveys. 



Previous comparisons have been made of sighting 

 efficiency from aerial and surface vessel platforms 

 (Kraus et al. 1983). They found that observers on 

 boats saw 52% of the harbor porpoise groups seen 

 by shore-based observers, whereas aerial observers 

 saw only 14%. Based on this, density from aerial 

 surveys might be expected to be only 27% of that 

 from ship surveys. In the present study, unadjusted 

 density based on aerial surveys is 32% of the den- 

 sity from ship surveys. The two studies are not 

 directly comparable, however. Weather conditions 

 are not reported by Kraus et al. (1983) and may have 

 included less than optimal sighting conditions. It 

 should be noted that the fraction of harbor porpoise 

 groups seen by aerial observers in their study (14%) 

 is even lower than the fraction of harbor porpoise 

 we assumed would be in surfacing series and hence 

 near the surface (23.9%). It is possible that 

 behavioral differences between harbor porpoise 



from the two coasts (such as travelling or behavior 

 mode "A" noted by Watson and Gaskin (1983)) could 

 account for some of the differences noted above. 

 Estimates based on ship surveys cannot, of course, 

 be considered the true density of harbor porpoise. 

 The overall estimate based on ship surveys is rela- 

 tively imprecise (C.V. = 49%) and may be biased 

 (Barlow 1988). The ship survey estimate is, however, 

 superior to current estimates from aircraft for sev- 

 eral reasons. Line transect methods were used on 

 the ship surveys, and the principal assumption of 

 this method (that 100% of the animals in the imme- 

 diate vicinity of the trackline are seen) is more easily 

 met than the comparable strip transect assumption 

 (that 100% of the animals within a strip are seen). 

 Acceptable sighting conditions for ship surveys in- 

 cluded Beaufort & 1 , and 2 and were not restricted 

 by cloud cover (Barlow 1988). This allowed more 

 complete geographic coverage than did aerial 

 surveys. Also, the ship travelled much slower than 

 the aircraft (10 knots vs. 80-90 knots), thus the prob- 

 ability of missing a diving individual was much less. 

 Barlow (1988) calculated that diving animals located 

 near the trackline would be missed by observers on 

 ships only if dive times exceeded 2 minutes. Final- 

 ly, estimates of the correction factor to account for 

 submerged animals is relatively imprecise. Addi- 

 tional observations on ventilation patterns may 

 allow further refinements in density estimates based 

 on aerial surveys. 



RECOMMENDATIONS 



The design of future surveys for harbor porpoise 

 could be improved based on the results obtained 

 from our aerial surveys. We found that sighting con- 

 ditions deteriorated rapidly with both increasing 

 cloud cover and rougher sea states. To the extent 

 that is possible, aerial surveys for harbor porpoise 

 should only occur on clear days with little wind. Ob- 

 servations made by the data recorders indicate that 

 some harbor porpoise will be missed even in good 

 sighting conditions. If strip transects are used, ex- 

 periments with two independent teams of observers 

 searching at the same time could be used to quan- 

 tify the fraction of animals that are missed by using 

 just one team. Given that fewer harbor porpoise 

 were seen between 400 and 500 m of the track line, 

 we also suggest that, when surveying at 213 m 

 altitude, the strip widths should be decreased to only 

 include the area between 100 and 400 m. 



We believe, however, that the problem of miss- 

 ing harbor porpoise could be reduced if line transect 

 methods were used in place of strip transects. Line 



443 



