BARLOW ET AL.: AERIAL SURVEYS OF HARBOR PORPOISE 



shore (Table 2), and neither was statistically signif- 

 icant. For 1985, the opposite was seen; the density 

 on the inshore side was greater than on the offshore 

 side and this difference was less than the difference 

 between observers (Table 2), but again, neither dif- 

 ference was statistically significant. In 1984, both 

 observers had previous experience doing aerial 

 surveys, but in 1985, the observer with the lower 

 density estimates had no previous experience in 

 cetacean surveys. Experience may be a factor in 

 density estimates from strip transects, but since the 

 inexperienced observer was always on the same side 

 of the plane, it was not possible to test this with a 

 factorial design. 



Area and Sea State 



Analysis of research vessel data (Barlow 1988) in- 

 dicated two geographic regions in California with 

 low porpoise density (regions 1 and 3 in Figure 2). 

 We tested whether density observed from aircraft 

 are also lower in these areas. Because observation 

 conditions may have differed in the two areas, we 

 included sea state as a second factor in a two-way 

 ANOVA. Mean values are presented in Table 3. For 

 1984, three categories of sea state were used: 

 Beaufort & 1, Beaufort 2, and Beaufort 3 and 

 greater. The effect due to area was significant (P 

 < 0.001), with the area that showed low density in 

 the ship surveys also showing lower density in the 

 aerial survey. For 1985, only the first two categories 

 of sea state were used due to insufficient data at 

 Beaufort 3 and greater. Again the effect of area was 

 significant (P < 0.025), and the same trends were 

 seen. In neither case were the effects of sea state 



Table 3. — Uncorrected harbor porpoise densities (km' ) for 

 the two-way comparison of area and sea state. Low-density 

 areas refer to two regions in California that were found to have 

 much lower than average density in previous ship surveys (see 

 text). High-density areas include all other regions. Only inside 

 swaths were included. Numbers in parentheses refer to area 

 (km^) surveyed under the given condition. Densities in 

 brackets were excluded from two-way comparisons due to the 

 small area covered in one cell. 



or the interaction effects significant. To eliminate 

 area effects from confounding statistical results, 

 only data for the larger, high-density area were in- 

 cluded in subsequent tests. The low-density area was 

 included in later estimates of overall harbor porpoise 

 density. 



Sea State and Cloud Cover 



Both sea state and cloud cover can affect sighting 

 conditions. Because both are affected by local 

 weather, the effects of these are likely to be con- 

 founded. These two factors were therefore tested 

 simultaneously in a two-way ANOVA. We con- 

 sidered only the inside swath and excluded the 

 two low-density regions. We used the same sea 

 state categories as above. The sky was categorized 

 as clear if cloud cover was <25% and cloudy if 

 >25%. Mean porpoise densities for each category 

 are given in Table 4. It was necessary to exclude 

 the Beaufort & 1 category for 1985 because only 

 52 km were surveyed in cloudy conditions for these 

 sea states. The effect due to cloud cover was signifi- 

 cant for 1984 (P < 0.025) and 1985 (P < 0.05). The 

 effect of sea state and the interaction effect of 

 sea state and cloud cover were not significant for 

 either survey (P > 0.10). Transect segments with 

 >25% cloud cover were excluded from subsequent 

 analyses. 



Table 4. — Uncorrected harbor porpoise densities (km"^) 

 for the two-way comparison of sea state and cloud cover. 

 Clear refers to <25% cloud cover, and cloudy refers to 

 >25%. Data include only inside swaths in high-density 

 areas. Numbers in parentheses refer to area (km^) sur- 

 veyed under the given condition. Densities in brackets were 

 excluded from two-way comparisons due to the small area 

 covered in one cell. 



Sea State 



The effect of sea state was tested alone using only 

 the transect segments which occurred under clear 

 skies (<25% cloud cover) and within the high-density 

 areas. For 1984, only the inside swath was included. 



439 



