180 



Fishery Bulletin 94(1). 1996 



cision of each method. Three independent trials were 

 made during two days at the rookery located at 

 Piedras Blancas, California (a 1-1/4 hour interval 

 separated trial 2 and trial 3 on the second day). Three 

 distinct groups of elephant seals were photographed 

 from the air while they were counted on the ground by 

 three persons (each person made two counts of each 

 group). The seals in the photographs were also counted 

 by three persons (each person made two counts of each 

 group). Total counts of all seals were made to elimi- 

 nate problems associated with classifying a seal to an 

 age and sex class category. The ground count was made 

 from an unobstructed view atop cliffs or sand dunes 

 that were approximately 10—15 meters above the seals. 



The Wilcoxon paired-sample test was used to com- 

 pare CV's obtained from ground counts with photo 

 counts at Piedras Blancas to determine whether the 

 precision of the two methods were equal. The counts 

 were then log 10 -transformed to meet requirements 

 of homoscedasticity. Analysis of variance ( ANOVA) 

 was used to determine whether ground counts and 

 photo counts were equal; the factors tested were 

 method (ground vs. photo), person, group (of seals), 

 trial, and their interactions. An ANOVA was also 

 performed on ground counts and another on photo 

 counts; the factors tested for each test were person, 

 group (of seals), trial, and their interactions. 



A comparison of elephant seal counts made at 

 Santa Barbara Island was conducted in 1995 to ex- 

 amine aerial photographic counts, small-vessel and 

 ground counts, and estimates of births for this is- 

 land. Elephant seals were counted from two aerial 

 photographic surveys (Table 1) and from a small-ves- 

 sel and ground survey conducted two days after the 

 second aerial survey. 



Estimate of births 



The number of northern elephant seal births was 

 estimated by summing the counts of pups that were 

 alive or of unknown status and the counts of decom- 

 posed pup carcasses. When more than one survey 

 was made each year, the greater count of each cat- 

 egory (from either survey) was used. Estimates of 

 births for San Miguel, San Nicolas, and Santa Rosa 

 Islands made by Stewart ( 1992 ) were compared with 

 those from this study Long-term changes in birth 

 rates were examined for San Miguel Island for 1985 

 through 1991 (data for the other islands were insuf- 

 ficient for this comparison) with log-linear regres- 

 sion analysis (Eberhardt and Simmons, 1992). 



Estimates of births at Santa Barbara Island were 

 made from the number of pups counted. The number 

 of adult females counted during the peak breeding-sea- 

 son survey was also compared with the number of pups 



counted during the three surveys because Stewart 

 (1989) used the count of adult females during peak- 

 breeding season to estimate births for this rookery. 



Results 



Total or near total survey coverage was achieved for 

 most surveys (90%, Table 1). No counts are avail- 

 able for San Miguel Island in 1987 or for the peak 

 breeding-season survey at San Miguel Island in 1992 

 owing to missed photographic coverage. 



Counts of northern elephant seals at Piedras Blancas, 

 California, from vertical 126-mm-format photographs 

 were more precise than counts made by people at 

 ground-level (Table 2). The CV's of photo counts (be- 

 tween 0.002 and 0.009) and ground counts (between 

 0.054 and 0.231 ) were significantly different (P<0.008). 



The counts obtained from 126-mm-format photo- 

 graphs at Piedras Blancas were significantly differ- 

 ent from counts obtained by people on the ground 

 (P<0. 001, Table 3). There was a significant difference 

 between persons for the counts obtained on the 

 ground (P<0.001) but not between persons counting 

 the seals from aerial photographs (P=0.065). Each 

 method detected a significant difference between the 

 size of the three groups of seals and between the three 

 trials (P<0.002). 



The lowest CV obtained by counting northern el- 

 ephant seals at Piedras Blancas was 0.054 and the 

 highest CV was 0.231 (Table 2). By our definition, 

 excellent precision would be a CV of <0.027, good 

 precision would be between 0.027 and 0.054, accept- 

 able precision would be between >0.054 and 0.231, 

 and unacceptable precision would be >0.231. Thus, 

 the counts of elephant seals from aerial photographs 

 taken at Piedras Blancas had excellent precision. The 

 replicate counts from the photographic surveys of San 

 Miguel Island in 1988 and 1989 also had excellent pre- 

 cision for counts of pups that were alive or of unknown 

 status, total number of pups, adult females, subadult 

 and adult males, and for total counts of all northern 

 elephant seals (Table 4). The precision for counts of 

 decomposed carcasses of pups was good or acceptable. 



Estimates of northern elephant seal births from 

 large-format aerial photographic surveys differed 

 from estimates made from ground surveys at San 

 Miguel and San Nicolas Islands by -8.4% to +11.2% 

 and from small-format aerial photographic surveys 

 at Santa Rosa Island by -14.8% to +17.8% (Table 5). 

 Estimates from aerial photographic surveys were 

 higher than estimates from ground surveys for six of 

 ten comparisons, but the differences were not statis- 

 tically significant (P>0.05). At Santa Barbara Island, 

 the number of pups counted from large-format aerial 



