FISHERY BULLETIN: VOL. 82, NO. 3 



FIGURE 1. — Study area showing the two 

 major haul out sites, Kent Island (KI) 

 and Pickleweed Island (PWI), Calif., 

 (small solid circles), and the location of 

 the two cameras (large solid circles). 



"814 XL Super 8" with a 7.5 to 60 mm zoom lens 

 was focused on KI from a private residence on the 

 sand spit about 400 m away. A photocell activated 

 the camera at day break and deactivated it at 

 sundown; an intervalometer exposed one frame of 

 film every minute. Another camera, a Eumig "880 

 Super 8" with a 7 to 56 mm zoom lens, positioned in 

 a weatherproof box on private property along 

 State Route 1, was focused on PWI, about 300 m 

 away. The built-in intervalometer also triggered 

 1 frame/min. An electrical timer activated the 

 camera during daylight hours. The film used 

 (Kodachrome ASA 40) contained 3,000 frames/roll 

 and lasted a week except during summer, when 

 film was changed every 4 d due to increased day 

 length. 



We used a Kodak "Moviedeck model 475" projec- 

 tor to analyze film. We noted time, tide level, and 

 the number of seals once every hour of photo- 

 graphic time. Any major change in seal numbers 

 within the 1 h interval was also noted, as well as 

 any disturbance. A stake marked with 0.3 m in- 

 crements was placed near the major channel by 

 KI and in line of the Canon's viewfinder. This 

 provided a photographic record of tidal change. 

 Actual time for the first frame each day was ex- 

 trapolated from tables for sunrise and sunset; on 

 several mornings and evenings we compared these 

 times with the actual time that camera operation 



began or ceased. Extrapolated times were accu- 

 rate to within 1.0 h. 



Twelve (once per month) day-long watches and 

 211 (124 breeding, 47 summer, and 40 winter) inci- 

 dental sightings validated camera counts, de- 

 tected sources of disturbance outside the camera 

 field, estimated disturbance distance from the 

 herd, identified additional haul out sites, and ac- 

 curately counted pups, which were difficult to de- 

 tect on film. 



When analyzing data, the year was divided into 

 three seasons: winter (November through Feb- 

 ruary), breeding (March through June), and 

 summer (July through October); seasonal aver- 

 ages are expressed with ±1 standard deviation. 

 Correlation coefficients were calculated to com- 

 pare camera and field counts as a test for camera 

 reliability, and to examine the relationship of seal 

 numbers to tide level within each hour (Snedecor 

 and Cochran 1967). For daily use per season 

 graphs of hourly means were compared; the "runs 

 up and down test" (Bradley 1968) was used to test 

 the sequence of hourly means for randomness. 



All human activities ( including dogs off leashes) 

 in the area were divided into two types: actual 

 disturbance and zero-seal disturbance. Actual dis- 

 turbances, or any activity occurring when at least 

 one seal was present, were further subdivided into 

 type I, where at least one seal left the area, and 



494 



