ALLEN ET AL.: EFFECT OF DISTURBANCE ON HARBOR SEALS 



10 



tr 



z> 

 o 



I 



Or 



UJ 

 Q. 



< 



tr 



LU 4 

 CO 



Z 



Z 3 



< 

 UJ 



2 

 2 



PICKLEWEEO ISLAND 



• = WINTER 

 ' = BREEDING 

  = SUMMER 



10 II 12 13 14 15 16 17 18 19 20 

 TIME OF DAY (PST) 



FIGURE 3. — Graph of time of day and the mean number of seals 

 hauled out per time period for winter, breeding, and summer 

 seasons at Pickleweed Island. 



0.7% helicopters. The camera did not record the 

 cause in 40% of the disturbances. During De- 

 cember, January, and February, commercial bait 

 harvesters accounted for 25.7% of the distur- 

 bances. Disturbances from aircraft were detected 

 only in field observations. The seals were dis- 

 turbed at least once on 71% of 356 d when the KI 

 camera was functioning, and during these days, 

 72.7% of 539 disturbances caused the seals to dis- 

 perse (Table 3). On 211 d during which the PWI 

 camera was functional, seals reacted to 57% of 

 236 disturbances (Table 3). The frequency of ac- 

 tual disturbances per day averaged highest during 

 the summer for both KI and PWI. 



Of the actual disturbances of known cause, most 

 occurred within 100 m of the KI site and resulted 

 more from nonpower boats than from any other 

 source (Table 4). The deletion of terms [13] and [14] 

 from the log-linear model in backward selection, 

 however, indicated that only the distance of a dis- 

 turbance significantly affected seal behavior (Ta- 

 ble 5). Seals did not react differentially to any 

 disturbance type and did not react more to distur- 

 bances during weekend/holidays than during 

 weekdays. The relative magnitude of the weights 

 associated with the distance/seal response in- 

 teraction term [12] denoted that seals responded to 



TABLE 2. — Correlation coefficients between tide level and the 

 number of seals hauled out at hourly intervals per season at Kent 

 Island (Kl)and at Pickleweed Island (PWI), from camera data; n 

 = number of censuses. Insufficient data were available for PWI 

 during the winter. A positive correlation indicates a positive 

 relationship between seal number and low tide, and a negative 

 correlation indicates a negative relationship between seal 

 number and high tide. 



TABLE 3. — The frequency of disturbances on Kent Island (KI) 

 and Pickleweed Island (PWI) by season; n is the number of days 

 the camera was functional, A is the number of actual and zero- 

 seal disturbances combined, A 1 is the number of actual distur- 

 bances type I, A In is the mean number of all disturbances per day, 

 and A \ln is the mean number of the actual disturbances per day. 



TABLE 4. — Data from the Kent Island camera used in log-linear 

 model analysis (Bishop et al. 1975). The numbers in the table are 

 the number of disturbances where seals were present for each 

 category. Y = response (at least one seal left site); N = no 

 response (no seals left site). 



Weekday 



Weekend/holiday 



Distance: 



100 m 



101- 

 200 m 



201- 

 300+ m 



:100 m 



101- 

 200 m 



201- 

 300+ m 



TABLE 5. — Backward selection of log-linear model using data in 

 Table 4; model variables are 1 (response), 2 (distance), 3 (day of 

 week), and 4 (disturbance type). An asterisk indicates a term 

 that is significantly different from zero. 



497 



