336 



Fishery Bulletin 97(2), 1999 



350-1 



Before Initial Harassment 



Alter Initial Harassment 



Deteelion 



Alarm 



Harassment 



Figure 2 



Mean distance (+SEl between harbor seals and powerboat for disturbances occurring 

 before and after the initial harassment during a haul-out period. An asterisk (*) indi- 

 cates a significant (P<0.05) difference between before and after initial harassment data 

 for that disturbance category. 



There was no apparent difference in the propor- 

 tion of harbor seal pups onshore before and after 

 harassments (x"=0.183, n=27, P>0.50). Pups did not 

 appear to be affected disproportionately to subadults 

 and adults. 



The percentage of vigilant harbor seals varied with 

 each comparison (due to random subsampling of data 

 to control for group size), hence there was no single 

 value for percentage of vigilant harbor seals at each 

 haul-out site. The percentage of vigilant harbor seals 

 was significantly greater at Puffin Island compared 

 with Clements Reef (P=0. 003) and Skipjack Island 

 (P=0.002; Table 3). This finding is likely due to the 

 greater percentage of female-and-pup pairs at Puf- 

 fin Island and, in addition to lower recovery, indi- 

 cated seals were more susceptible to disturbance. 



Powerboat speed did not significantly infiuence 

 distance of disturbance. Lack of statistical signifi- 

 cance was likely due to small sample sizes after sub- 

 dividing data by approach speed and initial versus 

 subsequent harassments. There was also no signifi- 

 cant difference in distance of .disturbance between 

 sites (data were not collected at Skipjack Island). 

 Given the large variability of the data and small ef- 

 fect size, greater than 450 observations would be re- 

 quired for a power of 0.80 to detect a significant dif- 



ference between sites. Data, therefore, were pooled 

 for further analysis. Twenty-five percent of harass- 

 ments occurred when vessels were < 100 m from seals, 

 50% occurred at 100-200 m, and 25% at 200-300 m. 

 After detection by seals, powerboats were able to 

 approach significantly closer (P=10. 51, P<0.001) be- 

 fore causing alarm or harassment of seals (Fig. 2). 

 Interestingly, we found that distances of initial dis- 

 turbance were significantly greater than subsequent 

 disturbances (Fig. 2). This indicated that seals re- 

 maining or returning to shore following the first ha- 

 rassment were less easily disturbed. 



Discussion 



Many islands in the San Juan Archipelago are state 

 parks or have resort harbors that attract numerous 

 boaters during the summer. Clements Reef is located 

 near (0.6 km) Sucia Island, which is the most heavily 

 visited state park in the northern islands. It was not 

 surprising, therefore, that most disturbances at 

 Clements Reef were caused by boaters. Relatively 

 few kayakers ventured out to Clements Reef, Puffin 

 Island, or Skipjack Island. Kayakers typically travel 

 along the shoreline and have been shown to cause 



