Burn and Doroff: Decline of Enhydra lutris along the Alaska Peninsula 



273 



again assuming no significant difference 

 in the proportion of sea otters detected 

 between surveys, we computed the propor- 

 tional change in density between survey 

 periods UD,,-D n )/D n ) of sea otter den- 

 sity at each island within the study area 

 between 2001 and 1986 (Brueggeman et 

 al. 1 ) and each Alaska Peninsula coastline 

 segment between 2001 and 1989 (DeGange 

 et al. 2 ). 



Results 



Offshore surveys 



In 1986, Brueggeman et al. 1 flew four sur- 

 veys and an average of 3676 km of transect 

 effort per survey. The majority (599? ) of the 

 1986 survey effort was conducted in Beau- 

 fort sea state 2 (wind less than 7.4-11.1 

 km/h, no whitecaps) or less; and 95% of the 

 survey effort was conducted with visibility 

 categorized as good or better. In May 2000 

 and April 2001, we flew 6334 km of tran- 

 sects and 56% of our effort was conducted 

 in Beaufort sea state 2 or less; 97% of our 

 effort was conducted in visibility catego- 

 rized as good or better. 



In 1986, sea otter detection probability 

 was not uniform between sighting zones 

 (X 2 = 1796, df=2, P<0.0001) and substan- 

 tially more sea otters were observed than 

 expected in the 0.0-0.23 km distance zone 

 (Fig. 2A). As a result, Brueggeman et al. 

 (1988) used only this zone in their cal- 

 culation of sea otter abundance. In our 

 2000-01 surveys, sea otter detection prob- 

 ability was also not uniform <x 2 =217, df=5, 

 P<0.0001i. The observed frequency of sea 

 otter sightings exceeded the expected val- 

 ue in our second (0.115-0.230 km) and 

 third zones (0.230-0.345 km) but in the 

 first zone (0.0-0.115 km) we recorded on- 

 ly half as many sea otter sightings as in 

 the second zone (Fig. 2B). Therefore, only 

 sightings from the second and third zones 

 were used in our calculation of sea otter abundance. 

 As a result, the overall width of the survey strip was 

 the same for the 1986 and the 2000-01 surveys (0.46 

 km), but our strip was offset by 0.115 km from the 

 trackline. The proportion of all sea otter sightings was 

 similar between the usable zones in 1986 (62.8%) and 

 2000-01 (62.4%). 



Sea otter encounter rate (otter groups/km) decreased 

 as wave height increased and visibility conditions be- 

 came worse in both the 1986 and 2000-01 surveys. As 

 noted by Kenyon (1969), wave height has a profound 

 influence on the ability of observers to detect sea ot- 

 ters. Prior to calculating abundance estimates for both 



000 



0.23 



46 



093 



0.00 012 0.23 035 46 0.58 



Distance from trackline (km) 



0.93 



Figure 2 



Distribution of sea otter (Enhydra lutris) sightings in offshore areas 

 grouped according to perpendicular distance from the survey track- 

 line. (A) 1986 data from Brueggeman et al. 1 (Bl 2000-01 data from 

 this study. Values above bars represent total number of sightings in 

 each zone. 



1986 and 2001, we subset both data sets to include 

 only those transects where Beaufort sea state was s2 

 and visibility was categorized as good or excellent for 

 counting sea otters. This procedure reduced the 1986 

 usable survey effort by 42%, and the 2000-01 survey 

 efforts by 44%. 



Sea otter encounter rates did not differ significant- 

 ly between rest and nonrest periods in the 1986 data 

 (£=1.63, df=79, P<0.1064). Likewise, there was no differ- 

 ence in encounter rates for rest and nonrest periods in 

 2000-01 (*=-0.79, df=71.6, P<0.4327). As a result, we 

 did not exclude any survey effort and sea otter sightings 

 based on time of day. 



