272 



Fishery Bulletin 103(2) 



_Ukolnoi wosnesensk. (^opoP 



Korovm 



^ Andronica 



Spectacle 

 UiLBig Koniuji 



-&endfi ? - l " l fi KQn ' " JI 

 TiJrner *- / V. o 



£? Simeonof 



Bird*** £j Chernabura 



-Sanak/ 



A 



25 50 100 

 I I I I I I I I I 



Kilometers 



Figure 1 



Sea otter {Enhydra lutris) survey areas along the Alaska Peninsula. (Al 

 Offshore areas. (Bl South Alaska Peninsula Islands. (C) Alaska Peninsula 

 coastline. Surveyed areas in (B) and (C) include a 0.46-km zone adjacent 

 to shore. 



for estimating abundance. We calculated an 

 encounter rate as the number of sea otter 

 groups per km of survey effort and used 

 this rate to examine the effects of time of 

 day and environmental conditions (wave 

 height, and visibility) on detectability of 

 sea otters. 



At the time of the surveys in 1986, re- 

 searchers had documented a core resting 

 period for sea otters which occurs about 

 mid-day (Garshelis and Garshelis, 1984; 

 Estes, 1977). As a result, Brueggeman et 

 al. 1 subset the 1986 data using only effort 

 and observations recorded between 0830 

 and 1430 hours local sun time for their 

 abundance estimates. Recent studies in- 

 dicate activity patterns for sea otters are 

 strongly linked to sex, age, weather condi- 

 tion, season, and time of day (Gelatt et al., 

 2002). We tested the assumption that sea 

 otters were more visible during the core 

 resting period using a /-test of the encoun- 

 ter rate for each transect during presumed 

 rest and nonrest periods for the 1986 and 

 the 2000-01 data. 



We measured the area of the NAP and 

 SAP study areas using a geographic infor- 

 mation system (Arc/Info). Our measure- 

 ments differed from those of Brueggeman 

 et al., 1 presumably because the original 

 researchers had not used an equal-area 

 map projection in their calculations. Like 

 Brueggeman et al., 1 we estimated abun- 

 dance of sea otters in the Alaska Peninsula 

 offshore areas using the modified ratio of 

 means estimator (method I) of Estes and 

 Gilbert (1978). Noting computational er- 

 rors in the original analysis, we recalcu- 

 lated abundance estimates from the origi- 

 nal 1986 data of Brueggeman et al. 1 The 

 proportion of sea otters within the survey 

 swath that went undetected by observers 

 was not estimated in either our survey or 

 the surveys of Brueggeman et al. 1 ; there- 

 fore all abundance estimates were biased 

 low to an unknown degree. We computed 

 the proportional change in abundance be- 

 tween survey periods ((N t2 —N n )IN n ) as a 

 range, using the minimum and maximum 

 estimates from 1986 as a baseline and 

 assuming no significant difference in the 

 proportion of sea otters detected between 

 surveys. 



Coastline survey analyses 



We calculated the area surveyed as the 

 product of the coastline length and the 

 survey strip width and calculated the den- 

 sity of sea otters per km 2 surveyed. Once 



