cent cover) of plant species in each community was deter- 
mined quantitatively by means of a two-stage systematic 
sampling design. Strip transects 1 m wide were located ran- 
domly in each community from a grid placed over the vege- 
tation map. Randomly chosen 1- x 1-m secondary sam- 
ple units were replicated as many times as the transect 
length dictated. 
Within each sample unit the vegetation was stratified 
according to relative height. Overstory was defined as the 
layer of plants that overshadowed all layers below it. The 
middle story was the level of plants that overshadowed the 
ground layer plants, but which was overshadowed by taller 
plants. Generally, only those plants overshadowed by all 
other layers were considered in the ground story. The per- 
cent coverage of each species was estimated visually in each 
story. All plant nomenclature followed Hulten (1968). 
Vegetation Around Goose Nests 
At each goose nest the percentage of coverage of each 
plant species in each of the height categories was recorded 
in two quadrats, 1 x 1 mand5 x 5m, with the nest as 
their center. Cover values in the 1- x 1-m nest plots were 
expanded proportionately to total 100 % to compensate for 
the opening created by the nest. This was necessary to com- 
pare data from the 1- x 1-m plots with data from the 5- x 
5-m plots around nests and with the 1- x 1-m plots in the 
communities. 
Physical Characteristics of Nest Sites 
The slope at each nest was measured with a device called 
a “distance-height measurer.” Physical surroundings were 
noted, aspects were measured with a compass, and eleya- 
tion was determined with a pocket altimeter. 
Population Size and Structure 
Banding 
Conventional banding drives (Cooch 1953, 1955) were 
not possible at Buldir because flightless geese seldom formed 
large flocks. Instead, birds were in isolated family groups, 
occasionally including yearlings, or in groups of two to four 
adults, presumably non-breeders. Geese were captured by 
hand or with long-handled nets, usually after foot chases. 
Birds were marked with standard U.S. Fish and Wildlife 
Service size 7B leg bands and plastic leg bands (34 mm high, 
14 mm inside diameter) with color and numerical codes, 
Population Estimation 
Each fall from mid-August until our departure, we re- 
corded the ratio of banded to unbanded geese and the fam- 
ily group sizes. Plastic leg marker codes were read to deter- 
mine the presence of known individuals. In 1976 we re- 
mained at Buldir later than in previous years, and a Questar 
field telescope (60 x to 130x) made it possible to read 
plastic leg band codes from a distance of 250 m. Breast and 
neck plumage (Hanson 1962; Higgins and Schoonover 1969) 
were used to identify fledglings (fledglings had particularly 
rufous breasts and small white neck rings). 
In 1977 a stratified random sampling design was em- 
ployed to estimate the number of breeding pairs of geese 
at Buldir. The goose nesting habitat was divided into four 
strata (south-facing sea slopes, north-facing sea slopes, in- 
land slopes, and uplands covered by the moss—willow 
tundra community). Habitats obviously not used for breed- 
ing, such as large slide areas, the sparsely vegetated high- 
est elevations, and Camp Valley were excluded from the 
area sampled, A total of 30 randomly selected 200- x 200-m 
sample plots were examined. The allocation of plots among 
strata was weighted to reflect the relative variability of 
goose nesting density. 
The plots were located by visual orientation from map 
reference and altimeter readings. Boundaries were deter- 
mined with a compass and a 50-m steel tape. Researchers 
walked abreast 5-15 m apart over each plot to locate all 
nests. To aid in future plot location, a 2-m-long aluminum 
marker, tipped with “day-glow” orange paint, was placed 
in the most conspicuous corner of the plot. 
In addition to surveying the sample plots, we censused 
two areas, a seaslope near Gull Slide and the Northwest 
Point peninsula, because their irregular shapes made it 
impossible to use normal sample plots. 
Results and Discussion 
Breeding Biology 
Arrival of Geese at Buldir 
Aleutian Canada geese began arriving at Buldir during 
the first week of May based on the following observations: 
(1) in 1974, no geese were seen on 30 April during our ini- 
tial circumnavigation of the island; (2) on 9 May, our next 
chance to survey Buldir, 44 geese were counted in an in- 
complete survey; (3) in 1975 and 1976, geese were present 
when we arrived on 17 and 19 May, respectively. In the 
only previous spring survey at Buldir, King (K, W. Kenyon 
and J. G. King, unpublished report) counted 52 geese on 
5 May 1965, on an aerial count he called incomplete. 
The early May arrival at Buldir suggests a rapid move- 
ment from spring staging grounds. From 1975 to 1978, geese 
left Crescent City, California, during the last half of April 
(Woolington et al. 1979), 
Most breeders presumably are present by mid-May, 10 
to 14 days before laying, The average time between peaks 
of arrival and nest initiation is 12 days for B. c. interior 
(Raveling and Lumsden 1977) and 10 to 15 days for B. c. 
minima (Raveling 1978; Dau and Mickelson 1979). 
