access to the refuge interior permitted. Counts were made 
from a vehicle with binoculars or a spotting scope. 
Although all-weather roads on the refuge provided a route 
from which most eagles could be counted, some parts of 
the refuge were not accessible for census. Most counts were 
made between 0630 and 1100 hours, when eagle congre- 
gations were usually largest. Counts were not made if 
weather was severe or if visibility was limited by fog or 
precipitation. If the entire head and tail of a bird were 
white, it was classed as an adult; all others were classed as 
immatures. We made 22 counts of eagles on Swan Lake 
NWR in both 1975-76 and 1976-77, and 17 in 1977-78. 
No counts were made from late October to mid-December 
1977 because of limited access to the refuge interior. 
Aerial waterfowl inventories were conducted periodi- 
cally during the three winters by the Missouri Department 
of Conservation. Surveys of Canada goose numbers in the 
Swan Lake Zone were made between September and Jan- 
uary. 
In 1975-76 and 1976-77, we made counts of waterfowl 
and fish carcasses along shoreline transects on the refuge. 
In 1977-78, early freeze-up on the refuge, wide dispersal 
of waterfowl in the Swan Lake Zone,-and limited access to 
the interior of the refuge during the hunting season, pre- 
vented us from making adequate waterfowl carcass 
counts. Additionally, no fish kill occurred on the refuge in 
1977-78. All waterfowl and fish count transects extended 
4.2 km along the west shore of Silver Lake, and a fish 
transect also extended 1.3 km along the northwest shore of 
South Pool (Fig. 1). Due to waterfowl use of ice-free 
openings in the middle of refuge impoundments, transect 
counts did not always accurately reflect relative numbers 
of waterfowl carcasses available to eagles. Transects were 
selected for their accessibility and abundance of carcasses. 
When possible, transects were systematically examined 
from a vehicle and on foot on the same day on which 
eagles were counted. Carcasses thoroughly stripped of 
meat were not counted. 
Feeding Habits 
Feeding activities of eagles were observed throughout 
the study period. Roosting sites and feeding perches were 
periodically searched for discarded food and cast pellets. 
Species represented among waterfowl and fish carcasses 
along carcass transect lines were identified. Cast pellets 
were analyzed by methods described by Errington (1932). 
Mammalian prey items in pellets were identified by using 
the key to guard hairs published by Stains (1958). 
Capture 
Eagles were trapped in the field seasons of 1975-76 and 
1977-78 with specially designed leg-hold traps, concealed 
around anchored Canada goose carcasses; a live decoy 
3 
bald eagle was often used (Griffin 1978). Trapping sites 
were in fields and on frozen impoundments adjacent to 
eagle feeding areas and perches. All captured unbanded 
eagles were banded with U.S. Fish and Wildlife Service 
rivet-type bands. 
Radiotelemetry 
Radio transmitters were placed on 6 captured bald 
eagles in 1975-76 and 12 in 1977-78. Sixteen of the trans- 
mitters (manufactured by AVM Instrument Company, 
Champaign, Illinois) were powered by lithium batteries 
(SB-2), and the other two (Telemetry Systems, Inc., Mil- 
waukee, Wisconsin) by solar cells and rechargeable mer- 
cury batteries. Transmitters operated in the 150-151 MHz 
band and each was identifiable by a distinct frequency 
and pulse rate (48 to 120 pulses per min). The six transmit- 
ter units placed on eagles in 1975-76 were encased in blue 
Herculite plastic to increase their visibility and protect 
them from damage by the eagle and weather. The Hercu- 
lite casings covered the entire transmitter unit, including 
the antenna. The size (42 g) of the units necessitated a 
back-mounted harness system on the eagles. 
A 12-channel receiver (AVM Instrument Company) was 
used with either a hand-held four-element yagi antenna or 
a four-element yagi null-peak antenna system attached to 
a vehicle. When daily ground searches by vehicle failed, 
we made aerial searches using two yagis, one on each strut 
of a Cessna 172. 
Depending on weather and road conditions, radio- 
tracking was attempted at least twice a day by ground tri- 
angulation. Ten aerial searches were made. Locations 
were made from 0600 to 2200 hours; most were made 
from 0700 to 1100 and 1500 to 1900. All available radio 
fixes and visual sightings were plotted on acetate laid over 
aerial photos of the study area. Minimum local winter 
range size was determined by a modification of the “mini- 
mum home range” method of Mohr (1947). 
Results 
Eagle Populations 
Numbers and Annual Chronology 
Bald eagles occurred in the Swan Lake Zone from Octo- 
ber through March. In each of the three fall and winter 
seasons, numbers fluctuated considerably; the population 
was characterized by three or four distinct peaks, 
separated by lows of varied length and degree. 
In 1975-76, the first bald eagle was sighted on the 
refuge on 2 October. Four major peaks occurred during 
fall and winter (Fig. 2). The first and highest (66 birds) 
occurred on 2 December. Numbers declined to 37 on 
8 December, increased to a second peak of 58 on 18 
— 
