On 1 May 1967, a severe blizzard occurred in North Dakota 
but little snow fell at Minnedosa. Nearly all wetlands on 
the study area were still frozen over on 3 May and canvas- 
back migration and nesting were considerably delayed. In 
general, production was best during years when water levels 
remained fairly constant and no extremely cold weather 
occurred after incubation had begun. 
Nesting canvasbacks, along with mallards and pintails, 
began searching for nest sites soon after arriving on the 
nesting grounds. H. A. Hochbaum (personal communica- 
tion) reported that canvasbacks arrived at the Delta Water- 
fowl Research Station, south central Manitoba, on 24 April 
in 1965. During the night of 25-26 April, large numbers 
of many species of ducks arrived, including many canvas- 
backs. The Delta Station lies 113 km east of the Minnedosa 
study area and the phenology at Delta differs by a few days 
from that at Minnedosa. On 30 April 1965, canvasback hens 
were seen searching for nest sites on the study area and the 
first egg was laid on 1 May. In 1966, all wetlands and even 
large lakes were frozen on 1 May but by 9 May the sight- 
ing of canvasback hens searching for nest sites was com- 
mon. The first egg was laid on 6 May. 
Hochbaum (1944:62) stated that hen canvasbacks select 
the breeding territory. This observation was corroborated 
during the present study. Hens selected the nest wetland 
and drakes either waited on the wetland or flew to nearby 
wetlands to join other drakes and pairs. During nest search- 
ing and early stages of egg laying, males usually waited near 
the hens. Many nests were found after a lone drake was 
observed on a wetland, and the hen was flushed from the 
cover. If the hen was merely searching for a nest site but 
not actually laying eggs or sitting on the nest, disturbance 
usually caused her to abandon further use of that wetland. 
Searching for nest sites took several days, during which the 
hens were wary and easily frightened. 
Nest success was poor in 1961, 1965, 1967, and 1968, and 
though most of the nests were actually destroyed by preda- 
tors, weather played an important role in each of those 
vears. Most nest abandonment was caused by weather but 
amounted to less than 15% of total nest failures. Stable or 
gradually lowering water levels like those in 1968, 1970, 
and 1971 were usually conducive to good nesting success 
for overwater nesters. However, burning of pothole edges 
and predation upset this generality in 1968. Abruptly rising 
water levels in July 1965 caused an appreciable nest loss 
because of an extensive late-nesting effort brought about 
by loss of early nests resulting from low temperatures and 
snow in May. 
Extreme rainfall affects pond levels and nest success, On 
2() July 1965, more than 15.2 em of rain fell during a few 
hours. Wetlands filled and spilled over, flooding many over- 
water nests of redheads, ruddy ducks, and late-nesting can- 
vasbacks. Although flooding occurred in July 1969, early 
nests were very successful; consequently, the effect of the 
flooding on total nest success in that year was not severe. 
Years in which rainfall was lacking during the nesting 
25 
season were also poor for duck production. In May and June 
1967, for example, wetland levels dropped rapidly and 
much overwater nest cover became unsuitable. Predators 
took a heavy toll of nests stranded on dry ground, and re- 
nesting was greatly inhibited. Olson (1964) also reported 
poor nest success and high predation rates during drought 
years. 
Both drought and excessive rainfall appeared to have 
disastrous effects on nesting. Drought was the most serious, 
however, as it usually had the longer and more critical 
effects. In many instances when wetlands became dry, they 
were plowed and permanent drains established. Even 
where no drainage occurred, cultivation of a dry wetland 
temporarily destroyed aquatic and semi-aquatic vegetation 
(Fig. 16). 
Heavy predation on early nests occurred during the 1961 
drought. No strong renesting effort ensued because of dry 
weather and resultant rapid lowering of water levels. Over- 
water nest cover was rapidly made untenable and renesting 
was curtailed. In 1965, a severe cold front moved into the 
area on 23 May. Rain was followed by snow and subfreez- 
ing weather. Many nests were deserted and in some nests 
eggs were inadvertently buried by hens trying to keep nests 
and eggs above rising water levels. Nest data indicated that 
19% of the first nesting attempts succeeded and, although 
47% of the renests also succeeded, total success was below 
average for the area. Precipitation during the nesting season 
in 1967 was the lowest for the 12-year period, water levels 
receded, canvasback nests were stranded, and predation 
was severe. The weather was cool enough to stimulate a 
good renesting effort but nest success was poorer than for 
the first effort. Renests are usually more successful than first 
nests but this was not true in 1967. Although overwater nest 
cover appeared to be adequate for the number of canvas- 
back pairs present on the area, nest success (including 
renests) was low in 1968. 
Relations between nest success and habitat types were 
often difficult to determine. Nest success for any species of 
waterfowl is in part related to the species composition and 
density of predators and size of their home ranges. If a nest 
was located in a comparatively predator-free area and was 
not subjected to severe weather or farming operations, 
chances for its success were high, irrespective of the quality 
of the cover. The local influence of predation on nests was 
believed to be one of the main reasons for the variable use 
of ponds by broods, even though habitat conditions re- 
mained the same. To accurately assess the relative value 
of various habitat types, one would need a precise estimate 
of predator populations and size of their home ranges, along 
with other basic information. 
When raccoons became more common, overwater nest 
success decreased. Numbers of raccoon pelts received by 
the Hudson Bay Company in Winnipeg, Manitoba, in- 
creased from 65 during the 1953-54 season to 146 in 
1954-55; by 1960-61 the number had risen to 871 (E. G. 
Bossenmaier, personal communication). In 1959, Olson 
