of Minnedosa was not limited to 1968 but was observed in 
other years as well. Field surveys and related observations 
suggested that raccoon populations were higher adjacent 
to the Minnedosa River and more wetlands and abandoned 
farm buildings east of the river might have provided food 
and shelter for raccoons. 
Canvasback nest success (unadjusted for renesting) on the 
Redvers study area averaged 70% (Stoudt 1971) compared 
with 45 % in the present study. Of the nest failures at Red- 
vers in 1963, 93% were due to flooding. The wide disparity 
of nest success between Redvers and Minnedosa for other 
years perhaps can be attributed to the relative abundance 
of raccoons in the Minnedosa area. Other limiting factors 
appeared to be about equal on the two areas. Crows were 
not generally serious predators on canvasback nests, except 
during periods of rapidly receding water levels. 
The percentage of canvasback nests destroyed by rac- 
coons varied to some extent according to weather condi- 
tions. It appeared to be high in wet years and low in dry 
years. During dry years when wetland levels receded 
rapidly, many overwater nests were stranded before hatch- 
ing and were vulnerable to skunks, crows, fox, and other 
predators. Nest losses for most species were high in 1968 
because spring burning of nest cover was extensive and 
water levels were low and receding, increasing the vulnera- 
bility of overwater nests to predation. During the 12-year 
study, nest success was poorest in 1968 but best in 1969. 
Probable reasons for good nest success in 1969 were a de- 
crease in numbers of red fox and raccoons, abundant preda- 
tor food, and abundant upland cover that provided habi- 
tat for other nesting birds and for mice, both of which were 
buffer foods for predators, 
Olson (1964) found that canvasback clutch sizes in south- 
western Manitoba ranged from 9.7 to 10.1 for 53 complete 
clutches found during 1959-61. These clutches included 
redhead eggs. The number of canvasback eggs per clutch 
ranged from 6.7 to 8.5 and averaged 7.6 during the 3-year 
period of Olson’s study. This number is slightly less than 
the 1961-71 mean of 7.7 at Minnedosa. 
Clutch size averaged 8.3 eggs for 207 canvasback nests 
(1952-70) on the Redvers study area in southeastern 
Saskatchewan (J. H. Stoudt, unpublished data). Mean 
clutch size was lowest (7.9) in 1967 and highest (8.7) in 
1958. During the 18-year period, the mean number of 
redhead eggs per canvasback nest was 0.2; thus the mean 
clutch size for canvasback eggs was 8.0. The slightly lower 
mean clutch size at Minnedosa may have been due to red- 
head parasitism, which sometimes caused canvasback eggs 
to be rolled out of the nest. Erickson (1948) found a dif- 
ference of 2.2 eggs between parasitized and unparasitized 
nests, whereas at Minnedosa the difference was slightly less 
than 1.0 egg. 
Erickson (1948) found an average clutch size of 9.9 in 
unparasitized canvasback nests and 7.7 in parasitized nests 
in southeastern Oregon. Weller (1959) found that the aver- 
age number of redhead eggs per parasitized nest ranged 
27 
from 4.2 to 6.5 in 48 canvasback nests on the Delta Marsh 
in Manitoba. Olson (1964) stressed the importance of para- 
sitism as a means by which redheads were able to compete 
with canvasbacks and increase their numbers at the expense 
of canvasbacks. He stated, “Increased productivity is great- 
est where redhead numbers are small in comparison to their 
host species, as in pothole habitats. Here redhead produc- 
tion may be increased two-fold by their parasitic habit.” 
Olson also wrote, “... drainage tends to increase the 
average size of the remaining wetlands even though de- 
creasing the total amount of breeding habitat. Where this 
occurs the remaining nesting cover for both species is in the 
larger areas, which contain proportionally more redheads, 
and the result appears to be decreased canvasback pro- 
duction.” 
Larger clutch sizes of early nests were probably due in 
part to fewer eggs laid by renesting hens and more eggs 
rolled out of nests by redheads when parasitism was more 
prevalent. On Delta Marsh, Weller (1959) found that red- 
head parasitism decreased as the season progressed. This 
was not true on the Minnedosa study area, and the dif- 
ference might be related to differences in habitat between 
the large marsh and the pothole area. At Minnedosa, red- 
head parasitism was considerably greater during the later 
canvasback nesting effort. The lower clutch size for renest- 
ing attempts illustrated one important reason why produc- 
tion from renesting never equaled a successful initial nest- 
ing effort. 
Broods 
Brood use of various habitat types was probably more 
closely related to availability of food and escape cover than 
to other factors. Escape cover may have been related to both 
emergent vegetation and water depth. For example, a small 
wetland had to be deep to allow escape by diving, and thus 
water depth may have been more important than abun- 
dance of emergent vegetation. Higher use of large, deep 
wetlands was probably due to type and abundance of food 
available as well as the safety of deep water. 
Possibly the heavy use of wetlands in pasture was the 
result of high fertility and resultant high production of plant 
food and invertebrates. One pasture wetland on the study 
area that was favored by broods and intensively studied by 
Bartonek (1968) had a heavy growth of sago pondweed and 
an abundance of invertebrates. Both are important foods 
of ducklings (Bartonek and Hickey 1969b). The lower-than- 
average use of wetlands in ungrazed grasslands was proba- 
bly a resuit of a large proportion of these areas being adja- 
cent to a road where vehicle traffic may have been a deter- 
rent to broods. 
Most canvasbacks fed in open water. Stands of hardstem 
bulrush were relatively easy for broods to move through 
and probably provided better resting and escape cover than 
did dense stands of cattail and whitetop. Wetlands devoid 
of emergent vegetation were frequently used by canvasback 
