28 
broods if adequate food was present, if open spots were 
available along the shore where broods could roost with- 
out being disturbed, and if water was deep enough to allow 
escape by diving. 
The annual variation in brood use suggests that the 
species of emergent vegetation may not be as important as 
the percentage of water surface covered with emergents. 
The use of bulrush and cattail wetlands appeared to be im- 
portant but most of these wetlands were less than one-third 
vegetated, There may be a positive relation between food 
production for canvasbacks and emergent cover. The entire 
subject of brood use of various habitat types must be further 
investigated to provide more specific information, especially 
about the relative abundance of various species of aquatic 
invertebrates. 
Annual Production 
The number of juvenile ducks in the fall flight is obviously 
less than the number of eggs hatched; therefore, brood sur- 
veys are essential for research and management. Observa- 
tions along roadside transects yielded reliable trend data 
for canvasbacks, but beat-out surveys were necessary for 
a more precise determination of production per unit of area. 
An average of 60% of canvasback broods counted during 
beat-out surveys were found within the 0,4-km roadside 
transect. A visibility rate of 68% was found on the Red- 
vers study area from 1952 to 1958 where the roadside tran- 
sect was 0.2 km wide (Stoudt 1959:80). 
Mean brood sizes were average in 1961, 1962, 1964, 1969, 
and 1971: above average in 1965 and 1966 (probably be- 
cause average clutch sizes were large): and small in 1963 
and 1967 (probably because clutch sizes were small), The 
fact that the average size of Class I broods in 1967 was con- 
siderably below the mean, whereas that of Class I] broods 
was above the mean, is a strong indication that some of 
the early broods died soon after hatching. However, the 
mean size of Class I canvasback broods in most years fol- 
lowed the same trend as the mean number of canvasback 
eggs hatched per clutch. 
Except in 1966, both the frequency and number of red- 
head ducklings per canvasback brood increased as the sea- 
son progressed, These data parallel nest data that show that 
parasitism by redheads was greater during the renesting ef- 
fort. Brood data from older age classes, especially those of 
Class III, were often unreliable because of the tendency 
of redhead ducklings in this age class to separate from 
broods. This is even more prevalent when broods are de- 
serted by the hen. 
Juvenile mortality in diving duck broods in the Minne- 
dosa area probably varied with the amount of time they 
spent on land, either traveling between wetlands or rest- 
ing on shore, Canvasbacks were vulnerable to many up- 
land predators. Hens and broods spent many of the day- 
light hours sleeping on water or in the pseudo-sleeping atti- 
tude described by Cornwell and Bartonek (1963), but some 
broods were observed in open areas on sand bars or mud 
flats. If these locations were near cover, predation was 
probable and was actually observed in the present study, 
Overland travel, which might have been in search of food, 
as a result of disturbance by man, machinery, or preda- 
tors, or merely as a result of the propensity of hens to change 
wetlands, was common, On 21 July 1961, three canvasback 
broods were observed on one wetland. Two of the broods 
spent 80% of this time on open water but the third hen 
took her brood across a gravel road to another wetland and 
returned later. 
During wet vears when July water levels were high and 
wetlands numerous, overland travel between wetlands was 
frequent and comparatively safe. In dry years such as 1961, 
1967, and 1968, wetlands were low and less numerous and 
travel between them was farther and more hazardous. 
In 1967, there was wide disparity between nest success 
and pair-brood data. Broods were so scarce that none were 
found on aerial flights outside the beat-out areas and road- 
side transects. Several broods from nests known to have 
hatched later disappeared from brood wetlands, and re- 
peated attempts to relocate them were unsuccessful, Brood 
movement between wetlands in 1967 was probably in 
response to low water. W hatever the reason, increased 
mobility of broods could easily have contributed to a higher- 
than-normal juvenile mortality. Another factor that could 
have contributed to the loss of ducklings and entire broods 
was the frequency of smaller broods. 
Juvenile mortality for 1963 seemed abnormally low and 
the number of eggs that hatched might have been under- 
estimated. Some nests were flooded after hatching but be- 
fore they were checked. The number of eggs that hatched 
in some of these nests might have been higher than esti- 
mated. Mortality in 1970 was low, probably because water 
levels were stable and brood movement negligible. 
Poor production in 1961, 1967, and 1968 was a result 
of a combination of low breeding populations and poor nest 
success, Production in 1963, 1964, and 1966 was good pri- 
marily because average breeding populations were large 
and nest success was higher than average. Good produc- 
tion in 1969 was mainly a result of excellent nest success. 
In general. brood mortality had less effect on total produc- 
tion than did nest success. After canvasback broods move 
from a small nesting pond to a larger brood pond, broods 
are comparatively safe from most predators. 
Acknowledgments 
I thank my former supervisors C, Williams, E. G. Wel- 
lein, H. kK. Nelson, and the late H. W. Murdy for their 
direction and encouragement. To the many Canadian and 
American field personnel, I express deep personal gratitude. 
I acknowledge the cooperation of the many landowners on 
whose property these studies were conducted. I also thank 
C. Bolin, K. F. Higgins, J. R. Serie, D. L. Trauger, and 
