Elder (1946) has already shown that a state refuge can be a death 
trap for Canada geese. The fluctuations in rate shown here appear 
to have no relationship to the number of hunters or recent hunting 
regulations. 
Because many duck-banding operations have been carried on 
during the course of a hunting season, Thompson and Jedlicka (19),8) 
and Bellrose and Chase (1950) have worked out several formas to 
compensate for the lapse of time between the opening date of hunting 
and the subsequent dates on which the ducks are banded. It is 
hardly fair to these fine intellectual contributions to point cut 
that in each case no adjustment has yet been made to account for the 
marked early-season vulnerability of young birds to the gun. In two 
of the formulas a second limitation may appear in occasional years 
when some extraordinary changes occur in the monthly variation of 
waterfowl mortality. (These apparently took place in both 1939 and 
19L6 in the Mississippi Valley when the percentage of adult mallards 
taken in October significantly exceeded that taken in November.) 
Reccvery Rates and Mortality Rates 
Are these fluctuations in seasonal recovery rates, described 
above, actually correlated with annual mortality rates? To answer 
this question, mortality curves available for this period were com 
piled from birds banded in Manitoba, Saskatchewan, and the various 
states bordering the Mississippi River. Data obtained by the Illinois 
Natural History Survey were omitted, as this sample was being analyzed 
by Bellrose and Chase (1950). The hunters' reports involve only birds 
shot as adults and only those banded before September 1 of the year 
in which they were shot. While I have termed these birds "Mississippi 
Valley" mallards, one must remember that this term is an arbitrary 
One « 
Computations of mortality rate were carried out for these 
adult birds on a time-specific basis. The limitations of this analytical 
method have been described in Chapter III, and the results given in 
figure 15 are therefore termed an index rather than a series of 
mortality rates. A fair degree of correlation with adult-recovery 
rate occurs in five out of the six years studied. In view of both 
the unreliability of time specific calculations and the small samples 
from which the recovery rates were taken, the correlation seems to 
me to be surprisingly close. 
The accuracy of a Canadian first~season recovery rate as 
arn. index of hunting mortality in mallards is not readily susceptible 
of proof, Whatever its reliability, it is obvious that this rate 
can at best only concern the regional populations from which the 
birds were trapped and banded. It does not follow that mallards 
from Alaska, Yukon, Mackenzie, and British Columbia have recovery 
rates of a similar character. I do think we can conclude that, 
given sufficiently large samples and adequately randomized banding 
operations, seasonal recovery rates will fluctuate directly with 
the mortality rates in a waterfowl population. 
130 
