that the prairie regions contribute primarily to the kill in the 
central and southern portions of the Atlantic Flyway. The prairie 
regions do not appear to contribute significantly to the waterfowl 
harvest in the New England States. 
PART 2--DISTRIBUTION OF 
RECOVERIES FROM MALLARDS BANDED AS LOCALS 
The mallard is by far the most important species of waterfowl 
in North America, making up nearly 50 percent of the total duck 
harvest in an average year. The importance of this species was 
recognized when the current program to band locals was first 
organized. The southern portion of the Prairie Provinces is con- 
posed of approximately 80 one-degree blocks of latitude and 
longitude. Within this area, the objective was to band a sufficient 
number of mallard locals to obtain 30 direct recoveries from each 
one degree block of latitude and longitude. With certain exceptions, 
this banding was to be accomplished gradually over a period of 5 
years. The purpose was to obtain an average measure of distribution 
under a variety of conditions. Exceptions were made in several 
areas where it was feasible to band large numbers of birds annually. 
The purpose here was to determine annual distribution from a given 
location and thus determine any changes in distribution that might 
result from differences in weather and other factors. 
In fair measure, the original objectives of the program to band 
mallard locals has been met. Therefore, it seems appropriate to 
examine mallard recoveries in more detail. 
Method of Analysis 
In this section, neither aerial survey strata nor political 
subdivisions were used to accumulate banding locations for purpose 
of analysis. Rather, the distribution of first hunting season re¬ 
coveries from each banding location was examined and those with 
similar recovery distribution patterns were grouped together. 
Generally speaking, all banding locations within each one-degree 
block were arbitrarily totaled and examined together. Degree 
blocks with similar characteristics were then grouped into sub¬ 
divisions of the breeding range which shall be referred to as 
reference areas . If the number of mallards banded within a State 
or Province was small, the State or Province was considered as a 
single reference area even though one might suspect that additional 
bandings will indicate that one or more subdivisions might be needed. 
Political boundaries were used to separate reference areas where 
practical because they best fit available data on size and distri¬ 
bution of the breeding population. Reference areas were grouped, 
for purposes of presenting the tabular material in this report, into 
8 regions. The 8 regions and the reference areas contained within 
each are delineated in figure 3. 
11 
