The exact statistical relationship between these two kinds 
of rates appears to require some detailed studies that could not be 
pursued within the time limits of my own project. If annual mortality 
rates can be estimated within reasonable limits by analyses of seasonal 
recovery rates, waterfowl managers have a life-table short cut of 
considerable importance in their evaluation of hunting regulations. 
Some crude data on the subject were available near the end of my study. 
These are given in a subsequent chapter. 
Relative Geographic Abundance 
Under certain conditions, the mortality data obtained from 
bird banding may be converted into an index of the relative abundance 
of two geographic populations of a migratory species. One might, 
for instance, attempt to determine that the breeding mallard popula- 
tion of Manitoba is x times as large as the breeding mallard popula- 
tion of Saskatchewan Perce Sl a as large as that of Alberta. A 
similar comparison might involve wintering populations in nearby 
regions. The conditions required for the calculation appear at this 
time to be as follows: 
(1) The two geographic populations to be compared must reside at one 
time of the year in nearby regions, such as Alberta and Saskatche- 
wan; or in Texas and Louisiana. 
(2) During this period of residence, samples of the two populations 
need to be banded in sufficient size and apparently in a random 
manner. 
(3) After the birds have migrated, samples of the two populations 
must again be banded in some third region in sufficient numbers 
and apparently in a random manner. I term this third region 
the area of reference. 
(4) The species studied mst, of course, possess a satisfactory 
recovery rate for each banding operation in order to provide 
enough recoveries for analysis. 
In my work on the North American banding files, I encountered 
no banding operations that met these conditions when rigidly applied. 
Some of the mallard work may, however, serve as an illustration of 
how such a geographic index can be computed and perhaps emphasize 
its use in a carefully planned international system of bird-banding 
stations. 
We can begin by comparing the breeding mallard populations 
of Manitoba and Saskatchewan. (Manitoba contains 216,512 square miles, 
Saskatchewan 251,700 square miles.) Adult mallards were handed from 
May or June through August 31 in the southern parts of these two 
prairie provinces; they roughly totaled 1500 for Manitoba and about 
8000 for Saskatchewan, and at the present time cannot be safely 
regarded as randomized samples of the breeding populations of these 
two areas. The Manitoba sample is weak because the 1936-6 recoveries 
132 
