92 THE RIDDLE OF MIGRATION 



out. The ones that go somewhat further south face 

 a diminished risk; those that go still further may run 

 no risk at all apart from exceptional winters. 

 There is thus set up at the hands of the winter fates 

 a system of selection. Resident mallards can never 

 become numerous. Partial migrants stand a better 

 chance while complete migrants get the best chance 

 of all. They will always be the dominating class as 

 long as the Alberta climate remains what it now is. 

 In particular sections of the north the case is 

 otherwise. Mallards are resident in Alaska and 

 Iceland, while Greenland actually boasts a distinc- 

 tive resident sub-species, quite within the realm of 

 legitimate expectation if the area involved (as is the 

 case in Greenland) supports a sufficiently large popu- 

 lation. Breeding isolation will probably result and 

 sub-specific differentiation become likely. The 

 mallard, like several other ducks, can feed at night 

 (and does so habitually in Alberta) and as long as 

 open water providing enough food for continued 

 sustenance, containing a sufficiency of vitamins, 

 is available and if the species can sustain the pre- 

 vailing temperatures, there would be nothing to 

 force migration on the race. They would survive 

 as residents. These conditions are but sparsely and 

 very locally fulfilled in Alberta. Numerical re- 

 strictions entirely preclude the development of a 



