Predators 231 



at the map will show that the roosts closely follow this distribution of abundance. 

 This can be explained on the assumption that there is some biological advantage 

 in a large roosting aggregation, but that when the distance which it is necessary 

 to travel to form a large aggregation is too great, this advantage is nullified and 

 no large roosts are formed. 



In general, roosts appear to be more frequent and larger on the richest land 

 or in the regions of the heaviest corn, which is simply another way of saying 

 that they occur where abundant food supplies induce heavy winter crow popula- 

 tions. 



Stability of Roosts. In Missouri former roosts, now abandoned, were 

 also mapped. These show as hollow squares. Where the date of abandonment 

 is known, it appears to the right of the roost symbol. 



If it be true that roosts are associated with abundance, then it may be true 

 that abandoned roosts constitute evidence of decrease. It is my impression that 

 abandoned roosts are much more frequent in Missouri than in Indiana or Wis- 

 consin. I am unable to suggest any reason why crows should decrease in Missouri 

 more than elsewhere. The amount of crow shooting there is if anything lighter 

 than in other States. 



Some crow roosts have maintained the same location for long periods of 

 time. Thus there is a roost near Stebbinsville in northwest Rock County, Wis- 

 consin, which local sportsmen say was occupied 30 years ago and has been used 

 continuously ever since. Near Spring Prairie in Walworth County, Wisconsin, a 

 single roost was used steadily for 20 years, but recently it has been shot out. 



Steady shooting, especially night shooting, forces a shift of roost location. 



In Pike County, Missouri, are two roosts known to have been occupied since 

 1900 and 1915, respectively. In southeast Bates County, Missouri, there is a roost 

 occupied since 1890. 



While roosts may persist through a long period of years, they are not always 

 stable through the various seasons of use. In northern Illinois, for instance, fall 

 roosts and spring roosts are said to often differ in location. Likewise the roost at 

 McFarland in east central Dane County, Wisconsin, is reported to be used in fall, 

 but not in spring. In the vicinity of Oberlin, Ohio, Dr. Lynds Jones told me 

 there is a winter roost of about 20,000 birds. About February 15 this roost 

 progressively breaks up and scatters into smaller units. By March 15 a dozen 

 roosts of 1,000 to 1,200 birds each are formed. In late March these dissolve into 

 nesting pairs, and there is no more gregarious roosting until the following fall. 



Radius of Roosts. There is room for some very interesting field research 

 to determine the radius of feeding territory covered from a single roost. This 

 could be ascertained by plotting on large maps the time and direction of evening 

 flights. Ordinary observation shows that the territory must often be as large as a 

 county. The last arrivals in the evening sometimes do not arrive at the roost until 

 after dark. At the Sun Prairie roost in Dane County, Wisconsin, which I visited 



