CHANGES IN THE WINTER DISTRIBUTION OF THE ROUGH-LEGGED HAWK 
Figure 1 . Ten regions and sites of Christmas Bird Counts used to evaluate changes 
in the numbers and distribution of the Rough-legged Hawk in North America. Black 
circles represent counts on which the number of Rough-legged Hawks per party hour 
increased from count years 79-92 to count years 93-106; white circles represent 
counts on which the number decreased. Circles are approximately to scale. 
Potential Predictor Variables 
To assess variables that might influence the Rough-legged Hawk’s winter 
range, we attempted to link climate factors, human-population changes, and 
Red-tailed Hawk abundance to each CBC circle. Because of lack of data 
we did not analyze other variables such as prey density or composition or 
specific land-use changes that likely also have an effect. 
We obtained climate data from the U.S. Historical Climatology Network 
web site (http://cdiac.ornl.gov/epubs/ndp/ushcn/usa_monthly.html). We 
assigned climate data to a CBC circle on the basis of the weather station 
closest to that CBC circle, provided that station was within 100 km of the 
center of the CBC circle. In 11 cases the closest weather station was at an 
elevation more than 400 m different from the CBC circle. In those cases, 
because of concern that the elevation difference would create too large a 
discrepancy in climate conditions between the CBC circle and the weather 
station, we used the next nearest weather station within 100 km of the 
CBC circle and similar in elevation. For our analyses, we used two climate 
variables: average December temperature (°F as reported by weather sta- 
tions) and the number of December days with snow cover >5 cm (2 inches 
as reported by weather stations). Although temperature data were complete, 
in some cases there were gaps in the snow-cover data. In cases where snow 
cover measured >5 cm on a given day and subsequent days had no data, we 
used daily high temperatures to determine whether snow cover remained. 
If the daily high temperatures remained at 0 °C or less through the gap, we 
assumed that the snow cover remained >5 cm during those days. 
Since historical land-use data for each CBC circle are either unavailable 
or impractical to obtain, we used population change between the 1980 and 
2000 U.S. censuses for the county constituting the largest fraction of a given 
CBC circle as a crude proxy for land-use change. 
212 
