Methods
Previous studies were based mostly on analyses of stomach contents. My study is based on the analysis of 617 regurgitated pellets collected from roosts and lookout posts. Fifty-three collections of pellets were made throughout the year at regular intervals, except that none was made in January, March, or May. The pellets were wrapped individually in paper or leaves as collected, and each was analyzed separately. The percentages by bulk of different food residues (excluding sand and other extraneous material) were estimated in each pellet and recorded.