Radar ring angels and the roosting behaviour of starlings

An account is given of the radar ring angel phenomenon; observations are described which prove that these rings are produced by the dispersal of starlings (Sturnus vulgaris) from their roost at sunrise. Radar film records of starling roost activity within an area of 5000 square miles centred on the radar station at Bushy Hill, Essex, have been compiled since August 1958. Analysis of these records establishes a seasonal cycle of activity which rises to a maximum in the late summer, followed by a reduction through autumn and winter, due to the coalescence of flocks to form larger roost communities. There is a complete cessation of ring activity during the breeding season. The radar method of observation has permitted detailed study of the starling’s roosting habits. Calm weather dispersals take the form of concentric circles with the birds flying at a mean altitude of 150 ft. In the presence of wind there is a tendency for the dispersal to be in the form of arcs moving downwind, thought to be partly a result of the reduced altitude of upwind flights rendering the birds invisible to the radar. Dispersals from widely separated roosts are approximately simultaneous and take place about sunrise but a significant correlation is found between the time of first flight and the local light intensity as affected by varying amounts of cloud. Detailed analysis shows a tendency for starlings to fly out earlier and return later, relative to the sun and hence the light intensity, in winter than in summer. The mean air speed of starlings leaving the roost was measured as 37 knots, with higher speed in winter (40 knots) than in summer (32 knots). This speed difference is presumably a seasonal effect, but could be due to the observation of different populations. The successive waves of birds from the roost were found to fly progressively about 1knot slower. The time interval between such departure waves varies in accordance with a log-normal distribution, the modal value of which is 3 min; there is a tendency for greater values in summer than in winter. Radar signal amplitude measurements are described which yield a population estimate of 500000 starlings for one large roost in February 1961. A density of one starling per acre is deduced, with the possible maximum value of three per acre. Examples are given of the interpenetration of radar rings derived from different centres, showing that feeding regions may be cropped by starlings from different roosts. Radar observations upon the flocking behaviour of the starling are described; this shows that a 360° dispersal may sometimes be followed by an evening assembly from one direction only. The apparently disciplined behaviour of the starlings during the dispersal process is discussed, while cyclic variation in the intensity of starling chatter, as recorded at the roost, is correlated with the periodic eruptions of the waves of birds which produce the rings upon the radar.