1. This paper is based on 34 roost-site and 76 roosting flight observations of the Grey Starling, Sturnus cineraceus, and the factors influencing the timing of roosting are considered. 2. Roost-site observation. Behaviours of preroosting assembly and roosting in this species is similar to those described by JUMBER in the European Starling, Sturnus vulgaris. The timing of roosting in the breeding season is solely subject to the degree of the parents' adherence to their nest, being latest in incubation and feeding periods (at least one parent retires to common roost in these periods), but by flying of chicks the roosting assembly becomes distinctly earlier. There is also noticed seasonal tendency that longer the daylength the earlier becomes the roosting. 3. Roosting flight observation. The following periods may be distinguished in the"feeding ground assembly" (of JUMBER) before starting to roosts : 1) Joint feeding period, 2) Pre-flight resting period and 3) Starting period. Observations are reported about the behaviour of flocks in these periods. The light intensity was measured by a portable TL-4 Lux meter. 4. The factors influencing the timing of start from the feeding ground to roosts are considered to be as follows : a) Distance. The longer the distance (but no difference up to 20 km) the earlier the start, but the shorter (or cut off) becomes the resting period, since they feed as long as possible. Thus the loss of feeding period owing to distance is kept at the minimum. b) Temperature. In winter, they feed actively and longer, therefore starting later (and without resting period) on severe cold days. On warm days, they finish feeding earlier, rest longer and start earlier. The higher rate of metabolism caused by low temperature would urge them to increase the amount of food intake, and may stimulate their social activity. Thus, active social bathing is often observed in the cold evening. c) Season. Roosting flight becomes earlier in early spring (though late in the breeding season), which is also supported by roost-site observation. This may be correlated not only with sufficient feeding time by longer daylength (as BROWN has suggested), but also with the increase of body fat in spring (proved by specimens), as well as lower rate of metabolism by the increasing temperature, d) Food supply. In autumn, it was observed that when local food supply was temporarily abundant, they left for roost earlier. Examples of this were recorded during the harvest season of rice plants, when the larvae of rice-moth were easily obtained from cut stems both on the ground and hung on poles. Also, where fruits of trees were abundant they left for roost earlier. e) Weather. In cloudy weather their start of roosting flight was earlier than on fine days. Considering also the reported early roosting in bad weather (JUMBER), this may be explained by that on bad or cloudy day the activity of birds is generally suppressed which decreases their metabolic rate, and therefore, the necessary amount of food intake is gained earlier than on fine days. f) Light intensity. The roosting flight releasing light intensity may be classified into : 1. Highest (or initial) light intensity, which is reacted by those birds which have taken enough food for the day (therefore ready for roosting), 2. Average (or most frequent) light intensity, to which most birds of the flock respond ; 3. Lowest (or final) light intensity, which is reacted by latest birds. For statistical comparison of roosting flight releasing light intensity, 'the number index' of roosting birds was used, which is calculated by : Σ number of birds starting or flying to roost at a certain light intensity × frequency of observation, or number of flocks observed, from winter observation, the highest light intensity was 28,000 Lux (58 minute before sunset) at 40km from the roost ; 15,000-7,000 Lux (49-38 minute before sunset)
(View PDF for the rest of the abstract.)