1. Roost distribution, feeding dispersal and density and behaviour of feeding flocks before roosting flight in the Grey Starling, Sturnus cineraceus
, were studied in Kanto Plain from 1953 to 1962. This paper describes the results of some 45 winter observations. The roost distribution will be further iuvestigated.
2. In Kanto Plain, extensive feeding range of big Koshigaya roost (about 50, 000 birds) north of Tokyo occupies 30-40km radius circle of the plain. Other smaller roosts are situated outside of this feeding range.
3. Daytime (late afternoon observation) feeding flocks usually consisted of several to 30, sometimes 100 birds and each flock was more or less 500m apart. This spacing by small flocks should be a response to the food availability which are minute mud fauna and rather sparcely distributed larger larvae, etc.
4. In feeding ground assembly towards evening, up to 300, rarely 500 or exceptionally 1, 000 birds gathered into a flock, the average flock size being 176 birds. Thus an area where a flock of over 150 birds was found could be regarded as a good feeding place. Evening flocks gradually moved always towards the direction of main roost (to which most of the flock members belonged). The distance of this late afternoon movement to final gathering place was usually 1.5-2.0km.
5.The density, the economic density, of daytime feeding flocks in the field was 30-500 (average 160) birds per 1km2
, therefore 0.3-5.0 (average 1.6) birds per 1 ha, which coincided with previously calculated data. This may suggest that a space of about 1 ha is wanted by one starling.
6. Two types of feeding dispersal from a roost were noticed. In type A, the flock size decreased with greater distance along a certain direction and under similar conditions of the feeding environment (see NNW direction in the map). In type B, concentration to a certain feeding area was found. A remarkable concentration was at 40km from the roost which was about the maximum distance of feeding dispersal. On this feeding flight line, the size of feeding flocks decreased with shorter distance from the roost, i. e. greater flock size towards the concentration area. This concentration was possibly due to higher food resources plus good breeding envirnment (old trees with holes).
7. The middle area on the B type dispersal line was apparently occupied by a big feeding flock of another roost (confirmed by roosting flight direction). If this were a competitve segregation of feeding grounds by two different roost-flocks, it is a remarkable instance.
8. It commonly happens that at the peripheral feeding area, the feeding flocks make a joint flock with birds of other roost or roosts. Thus on starting for roosting flight they take off to different, or to each determined, directions.
9. Distribution and observation data at eleven roosts in the study asea (see fig. 1) are described.