Predicting the Habitat of Great Tits (Parus major) in a Large City Using a Logistic Regression Model

Abstract

We created habitat models for Great Tits, (Parus major) in urban parks of Osaka, Japan. A survey of the presence of Great Tits was conducted in 85 parks during the breeding season (April-July) of 2000. The following variables were measured to model the habitat of Great Tits by a logistic regression model with GIS: park properties (area, perimeter, shape, age, distance to nearest mountains, distance to the nearest 10ha forest), area of tree cover (within and surrounding parks) and proportion of surrounding land use types. Surrounding environmental variables were measured by two methods, (a) 5 types (0, 50, 100, 150, 200m) of buffer zones from park border, and (b) 4 types (radius of 50, 100, 150, 200m) of circles from the park centre. From the two models obtained, one model suggests that tree area in a park and its 50m buffer zone affects positively the habitat quality of Great Tits, and round shaped parks are better. The other model shows that tree area within a radius of 200m from the centre of a park has a positive effect on the occurrence of Great Tits. From the latter model, about 4.0ha (32%) of tree area is required in a radius of 200m to achieve a probability of above 0.5 for occurrence of Great Tits. These models will support urban greenery planning for birds.