Date of Award


Document Type



Edith Cowan University

Degree Name

Master of Science (Biological Sciences)


School of Science

First Supervisor

Professor William D Stock

Second Supervisor

Dr Dave Blake

Third Supervisor

Dr Robert Davis

Fourth Supervisor

Dr Geoffrey Barrett


There is limited information on communal roosting in parrot species of Western Australia and other parts of the world. Carnaby’s Black-Cockatoo is an endangered species that forms large nocturnal communal roosts, and for this reason they are an ideal model species to test the characteristics or factors that are associated with roost sites. Known roost sites distributed across the Swan Coastal Plain were identified and selected through the Great Cocky Count project. A minimum of five and maximum of ten individual trees were assessed at 11 roost sites with an overall total of 95 roost trees sampled. I determined the tree species composition and vegetation structural arrangement, as well as the roost site location properties of each of the selected roosts. A total of 18 widely dispersed roost sites on the Swan Coastal Plain were chosen for spatial analysis of landscape characteristics at three scales, namely 1, 6 and 12 km radii around each roost. Landscape characteristics were derived from data layers using a geographical information system. Generalized linear modelling was used to investigate which landscape variables best explain the roost count numbers and fidelity of Carnaby’s Black-Cockatoo at the three spatial scales. Landscape variables were broadly categorised into urban pressures, tree characteristics, and, food and water availability.

I found that at the roost tree scale, the cockatoos utilised a wide range of native and non-native trees, situated within a variety of land-use types. Results showed that bird’s roosted in tall (average of > 25 m) tree species that have relatively thick trunks (average DBH of 1 m) and medium foliage density (average of 50 %), and that are not too densely forested amongst other trees (average tree crown connectivity was 20.58 %). Carnaby’s Black-Cockatoos showed no preferences for any particular tree species or bark colour across the study sites. At the landscape scale, models (based on Akaike Information Criterion) showed that variables associated with bird abundance and roost fidelity varied with scale. The models highlighted the importance of a 1 km radius of potential roost trees (tall trees i.e. ≥ 8 m) across all scales, and food (Banksia and pine) and water sources, particularly within a 1 and 6 km radius. Study sites surrounded by more urban pressure may be driving greater numbers of Carnaby’s Black-Cockatoo at such sites because the formation of larger flocks can increase resource location in fragmented landscapes. Reduced density of roads and non-native ground cover vegetation, over the greater landscape, indicated that restricting the amount of densely urbanised structures should be considered when further developing around roost sites across the Swan Coastal Plain.

The nocturnal roost study sites had greater fidelity and numbers of cockatoos with a combination of landscape variables at different scales, which are based on habitat structure, food availability and water availability. The understanding of the characteristics of communal roosts, roost site choice, and the surrounding matrix developed in this thesis provides managers with insights on how best to conserve this species.


Paper Location