Bird community responses to habitat fragmentation: how consistent are they across landscapes?

Aim The woodland ecosystems of south-eastern Australia have been extensively disturbed by agriculture and urbanization. Herein, the occurrence of birds in woodland remnants in three distinct landscapes was analysed to examine the effects of different types of landscape matrices on species richness vs. area and species richness vs. isolation relationships and individual species responses to woodland fragmentation. Location The study system comprised three distinct woodland landscapes of the northern Australian Capital Territory and bordering areas of New South Wales. These landscapes (termed agricultural, peri-urban and urban) are located within 50 km of each other, have remnant fragments of similar age, size, isolation, woodland cover, elevation and climates. The major distinguishing feature of the three landscapes was the properties of the habitats surrounding the numerous woodland remnants. Methods Bird surveys, using an area-search methodology, were conducted in 1999 and 2000 in 127 remnants in the three landscapes to determine bird species presence/absence. Each remnant was characterized by measures of remnant area, isolation and habitat complexity. To characterize differences between each landscape, we conducted an analysis of the amount of tree cover and human disturbance in each landscape using SPOT imagery and aerial photographs. Linear regressions of woodland-dependent species richness vs. remnant area and remnant isolation for the three different landscapes were calculated to see if there were any apparent differences. Binomial logistic regressions were used to determine the relationships between the occurrence of each species and the size and isolation of woodland habitat, in each landscape. Results All the landscapes displayed a significant (P < 0.01) species vs. area relationship, but the slope of the urban relationship was significantly greater than those of the other landscapes. In contrast, only the agricultural landscape displayed a significant (P < 0.01) species richness vs. isolation relationship. When individual species were investigated, we found species that were: (1) apparently insensitive to reduction in remnant area and increase in isolation across all landscapes, (2) absent in small remnants in all landscapes, (3) absent in small remnants in all landscapes and also absent in isolated remnants in the agricultural landscape, (4) absent in isolated remnants in the agricultural landscape, and (5) absent in small remnants in the urban landscape. Threshold values (50% probability of occurrence) for area and isolation for individual species were highly variable across the three landscapes. Main conclusions These results indicate that woodland bird communities have a varying response to habitat fragmentation in different landscapes. Whilst we cannot be sure how representative our chosen landscapes are of other similarly composed landscapes, these results suggest that the type of landscape matrix may have a considerable influence on how bird species are affected by woodland fragmentation in the region. For instance, the properties of a matrix may influence both the resources available in the landscape as a whole for different bird species, and the connectivity (dispersal of birds), between woodland remnants. We encourage further research that examines these hypotheses and argue that the management of the matrix should be included in conservation strategies for fragmented landscapes.