Anthropogenic landscape change results in the loss and fragmentation of habitat, and is recognised as the major contributor of species declines and extinctions, producing challenges for biodiversity conservation and management. For species with large biogeographic ranges, conservation is particularly complicated, because they occupy different habitat types, across multiple climatic zones and often across political boundaries. Furthermore, species-habitat relationships can differ from region to region, and transferring the same management strategies from one region to another can produce sub-optimal outcomes. Therefore, determining spatially explicit species-habitat relationships is important for effective conservation management of widely distributed species.
This thesis addressed this problem by investigating how landscape change impacts the multiple-scale species-habitat relationships for a widely distributed species, the koala (Phascolarctos cinereus), at the edge of its biogeographic range in the semi-arid Mulga Lands Bioregion of eastern Australia. The koala is a specialist folivore that is distributed across more than 30 bioregions. However, the current ecological knowledge of koalas is biased towards populations in mesic regions, with comparatively few studies conducted in semi-arid regions. The research in this thesis involved developing spatially explicit models in an information theoretic approach framework to investigate four main objectives: (1) variation in resource selection by koalas at multiple scales in the Mulga Lands; (2) the effects of landscape change on koala resource and habitat use at multiple scales in the Mulga Lands during and after a prolonged drought; (3) the presence of critical thresholds in the amount of habitat for koala persistence in the Mulga Lands; and (4) comparisons of spatial patterns of koala habitat use in the semi-arid habitats with habitat use in regions with more mesic climates.
In chapter 2, I demonstrate a method that explicitly tests at which scale koala tree resource selection varies. Scale is a fundamental aspect of ecological theory, but few studies have explicitly tested whether resource selection varies among scales. I found that the tree species and tree height were the most important factors influencing tree selection and this did not vary among scales. However, koalas were selecting trees with canopies of different condition, which was a measure of resource quality, across landscapes but not at the site-scale (i.e., a stand of 30 trees, each with DBH >10 cm).
In chapter 3, I applied a hierarchical mixed-modelling approach to examine the pattern of occurrence by the koala in anthropogenically modified and drought affected semi-arid landscapes. Koalas in the Mulga Lands depend on river red gums (Eucalyptus camaldulensis) as their primary resource. These trees occur in sites adjacent to watercourses. Consequently, riverine habitat is primary habitat, providing refugia in times of drought and extreme heat. However, the amounts of primary and secondary habitat in the landscape were not important on their own. There was a strong positive interaction between the amounts of primary and secondary habitat, suggesting that koalas require both habitat types and are more likely to occur in more intact landscapes. However, there was no difference in habitat use between dry and wet years.
In chapter 4, I tested for critical thresholds on a site-scale variable and landscape-scale habitat variables at four landscape extents: 500 m, 1000 m, 3000 m and 5000 m radii. I found there was a critical threshold on the percentage of primary trees (18.5%) required for koala persistence at the site-scale. A threshold relationship was also found on the percentage of primary habitat (1.9%), but this was not supported by plots of the spread of data. There were no further critical thresholds at the four landscape extents on amounts of primary and secondary habitats, or their interaction at varying extents of landscape-scale.
In chapter 5, I explicitly tested for differences in spatially explicit koala-habitat relationships among regions, and found that there was considerable regional variation in the parameter estimates for habitat predictor variables. Regional differences in site-scale distance from watercourse, plus the landscape-scale amounts of primary and secondary habitat, and road density variables, were deemed most important, and the regional differences in the site-scale proportion of primary trees and the soil type were least important.
This study makes an original contribution to the fields of landscape ecology and wildlife ecology by demonstrating how resource and habitat selection by animals can vary among scales and across landscapes and regions. It advances the understanding of how widely distributed species are impacted by anthropogenic and natural processes at the edges of their range. My research expands the knowledge of koala ecology in semi-arid extremities of its range, demonstrating koalas use resources and habitat differently in semi-arid landscapes compared to the more mesic landscapes. It is the first multiple-scale study of koala-habitat relationships in a semi-arid region, and is highly relevant for conserving and managing koalas in semi-arid landscapes.