Understanding the relationship between organisms and their environment is particularly important for the conservation and management of endangered species. The kakapo (Strigops habroptilus, Gray 1845) is a critically endangered, lek breeding, flightless nocturnal parrot endemic to New Zealand. In April 1998, a total population of fifty-six kakapo was known to survive on offshore islands. Twenty-six kakapo, thirteen males and thirteen females, were temporarily transferred to Pearl Island (518 ha), southern Stewart Island, from April 1998 to April 1999. The translocation of kakapo to Pearl Island, and subsequent breeding season, provided an ideal experimental framework to study kakapo dispersal, movement patterns, home range development, habitat selection, and lek development during the non-breeding and breeding seasons. A total of 4425 radio locations were analysed for all twenty-six birds, with a mean error polygon of 0.03 ha and an estimated average radio telemetry error of 21.6 m. Various home range analysis techniques were used to estimate kakapo home range size and overlap including: minimum convex polygons (MCP), modified minimum convex polygons (MMCP), harmonic mean analysis, adaptive kernel methods and cluster analysis. Estimates of kakapo home range size differed significantly depending on the method used (ANOVA, general linear model: F₁₃, ₁₀₇₆ = 63.99, p < 0.0001) and the season (F₂, ₁₀₇₆ = 160.75, p < 0.0001). Breeding home range size was significantly larger than non-breeding range size (mean difference = 67.6 ha, t₂₅ = 15.27, p < 0.0001). Calculations from 100% MCP and 95% harmonic mean analysis resulted in larger estimates of home range size and overlap compared to other methods. Cluster and kernel analyses appeared to give the most accurate home range representation for kakapo. Core home range areas showed a greater degree of similarity between methods. Male and female mean annual home range size did not differ significantly, whereas males had significantly (p < 0.05) larger home ranges than females during the nonbreeding season. Minimum convex polygons and harmonic mean analysis suggested that there was no significant difference in the way in which males and females interacted with each other. Kernel and cluster analyses indicated that females would overlap a greater proportion of another bird�s home range than males would. Cluster analysis also indicated that a female would have more of her home range occupied by another bird than a male would. The fact that different methods produced different quantitative results is an important consideration when using home range analysis to make conservation management decisions. Researchers must determine which method is the most appropriate for a particular research objective, species, or study area. The application of geographical information systems, ERDAS image classification techniques and global positioning systems was an integral part of this study. A large-scale vegetation classification map of Pearl Island was produced in order to quantify habitat selection by kakapo. The unsupervised classification technique produced the least accurate vegetation map, with an accuracy measure of 17-23%, compared to 52% for the supervised classification. The highest accuracy was obtained using an integrated approach involving inductive classification and deductive mapping, resulting in a vegetation classification map which correctly classified 95% of vegetation samples. Thirty-seven ecotone classes were identified and a total ecotone length of approximately 124 km was detected. Resource selection ratios and resource selection functions were estimated using a combination of discrete, continuous and area-based habitat variables. Circular buffers around used and available point locations were generated to determine whether kakapo selectively use vegetation mosaics. The probability of selection increased with increasing species diversity in each 75-metre radius buffer. Kakapo selected habitat mosaics and vegetation types with higher species diversity and moderate to high abundance of mature rimu and yellow silver pine trees.

Understanding the relationship between organisms and their environment is particularly important for the conservation and management of endangered species. The kakapo (Strigops habroptilus, Gray 1845) is a critically endangered, lek breeding, flightless nocturnal parrot endemic to New Zealand. In April 1998, a total population of fifty-six kakapo was known to survive on offshore islands. Twenty-six kakapo, thirteen males and thirteen females, were temporarily transferred to Pearl Island (518 ha), southern Stewart Island, from April 1998 to April 1999. The translocation of kakapo to Pearl Island, and subsequent breeding season, provided an ideal experimental framework to study kakapo dispersal, movement patterns, home range development, habitat selection, and lek development during the non-breeding and breeding seasons. A total of 4425 radio locations were analysed for all twenty-six birds, with a mean error polygon of 0.03 ha and an estimated average radio telemetry error of 21.6 m. Various home range analysis techniques were used to estimate kakapo home range size and overlap including: minimum convex polygons (MCP), modified minimum convex polygons (MMCP), harmonic mean analysis, adaptive kernel methods and cluster analysis. Estimates of kakapo home range size differed significantly depending on the method used (ANOVA, general linear model: F₁₃, ₁₀₇₆ = 63.99, p < 0.0001) and the season (F₂, ₁₀₇₆ = 160.75, p < 0.0001). Breeding home range size was significantly larger than non-breeding range size (mean difference = 67.6 ha, t₂₅ = 15.27, p < 0.0001). Calculations from 100% MCP and 95% harmonic mean analysis resulted in larger estimates of home range size and overlap compared to other methods. Cluster and kernel analyses appeared to give the most accurate home range representation for kakapo. Core home range areas showed a greater degree of similarity between methods. Male and female mean annual home range size did not differ significantly, whereas males had significantly (p < 0.05) larger home ranges than females during the nonbreeding season. Minimum convex polygons and harmonic mean analysis suggested that there was no significant difference in the way in which males and females interacted with each other. Kernel and cluster analyses indicated that females would overlap a greater proportion of another bird�s home range than males would. Cluster analysis also indicated that a female would have more of her home range occupied by another bird than a male would. The fact that different methods produced different quantitative results is an important consideration when using home range analysis to make conservation management decisions. Researchers must determine which method is the most appropriate for a particular research objective, species, or study area. The application of geographical information systems, ERDAS image classification techniques and global positioning systems was an integral part of this study. A large-scale vegetation classification map of Pearl Island was produced in order to quantify habitat selection by kakapo. The unsupervised classification technique produced the least accurate vegetation map, with an accuracy measure of 17-23%, compared to 52% for the supervised classification. The highest accuracy was obtained using an integrated approach involving inductive classification and deductive mapping, resulting in a vegetation classification map which correctly classified 95% of vegetation samples. Thirty-seven ecotone classes were identified and a total ecotone length of approximately 124 km was detected. Resource selection ratios and resource selection functions were estimated using a combination of discrete, continuous and area-based habitat variables. Circular buffers around used and available point locations were generated to determine whether kakapo selectively use vegetation mosaics. The probability of selection increased with increasing species diversity in each 75-metre radius buffer. Kakapo selected habitat mosaics and vegetation types with higher species diversity and moderate to high abundance of mature rimu and yellow silver pine trees.