Habitat selection by feral cats (Felis catus) and three rat species (Rattus spp.) on Stewart Island (Rakiura) and their impacts on native birds

Populations of many native bird species on Stewart Island are in decline, despite the absence of mustelids, which are known bird predators on the mainland of New Zealand. Other bird predators, principally ship rats, possums, and feral cats, are therefore likely to be responsible for the extinction and reduction in numbers of native bird populations on Stewart Island. Feral cats, as efficient predators of ground nesting birds, are thought to be the principal reason for the steep decline in numbers of the endangered southern New Zealand dotterel, that nest in alpine heath on the mountain tops of Stewart Island. Feral cats are being controlled to reduce predation and thus increase the numbers of dotterels. To improve the current control of cats, the habitat selection of feral cats on Stewart Island was investigated. Cats are known to select habitat depending on where their prey is located, so trapping of three rat species, which are cats' principal prey, was carried out in four forest types to examine rats' habitat selection and seasonal changes in relative abundance. Ship rats were found in all forest types, and were numerically dominant in two. Norway rats dominated in subalpine shrubland and Pacific rats dominated in Leptospermum scoparium shrubland. Overall, rat numbers fluctuated seasonally, with a low in relative abundance occurring over the late summer and early autumn. When rat abundance was reduced cats were more likely to leave established home ranges or die, probably through starvation. As rats became less abundant cats did not apparently 'prey-switch' to birds, as secondary prey, but cats did eat proportionally more birds by weight as rat abundance declined. Rats formed 81 % of cats' diet by weight and seasonal depressions in rat abundance every year were limiting cat numbers. Cats were radio-collared and radio-tracked to investigate home range sizes and habitat selection. The home ranges of cats on Stewart Island, measured using the minimum convex polygon method, were the largest recorded for females and the second largest recorded for males. The large home range sizes were probably due to the seasonal depressions in primary prey abundance. Habitat selection by cats was measured using the kernel method. Cats used podocarp-broadleaf forest more than was expected by availability, and used subalpine shrubland significantly less than podocarp-broadleaf forest. Selection of the forest types by cats was influenced by the need for shelter from wet weather. Similarly, it appears that dotterel recruitment is adversely affected in dotterel breeding seasons with very wet weather, despite the apparent success of the current cat control programme in reducing predation of adult dotterels. Habitat selection of the New Zealand robin was thought to be influenced by known nest predators, such as ship rats and possums, on Stewart Island, and was investigated using trapping records and artificial nest predation experiments in three forest types. Ship rats and possums were trapped significantly less in Leptospermum shrubland than in podocarp-broadleaf forest, and more artificial nests were successful in Leptospennum shrubland than the other forest types in spring, and more successful than artificial nests in podocarp-broadleaf forest in summer. These results support the theory that robins are restricted to Leptospermum shrubland by the habitat preferences of nest predators. As a result of the conclusions from this thesis it is suggested that control of cats continue in the subalpine shrubland, until large-scale control of cats is possible. Any large-scale control would need to be done in concert with rat control due to the probable problems of mesa-predator release, of rats, for nesting birds. Large-scale cat control would also be more effective if it was carried out in lowland forest, and in the summer, when cats' prey abundance is low. Monitoring of dotterel recruitment, in relation to seasonal weather means, requires more investigation, as does the population dynamics of rats and cats during podocarp mast-seed years, the status of ground-nesting birds on Stewart Island, and the reasons for the observed distribution of robins.