Rabbits to control cats: open landscape conservation of reintroduced western quolls

Invasive alien predators are causing havoc on Australian ecosystems. One species causing especial damage due to their adaptability is the feral cat (Felis catus). Cats are considered to have contributed to the extinction of 22 species of small mammals on the Australian continent and are threatening another 75 species. Additional to the threat of predation, they directly affect native animals through resource competition and disease transmission. The devastating impact of feral cats has led conservation managers to implement a multitude of methods with the intention to minimise their impact. Conventional methods such as regulating their population through poison baiting or by shooting are inefficient at large scales, though, due to the high reproductive and movement capability of cats. Cats have been known to quickly reinvade from surrounding areas, repopulating the treated area. Other methods to conserve threatened species and their ecosystems include the establishment of predator-free islands and fenced reserves, of which a multitude have been established across the Australian continent. The high maintenance costs of fences, however, make predator exclusion fencing unfeasible at larger scales. Furthermore, protection in these fenced areas focusses on a few selected animal species and is limited to the population number that can be contained in the area.
If conservation on the Australian continent is to be successful in reducing the impact of feral cats, solutions need to be found that are effective beyond fenced reserves and islands. Such open landscape conservation has challenges of its own. A deep understanding of the ecological processes, landscape dynamics and species interactions of the landscape to be managed is necessary as animal movements and therefore reinvasion of feral predators cannot be contained or controlled. With detailed knowledge of habitat preferences, movement patterns and animal abundances, ecological levers can be identified and tested to tilt the system in favour of native animals. Testing and subsequently improving these methods in open landscapes will increase the understanding of the underlying processes that drive the dynamics and interactions between animals and their environment.
In this thesis, I investigate the interactions amongst feral cats, European rabbits (Oryctolagus cuniculus) and re-introduced western quolls (Dasyurus geoffroyi) in the Flinders Ranges in South Australia, particularly in the 950km2 Ikara-Flinders Ranges National Park. The study area is part of an open landscape conservation project extending over multiple National Parks, private lands and private protected areas. The aims of this study are twofold: 1) to gain a better knowledge of the ecological processes driving current patterns of distribution and abundance of rabbits, cats, and quolls; and 2) to test an ecologically based method applicable to open landscapes to reduce the predatory impact of feral cats on native fauna including the quolls. I therefore apply several methods to extend current knowledge of the habitat preferences, movement patterns and abundances of these three species. The results of this research are tailored to guide conservation management to a more cost-efficient approach and to build a foundation for future research to enhance the scale and benefit of this open landscape method for wildlife conservation and ecological restoration.
In the first chapter, we report on long-distance movements made by feral cats in the arid and semi-arid zones of Australia, based on data from cats in this study, and other published and unpublished studies. That only a fraction of the GPS-VHF collared cats showed this behaviour raises the question of just how common this behaviour is occurring. Reading through the literature of GPS collaring studies, a reasonable proportion of feral cats that are collared are either lost or their data excluded from analysis because their wide-ranging movement pattern is considered an outlier that will distort the results. Large displacements and long-distance movements could therefore be a regular occurrence and an underestimated part of the behaviour of feral cats. We highlight this issue as a reminder for conservation managers and investigators working with cats as it shows the capability of these animals to cover huge distances in the search for suitable habitat and to re-invade managed areas.
In the second chapter, we map the distribution of European rabbits in relation to biotic and abiotic features, using a novel transect method developed for ocean floor surveys and applying it for the first time in a terrestrial environment. We surveyed an area of 650 km2 using 95 transects (1.4 km length by 50 m width) that were placed randomly but spatially balanced across the survey area. Using this method, we can predict the number of rabbit holes with up to 96.4% accuracy, providing a significant improvement in detail and resolution to previously available maps. Our results show that the presence of rabbit warrens is influenced by geology, vegetation cover, and the topographic wetness index, whereas the number of holes per warren depends on the ruggedness of the terrain, the vegetation cover and whether the warren has previously been destroyed during pest management. With this survey design, we were able to create a high-resolution map of the distribution of this feral herbivore across the entire survey area and simultaneously reduce the effort necessary to obtain this detailed information. This map can now be used to direct conservation management and provides a base environmental layer to include in the analysis of cat and quoll habitat preferences in the upcoming chapter.
In the third chapter, we use data collected by GPS/ VHF collars to investigate the movement dynamics and habitat selection of feral cats and western quolls in relation to rabbit abundance and other biotic and abiotic factors across an ecological gradient, from rocky ranges over forested hills to open and flat grassland. We show that there are differences in the patterns of movement and habitat preferences between the two predator species suggesting that they possibly occupy distinct niches. Cats select Shale geology, grassland, and higher rabbit densities and select against Siltstone, forest, and very low rabbit densities, whereas quolls select Siltstone layers, forest, and very low rabbit densities, and select against Limestone and Shale layers, grasslands, and higher rabbit densities. There is very little to no overlap between the two species in individual core home ranges, resulting in almost complete separation in space use in the landscape. We discuss several alternative hypotheses to explain these patterns: 1) do quolls and cats have fundamentally different habitat requirements, 2) is quoll occupancy restricted by predatory or competitive exclusion by cats, 3) are quolls restricted to refuge habitats that provide higher moisture and vegetation cover, in a landscape denuded by over-grazing through high densities of herbivores, particularly during periods of drought, 4) would a combination of high densities of herbivores, including rabbits supporting elevated numbers of cats, limit quolls to a relictual niche in the longer term, and 5) is the current distribution of cats and quolls a result of the momentary population densities? To test whether quolls are limited by cats, removing cats from patches of landscape could indicate whether quolls can use these spaces. As removing cats entirely from certain parts in open landscape is difficult due to their capability to move long distances and re-invade landscapes, we first need to test whether it is possible to change the movement and local activity of cats with management interventions.
In the last chapter, we experimentally test whether we can reduce the local activity of feral cats in the landscape by reducing the number and density of rabbits. We test three different methods of rabbit management: shooting of rabbits, the release of the calicivirus that causes Rabbit Haemorrhagic Disease (RHD), and the destruction of rabbit warrens by ripping. We deployed arrays of remote cameras in a before-after control-impact design, with the three treatments and four control sites, collecting data for three months before and after the manipulations. We show that ripping is the only method that achieved the desired results. By ripping large areas (about 5km2) of rabbit warrens, we were able to cause the cats present in the treated areas to move away, thus reducing local levels of cat activity. In the timescale of our study, we recorded cats moving into adjacent areas of high rabbit abundance. This demonstrates that it is possible to shift the activity centres of feral cats in open landscapes. This result provides a proof-of-concept that removing rabbits by mechanical destruction of warrens is an effective method to reduce local cat activity and represents a valuable step forward for conservation management in large open landscapes.
In conclusion, this body of work shows that given detailed knowledge of the habitat preferences and movement behaviours of interacting species in an ecosystem, we can identify and design ecologically based management actions that might feasibly shift a system in favour of the species of conservation interest. These are important findings that open up new solutions for open landscape conservation in the semi-arid and arid zone of Australia.