The available science often demonstrates the need for feral horse population control but not the degree of control required to achieve environmental conservation objectives. To better manage the influence of feral horses, we must first understand the relationship between feral horse density and environmental impact. We recorded vegetation and soil disturbance, and the sign of potential causes of this impact in two parts of the Australian Alps, the Bogong high plains (BHP) and the Eastern Victorian Alps (EVA). We calculated density‐impact functions to assist managers with determining feral horse density targets for control programmes. Minimal sign of feral horse impact was detected on the BHP, with no impact of feral horses observed along 99% of the length of transects. In contrast, impacts assigned to feral horses were significantly higher in the EVA, where a larger, higher‐density population of feral horses existed. However, greater than 83% of the walked transect length was still undisturbed by feral horses in the EVA. We detected a threshold of horse impact at ~ 250 horse faecal piles per ha. Above this threshold, a slight increase in horse density resulted in a disproportionately large increase in impact. In this context, a relatively small population control effort may substantially reduce direct horse impact. But where horse densities exist below this threshold, considerably more expense and control effort (resulting from the difficulties related to control at low density) is likely to make very little difference to an already low level of direct impact. The combined impacts associated with the sign of deer, feral pigs, fire and humans were large compared to that of feral horses. Management of feral horses to reduce their direct impact is unlikely to be beneficial without complementary management to reduce the effects of these other agents of impact.

Context European rabbits have a great impact on native vegetation and small vertebrates in Australia. Rabbits consume vegetation and promote invasive plants and invasive predators, and compete directly and indirectly with native animals suppressing those populations. Aims We explored the changes in small native vertebrates and invertebrates following the removal of rabbits. Methods Warren ripping was undertaken on a property in south-western Queensland at four sites and the results of pitfall trapping were compared with four nearby paired control sites. Invertebrates and small mammals were counted in pitfall traps, and bird surveys were conducted in all treatment and control sites. Key results Following a rabbit-control program, we observed a four-fold increase in the number of dunnarts trapped in treatment plots, whereas no change was observed in control plots. The spring following the rabbit-control program also saw an increase in some lizards in treatment plots. Conclusions The presence of rabbits in arid-zone Australia can suppress native animal populations. Implications Many species of small native mammals and lizards rely on food sources that fluctuate greatly with environmental conditions. The presence of rabbits altering the landscape, supporting introduced predators, reducing vegetation and, therefore, insects, adds increased pressure for insectivorous species. Rabbit control through warren ripping in arid-zone Australia is an effective method to reduce rabbit numbers, and allowed for an increase in small vertebrates in treated areas.

Context Understanding population dynamics of invasive species is crucial for the development of management strategies. Feral horses (Equus caballus) are a growing problem in the Tuan–Toolara State Forest (TTSF), a coniferous plantation in south-eastern Queensland, Australia. Aim The population dynamics of the TTSF feral horses was not known. Therefore, the study was designed to characterise the major vital parameters of this population and, using these data, develop a long-term management plan. Methods The study was conducted over 3 years (2011–14) involving 522 individually identified horses. Foaling rates were used to calculate fecundity. Body fat distribution was estimated using body condition score (BCS), which reflects the nutritional, metabolic and general health of individual animals. Multi state mark–capture population models were used to estimate age-specific survival, and the Leslie age-structured projection matrix model was used to determine the annual rate at which the population increased. Key results The mean annual fecundity was low (0.23 ± 0.07 s.d.). The mean BCS of the population was mid-range (2.55 ± 0.51 s.d.) with adult females having lower scores than other age and gender groups. Survival estimates were consistently high (0.92–0.95) across all age groups. The average annual finite rate of population increase (λ) for the 3 years of the study was 1.09. Sensitivity analysis demonstrated that the population growth rate was almost seven times more sensitive to changes in adult survival compared with juvenile survival, and almost twice as sensitive to changes in fecundity. Key conclusions Population dynamics of the TTSF feral horses were comparable to other feral horse populations similarly challenged by environmental nutritional limitations. Implications Defining population dynamics of the TTSF feral horses permits the formulation of management goals that can be audited and adapted as required. The most effective strategy for controlling population growth in the TTSF would involve the continuing removal of substantial numbers of adult females or manipulation of survival and/or fecundity. As selective removal will likely alter the adult sex ratio and age structure of the population, ongoing assessment is necessary to minimise adverse welfare outcomes.

Context Feral horses are a growing problem in Australia, despite implementation of management strategies. The incidence of horse sightings and horse-associated vehicle collisions within the Tuan and Toolara State Forest (TTSF), a coniferous plantation in south-eastern Queensland, has increased in the past decade, indicating an increase in population and the necessity to undertake an accurate appraisal of population abundance. Aim To determine the feasibility of using dung counts, defecation and dung-decay rates to determine the distribution and abundance of feral horse population in a managed coniferous plantation where dense vegetation prohibits the use of visual-based animal detection methods. Methods Population distribution was assessed by dung count incorporated into a vehicle strip-transect survey on 582 km of forest tracks. Population abundance was estimated from dung density, the mean defecation frequency of feral horses, and the mean number of days required for dung to decay. Dung density was obtained from on-ground distance line-transect survey that sampled all representative plantation habitats and consisted of 111 transects totalling 44.3 km. Key results The strip-transect survey clearly showed that although feral horses were dispersed across the plantation, the distribution was uneven, with the central region of the plantation being the most heavily populated. The combination of dung counts, defecation rate (mean ± s.d., 7.97 ± 8.74) over 24 h and dung-decay rate (444 ± 150.7 days) provided an estimation of the density of feral horses in various habitats and indicated that the plantation was occupied by 1321 (95% CI 940–1965) horses, which corresponded to an average density of 1.8 horses km–2. The method clearly identified variations in horse abundance among the various habitats within the surveyed areas. Open habitats, created following harvesting, showed higher occurrence rates, whereas the habitats of mature forest were scantily occupied. Key conclusions Dung counts are a simple, effective and practical technique that can provide information on distribution and abundance of feral horse population in densely forested habitats where visual-based techniques are not applicable. However, unbiased and precise defecation and decay rates must be estimated. Implications The study validated the use of dung counts to provide information on feral horse distribution and abundance in densely forested environments where direct methods of census may be difficult to obtain. The methods are applicable to a range of ecosystems, but defecation and dung-decay rates must be determined separately for each ecosystem.

Engaging stakeholders and encouraging community engagement is fundamental for adoption of model results. Engagement benefits from starting from model development and continuing through to model completion and roll-out. This is especially the case when managing invasive species, where a coordinated management strategy between stakeholders works best. We use a participatory approach that combines expert knowledge and spatial data in a Bayesian network to develop risk maps of potential threats. Of particular relevance is the inclusion of management scenarios that can be manipulated to determine the effects of different strategies and combinations of coordinated strategies. Scenarios are built on expert opinion, guided from years of experience in research and management of the invasive species under field conditions. We capture their knowledge and understanding of the study system allowing us to independently validate how the model and scenarios replicate reality. Stakeholders are trained to run scenarios themselves at field days to encourage community engagement. Model results are presented as maps that visualise risk after applying different management scenarios. Maps facilitate interpretation of model results in a spatial context relevant to local communities. Capturing impacts of species invasions by running scenarios specific to targeted community groups, such as land managers, allow for increased adoption when economic incentives are introduced.We demonstrate this with a case study using the European rabbit (Oryctolagus cuniculus), which is a serious agricultural pest in southern Queensland. We model the economic costs and effects of different rabbit management scenarios with the impacts of varying degrees of rabbit damage on different agricultural commodities. Our model allows end-users to choose the management strategies and the desired agricultural commodity, such as broccoli, spatially limited to the areas of suitable growth across the region. Model results show a distribution of costs for each modelled commodity scenario, giving end-users a range of costs unique to each agricultural region. Using ecological knowledge to understand the species' interaction within the environment and economic drivers to capture impact, our scenario modelling allows for immediate relevance to community users and ease of adoption with its participatory approach.

Context The study of any wild animal’s home range requires the collection of spatiotemporal data, obtained independently of climatic conditions or time of day. This can be achieved by the attachment of global positioning system (GPS) data loggers, which, in large species, is best achieved by remote immobilisation. Feral horses (Equus caballus) usually occupy remote areas of Australia; however, a considerable population increase has been observed in a close proximity to metropolitan areas of the Australian east coast, creating increasing conflict with human interests. Aim The aim of the present study was to investigate the efficacy of remote chemical immobilisation of feral horses with medetomidine combined with tiletamine–zolazepam to facilitate placement of satellite GPS collars. Methods Nine feral horses were darted from the ground with 60 mg (i.m.) medetomidine and 1500 mg (i.m.) tiletamine–zolazepam. The effects of medetomidine were reversed with 50–100 mg (i.m. or i.v.) atipamezole 30–40 min after induction (IV/IM). Physiological variables monitored during anaesthesia were heart rate, respiratory rate, temperature and oxygen haemoglobin saturation (Spo2). Key results All horses were successfully immobilised with between one and three darts (n = 9). The mean (± s.e.m.) dose of medetomidine was 0.15 ± 0.01 mg kg–1, whereas that of tiletamine–zolazepam was 3.61 ± 0.16 mg kg–1. Mean time from darting to lateral recumbency was 13.3 ± 2.7 min and mean recumbency time was 54 ± 13 min. Vital signs for all anaesthetised animals remained within the normal range during anaesthesia, with the exception of one animal exhibiting a transient drop in Spo2. There were no deaths. Key conclusions The combination of medetomidine and tiletamine–zolazepam provided adequate anaesthesia in feral horses in the field for application of GPS collars. Implications Although a limited number of horses was immobilised, the present study shows that the combination of medetomidine and tiletamine–zolazepam provides effective short-term anaesthesia for feral horses, affording a practical and field-accessible capture technique. This method could also be applied to other management actions requiring the safe and humane capture of feral horses.

Extensive resources are allocated to managing vertebrate pests, yet spatial understanding of pest threats, and how they respond to management, is limited at the regional scale where much decision-making is undertaken. We provide regional-scale spatial models and management guidance for European rabbits (Oryctolagus cuniculus) in a 260,791 km(2) region in Australia by determining habitat suitability, habitat susceptibility and the effects of the primary rabbit management options (barrier fence, shooting and baiting and warren ripping) or changing predation or disease control levels. A participatory modelling approach was used to develop a Bayesian network which captured the main drivers of suitability and spread, which in turn was linked spatially to develop high resolution risk maps. Policy-makers, rabbit managers and technical experts were responsible for defining the questions the model needed to address, and for subsequently developing and parameterising the model. Habitat suitability was determined by conditions required for warren-building and by above-ground requirements, such as food and harbour, and habitat susceptibility by the distance from current distributions, habitat suitability, and the costs of traversing habitats of different quality. At least one-third of the region had a high probability of being highly suitable (support high rabbit densities), with the model supported by validation. Habitat susceptibility was largely restricted by the current known rabbit distribution. Warren ripping was the most effective control option as warrens were considered essential for rabbit persistence. The anticipated increase in disease resistance was predicted to increase the probability of moderately suitable habitat becoming highly suitable, but not increase the at-risk area. We demonstrate that it is possible to build spatial models to guide regional-level management of vertebrate pests which use the best available knowledge and capture fine spatial-scale processes.

Increasing resistance of rabbits to myxomatosis in Australia has led to the exploration of Rabbit Haemorrhagic Disease, also called Rabbit Calicivirus Disease (RCD) as a possible control agent. While the initial spread of RCD in Australia resulted in widespread rabbit mortality in affected areas, the possible population dynamic effects of RCD and myxomatosis operating within the same system have not been properly explored. Here we present early mathematical modelling examining the interaction between the two diseases. In this study we use a deterministic compartment model, based on the classical SIR model in infectious disease modelling. We consider, here, only a single strain of myxomatosis and RCD and neglect latent periods. We also include logistic population growth, with the inclusion of seasonal birth rates. We assume there is no cross-immunity due to either disease. The mathematical model allows for the possibility of both diseases to be simultaneously present in an individual, although results are also presented for the case where co-infection is not possible, since co-infection is thought to be rare and questions exist as to whether it can occur. The simulation results of this investigation show that it is a crucial issue and should be part of future field studies. A single simultaneous outbreak of RCD and myxomatosis was simulated, while ignoring natural births and deaths, appropriate for a short timescale of 20 days. Simultaneous outbreaks may be more common in Queensland. For the case where co-infection is not possible we find that the simultaneous presence of myxomatosis in the population suppresses the prevalence of RCD, compared to an outbreak of RCD with no outbreak of myxomatosis, and thus leads to a less effective control of the population. The reason for this is that infection with myxomatosis removes potentially susceptible rabbits from the possibility of infection with RCD (like a vaccination effect). We found that the reduction in the maximum prevalence of RCD was approximately 30% for an initial prevalence of 20% of myxomatosis, for the case where there was no simultaneous outbreak of myxomatosis, but the peak prevalence was only 15% when there was a simultaneous outbreak of myxomatosis. However, this maximum reduction will depend on other param- eter values chosen. When co-infection is allowed then this suppression effect does occur but to a lesser degree. This is because the rabbits infected with both diseases reduces the prevalence of myxomatosis. We also simulated multiple outbreaks over a longer timescale of 10 years, including natural population growth rates, with seasonal birth rates and density dependent (logistic) death rates. This shows how both diseases interact with each other and with population growth. Here we obtain sustained outbreaks occurring approximately every two years for the case of a simultaneous outbreak of both diseases but without simultaneous co-infection, with the prevalence varying from 0.1 to 0.5. Without myxomatosis present then the simulation predicts RCD dies out quickly without further introduction from elsewhere. With the possibility of simultaneous co-infection of rabbits, sustained outbreaks are possible but then the outbreaks are less severe and more frequent (approximately yearly). While further model development is needed, our work to date suggests that: 1) the diseases are likely to interact via their impacts on rabbit abundance levels, and 2) introduction of RCD can suppress myx- omatosis prevalence. We recommend that further modelling in conjunction with field studies be carried out to further investigate how these two diseases interact in the population.

Context. For over 100 years, control efforts have been unable to stop rabbits causing damage to cattle production and native plants and animals on large properties in arid parts of Australia. Warren destruction by ripping has shown promise, but doubts about long-term success and the perceived expense of treating vast areas have led to this technique not being commonly used.Aims. This study measured the long-term reduction in rabbit activity and calculated the potential cost saving associated with treating just the areas where rabbits are believed to survive drought. We also considered whether ripping should be used in a full-scale rabbit control program on a property where rabbits have been exceptionally resilient to the influence of biological and other control measures.Methods. Rabbits were counted along spotlight transects before warrens were ripped and during the two years after ripping, in treated and untreated plots. Rabbit activity was recorded to determine the immediate and long-term impact of ripping, up to seven years after treatment. The costs of ripping warrens within different distances from drought refuge areas were calculated.Key results. Destroying rabbit warrens by ripping caused an immediate reduction in rabbit activity and there were still 98% fewer rabbits counted by spotlight in ripped plots five months after ripping. Seven years after ripping no active warrens were found in ripped plots, whereas 57% of warrens in unripped plots showed signs of rabbit activity. The cost of ripping only the areas where rabbits were likely to seek refuge from drought was calculated to be less than 4% of the cost of ripping all warrens on the property.Conclusions. Destroying rabbit warrens by ripping is a very effective way of reducing rabbit numbers on large properties in arid Australia. Ripping should commence in areas used by rabbits to survive drought. It is possible that no further ripping will be required.Implications. Strategic destruction of warrens in drought refuge areas could provide an alternative to biological control for managing rabbits on large properties in the Australian arid zone.

This datasheet on Oryctolagus cuniculus covers Identity, Overview, Distribution, Dispersal, Biology & Ecology, Impacts, Uses, Prevention/Control, Further Information.

Wild European rabbits are a serious problem to agriculture in Australia, with an estimated annual cost of A$ 113 million. Biological control agents (myxomatosis and rabbit haemorrhagic disease virus) have caused large and sustained declines in rabbit populations throughout Australia. A simulation model incorporates these diseases as well as warren destruction as methods of controlling rabbit populations in Queensland, north eastern Australia. These diseases reduced populations by 90-99% and the combination of these and warren destruction led to 100% control in simulations at six sites across southern Queensland. increasing monthly pasture growth by 15% had little effect on simulated populations whereas a 15% decrease reduced populations by 0-50%. An increase in temperature of 2.5 degrees C would lead to a 15-60% decrease in populations. These effects suggest that climate change will lead to a decrease in the population of rabbits in Queensland and a retraction in the northern limit of their distribution in Australia.

Release of virulent myxoma virus has been a key component of rabbit-control operations in Queensland, Australia, since the 1960s but its use rests on anecdotal reports. During a routine operation to release virulent myxoma virus we found no evidence to support the continued regular use of the technique in south-west Queensland. Radio-tagged rabbits inoculated with virulent myxoma virus contracted the disease but failed to pass enough virus to other rabbits to spread the disease. Rabbits with clinical signs of myxomatosis that were shot were infected with field strain derived from the original laboratory strain released in 1950 rather than the virulent strain that has been released annually. There was no change in rabbit survival or abundance caused by the release. Nevertheless, the release of virulent virus may be useful against isolated pockets of rabbits mainly because field strains are less likely to be present. Such pockets are more common now that rabbit haemorrhagic disease virus is established in Queensland.

Rabbit haemorrhagic disease virus (RHDV) escaped from quarantine facilities on Wardang Island in September 1995 and spread through South Australia to Queensland by December 1995. To determine the impact of this biological control agent on wild rabbit populations in Queensland, shot sample and spotlight count data were collected at six sites. RHDV spread across Queensland from the south-west to the east at a rate of at least 91 km month–1 between October 1995 and October 1996. The initial impact on rabbit density appeared highly variable, with an increase of 81% (255 ± 79 (s.e.) to 385 ± 73 rabbits km–2) at one site and a decrease of 83% (129 ± 27 to 22 ± 18 rabbits km–2) at another during the first outbreak. However, after 30 months of RHDV activity, counts were at least 90% below counts conducted before RHDV arrived. Using a population model to account for environmental conditions, the mean suppression of rabbit density caused by rabbit haemorrhagic disease (RHD) was estimated to be 74% (ranging from 43% to 94% between sites). No outbreaks were observed when the density of susceptible rabbits was lower than 12 km–2. Where rabbit density remains low for long periods RHDV may not persist. This is perhaps most likely to occur in the isolated populations towards the northern edge of the range of rabbits in Australia. RHDV may have to be reintroduced into these populations. Further south in areas more suitable for rabbits, RHDV is more likely to persist, resulting in a high density of immune rabbits. In such areas conventional control techniques may be more important to enhance the influence of RHD.