Context. Night parrots (Pezoporus occidentalis) are one of Australia’s most endangered birds, and there is evidence suggesting feral cats (Felis catus) are a major cause of decline. However, because night parrots currently have a restricted distribution, little is known of the ecology of feral cats around their remaining populations. This limits the development of effective management strategies.

Aims. The aims of this study were to understand feral cat movement and habitat selection around night parrots, and to then estimate the effectiveness of possible management actions.

Methods. Research was conducted around the only confirmed night parrot population in eastern Australia. In 2019 and 2020, we obtained GPS data from nine feral cats, and used step selection functions to assess preferred habitats. Management options were then simulated based on cat movement data, including altering trap numbers and layout, and changing routes for night spotlight shooting (using existing roads, random walking or creating new roads in preferred habitats).

Key results. Feral cats preferred alluvial and riparian habitats and avoided rocky woodlands and roads. Simulated control efforts were more successful if traps are placed at ‘pinch points’ where drainage lines converged, and if new roads were created near to creek lines and alluvial habitats.

Conclusions. Feral cats move around the last known population of night parrots in eastern Australia, travelling through and using many shared habitats. Targeting creek lines and alluvial areas in cat control operations would improve effectiveness and potentially reduce predation impacts on night parrots.

Implications. Conservation of endangered birds like night parrots can be enhanced through understanding the ecology of threats such as feral cats to develop locally tailored control operations.

Context. Recent studies have shown that the sensitivity of wild house mice to zinc phosphide (ZnP) in Australia is significantly lower than previously assumed, which may account for the reported variability in efficacy of ZnP baits used for broadacre control of house mice in grain-growing regions. Under laboratory conditions ZnP-coated grains with a new higher dose (50 g ZnP/kg grain) were readily consumed but the efficacy of using grains with this higher dose under natural field conditions has not been tested.

Aims. To test whether the newly derived ZnP50 (50 g ZnP/kg grain) was more effective under field conditions than the currently registered ZnP25 (25 g ZnP/kg grain) in reducing populations of house mice during a mouse population irruption.

Methods. We used a before–after-control–impact (BACI) design to assess changes in mouse population size under different baiting treatments in a replicated field trial. We assessed changes in mouse abundance in recently sown paddocks with either ZnP50 (n = 3) or ZnP25 (n = 3) compared with unbaited control sites (n = 3).

Key results. Baiting with ZnP50 led to a median reduction in mouse numbers of >85%. Our modelling showed that under similar circumstances, using the ZnP50 formulation should deliver >80% reduction in population size most (>90%) of the time. In contrast, the current registered bait (ZnP25) achieved approximately 70% reduction in population size, but with more variable results. We would be confident of getting an 80% reduction in population size only 20% of the time by using the currently registered ZnP25 bait under similar field conditions.

Conclusions. Consistent with laboratory studies, this study demonstrated the higher probability of achieving a consistently high kill rate under field conditions with the new ZnP50 bait compared with the currently registered formulation (ZnP25).

Implications. By using the new ZnP50 bait, farmers are far more likely to get good kill rates, thereby reducing the need for repeated baiting (which is costly and generally ineffective at protecting newly sown crops). Using the new bait should result in lower control costs for farmers and fewer toxic grains being spread to control mice.

Context. House mice (Mus musculus) are the main drivers of biodiversity declines on Gough Island (6500 ha; 40°21′S, 009°53′W), central South Atlantic. A mouse eradication operation was planned, the largest global attempt targeting only this species. Understanding and managing challenges of operating at such scales are crucial for maximising the chance of eradication success. The Gough Island mouse eradication attempt was implemented between June and August 2021, after years of planning and trials. We expected poor weather and negligible non-target bait consumption.

Aims. We aimed to assess the impact of expected and unexpected challenges faced during the eradication operation on Gough Island, namely poor weather and rapid bait disappearance.

Methods. We set up bait degradation plots across the primary habitats to monitor the impact of expected heavy rain on bait pellets. In contrast, bait availability monitoring and slug laboratory trials were set up ad hoc in response to unexpected observations of high bait consumption by invasive slugs in the lowlands, where both slugs and mice are more abundant.

Key results. Bait degradation rates were very different between the highlands and the lowlands, with bait in the highlands lasting about six times longer, despite bait pellets receiving more precipitation and the highlands being persistently under cloud. Bait availability in the lowlands dropped by >80% within a few days of the second and third bait application, down to critically low levels (~2 kg ha⁻¹). Importantly, mouse activity was negligible by this time. Non-native slugs appeared to be the main cause of such a sudden drop in rodent bait availability.

Conclusions. The expected rainy weather was not a significant direct cause of bait degradation in the short term. In contrast, the unexpected slug interference, overlooked in earlier planning trials, resulted in major adjustments of the baiting strategy. Indeed, the rapid bait disappearance in the lowlands triggered the third bait application over this area, at a higher rate. This was not enough, as mice are still present.

Implications. This is the first report of slug interference during aerial rodent eradications. Our results illustrate how interference by non-target species could affect future pest eradications using baits and should, as far as possible, be assessed early during planning.

Context. The Felixer grooming device (‘Felixer’) is a lethal method of feral cat control designed to be cost-effective and target specific.

Aims. This study aims to test the target specificity of the Felixer in Tasmania, with a particular focus on Tasmanian devil and quoll species due to the overlap in size, habitats and behaviour between these native carnivores and feral cats.

Methods. Our study deployed Felixer devices set in a non-lethal mode in nine field sites in Tasmania, one field site in New South Wales and two Tasmanian wildlife sanctuaries.

Key results. Our study recorded 4376 passes by identifiable vertebrate species including 528 Tasmanian devil passes, 507 spotted-tailed quoll passes and 154 eastern quoll passes. Our data showed that the Felixer can successfully differentiate quoll species from feral cats with spotted-tailed quolls and eastern quolls targeted in 0.19% and 0% of passes, respectively. However, Tasmanian devils and common wombats were targeted in 23.10% and 12% of passes, respectively, although sample size was low for common wombats (n = 25).

Conclusions. The Felixer could not reliably identify Tasmanian devils and possibly common wombats as non-target species. Further data is needed to confirm the potential for impacts on the common wombat and other potential non-target species in Tasmania, and the likelihood of the toxin being ingested by falsely targeted individuals.

Implications. Our study suggest that the Felixer device is safe for use in the presence of two species of conservation concern, the eastern and spotted-tailed quoll. It also supports evidence from previous studies that the Felixer is unlikely to impact bettongs and potoroos. Use of Felixer devices across much of Tasmania would have to balance the conservation or economic benefits of cat control against potential impacts on Tasmanian devils. We suggest that active Felixer deployments be preceded by surveys to establish the range of species present at the control site, and the season of control considered carefully to minimise potential impacts on more susceptible juvenile animals. In addition, modifications to the Felixer device such as the proposed incorporation of AI technology should be tested against the Tasmanian devil and other non-target species.

Context. The introduced red fox has driven the decline or extinction of numerous wildlife species in Australia, yet little information exists on the population densities of foxes in most ecosystems. Fox monitoring programs will differ widely depending on the goals of management, which, in turn, will determine whether the appropriate metric is a density estimate, or some proxy thereof, and the time and resources required.

Aims. This study aims to assist wildlife managers to design fit-for-purpose monitoring programs for foxes by providing a better understanding of the utility and precision of various monitoring methods.

Methods. We surveyed foxes monthly over four consecutive years in a semi-arid region of Australia by using sand plots, camera traps and GPS telemetry. The resultant data were used to produce population estimates from one count-based method, two spatially explicit methods, and two activity indices.

Key results. The incorporation of GPS-collar data into the spatial capture–recapture approaches greatly reduced uncertainty in estimates of abundance. Activity indices from sand plots were generally higher and more variable than were indices derived from camera traps, whereas estimates from N-mixture models appeared to be biased high.

Conclusions. Our study indicated that the Allen–Engeman index derived from camera-trap data provided an accurate reflection of change in the underlying fox density, even as density declined towards zero following introduction of lethal control. This method provides an efficient means to detect large shifts in abundance, whether up or down, which may trigger a change to more laborious, but precise, population monitoring methods. If accuracy is paramount (e.g. for reintroduction programs) spatially explicit methods augmented with GPS data provide robust estimates, albeit at a greater cost in resources and expertise than does an index.

Implications. Our study demonstrated that the shorter the survey period is, the greater is the likelihood that foxes are present but not detected. As such, if limited resources are available, longer monitoring periods conducted less frequently will provide a more accurate reflection of the underlying fox population than do shorter monitoring periods conducted more often.

Context. House mice (Mus musculus) on temperate Gough Island (6500 ha) are known for their large size, boldness, and tendency to kill large prey such as albatross chicks and even adults. To remove this threat, a mouse eradication operation was implemented in June–August 2021. How mice react to bait during eradications is not well understood, so we capitalised on this operation and conducted the first study with wild house mice during an actual eradication.

Aim. To document how rapidly mouse activity declined after application of rodent bait, to improve eradication guidelines.

Methods. We set up a monthly monitoring regime using 10 trail cameras without lures, active for three nights in various habitats around a research station, because this area supported the highest abundance of mice and was logistically feasible. Monitoring commenced before the mouse eradication operation (January–May 2021), and continued when rodent bait was spread (from June 2021), when mouse activity was monitored for 17 consecutive nights, starting the day before baiting. In addition, an increasing number of cameras (up to 15) associated with lures were set further afield in July–August to detect survivors.

Key results. In the months before bait application, mean daily mouse activity was 3.2 detections/camera (range: 0–56 detections/camera). Immediately after the first bait application, detection rates declined dramatically, from 9.6 to zero detections/camera per day on Day 4 post-baiting. From 1 week post-baiting, mouse detections were extremely rare on both cameras with and without lures. Our last mouse record, 27 days after the first bait application, may be related to initial rapid bait disappearance. Opportunistic camera traps first detected surviving mice 6 months after the first bait drop.

Conclusions. The rapid decline in detections suggests that most mice consumed bait as soon as it became available, which is faster than what laboratory trials suggest. Future similar operations can expect that mouse activity will decline sharply within 1 week, although some mice may survive longer.

Implications. Documenting similar declines in mouse activity using cameras could inform operational decisions such as timing of a second bait application or non-target monitoring on future eradication projects. Cameras, particularly with attractive lures, are an effective addition to the mouse detection toolkit, and facilitated a timely confirmation of eradication outcome.

Context. Aerial surveys are widely used for estimating the abundance of wildlife over large areas. The failure to count all animals within survey transects is commonly acknowledged and there are many techniques to measure and correct for underestimation. However, the possibility of animals being counted more than once in intensive surveys, which leads to overestimation, is rarely examined. Animals can move in response to observers or vehicles, and bias can occur when animals move before or after detection. Movement of animals immediately prior to and associated with observation processes is methodologically accommodated in distance sampling but bias attributable to responsive movement after observation platforms have passed requires investigation.

Aims. We sought to investigate potential biases caused by animal movement during intensive helicopter surveys of feral goats, and to quantify the probability that animals are available for recounting because of their responsive movements.

Methods. Using ground-based behavioural studies simultaneous with intensive helicopter strip surveys of feral goats, we measured the extent of responsive movement, distances and directions moved, and sampling design parameters, and contrasted those with random movements.

Key results. Feral goats did not move randomly in response to helicopters. Animals within the transect strips, and therefore potentially visible from the aircraft, were more likely to move than those outside the transect. Considerable responsive movement (flushing) occurred between transects and more animals (64%, n = 448) moved towards unsampled transects than towards transects already sampled. Because of the spatial separation of transects, 21% of goats were available for recounting in adjacent transects, leading to potential overestimation.

Conclusions. Although most extensive surveys of macropods and other wildlife in Australia account for overestimation in their design, surveys that sample intensively and apply valid corrections for undercounting are likely to produce positively biased estimates of abundance where flushing occurs. Likewise, intensive thermal surveys could be subject to positive bias for animals prone to flushing. This is routinely ignored in wildlife management and research where close transects are used to estimate abundance.

Implications. Responsive movement requires consideration when designing intensive aerial surveys of wildlife. Randomised transects without replacement or larger distances between transects will counteract recounting bias.

Context. Long-term monitoring is essential for control and eradication of invasive mammalian predators. Relative abundance indices are increasingly used when assessing population changes. However, indexing assumes constant detectability, when, in fact, it varies depending on numerous factors, including the type and spacing of monitoring devices, seasons, vegetation types, and inter- and intra-specific interactions.

Aims. We studied a population of Pacific rat (Rattus exulans) and examined the influence of vegetation types, seasons and inter-specific interactions on their detection.

Methods. We deployed tracking-tunnels, live-traps, chew cards, and trail cameras in three vegetation types during summer and winter. Detection rates of Pacific rats, mice (Mus musculus), stoats (Mustela erminea) and weka (Gallirallus australis) were calculated and compared among vegetation types, seasons and devices.

Key results. Pacific rats were not detected by any monitoring devices in the farmland, despite their presence in this habitat. In the forest and shrubland, cameras had the highest detection rate among all of the monitoring devices, whereas live-trap and chew cards failed to detect rats. Tracking tunnels detected Pacific rats only in the shrubland. Camera detections of Pacific rats were lower in winter than in summer, and detections were dominated by weka and stoats for both seasons. The seasonal effects may have reflected the population cycle of Pacific rats. Weka and stoats caused substantial physical interference to the tracking tunnels, live-traps and chew cards, which is likely to have deterred Pacific rat interactions through imposing high predation risks.

Conclusions. Cameras were the most effective device at detecting Pacific rats in low density and under high predator disturbances. Tracking tunnels and chew cards that are successful at detecting other Rattus spp. might not be reliable for detecting Pacific rats.

Implications. We recommend using camera monitoring for rodents where population density is low, or predator disturbance is high, and species are identifiable from pictures. Studies that draw inferences from relative abundance indices on rodents should exercise caution when rodent detectability is unknown.