Introduction

The damage caused to island ecosystems by invasive species is well documented (Angel et al. 2009; Banks and Hughes 2012; Shiels et al. 2014; Towns et al. 2006). Rodents, and rats in particular, are among the most harmful, causing a multitude of extinctions and suppressions across a range of taxa, worldwide (Atkinson 1985; Auld et al. 2010; Campbell and Atkinson 2002; Cuthbert and Hilton 2004; Pender et al. 2013; Shaw et al. 2005; Shiels and Drake 2011; Smith et al. 2002). Because of these impacts, conservation practitioners have increasingly used eradication of invasive rodents as an important tool in restoring degraded island ecosystems. There have been more than 700 successful rodent eradications on islands worldwide and the benefits are demonstrated in the often-remarkable recovery of flora and fauna that follows rodent removal (Croll et al. 2016; Le Corre et al. 2015; Marie et al. 2014; St Clair et al. 2011).

Lord Howe Island, located off the east coast of Australia, is a World Heritage Area boasting outstanding natural heritage values, including a suite of endemic flora and fauna species. The island has a permanent population of approximately 350 people and is a well-known and popular tourist destination, with up to 400 visitors on the island at any one time. Ship rats (Rattus rattus) were introduced to the island in 1918 when the supply ship SS Makambo ran aground, allowing rats to come ashore (McCulloch 1921). Numerous detrimental impacts soon followed, with the loss of five endemic bird taxa (Hindwood 1938) and the near extinction of the Lord Howe phasmid (Dryococelus australis) (Priddel et al. 2003) among the most notable. Rats also likely contributed to the extinction of an endemic shrub (Solanum bauerianum) (Auld and Hutton 2004), and consumed the fruits of two endemic palm species, increasing their extinction risk (Auld et al. 2010). House mice (Mus musculus) were introduced sometime earlier than rats and while their impacts are less well documented, this island is unlikely to be immune from the destructiveness that mice are known to cause elsewhere (Angel et al. 2009; Cuthbert and Hilton 2004; Smith et al. 2002).

Brodifacoum, a second-generation anticoagulant, is highly toxic to rodents (Redfern et al. 1976) and has been used effectively in many successful rodent eradications (Parkes et al. 2011). However, because brodifacoum lacks specificity (Eason and Spurr 1995; Rattner and Mastrota 2018) many other animals, particularly mammals and birds, are also vulnerable to poisoning during baiting operations (Hoare and Hare 2006). Some losses of non-target species are expected and unavoidable during eradications (Innes and Barker 1999), and following successful removal of rodents, non-target species generally recover rapidly, often exceeding pre-eradication population levels (Croll et al. 2016). Thus, because impacts are only temporary and the potential benefits are ongoing, some non-target losses may be viewed as an acceptable consequence of island restoration. However, where there are particularly vulnerable species or species of conservation concern, large numbers of non-target losses may be unacceptable (Walsh et al. 2019). Thus, a key challenge faced by eradication managers is to develop mitigation actions to minimise impacts on non-targets while not jeopardising the main goal of eradicating the target invasive species. Moreover, a recent review by Segal et al. (2021) found that less than 13% of eradication projects have reported findings on non-target mortality. Therefore, there is a need for greater understanding of the impacts of eradications on non-target wildlife.

In the latter half of 2019, a rodent eradication program was carried out on Lord Howe Island. The operation involved the use of cereal bait containing brodifacoum distributed across the entire island through a combination of aerial dispersal via helicopter, hand broadcasting and bait stations. This program was the largest of its kind on an inhabited island anywhere in the world and thus attracted considerable attention from conservation managers and the media.

Due to the high conservation value of many non-targets on the island, and because of New South Wales and Australian government regulatory requirements, a comprehensive Non-target Species Monitoring and Mitigation Plan was developed for all endemic and some non-endemic native bird species. This paper describes the implementation of that plan during the eradication. The objective of the plan was to identify at-risk species, employ mitigation measures to protect those species, to monitor deaths of free-living non-targets during the eradication, and to assess potential impacts on non-targets through population surveys before and after the eradication. The overall goal for the eradication was to have no long-term impact on any non-target species at the population level. A key outcome in achieving this goal was that post-baiting populations should be equal to or exceed the pre-baiting population within two years following the eradication.

Location description

Lord Howe Island (LHI) is in the South Pacific approximately 780 km north-east of Sydney, Australia. The crescent-shaped island covers nearly 1400 ha, is 12 km long, and varies in width from 1.0 to 2.8 km. A coral reef encloses a lagoon on the western side. Mount Gower (875 m), Mount Lidgbird (777 m) and Intermediate Hill (250 m) form the southern two-thirds of the island; 200-m high sea cliffs border the northern end. The permanent settlement covers about 15% of the island and associated grazing land a further 10%. Around 75% of LHI together with all outlying islands, islets and rocks are protected under a Permanent Park Preserve, which has similar status to that of a national park. The park preserve on the main island comprises a range of vegetation types and is dominated by oceanic rainforests and palm forest. The top of Mount Gower is vegetated with Gnarled Mossy Cloud Forest, listed as a Critically Endangered Ecological Community under the NSW Biodiversity Conservation Act 2016 (NSW Department of Planning Industry and Environment 2019). The park preserve is intersected by a number of public access walking tracks and a network of less well-defined routes used for the management of invasive plants. The LHI Group was listed as a World Heritage Area in 1982 and was placed on the National Heritage List in 2007 (Australian Government 2020).

For the eradication, bait (Pestoff 20R®) was applied by three methods: (1) Bait stations (19,000 externally and 3,500 in dwellings) throughout 190 ha of the Settlement Area, (2) Two applications (12 kg/ha and 8 kg/ha) of aerial baiting over the park preserve, and (3) Two applications of hand broadcasting (12 kg/ha and 8 kg/ha) in the buffer zone (163 ha) between the park preserve and the bait station zone (Harper et al. 2020).

Methods

Non-target species

In line with both State and Federal regulatory requirements, a comprehensive assessment was performed prior to the eradication to identify non-target species that would potentially be placed at risk from the baiting program. A previous bait-uptake trial (Department of Environment and Climate Change 2007), that used fluorescing non-toxic bait to identify species that consumed bait, together with a desktop analyses of behaviour and diet of island species were primarily used to make this assessment.

Two endemic bird species were identified as being at potentially high risk and in need of mitigation measures prior to the baiting program—the flightless Lord Howe woodhen (Hypotaenidia sylvestris) and the Lord Howe currawong (Strepera graculina crissalis). The Lord Howe woodhen is listed as an Endangered species on the IUCN Red List of Threatened Species, and under state (NSW) and Australian Legislation; the Lord Howe currawong is listed under NSW legislation as a Vulnerable subspecies of Pied Currawong, which is of Least Concern on the IUCN Red List. The population of both species was estimated to be between 200 and 250 individuals. Woodhen were found to consume bait during the non-toxic bait-uptake trial, either through direct consumption of pellets or through eating invertebrates that had previously eaten bait. The currawong was not observed to consume bait during the trial, but as rodents make up a significant portion of currawong diet (11% in litt.) there was concern that they could eat poisoned rodents and thus suffer secondary poisoning. To mitigate these potential risks, approximately 90% of the woodhen population (based on the 2018 population count of 221 individuals; Portelli and Carlile 2019) and 50% of the currawong population (based on 200–250 individuals from mark-recapture surveys) were taken into captive management during the eradication, where they remained until it was deemed that the potential risk of poisoning had passed. A larger proportion of the woodhens were taken into captivity because they were deemed to be at greater risk than currawongs. While it would have been ideal to capture 100% of the woodhen population, a proportion of the population occurs (albeit in small numbers) in largely inaccessible parts of the island. As only a small proportion of free-living currawongs were expected to die from consuming poisoned rodents, 50% of the existing population was deemed to be a sufficient insurance population to preserve the population’s genetic diversity (Tracy et al. 2011).

A further four species were identified as being at high risk but with high potential for re-introduction if they suffered high impacts from the eradication (Table 1). Thus, it was assessed that captive management was not necessary for these species. They were, however, monitored during the baiting program. Low and medium-risk species, and the free-living population of currawongs, were also monitored during the eradication. The free-living woodhens were not monitored because they occurred in remote areas of the island where regular monitoring was impractical.

Table 1 Non-target species considered in the Monitoring and Mitigation Plan
Full size table

To monitor the number of bush birds killed during the eradication, 20 km of tracks, sampling all sections of the park preserve, were walked in search of dead birds (Fig. 1). A width of 5 m, centred on each track, was searched, giving a total search area of 10 ha (approximately 1% of the forested area of LHI). Searches involved two people walking at a slow pace (approximately 2 km per hour) along the designated tracks with each person searching either the left or right side only. Searches for non-targets were conducted every second day between the first and second bait drops and every second day for 3 weeks after the completion of the second bait drop. Searches were then conducted weekly for 14 weeks, based on a study of bait breakdown (Department of Environment and Climate Change 2007) which showed that bait dissipated completely within 100 days.

Fig. 1
figure 1

Locations of main Lord Howe Currawong trapping sites for population estimates and captive management

Full size image

In addition, 190 ha of the island was searched for bird carcasses during servicing of bait stations in the Settlement Area (Fig. 1). Because Pacific Black Ducks and Purple Swamphens are restricted to a few relatively small wet areas (32 ha), specific searches of those areas were carried out.

Whenever a dead bird was found, regardless of its state of decomposition, it was placed in a plastic bag with the GPS location recorded. The bird was then returned to the laboratory to determine cause of death if possible. Necropsy was performed as per the Department of Conservation (New Zealand) training and avian necropsy protocol (Ward et al. 2013). Necropsy aimed to establish whether brodifacoum toxicity was the likely cause of death, as indicated by blue/green staining of the digestive tract and blood in the body cavity and joints, but not to identify all potential causes of death via thorough post-mortem analysis.

Captive management

Protocols for capturing and housing woodhens were informed by a previous captive breeding and release program (1981–1983) in which three captive breeding pairs produced 82 released birds (Miller and Mullette 1985). In addition, a captive management trial was undertaken on LHI in 2013 for both woodhens and currawongs at a cost of AUD$363,000 (Taronga Conservation Society Australia 2014). The 100-day trial (i.e. the length of time for exposed bait to break down and thus the expected time birds would be held in captivity during the eradication) demonstrated that both species could be held in captivity for extended periods without discernible harm. At the end of that trial all woodhens (n = 20) and all currawongs (n = 10) were released at their point of capture. The success of this trial contributed to the decision to have the Captive Management Facility (the facility) on-island rather than move birds to the mainland. The facility was built specifically for the eradication, within the Settlement Area on the island. The woodhen enclosures were built with a rodent-proof 1800 mm high external wall of ribbed metal sheeting. Up to six individual (100 m2) enclosures with a 1200 mm wall were contained within the external perimeter. There were also several quarantine pens within the external perimeter to isolate unwell or aggressive birds. The currawong enclosures were contained in similar externally walled enclosures but contained up to 60 (5 m × 1.5 m) flight cages. Two currawongs were housed per cage. The total cost of the captive management program for the eradication was AUD$1.85 million.

Woodhens were caught with hand-held nets by a team of experienced netters at various locations throughout the island. However, the majority (> 75% prior to the eradication) of birds occur in the Settlement Area/lowlands where they are visible and easily captured and all but a few lowland birds were taken into captive management. Birds captured in the lowlands were placed in ventilated plastic containers and transported by road to the facility. If not already banded, they were fitted with a metal numbered band and a unique numbered plastic band on the other leg. To capture mountain birds, netters were transported to Mts Lidgbird and Gower, and Big Slope (southern scree slope below Mt Gower) by helicopter. Birds captured on Mt Gower and Big Slope were carried in boxes to a central location and loaded into a purpose-built timber pod slung below a helicopter. Birds captured on Mt Lidgbird were transported directly back to the LHI airport. From the airport they were transported in their boxes by road to the facility. Over 7 days, 50 birds were removed from the mountains and, while the mountain population is not as well-known as the lowlands population, this number represents the majority of birds typically counted on the mountain in annual surveys. In the following 2 weeks prior to the commencement of the eradication, a further 183 woodhens were captured in the lowlands and taken to the facility. The total number caught exceeded the 2018 population estimate and represents approximately 90% of the total population rather than 90% from each area. Once in the facility, woodhens were housed in groups of approximately 20 individuals where they typically showed little signs of aggression. The occasional aggressive bird was isolated from the others in individual holding pens. To manage this facility, 8–12 professional animal husbandry staff were permanently located on the island during the period of captivity.

Currawongs were trapped using a line-triggered stainless steel clap-trap baited with raisins or bread as a lure. Trapped birds were banded with a metal numbered band as well as a unique combination of three coloured plastic bands to allow identification of individual birds. The majority of trapping occurred at sites across the island where feeding stations had been established and where birds were known to congregate in reasonable numbers (Fig. 2). Trapping was also carried out in previously identified breeding territories. Birds were placed in purpose-built boxes and transported to the facility by vehicle. Over the 4 weeks prior to the commencement of the eradication, 129 currawongs were captured.

Fig. 2
figure 2

Location of non-target species (NTS) search tracks (solid lines) and the bait station grid within the Settlement Area (crossed shading)

Full size image

Woodhens and currawongs were given a veterinary check before being released into their temporary housing. Birds remained in the facility until it was deemed safe for them to be released (details of criteria described below).

When it was assessed that bait and rodent carcasses were not available to currawongs, and prior to the bulk release of captive currawongs, 10 individuals (five breeding pairs) were fitted with radio-transmitters (Lotek® Ag393 glued to the tail-feathers) and released into their known territories to assess whether birds were able to re-claim and retain those areas. The location of birds was checked twice daily from several vantage points from up to 300 m away adjacent to their territory until an active signal was located or it was deemed the birds were not in the vicinity. Often a signal was located within minutes, but searches took up to 30 min per breeding pair if several vantage points were required to locate both birds.

Testing of brodifacoum residue in blood was used to confirm the environment was safe to release lowland woodhens from the facility after the eradication. It was assumed that woodhen in captive management would have no brodifacoum in blood due to their isolation. Thus, 2 weeks after bait stations had been removed, a group of 10 ‘sentinel’ woodhen (adult but of unknown age or sex) were taken from the captive management facility and released at their site of capture at the island’s waste management facility (dump). After 14 days, birds were re-caught and a blood sample was taken from each bird. The blood samples were sent to a certified laboratory for brodifacoum residue analysis. The return of a positive sample (≥ 2 µg/kg MDL) indicated that birds were being exposed to brodifacoum. Therefore, captive woodhens remained at the facility while the process was repeated with another group of 10 sentinel birds released from the facility at a different location. No detection of brodifacoum in the blood of these birds allowed the release of the remaining woodhens.

To assess whether woodhens had exposure to brodifacoum prior to the eradication (through residential use of commercial rodenticide), the plasma from blood samples taken from birds during health checks when entering the facility was sent for brodifacoum residue analysis. Because of the small size of these samples, 32 samples were combined (based on capture location) to make four batched samples. In addition, the livers of 13 individuals that had been collected dead or moribund (then euthanised) during 2017 were sent for brodifacoum residue analysis.

Pre-eradication population estimates

Woodhens have been surveyed annually since 1985. These surveys, using a team of 8–10 people over a 2-week period, repeatedly sweep the Settlement Area to attempt to catch and band (with individual coloured or numbered bands) all sub-adult and adult members of the population. Birds are relatively tame and they respond to loud hand-clapping and call play-back. Once located they are easily surrounded and caught. Woodhens are also territorial so if a bird evades capture, it will usually be caught on a follow-up visit to the same location. Thus, there are very few lowland birds that are not banded. A full day is spent on Mt Gower where all birds encountered during the ascent and a sweep of the summit are banded. To assess changes in the population, pre-eradication survey results (2016–2018) and the number of birds released from the facility were compared to post-eradication survey results which used the same methodology with the exception that three weeks were required in the lowlands to catch and band all birds.

Estimates of the currawong population were based on mark-recapture studies performed in November 2018 and November 2020. To estimate the size of the population (with standard errors), we used the mark-recapture formulae suggested by Bailey (1951, 1952) where M animals are marked in a population of size N (N being unknown) and m marked animals are recaptured or resighted in a subsequent sampling of n animals.

$${\text{N }} = {\text{ M}}\left( {{\text{n}} + {1}} \right)/\left( {{\text{m}} + {1}} \right)$$

This estimate has a standard error (SE) of

$${\text{SE }} = \, \surd \left( {{\text{M 2}}\left( {{\text{n}} + {1}} \right)\left( {{\text{n}}{-}{\text{m}}} \right)/\left( {{\text{m}} + {1}} \right){2}\left( {{\text{m}} + {2}} \right)} \right)$$

Currawongs were trapped and banded over a four-day period and then visually re-trapped at the same sites over an additional four-day period. Results from 2018 were compared to the 2020 results. Observations were also made of free-living currawongs during the eradication to gather further information about the potential impacts of the eradication on this species.

There have been no population estimates made for the remaining eight species of interest other than for purple swamp hens with an estimate of ‘about 50’ in 2004 (McAllan et al. 2004; Table 1). However, surveys for all species were carried out by the Canberra Ornithologists Group in September 2017. Point counts (see Pascoe et al. 2019) were undertaken at 96 points by a stationary observer who recorded all birds detected from their position over a 10-min period within a set radius of 50 m. Each point was surveyed twice before 9 am by two different observers. The process was then repeated with a gap of at least one day between surveys (a total of four surveys at each point). COVID-19 travel restrictions prevented the Canberra Ornithologists Group from repeating these surveys in full in 2021, however, all 96 sites were surveyed twice on non-consecutive days (rather than four times) in September/October by island-based personnel with suitable ornithological experience. Because there were zero birds in many plots, data were not normally distributed, making a comparison of mean number of birds per plot inappropriate. Thus, to examine changes in the abundance of each species, the relative abundance (i.e. the total number of individuals recorded across all plots divided by the number of surveys completed) was compared between 2017 and 2021 using Z-tests.

Bait breakdown and rodent carcasses

Release of Settlement Area woodhen was based on testing of sentinel birds that were released following the removal of bait stations. However, because the southern mountains were aerially baited, the timing of release of currawongs and mountain woodhens was based on the absence of rodent carcasses and bait in the environment. This release was informed by monitoring the breakdown of bait and rodent carcasses. To monitor state of decomposition of poisoned rodents, the numbers of recovered rodent carcasses were documented immediately after the conclusion of the second bait drop. As the numbers recovered declined to almost zero, a selection of defrosted fresh rodent carcasses were placed beneath 1 cm2 wire mesh to allow contact with the ground and invertebrate access but excluded birds. Rodent carcasses were placed in five environments across the island (forest shade, forest sun, woodland shade, woodland sun and exposed headland). Monitoring of cages occurred at 7-day intervals until carcasses were assessed to be no longer of food value for scavenging currawongs due to the advanced state of decomposition and lack of flesh and organs.

Bait breakdown was monitored in two environments: natural forest and pasture. Forest sites were located in the Northern Hills (four sites; two on ridges, two in gullies), on Transit Hill in the centre of the island (four sites; two on ridges, two in gullies), and in the Southern Mountains (four sites, two on ridges, two in gullies). In pasture, three sites were located in the south of the island and three sites were located in the north of the island. At each site, 20 bait pellets were scattered within a wire rodent-proof cage on the day of the second bait drop in each area. Bait breakdown was monitored on day 30, day 60, and then every 10 days until bait had dissipated.

Additional deaths

As well as the non-target species that were officially monitored, other avian species occur on the island either as permanent residents or itinerants. Whenever a carcass of any of these species was discovered, it was also collected and necropsied for cause of death. Of particular interest among these species was the masked owl. This species was introduced to LHI in the 1920s for rodent control and was originally thought to be derived from the Tasmanian subspecies (Tyto novaehollandiae castanops). However, genetic analysis has shown that it is a hybrid of Tasmanian and Australian mainland populations (Hogan et al. 2013) from several early introductions and is thus of no conservation value. This bird preys on native birds as well as rodents and because predation on native species was expected to increase in the absence of rodents, the masked owl was a target for eradication through secondary poisoning and subsequent hunting.

Results

Non-target species

Searches for non-targets in the Settlement Area began when bait stations became operational from 20 May 2019 and in the park preserve with the progression of the first aerial bait drop on 13 June 2019 and continued for 98 days. In the Settlement Area, the same 190 ha was searched each 7–10 days during bait station servicing by 55 to 60 people until 22 Nov (187 days), resulting in more than 9000 person hours of search in the Settlement Area. In the park preserve, total searches by two persons covered a cumulative 243.5 ha and involved 565.5 person-hours.

Table 2 Numbers of dead individuals of each non-target species considered in the Monitoring and Mitigation Plan found in search areas whose death was attributable to brodifacoum toxicosis
Full size table
Table 3 Summary of results of blood testing of woodhens for brodifacoum residues
Full size table

The non-target species with the highest number of deaths attributable to brodifacoum toxicosis was the buff-banded rail (49 deaths, all within the Settlement Area). In addition, four Lord Howe currawongs, two purple swamphens, two silvereyes, and one golden whistler were found to have died from brodifacoum toxicosis (Table 2). A further two dead currawongs were found but were in such an advanced state of decomposition that necropsy was not possible. These birds had green koilin (the membrane lining the gizzard) which was a likely indication of bait consumption. If the deaths of these birds were attributable to brodifacoum toxicosis, it would bring the total number of currawong deaths to six. Trapping and banding of currawongs also continued throughout the eradication leaving very few unbanded birds in the free-living population. This monitoring and banding resulted in the identification of 126 free-living individuals (both adults and juveniles), putting the total overall population prior to the eradication at a minimum of 255 birds.

Table 4 Relative abundance and available population estimates of avian non-target species in 96 survey plots across Lord Howe Island before and after the eradication
Full size table

Several observations were made of wild birds consuming bait. Buff-banded rails and some of the few remaining free-living woodhens were observed accessing bait from bait stations by inserting their head into the access tunnel and removing bait from internal bait trays. These two species were also seen eating bait off the ground outside bait stations. This bait may have been removed from bait stations by rats or by the birds themselves. While a concerted attempt was made to clean up such bait, many pellets were inevitably missed. The two woodhens observed eating bait were caught and brought into the facility and treated with Vitamin K, only one of which survived. Currawongs were also observed eating bait from the ground in areas where bait was spread aerially or hand broadcast. Others were seen ‘playing’ with bait pellets but not consuming them.

Captive management

Currawongs

It was originally intended that currawongs would be released once rodent carcasses had decomposed and were thus unavailable for currawongs to consume. However, because some free-living currawongs were observed consuming bait pellets, the release of currawongs was delayed until aerially spread and hand-broadcast bait had also disappeared sufficiently. The continued presence of active bait stations was not seen as essential for the release of currawongs as there was no evidence of free-living currawongs accessing bait from bait stations. The first release of currawongs occurred on 2 September 2019, when the 10 currawongs fitted with radio-transmitters were liberated into their known breeding territories. Radio-tracking showed that three pairs remained in their breeding territories; a fourth pair were pushed out by a free-living pair that had taken up residence since the original pair had been removed. The final pairing split, with one bird retaining the territory with a free-living bird. Observations of the paired birds suggested that breeding activities (nest building, territorial aggressive behaviours) began within days of release. All remaining currawongs (115 individuals) were released at their capture location 21–24 September 2019.

Within three months of release (December 2019), surveys of currawong breeding territory identified 80 of the 126 known marked birds that had remained in the wild since the eradication commenced (Melanie Massaro in litt.). Mark-recapture surveys and additional surveys of breeding territories post-eradication (September–November 2020) suggest that the currawong population was 213 ± 15 individuals—a slight decrease on pre-REP numbers (235 ± 11). The 2020 population included at least 93 birds that were banded before the eradication (34 of which were free-living during the REP and 59 of which were in captive management). Also, in September/October 2020, 44 juvenile birds (identified by their yellow gape or pale bill) were banded, indicating that currawongs were productive in the year following the eradication. With similar banding effort in 2016, 2017 and 2018, 38, 38 and 27 juveniles were banded, respectively.

Woodhens

The livers of dead woodhens collected prior to the eradication all had detectable brodifacoum residues. Additionally, some of the combined blood samples taken from birds entering captive management had low levels of brodifacoum in their blood, indicating that at least some birds were exposed to brodifacoum prior to the eradication.

Woodhen from the Southern Mountains were released on 14 November 2019 when it was deemed that all bait on the mountains had broken-down. These birds (50) were transported back to the mountains by helicopter. Up to eight wild surviving birds were identified from contact calls heard ahead of the release of the captive birds.

On 18 November 2019, when bait stations were no longer active, and potential residue loads were expected to have dissipated in ground dwelling invertebrates (De et al. 2013), ten lowland-captured woodhen were released at the dump within the Settlement Area. This site was chosen because it had a high likelihood of residual bait due to soiled bait from bait stations being disposed of here in a composting unit. Only seven of these birds were able to be relocated and re-caught 14 days later. A blood sample was taken from each of these birds for brodifacoum residue analysis. The results showed that these birds had low levels of brodifacoum in their blood (Table 3). While some birds entering captive management also had low levels of brodifacoum in blood, it was assumed that the birds released at the dump had consumed brodifacoum during the preceding 14 days because brodifacoum is not expected to persist for an extended period in blood (Fisher 2009). Thus, the bulk release of birds was delayed while an additional 10 birds were released at another site in the southern-most part of the Settlement Area on 15 December 2019. Only five of those birds were relocated and re-caught after 14 days. Blood residue analysis showed no detectable levels of brodifacoum (< 2 µg/kg). Five of the birds previously released at the dump were also re-caught 30 days after their original release and tested; none of those birds showed any signs of ill-health. The proportion of those birds found with brodifacoum residues in blood had reduced and the average concentration in blood had also decreased (Table 3). All remaining lowland woodhen were released 9–10 January 2020.

Table 5 List of other species not considered in the Monitoring and Mitigation Plan found during non-target species searches
Full size table

In December 2018, the known minimum population of woodhens was 221 based on annual banding surveys (Portelli and Carlile 2019). In 2019, 233 woodhens were taken into captive management with 228 birds (five died in captivity) released at the end of the project, including 179 birds in the lowlands. In November 2020, 11 months after the birds were released, 443 woodhens were counted (114 previously banded birds that were re-sighted and 329 freshly banded or seen birds) during a bi-annual survey. A further survey was carried out in March 2021 where 598 birds were counted (including a further 271 freshly banded birds). Surveys in December 2022 recorded a remarkable 1108 woodhens including 538 freshly banded birds. The previous maximum number of woodhens recorded in 35 years of annual surveys was 250, in 2016.

Post-eradication surveys for other non-target species included in the monitoring and mitigation plan

Post eradication point-surveys at the 96 plots across the island showed that the number of birds detected in survey plots had increased significantly for all species other than purple swamphen (Table 4).

Bait breakdown and rodent carcasses

Bait breakdown stations were established in the Northern Hills on 19 June 2019 and over the remainder of the island on 4 July 2019. After 30 days, the condition of bait in the forest sites was variable, with some cages still containing intact bait pellets. Bait in the pasture sites, however, had only pellet residue remaining in some cages—apparently due to consumption by invertebrates, including slugs (Families: Limacidae and Milacidae). By 28 August 2019 (after 56 days), bait had disappeared from all bait breakdown monitoring stations other than a few remnants of pellets remaining in the two ridge sites in the Northern Hills. By 18 September 2019 (after 91 days) bait had completely disappeared from all bait stations.

After caged rodent carcasses had been in place for 9 days, they were intact but bloated and with high fly maggot activity. After 16 days, carcasses had collapsed and decomposed with maggots pupated and emerged, making carcasses unfit for consumption by currawongs.

Additional deaths of species not included in the monitoring and mitigation plan

Necropsies revealed that brodifacoum toxicosis caused the death of eight other bird species that were collected during searches for non-target species. This list includes introduced species such as masked owls which were targeted with baiting as well as the introduced rock dove, mallard, and blackbird (Table 5). A boobook owl was also recovered but no necropsy was performed. Subsequent morphological and genetic analysis showed it to be a Tasmanian boobook Ninox leucopsis (Tsang et al. 2022).

During and following the eradication, 28 masked owl deaths have been recorded. Necropsies confirmed that 11 of those had died from brodifacoum toxicosis. A further 11 individuals were suspected to have died from brodifacoum toxicosis, but necropsies were not performed due to the advanced state of decomposition of carcasses. Another six masked owls were shot over 18 months after baiting was completed. The continued presence of masked owls is being monitored with acoustic recording devices (Wildlife Acoustics SM4®) located at various locations across the island’s forested areas with results still pending. The last reported sighting of a masked owl was on the island’s jetty in November 2021.

Discussion

The development and implementation of the Mitigation and Monitoring Plan showed that with careful consideration of the potential impacts on non-target species, and mitigation actions for species that are at high risk, an eradication using toxicants can be achieved with only minimal losses of non-target species. For the two species deemed to be most at risk, the Lord Howe currawong and the Lord Howe woodhen, the successful implementation of the captive management program, saw the release of healthy birds back into their environment once baiting had been completed. Recent surveys have shown no negative impact on population size of currawongs and a remarkable increase in the number of woodhens. For other native species assessed to be at lower risk, our surveys during the eradication throughout the park preserve and Settlement Area showed that there were negligeable losses to most species and demonstrates that the initial assessment was appropriate for these species. Subsequent surveys have shown that the population of all species, other than purple swamphen, have increased since the eradication.

The species with the highest mortality attributable to baiting was the buff banded rail with 49 dead birds being recovered in the Settlement Area, where this species predominantly occurs. The Settlement Area was thoroughly searched for the entire time bait stations were active. Thus, there is unlikely to have been significant deaths of this species going unrecorded here. Despite the high likelihood of this species taking bait, it was not considered for captive management. The species is widespread on mainland Australia and throughout the Pacific (Birds Australia 2017), and its occasional absence from LHI (McAllan et al. 2004) suggests that it readily re-colonises. Some losses of this species were deemed to be acceptable and the use of bait stations throughout much of this species’ habitat reduced access to bait. Losses of this species during eradications have been seen elsewhere and rails are recognised as a vulnerable non-target group during eradications (Ward et al. 2019). During the Henderson Island rodent eradication operation, there was high mortality (83–97%) in the Henderson crake Zapornia atra due to primary poisoning (Oppel et al. 2016)—much higher than the mortality of buff banded rails witnessed on LHI. On Henderson Island, brodifacoum bait was applied aerially across the whole island rather than in bait stations which suggests that although buff banded rails were able to access baits from stations on LHI, the impact was lower than might have occurred had bait been aerial or hand broadcast across the whole of the island. Despite the high mortality recorded on Henderson Island, crakes recovered to pre-eradication numbers within four years with relatively small supplementation from a captive management program. Recent surveys on LHI have shown that buff banded rails have also increased in the two years since the eradication.

Protection measures for woodhen and currawong included the AUD$1.85 million captive management project. This program was a success with 126 (97.7% of an original 129) currawongs released in November 2019 and 228 (98.5% of an original 233) woodhens released in January 2020. The deaths of three currawongs and five woodhens in captivity was not unexpected given estimated maximum longevities of more than 20 years for currawongs (New South Wales Office of Environment and Heritage NSW Office of Environemnt and Heritage 2019) and nine years for woodhens (Brook et al. 1997). Without this program there would likely have been unacceptable losses of woodhens as evidenced by observations of free-living woodhens eating bait from bait stations. Post-eradication surveys for woodhen found a dramatic increase from pre-eradication levels. While it was suspected that rodents were having some impact on this species, such a rapid increase in the size of the population was not anticipated and is likely due to significant increases in the quality of forage available.

Non-target species surveys during the REP found four currawongs within the search area that were confirmed to have died from brodifacoum toxicosis. Considering the proportion of the island that was searched, the number of birds found equates to an estimated total of 30 deaths out of the 120–130 free-living birds. Some currawong deaths due to secondary poisoning were predicted based on their occasional consumption of rodents, which is why approximately 50% of the population was taken into captive management as an insurance population. Currawongs were observed eating rodents on two occasions during searches for non-targets. However, currawongs were also seen to directly consume bait pellets. This observation was unexpected because there was no indication that currawongs consumed bait in the non-toxic bait trial. This may be due to the amount of time that bait was available on the ground. During the trial, rodents were present and they quickly collected and cached bait rendering it unavailable to currawongs. During the eradication, once rodents had been dramatically reduced, bait from the second aerial application was on the ground for a longer period of time thus providing more opportunity for currawongs to access it. Deaths of currawongs were not widespread, so it is unlikely that the majority of the free-living birds were consuming bait pellets. It was found in post-eradication surveys that at least 80 birds in 2019 and 34 birds in 2020 that were free-living throughout the eradication (n = 126) had survived, providing further evidence that mortality was not widespread. Post-eradication surveys estimated the population to be at 213—within the margin of error for pre-eradication population estimates. When birds were released from captive management, they almost immediately engaged in breeding behaviour and 44 juveniles were banded during post-eradication surveys. Banding effort was not consistent among years, so these numbers are not directly comparable with previous years, where a similar number of juvenile birds were banded, but it shows that there was successful breeding beyond the fledgling stage after the eradication. Whole-island surveys of land birds in 2021 indicate that the currawong population has increased post-eradication.

The release of captive woodhens was based primarily on the absence of bait in the environment (i.e. the breakdown of bait pellets in the park preserve for mountain birds and the removal of bait stations in the Settlement Area for lowland birds). Blood testing of sentinel birds provided useful addition information for this project but would need to be carefully considered for other projects based on individual circumstances and the species of concern. We were confident that birds could be re-caught for testing and were able to keep birds in captivity for an extended period if necessary. Moreover, we were able to observe sentinel birds’ behaviour and closely monitor their well-being. Luxuries such as these are not always going to be possible.

The first cohort of sentinel woodhens was released at the dump. This site was baited using bait stations but was used for the disposal of used bait and was thus deemed to be one of the more likely places for brodifacoum residues to persist. The rationale being that if birds released into this environment were not exposed to brodifacoum, then the remainder of the island having no current bait presence is likely to be safe. Once brodifacoum was detected in the blood of these birds, the decision was made to not proceed with the bulk release of woodhens. It was assumed that the sentinel birds had acquired brodifacoum during their period of release and that they were exposed to brodifacoum distributed during the eradication even though there was an indication that woodhens were being exposed to brodifacoum prior to the eradication. Captive managed birds were in rodent-proof pens and had no opportunity to consume poisoned rodents or bait directly. Moreover, any pre-eradication exposure to brodifacoum would not be expected to persist in blood for more than a few weeks (Fisher 2009). Thus, the sentinel birds were assumed to have no detectable brodifacoum residues in blood prior to their release. Taking a blood sample from sentinel birds prior to their release would have made the source of exposure clearer. However, this course of action was decided against from an animal welfare perspective due to the analysis requiring a relatively large amount of blood taken from the jugular vein. While the decision to delay the bulk release of woodhens meant an additional 10 days of captive management for the majority of birds, the finding of no detectable residues in the second cohort of sentinel birds and diminishing levels in the first cohort meant that we could be more assured of the bird’s well-being upon their eventual release.

The source of brodifacoum in the livers and blood of birds that were tested prior to the eradication is unknown. While there has been long term rodenticide (Warfarin, difenacoum and coumatetralyl) use by the LHI Board to control rodents in key biodiversity areas (Walsh et al. 2019), brodifacoum is not one of the poisons that has been sanctioned for use outside of buildings. Brodifacoum is however widely available and approx. 500 kg per annum was used on the island in commercial rodent baits (LHIB 2009). The evidence of brodifacoum residue in the liver and blood of woodhens suggests that some of these commercial brodifacoum baits have been accessed by woodhens (perhaps through off-licence use) and it is likely that other native bird species had also been previously exposed to brodifacoum. The one-off use of the poison for the total eradication of rodents (if proved successful) will eliminate the need to have further unmonitored amounts of poison entering the natural system here.

Surveys of 96 plots across the island in 2021 showed that the frequency of occurrence increased for all non-target species other than purple swamphen. For woodhens, the frequency of observations in survey plots increased four-fold post-REP compared to 2017 surveys, which aligns with the three-fold increase found during annual counting and banding surveys in April 2021. For currawongs, mark-recapture surveys 12 months after the release of birds from captive management showed a slight decrease in numbers. However, the increased frequency of occurrence in plot surveys performed a further 12 months later suggest that their numbers too have subsequently increased. Of note among the other non-targets assessed, was a tripling of sacred kingfisher observations, and a doubling of threatened silvereye observations. There was also a marked increase in observations of threatened golden whistlers in survey plots. There was a decrease in purple swamphen observations which, due to their fossorial foraging habits, may have been attributable to the eradication. Breeding is now occurring (NC and TO’D pers obs.) and there are no apparent ongoing eradication-related impacts on this species. During a visit to the island in June 2023, 43 swamphens were recorded, which is comparable to pre-eradication numbers of ‘about 50’ (McAllan et al. 2004).

There appeared to be an almost immediate change in behaviour of some bird species once rodents had been eliminated (or drastically reduced) in the park preserve. For example, currawongs were no longer visiting well-established feeding tables in large numbers just weeks after baiting. This change in behaviour was first perceived as a potential decline in the numbers of free-living currawongs and perhaps an under-estimation of the risk posed to this species. After further investigation it was determined that the lack of currawongs at feeders was more likely a result of birds taking advantage of additional resources in the native vegetation areas in the absence of rodents rather than their death due to baiting. The earlier misperception resulted in a delay of the release of currawongs while the impact of baiting on currawongs could be further assessed. This demonstrates the need to be ready to adapt survey and release plans in response to such changes as available food resources.

Following international best practice, the declaration of the island as rodent-free will not be taken until two years of no rodent detections following baiting. After almost 18 months of no rodent sightings, a recent detection of ship rats in April 2021 has delayed this declaration further. These rats were detected in a number of discrete locations within the Settlement Area. A rapid and intense response involving bait station deployment, trapping and detection using dogs and cameras resulted in 96 rats (44 adults and 52 juveniles) being found and eliminated (LHIB 2021). The source of the rodents has not been ascertained. However, the extensive monitoring during the response found no evidence of rodents outside the initial detection zone, suggesting that rats are not surviving elsewhere in the Settlement Area or park preserve. The last of these rats was detected in August 2021 and at the time of writing no rats had been detected for more than 600 days. In July 2023, a comprehensive check for rodents, alongside statistical modelling, concluded that rats or mice are no longer present on Lord Howe Island (Harper 2023). This finding indicates that the eradication was successful based on internationally adopted guidelines.

During the response to the recent re-discovery of rats, no non-target species were taken into captivity, and it is likely that some woodhen were exposed to bait, considering their ability to access bait stations. However, bait station deployment was restricted to a relatively small section of the Settlement Area and thus a relatively small number of woodhen were potentially at risk. If rodents were detected across a larger range, requiring a more widespread baiting response, then protection measures for woodhens would need to be considered before bait deployment. However, our monitoring during and after the eradication, some of which followed the re-discovery, has shown that despite some losses, non-target species survived the baiting program and the population numbers of most species now exceed pre-eradication numbers.