Maggie Nichols and Phil Bell – February 2019
INTRODUCTION
The Government announced, in July 2016, the ambitious goal for New Zealand to be predator free by 2050 (New Zealand Government 2016). Since then, there has been an emerging focus on the eradication of possums (Trichosurus vulpecula), rats (Rattus rattus, and R. norvegicus) and mustelids (Mustela furo, M. nivalis, M. erminea) from New Zealand’s mainland (Russell et al. 2015). There is substantial need for the research and development of new tools that will enable the success of the predator free goal (Norton et al. 2016).
Aerial 1080 (sodium fluoroacetate) operations are used over much of New Zealand as an effective tool for reducing mammalian pest populations of possums and ship rats (Rattus rattus) (Eason et al. 2011; Kemp et al. 2019) for the benefit of native biodiversity and to reduce the incidence of bovine Tb.
In consultation with experts from the Department of Conservation (DOC), OSPRI and Manaaki Whenua-Landcare Research, Zero Invasive Predators Ltd (ZIP) has developed a modified aerial 1080 technique known as ‘1080 to Zero’ that is designed to completely eliminate all possums from the treatment area - i.e. not just suppress them to low numbers (Bell 2017). The results of relatively small-scale trials (Bell 2019) led ZIP to conclude that the 1080 to Zero prescription shows potential to be able to completely remove possums from large areas of predominantly native forest. Consequently, ZIP is proposing to implement the 1080 to Zero technique to remove all possums from an 8,200-hectare research and development area in the Perth River valley (43.2616° S, 170.3590° E), in South Westland. One of the reasons this area was selected was that the rivers and streams and mountain tops are expected to be strong natural barriers to reinvasion by possums, rats and stoats, thus helping to maintain a predator-free state.
The Perth River Valley 1080 to Zero operation was originally scheduled to be completed in winter of 2018. Two pre-feeds were completed in May and June 2018. The remainder of the operation was however deferred in late July 2018 because weather and snow conditions were such that ZIP could not be confident that the 1080 to Zero technique could be fully implemented in a timely fashion, in order to completely remove introduced predators. The operation is now proposed to start in late March 2019.
The kea (Nestor notabilis) is a highly intelligent, charismatic, large parrot that is endemic to the South Island of New Zealand (Kemp et al. 2019; Orr-Walker et al. 2015; Van Klink and Crowell 2015). Kea are classified as a nationally endangered species, having a moderate population, but high ongoing or predicted decline (Robertson et al. 2013).
Kea are omnivorous and forage primarily on the forest floor (Greer et al. 2015), where aerial 1080 baits are typically distributed to target possums and rats. Kea are also highly curious birds that often investigate novel food objects (Kemp et al. 2019), thus they are at risk of direct poisoning by 1080 during aerial operations (Orr-Walker and Roberts 2009). Where kea have been killed by 1080 poisoning, the subsequent productivity of the local kea population generally outweighs these losses, albeit with some exceptions, as a result of the reduced predation pressure (Kemp et al. 2018)
Kea have been known to sometimes feed on carrion (Diamond and Bond 1999). Both single individuals and groups of kea have been observed feeding on tahr carcasses (Schwing 2010). Anecdotal evidence through communication with hunters and helicopter pilots suggest kea are regularly seen feeding on tahr carcasses left over as a result of hunting, and culling, operations in the Perth River valley.
Between February and May 2018, 55 kea - a mixture of adult males, females, juveniles, and sub-adults - were caught and banded in order to be individually identifiable, and monitored, throughout the originally-planned 1080 to Zero operation. The process of banding kea has indicated that the kea population at the Perth River valley research area totals 75-100 birds (I. Graham, DOC, pers. comm, 2018).
ZIP has undertaken trials to investigate two methods that have the potential to mitigate risk to kea from being poisoned as a result of consuming 1080 baits during the proposed operation, i.e. to use a bird repellent bait prior to sowing the toxic bait, and to provide fresh tahr as a more attractive preferred food source (i.e. than bait). The results of the ZIP bird repellent trial (Nichols and Bell 2019) can be found on the ZIP website.
This report describes two trials that we undertook during June to August and November to December 2018 to evaluate the potential of using tahr carcasses as a more attractive and preferred food source than cereal bait.
OBJECTIVES
The objectives of these trials were to:
ascertain whether kea activity is higher around tahr carcasses than at cereal bait (Phase 1), and
increase our confidence in tahr carcasses as a technique to mitigate the potential risk to kea in the Perth River valley from the proposed 1080 to Zero operation (Phase 2).
The trial covered a broad range of topics, including the: population demographics of kea seen at tahr carcasses and at bait, types of kea behavioural interactions seen with cereal bait, proportion of the banded population of kea visiting these sites, and average longevity of tahr carcasses as a food source for kea.
PHASE 1
Study area and methods
This trial consisted of three separate deployments of tahr carcasses at sites above the bush line (approximately 1,300-1,700m above sea level), within the Perth River valley, over the period of June-August 2018 (Fig. 1). A total of 67 tahr carcasses, all nannies (to ensure bulls were preserved for trophy hunting), were used for this trial. The tahr were provided by DOC, and only steel shot was used to shoot them.
At each site, we hand-laid a 1.5-kilogram pile of non-toxic cereal bait (Wanganui #7, dyed green, with a double dose of orange-scented lure), laced with the secondary bird repellent anthraquinone at a concentration of approximately 1.5% by weight. Anthraquinone is a widely-used secondary repellent for birds (Werner et al. 2011; Clapperton et al. 2014; Cowan et al. 2016). Once ingested, the repellent results in mild gut discomfort, leading to gagging behaviour, and sometimes vomiting (Werner et al. 2011). The repellent was used in an attempt to create a learned aversion in kea to green-dyed cereal bait as part of a kea risk mitigation strategy for the 1080 to Zero operation that was initiated in 2018.
The carcass/bait sites are shown on Fig. 1. A small number of deployment 1 carcasses were relocated to new sites owing to the inaccessibility and avalanche risk of the original sites. The number of sites reduced over each deployment as a result of the increasing avalanche risk towards the end of winter, i.e. from 21 sites in deployment 1, to 15 in deployment 2, and 9 in4 deployment 3. During deployments 2 and 3, most sites had two carcasses stacked, one on top of the other, to make them more visible in deep snow.
Figure 1: Phase 1 tahr carcass and repellent bait sites.
Trail camera traps were used to monitor kea at nine of the sites throughout Phase 1, as determined by the number of cameras available and by site accessibility. Each site was initially monitored with two Browning Strike Force BTC-5 trail cameras (Prometheus Group, Birmingham, Alabama), mounted on steel fence posts. One camera was positioned to record kea activity on and around the tahr carcass, and the other to record kea activity at the bait. Both cameras were set with a time delay of 5 seconds between being triggered. Tahr carcass cameras were set to record one image per trigger. Bait cameras were set to record videos of 10 seconds duration, in order to enable detailed examination of how kea interacted with the repellent-laced cereal baits.
The cameras were serviced six times; the period between services ranged from 7 to 10 days, depending largely on weather conditions.
Some cameras were damaged by kea during the early services, and were replaced with LTL Acorn 5210a cameras (Zhuhai Ltd Acorn Electronics Co. Ltd) until further supplies of the Browning model were available. All sites were re-equipped with Browning cameras by service 5. Unfortunately, there are no camera data from service 5 because the secure digital (SD) cards were corrupted due to both moisture damage and electronic malfunction arising from human error (i.e. failure to reformat the cards when the Acorn cameras were replaced with Browning cameras).
Consequently, the most complete set of camera data (i.e. the most functioning cameras at both tahr carcasses and bait), is from service 6, which covered nine tahr/bait sites, during which cameras were deployed for seven days (21 to 27 August).
We compared kea activity at tahr carcasses and at the bait based on the number of kea images recorded (i.e. not the number of birds recorded). We standardised the number of images by treating each 10 second video taken by the bait cameras as equivalent to 3 individual images taken by the cameras at the tahr carcasses (because, as a result of the camera settings used, 3 individual images could be taken by the tahr carcass cameras within an equivalent time period if set on the same photo mode as the bait camera used). This may be a slightly conservative estimate due to trigger time length; however, it results in a comparable estimate.
We report results in terms of average (i.e. mean) numbers of images recorded per day – i.e. an ‘operational camera day’ - because individual cameras were operational for varying periods, due to malfunctions and the SD cards reaching maximum capacity (owing to the high volume of images taken at some of the tahr carcasses).
Images from the Browning cameras were clear enough to identify individual kea based on their leg bands (Fig. 2), but the LTL Acorn model was unable to produce clear images of bands. Consequently, the estimate of kea population size is only based on images from the tahr carcass cameras.
Figure 2: Example of an image of a kea taken at a tahr carcass. Note the band ID (Black V on silver) (July 2018).
RESULTS
Kea activity and behaviour at sites In service 6, the average number of images of kea recorded per operational camera day totalled 6,821 for the tahr carcass cameras and 1,171 images for the bait cameras. Consequently, the overall ratio of the average number of kea images per operational camera trap day on tahr carcasses to bait was 6:1. The average number of kea images (corrected for operational camera days) for each of the nine tahr carcass/bait sites serviced during deployment 3 is shown in Fig. 3.
Figure 3: Average daily number of images of kea at tahr carcass/bait sites during service 6 (August 2018).
The different types of interactions kea had with bait were recorded during the analysis of camera footage from service 6. This was only done with the bait cameras, as all tahr carcass interactions were assumed to be consumption (as evident from both field observation and through camera footage).
Behaviours were categorised as follows: visual inspection, touch/pick-up (using beak or feet), consumption, and no interaction (walking past the camera/interacting with another kea). Visual inspection, touch/pick-up and no interaction behaviours were recorded at all bait sites. Consumption of bait was recorded at 4 of the 9 sites (Table 1).
Table 1: Kea behaviour by site with repellent bait in service 6 (August 2018).
Although consumption was recorded less often seen than the other behaviours, signs of the effect of the repellent were seen in multiple birds, such as gaping, head shaking, fluffed feathers, beak wiping and vomiting upon tasting bait. These signs are consistent with those seen in captive kea during recent repellent trials with anthraquinone-laced bait (Nichols and Bell 2019). Although anthraquinone is a secondary repellent, we believe the smell and taste of the double orange lure, as well as the consistent green colour, acted as primary cues for the repellent bait (as seen in recent ZIP captive trials) (Nichols and Bell 2019).
Kea population estimates
We recorded 31 of the 55 previously-banded kea at tahr carcasses over the course of the five usable camera services (Table 2). These sightings suggest a population of 59 kea visited the tahr carcass sites [95% C.I 46-86], using a Lincoln-Petersen population estimate in the program MARK (White and Burnham 1999). The sex and age of each of these individuals is shown in Table 2.
The most complete (and thus comparable) data set from both the tahr carcass and the bait cameras is from service 6. In service 6, a total of 17 individual birds were seen at the tahr carcasses, and 13 at the bait (Table 2).
Table 2: Banded kea seen at tahr carcasses during five camera services (shown in blue), and at bait during service 6 (green)
Note: Service 5 data is not included as the camera SD cards were corrupted. Kea on bait could not be reliably recorded for services 1-4 because of incomplete camera coverage as a result of camera damage by kea (July-August 2018).
Viability of tahr carcass for consumption by kea
We used camera footage and field team observation to estimate the average number of days a tahr carcass remains as a viable food source under high consumption by kea. During service 6, most carcasses were reduced to a skeleton after 6 days. The reduction in the quantity of food source was also reflected in a substantial reduction in the number of images of kea at tahr across the nine sites (Fig. 4).
Figure 4: Average daily number of images of kea taken from all tahr carcasses (pooled) during service 6 (August 2018). The '0' indicates the day tahr carcasses were deployed.
PHASE 2
Study area and methods
Phase 2 consisted of a trial to (1) more rigorously assess the attractiveness of tahr carcasses in relation to cereal bait; and (2) to refine tahr carcass deployment as a technique to mitigate risks to kea from the proposed 1080 to Zero operation.
Phase 2 took place during late spring/early summer 2018 (November-December). This trial utilised 11 of the carcass/bait sites used during the Phase 1 trials, some of which had not been used since the earliest service rounds then due to avalanche risk. Phase 2 consisted of two rounds of camera/tahr carcass/bait deployment.
This trial was designed as a mixed choice test, whereby birds could choose to attend sites that had a tahr carcass and a small pile of bait or sites that only had a pile of bait. For the first round of this trial, five of the sites were initially set-up with a tahr carcass and a 1.5-kilogram pile of cereal bait, and five other sites comprised bait-only. For the second round, the initial carcass/bait and bait-only sites were swapped, to try to ensure that any kea visits then were less influenced by their memory of the set-up during the first round. One site was a bait-only control site, which was retained at the same site for each round, in the event that kea showed a substantial change in behaviour towards bait only sites (which were formerly carcass/bait sites in round 1), which could be attributed to the memory of tahr carcasses at these sites.
We used RS5, cinnamon-lured, bait with no dye and no repellent, in order to remove any cues that kea could relate to the previous anthraquinone-laced repellent bait used in Phase 1 – to obtain a more realistic measure of kea activity around bait. RS5 was not used during the two prefeeds that ZIP sowed at the Perth River valley research area in May and June 2018.
Browning trail cameras were used to monitor bird activity at all eleven sites. Two cameras were deployed at the carcass/bait sites (one to record kea activity on and around the tahr carcass, and the other to record kea activity at the bait), and one camera at each bait-only site. The cameras were deployed for a total of 14 days during the first round, and for 10 days during the second round; the different periods reflected the influence of weather on our ability to access the cameras. All cameras were set to take one image per trigger, with a 5 second delay; in order to obtain the most direct comparison of kea activity between sites. All cameras were programmed to take images rather than videos because Phase 2 was focussed on kea activity between site types, rather than a close examination of interaction types.
We compared the average number of images of kea – i.e. all birds, both banded and unbanded – on both tahr carcasses and bait at the carcass/bait sites, and bait-only sites (including the control site). We also recorded banded individuals seen at each type of site, for each round.
One SD card failed at a tahr carcass during round 1, so this site was not included in the results.
Figure 5: Camera set up at a carcass/bait site. Cameras are circled in black, the tahr carcass in yellow, and the bait in blue (November-December 2018).
Figure 6. Five kea (circled in orange) consuming a tahr carcass during Phase 2 (November-December 2018).
Results
During round 1, cameras captured an average of 646.4 images of kea at tahr carcasses and 53.8 images of kea at bait at the carcass/bait sites, and 4.6 images of kea at the bait-only sites per operational camera trap day. During round 2, cameras captured an average of 464.7 images of kea at tahr carcasses and 46.6 images of kea at bait at the carcass/bait sites, and 3.7 images of kea at the bait-only sites, per operational camera trap day.
The control (bait only) site captured an average of 13.8 kea images in round 1, and 5.9 kea images in round 2.
The average number of kea images captured at tahr carcasses and bait at the carcass/bait sites, and at bait-only sites (corrected for varying camera deployment days in the two rounds) is shown in Fig.7. In round 1 the ratio of kea images captured on tahr carcasses to all bait sites (including the control site) was 5:1, and in round 2 the ratio was 8.3:1.
Figure 7: Average number of kea images captured at (i) tahr carcasses, (ii) bait at the carcass/bait sites, and (iii) bait-only sites, (iv) control bait-only, per round (November-December 2018).
In total, 17 individual banded birds were recorded, three of which had not been seen during Phase 1 (Table 3).
Table 3: Banded birds seen at tahr carcasses and bait during rounds 1 and 2.
DISCUSSION
The use of trail cameras in these trials has provided an abundance of information that helps to answer some fundamental questions underlying efforts to mitigate the potential risk to kea from being poisoned as a result of consuming 1080 baits.
The trials clearly answer the question of how attractive are tahr carcasses relative to cereal bait – the footage captured during both Phase 1 and 2 trials demonstrates a high level of kea activity at tahr carcasses relative to bait. We conclude that tahr carcasses are indeed a food source that kea find highly attractive, which may be unsurprising given past anecdotal accounts, but until now had not been quantified.
Another question the trials helped to answer is whether the presence of tahr carcasses attract birds from outside the valley – the results of the mark-recapture analysis in Phase 1, and coarse measurement of kea activity from both phases do not indicate that birds travelled from outside the Perth River valley to visit tahr carcasses. In part, that is because the population size estimates obtained from the mark-recapture analysis of banded kea seen at tahr carcasses are within the bounds of a recent population estimate by kea catchers (who banded and radio-tagged kea in the Perth River valley during the past summer field season). However, we acknowledge that we cannot account for the demographics of the individual birds without bands seen on cameras. Although juveniles are expected to range much wider than adults, they are not expected to disperse until late summer/early autumn; and we did not see an increase in activity between the two rounds during Phase 2. Thus, this information also helps us to conclude that it is unlikely that the trials have attracted birds from outside the Perth River valley.
The trials also helped us to answer the question of whether birds would continue to be active at a site where a tahr carcass was no longer present – the results of Phase 2 do not indicate that this is the case. Rather, the results indicate a substantial decrease in kea activity at the initial carcass/bait sites when they were swapped to bait-only sites. The decrease in bait activity is unlikely to be a learned aversion as a result of the repellent bait used in Phase 1, because a different bait was used in Phase 2 (i.e. W#7, double orange-lured, green-dyed, laced with anthraquinone repellent and RS5, cinnamon lured, with no dye respectively).
Another question the trials helped to answer is whether repeated exposure to cereal bait would result in greater consumption of baits by kea (as is the intended purpose with target species) – the decline in activity at bait-only sites in round 2 of Phase 2 does not support the idea that more bait exposures will naturally result in more consumption of bait by kea. Admittedly, however, kea activity at all bait sites was relatively low during all of these trials.
The exceptionally curious nature of kea generally drives them to interact with novel objects in their environment, thus we do not assume that the high levels of activity seen at tahr carcasses necessarily means birds will interact less with cereal bait. However, the decrease in kea activity at bait-only sites in round 2 of Phase 2, suggests that carcass/bait sites are more attractive than bait-only sites. Kea activity at bait at the Phase 2 carcass/bait sites remained nearly constant between rounds 1 and 2. This suggests kea are more likely to be drawn to a nearby site containing a tahr carcass than a pile of cereal bait alone; as demonstrated when sites containing tahr carcasses were swapped. The similar levels of kea activity at the bait at the carcass/bait sites throughout the two rounds of Phase 2 suggests that birds will indeed be exposed to repellent bait during implementation of aversion training in the Perth River valley research area, when repellent baits are placed alongside tahr carcasses, before and throughout the proposed 1080 to Zero operation.
The results, particularly from Phase 2, have helped inform our subsequent decision to make aversion training (repellent) bait available at multiple tahr carcass sites immediately adjacent to the treatment area of the proposed 1080 to Zero operation (i.e. at a higher altitude than the treatment area) both prior to and throughout the operation. We currently plan to use a concentration of 2.7% (by weight) anthraquinone, laced into the equivalent bait matrix for that phase of the operational sequence (i.e. W#7 double orange lured; then RS5 double cinnamon lured bait – all dyed green) in order to create (and potentially reinforce) a learned aversion to cereal bait.
Phase 1 demonstrated that it is possible to reliably identify individual birds through band information seen on camera trap footage with the Browning camera trap model, which enabled us to better understand the demographics of the kea population. The expectation during the design of Phase 1 kea monitoring was that cameras would only be capable of showing the presence of kea. This was true for the LTL Acorn camera trap model; however, the Browning camera trap model produced very clear images for which even the small letters on the birds’ colour bands were identifiable.
Camera footage and field observations provided other useful information to inform the development of an approach to mitigate the potential risk to kea from being poisoned as a result of consuming 1080 baits. First, we learned that a tahr carcass will last for approximately 7 days before kea consumption reduces it to a skeleton. We also observed evidence of the effects of anthraquinone bird repellent in wild kea after consuming cereal bait. The symptoms included gaping, head shaking and vomiting. The repellent effects seen were the same as those observed with captive kea, even though the concentration of anthraquinone used in Phase 1 with wild birds (approximately 1.5%) was less than that used with the captive birds (approximately 2.7%).
The results from both Phase 1 and Phase 2 trials do not provide evidence to substantiate two important concerns about the potential impact of the proposed Perth River Valley 1080 to Zero operation on kea, i.e. that increased availability and repeated exposure of tahr carcasses may bring kea from outside of the Perth River Valley, and the increased availability and repeated exposure of cereal bait raises the risk of cereal bait consumption by kea.
Overall, the results of the Phase 1 and Phase 2 trials have provided very useful information that is enabling ZIP to develop measures to minimise the potential impacts of the proposed 1080 to Zero operation on kea, by providing a highly attractive food source (i.e. tahr carcasses) to the local population of kea in the Perth River valley, and in doing so expose kea to repellent-laced cereal bait. It is not possible to guarantee zero mortality of non-target species during any predator control operation; however, we are confident that these measures will reduce potential consumption of toxin-laced cereal bait, and associated kea mortality during the 1080 to Zero operation.
ACKNOWLEDGEMENTS
Thank you to the ZIP field team members (Courtney Hamblin, Matt Chisnall, Chad Cottle, and Piper Douglas) for setting-up and maintaining the sites used in these trials, servicing the cameras, and reviewing an immense volume of camera footage. Thanks also to other ZIP team members Katie Coster and Becky Clements (Animal Behaviour Technicians) for reviewing camera footage, Helen Nathan (Predator Ecologist) for helping us to manage the data, Nick Mulgan (Predator Systems Modeller) for advice on statistical analysis, Nicholas Braaskma (GIS Data Analyst) for mapped tahr and bait sites, and Susannah Aitken and Joseph Arand for reviewing and publishing the final report.
We also acknowledge and thank: Ken Hutchins from Fox Franz HeliServices for safely transporting field teams to the trial sites; DOC Rangers Shane Cross and Gary Scott for shooting tahr and placing them at designated sites; the Game Animal Council and New Zealand Deerstalkers Association for supporting these trials; Geoff Kerr (Game Animal Council, and Lincoln University) and Nic Gorman (DOC) for providing feedback on the trial design for phase 2; Tamsin Orr-Walker, Nigel Adams, and Laura Young (Kea Conservation Trust) and Geoff Kerr (Game Animal Council; and Lincoln University) for their review of an earlier draft of this report.
REFERENCES
Bell P. (2017) 1080 to Zero: Jackson-Arawhata. Technical report #1. Zero Invasive Predators. Wellington. http://zip.org.nz/findings/2017/11/1080-to-zero-trial-in-south-westland Zero Invasive Predators Ltd.
Bell P., Nathan, H., Mulgan, N. (2019) ‘Island’ eradication within large landscapes: the remove and protect model'. In: C.R. Veitch, M.N. Clout, A.R. Martin, J.C. Russell and C.J. West (eds.) Island Invasives: Scaling Up to Meet the Challenge, pp. 600–606. Gland, Switzerland: IUCN.
Diamond J. & Bond A. B. (1999) Kea, bird of paradox: the evolution and behavior of a New Zealand parrot. Univ of California Press.
Eason C., Miller A., Ogilvie S. & Fairweather A. (2011) An updated review of the toxicology and ecotoxicology of sodium fluoroacetate (1080) in relation to its use as a pest control tool in New Zealand. New Zealand Journal of Ecology, 1-20.
Kemp J. R., Mosen C. C., Elliott G. P. & Hunter C. M. (2018) Effects of the aerial application of 1080 to control pest mammals on kea reproductive success. New Zealand Journal of Ecology 42, 158-68.
Kemp J. R., Mosen C. C., Elliott G. P., Hunter C. M. & van Klink P. (2019) Kea survival during aerial poisoning for rat and possum control. New Zealand Journal of Ecology 43, 0-.
New Zealand Government. (2016) New Zealand to be Predator Free by 2050. Speech by then Prime Minster, Hon John Key. https://www.beehive.govt.nz/release/new-zealand-be-predator-free-2050 (accessed on 20 February 2019).
Nichols M. & Bell P. (2019) Kea repellent for cereal bait: A captive study using anthraquinone. Technical report #4. Zero Invasive Predators. Wellington. http://zip.org.nz/findings/2019/2/can-kea-learn-to-avoid-eating-cereal-baits. Zero Invasive Predators Ltd.
Norton D. A., Young L. M., Byrom A. E., Clarkson B. D., Lyver P. O. B., McGlone M. S. & Waipara N. W. (2016) How do we restore New Zealand's biological heritage by 2050? Ecological Management & Restoration 17, 170-9.
Orr-Walker T., Kemp J., Adams N. J. & Roberts L. (2015) A strategic plan for kea conservation. A collaboration between the Department of Conservation (DOC) and the Kea Conservation Trust (KCT).
Orr-Walker T. & Roberts L. (2009) Population estimations of wild Kea (Nestor notabilis). Kea Conservation Trust.
Robertson H. A., Dowding J. E., Elliott G., Hitchmough R., Miskelly C., O'Donnell C. F., Powlesland R., Sagar P. M., Scofield R. P. & Taylor G. A. (2013) Conservation status of New Zealand birds, 2012. Publishing Team, Department of Conservation.
Russell J. C., Innes J. G., Brown P. H. & Byrom A. E. (2015) Predator-free New Zealand: conservation country. BioScience 65, 520-5.
Schwing R. (2010) Scavenging behaviour of kea (Nestor notabilis). Notornis 57, 98-9.
Van Klink P. & Crowell M. D. (2015) Kea (Nestor notabilis) survivorship through a 1080 operation using cereal baits containing the bird repellent d-pulegone at Otira, central Westland. Publishing Team, Department of Conservation.
White G. C. & Burnham K. P. (1999) Program MARK: survival estimation from populations of marked animals. Bird study 46, S120-S39.