A critique of the research used to justify kea poisoning

Kea (Nestor notabilis) have Absolute Protection under the Wildlife Act 1953. This review examines the research used to claim that aerial poisoning with 1080 benefits them. The references, appendices and the addendum on the history of kea and people, written by the late Bill Benfield, are in the PDF.

Download the full review (PDF)

Background

The number of kea (Nestor notabilis) remaining in NZ is unknown. Estimates as low as 1000 have been made since 1986 (1, 2). The kea’s official threat status has been worsening and reached Endangered in 2017 (2, 3). Declines and disappearances of kea have been very noticeable in the north and east alpine areas of the South Island (4). Kea have Absolute Protection under the Wildlife Act (1953), meaning it is illegal to kill or capture them or interfere with their nests (5).

A major cause of kea mortality, for decades, has been poisonous food bait distributed by aircraft to control unwanted mammals. Deaths of kea following aerial poisoning with 1080 (sodium monofluoroacetate) were noted in 1963 (6). Monitoring began in 2008 at Fox Glacier, where 7 of 17 kea were poisoned, indicating that 41% of the local kea had died (7). Recorded poisoning rates include: in 2020, 50% at Matukituki (8), also 50% at the remote Wet Jacket Peninsula (9); and in 2022, 22% at Arthurs Pass and 29% west of the pass at Otira-Taipo (10).

NZ’s Department of Conservation (DOC) claims that kea populations benefit from poisoning, because it kills nest predators: “Without pest control, typically about 60% of kea nests fail – mostly due to being preyed on by stoats or possums, and in some areas, feral cats” (11). The following three brief sections examine the evidence for this claim. Further details are provided in the Appendices. Lastly, the Addendum provides more on the history between kea and humans (written by the late Bill Benfield).

1. Radiotelemetry

Radiotelemetry is the basis of DOC’s, and charity (Kea Conservation Trust, KCT), studies on kea. The birds are caught and have a long-term backpack harness carrying a transmitter attached to them (335 kea reported by DOC up to 2021 (4); the KCT accounts for many more (12); animal control company Zero Invasive Predators is also now using the devices (13)).

In birds generally, those wearing backpack harnesses and transmitters are much more likely to die (14), e.g. annual survival rates of nearly 90% were found in falcons that managed to shed their backpacks compared to around 50% in birds that did not (15). Wearing telemetry equipment was found to affect most aspects of a bird’s life, and markedly increased energy expenditure (14). In the flightless takahe, this was attributed to heat loss, and the risk to alpine birds was pointed out: they may have little scope to forage more in winter to compensate (16). In pigeons (17) and ibises (18), flying was more difficult. Another effect of adding this equipment to birds, in general, is a marked decrease in nesting (14).

Reports from DOC’s use of this equipment on birds in general include: deaths through capture and handling stress, aspergillosis (a fungal disease brought on by capture and stress), hypothermia, entanglement and predation; loss of body condition; lack of breeding; and disappearances (Appendix I); plus injuries from long-term wear (19).

The only sign of consideration from the people using them, of effects of putting harnesses on kea, appears to be a feasibility study on GPS monitoring (20). Casual observations were made on 14 newly harnessed birds (mostly) recaptured after a median of 11 days. Although no negative effects were reported, it was noted the birds responded by trying to remove the harnesses (20).

Bird researchers were asked to consider whether their radio telemetry studies are worthwhile if they collect biased data from disadvantaged birds, which are being harmed as well (14).

2. Nest monitoring

Using the transmitter-wearing kea, low-flying aeroplanes, ground searches and playing recorded kea calls, DOC and KCT have been finding kea nests to monitor (196 reported by DOC from 1992 up to 2023 (2); many more by KCT (12, 21)). Humans have been climbing into the nest cavities, handling chicks (measuring, taking blood and feather samples, applying leg rings, putting telemetry gear on older ones), adding cameras in and outside nests and revisiting frequently (Appendix II, (12)).

The above manipulations are in direct conflict with collecting meaningful data on threats to kea, and the kea’s breeding opportunities. In the U.K. it is illegal to interfere at all with the nests or young of rare birds because of the high risk of causing failure (22). Even brief desertion can leave eggs cooling or chicks vulnerable, and the sights, noises and scents created and left behind (e.g. cameras) attract curious and opportunistic animals (23, 24, 25). A nest researcher in NZ observed that stoats were following her around (25). Disturbing older chicks can cause them to leave the nest early, when survival is poor (26). Using bird calls to find nesting adults is also condemned due to the distraction it causes (27). Aircraft are known to disturb breeding birds and were banned by DOC from the air space above nesting terns (28); they are likely to distract kea (29). Leg bands also harmed kea: “The large number of band injuries were caused not by the band being put on improperly but because a band prevents normal swelling of the leg, following a foot injury” (29).

Studies on kea at Arthurs Pass by J.R. Jackson in the 1950s–60s revealed the nests were worked on for years by the hen, then used for a lifetime. Nesting began in midwinter: “July: Heavy falls of snow cover much ground and Keas feed in the forest and on the forest floor. First eggs laid … October: Peak of laying. Adults very shy and quiet” (30). Jackson found that upset parent birds, other kea, gulls and falcons were all threats to eggs and chicks (31). Not rats, stoats or possums: “During the last hundred years Keas have shared their environment with rats Rattus spp. and stoats Mustela erminea. I have found no evidence of these animals affecting Keas … Twice I have found a dead possum Trichosurus vulpecula within five yards of a Kea nest. The opossum frequently chooses holes similar to a kea nest as a den and perhaps these two opossums prospected the Kea nests” (29).

In 1999 (32) and 2004 (33), DOC’s G. Elliott and J. Kemp confirmed Jackson’s assessment of a very low risk of predation at kea nests. Later, Kemp (with funding from the Animal Health Board (AHB), now OSPRI, a significant user of aerial poison to control bovine tuberculosis), began approving of ongoing poisoning of kea “to kill nest predators”, despite the very poor evidence he had (Appendix III). Even cats, noted as invading Arthurs Pass National Park from about 2015 onwards, left the nesting (but not the other) kea alone. Again, kea were put in harnesses with telemetry gear to see whether the cats would catch them: “This project aims to substantiate the importance of feral cats as a driver of native species declines for ground dwelling¹ birds in eastern forests” (34). It could be suggested that this study was unashamedly biased research and therefore has little merit, but it is another clear example of a poor method being used and placing endangered species at significant risk. Records from decades of kea nest monitoring repeatedly show great disruption: abandonment, disappearances, and unexplained deaths, and yet little evidence of predation by mammals (Appendix II, (12)). An Official Information Act query to DOC in 2026 drew the reply that over the many years a total of 8 instances of predation of these disadvantaged kea had been caught on in-nest cameras, involving a rat, a possum, and six stoats (35).

3. The experiment: Kemp et al., 2018

DOC’s kea review (2) and website (36) cite Kemp et al. (2018) (37) as the evidence that predator control increases kea nest survival. This study of kea nests, stoats and rodents in an area that was poisoned (Okarito) and one that was not (Fox-Paringa) ran from 2008–2013, and a final report was produced in 2015 (38, Appendix IV). Truncated (2009–2012), modelled data was used in the 2018 publication (in the New Zealand Journal of Ecology (37)).

The scientific rigor of the study was poor – the monitoring predisposed kea to trouble – but also, bias was clear from the outset, with Kemp stating in his 2011 Milestone Report to the AHB, “For nest survival in the treated area to be significantly higher at the p<0.05 level, then about 80% of the nests need to fledge at South Okarito during the 2011 season following aerial 1080 treatment. This is within the realms of possibility” (39). There was no replication (an attempt was abandoned), there was non-random assignment of treatment to area (the AHB intended to poison Okarito), and the observers were open to bias because they knew of the treatment (38, Appendix IV). The results from the study (preliminary in Table 1 of the PDF; final in Table 2 below and Appendix IV; published as truncated modelled data by Kemp et al., 2018 (37)) are the evidence used to back the claim (2, 36) that kea nest survival benefits from poisoning. (Neither of DOC’s other measures of kea productivity in the study (nest initiation and fledglings per nest) were considered to show any effect of the poisoning treatment.) The number of kea poisoned was not mentioned; in North Okarito, seven of nine monitored kea were killed (78% (41)). Table 1, results to 2012 from the Milestone Report to the AHB (Kemp, 2012 (40)), is an image taken from that report and is in the PDF. It carries the source document’s Official Information Act release mark, so the figures are best read there. It gives survival of kea nest contents through regular monitoring in two areas, one (Okarito) poisoned in 2011. Table 2. Final results to 2013: survival of kea nest contents throughout regular monitoring, in two areas, one (Okarito) poisoned in 2011. See Table 3, Appendix IV (Kemp et al., 2015 (38)), for full details.

Site	Season	No. nests observed	No. nests reared a chick
Okarito	2008	3	3
Okarito	2009	5	2
Okarito	2010	7	3
Okarito poisoned	2011	9	9
Okarito	2012	7	5
Okarito	2013	4	3
Fox-Paringa	2010	7	3
Fox-Paringa	2011	6	3
Fox-Paringa	2012	7	0
Fox-Paringa	2013	5	1

Why the poisoning is repeated

The rodent graphs from the 2015 report (38, Appendix IV) show why aerial 1080 is spread repeatedly: a large rise in mouse numbers occurred within three months, followed by steeply increasing rats, then stoats (truncated from the 2018 paper). This pattern, involving a large rat irruption, is a standard response to broadcasting 1080, fuelled by the resources left unused by dead animals (42–45). Poisoning then needs to be repeated frequently to control the rats (46).

Conclusions

There is no scientifically obtained evidence that supports poisoning kea to improve their reproduction. Studies have found remarkably little evidence of nest predation by mammals, given the opportunities created by monitoring.

Due to the unknown, but likely precarious state of the species, an urgent halt to killing, capture and all interference² is strongly indicated. The kea’s Absolute Protection status is not being acknowledged. Neither is its vulnerability to weather conditions, and starvation, which was the biggest natural killer of kea in Jackson’s studies (25):

“In the autumn as nesting finishes adult Keas may die, as they go into moult, perhaps in poor condition. The greatest mortality for all Keas is June to September when food is most scarce … tables show the large loss of Keas in their first year, especially soon after banding. Some of these Keas were inexperienced, recently fledged Keas. Others were banded and died in the spring when there is a large moving population, spurred by hunger …”

“The bad Season – October 1957–May 1958 … in each month the rainfall was greatly above average, especially from October to March, probably the significant months for the Kea … December was the worst month … There were frequent storms … This period caused a great reduction in the Kea population. Perhaps it was reduced to two thirds normal … Chicks were found dead in their nests, eating of eggs was first noticed, and many parent Keas disappeared. Usual foods failed … That summer there was little nectar, and the following winter few berries. Starvation was the immediate cause of death of many Keas that summer” (25).

Numbered references for the statements above are in the PDF, followed by Appendix I (DOC staff on telemetry), Appendix II (nest-monitoring records), Appendix III (internal reports on kea predation), Appendix IV (the 2015 Okarito report, released under the Official Information Act), and the addendum by the late Bill Benfield.

¹ This plan and the published results (4) repeatedly refer to kea as “ground dwelling”. However: “Young leaves and buds of forest trees form the main food of Keas. At all seasons they spend much time in the tree tops (at Arthurs Pass in mountain beech) steadily and slowly picking buds …” (30).

² Includes KCT: decades of blood sampling for lead; genetic sampling (500 samples taken); bait aversion trials. ZIP: bait aversion trials; communal feeding (which alters behavior and creates a disease transmission risk).

Related reading on this site

• Kea in Trouble
• Towards ethical and effective conservation of New Zealand’s natural heritage (peer-reviewed paper, Conservation, 3 September 2025)
• Frightening lack of science behind super-poisoning plans (July 2023)