This is an HTML version of an attachment to the Official Information request 'Post 1080 poison drop monitoring west coast 2017-2018'.
Kemp et al.: Kea survival during aerial 1080 dr
New Zealand Jour
nal of Ecology (2019) 43(1): 0-0 © 2018 New Zealand Ecological Society. 
Kea survival during aerial poisoning for rat and possum control
Joshua R. Kemp*, Corey C. Mosen, Graeme P. Elliott, Christine M. Hunter and Paul van Klink
Biodiversity Group, Department of Conservation, Private Bag 5, Nelson 7010, New Zealand
*Author for correspondence (Email: [email address])
Published online: 21 September 2018
Abstract: The kea (Nestor notabilis) is a highly intelligent and adaptable omnivorous New Zealand parrot. 
These traits potentially put kea at risk of poisoning during vertebrate pest poisoning operations. However, as 
kea fall prey to introduced pests, they also gain from pest control, creating a cost-benefit situation. Pest control 
in kea habitat is mainly by aerial 1080, the distribution of sodium fluoroacetate poison pellets by helicopter. 
Understanding the net outcome for kea of this pest control method is extremely important because kea are 
endangered and aerial 1080 use is controversial. We use 222 monitoring cases of individually marked kea at 19 
aerial 1080 operations to model kea survival of aerial 1080 operations with respect to five variables. Proximity 
to human-occupied sites where kea scrounge human food was inversely related to survival; the odds of survival 
increased by a factor of 6.9 for remote kea compared to those that lived near scrounging sites. High survival in 
remote areas is explained by innate neophobia and a short field-life of prefeed baits, which together preclude 
acceptance of poison baits as familiar food. Elevated risk to kea living near scrounging sites is explained by 
learned neophilia, possibly exacerbated by lead poisoning. Survival was also related to the history of aerial 1080 
treatment at a site; the odds of survival increased by a factor of 21.3 at sites with repeated operations compared 
with first time treatments. This effect is possibly due to selection for neophobic phenotypes. We suggest that 
1080 poisoning risk management for kea should focus on reducing human food availability through an advocacy 
campaign. If most kea have not been fed by humans, then the long term outcome of the South Island aerial 
1080 programme should be positive for the kea.
Keywords: aerial 1080; non-target risk; pest control; poisoning risk
New Zealand is to apply a single ‘prefeed’ of non-toxic baits 
1–6 weeks prior to poison baiting (Veltman & Westbrooke 
The use of aerial 1080 in New Zealand and consequences 
2011; Elliott & Kemp 2016) to improve kill rates (Coleman 
for native birds
et al. 2007; Nugent et al. 2011).
Aerial 1080 (sodium fluoroacetate) operations are increasingly 
Aerial 1080 operations can have both positive and negative 
used in New Zealand to control mammalian pests. Pest control 
effects on native birds. For example, Powlesland et al. (1999) 
campaigns in 2014 and 2016 saw an unprecedented 554 000 
found mortality rates of 9.7% and 55% for North Island 
ha and 729 000 ha, respectively, treated with aerially applied 
robins (Petroica australis longipes) following two aerial 1080 
1080 baits (New Zealand Department of Conservation unpubl. 
operations. However, overall robin abundance increased in both 
data). These operations target brushtail possums (Trichosurus 
cases one year after aerial 1080 baiting due to greatly improved 
vulpecula) and ship rats (Rattus rattus) to reduce predation 
recruitment in the absence of key mammalian predators. The 
on native and endemic species and/or to reduce the incidence 
Powlesland et al. (1999) study highlights the importance of 
of bovine tuberculosis in livestock (Coleman & Caley 2000). 
assessing the net effect of aerial 1080 for native species at the 
Aerial 1080 baits are cereal-based pellets, weighing 6 g 
population level over several years. Long term net outcomes 
or 12 g, with a field life ranging from a few days to a few 
are most often gauged by trends in relative abundance (e.g. 
weeks depending on rainfall (Bowen et al. 1995; Frampton 
O’Donnell & Hoare 2012; Greene et al. 2013; Van Vianen 
et al. 1999). Poison pellets contain 1.5 mg g−1 of 1080 toxin; 
et al. 2018) or by detailed studies of reproductive success and 
hence a 6 g pellet contains 9 mg of active ingredient and a 
survival (e.g. Powlesland et al. 1999, 2003). Understanding the 
12 g pellet contains twice this amount. The pellets are flung 
net effect of aerial 1080 for native species is of considerable 
from a hopper slung under a helicopter at coverage rates of 
conservation importance in New Zealand because it is one of 
1–3 kg ha−1 (Elliott & Kemp 2016). Most pellets fall to the 
few proven pest control methods that can be applied on a large 
ground, where they lie until they are consumed by animals 
scale in rugged terrain at reasonable cost (Wright 2011), and 
or degraded by moisture. Rodents and possums find pellets at 
because of public interest in the method (Green & Rohan 2012).
night by smell and directly consume them. Modern practice in 
DOI: 10.20417/nzjecol.43.2

New Zealand Journal of Ecology, Vol. 43, No. 1, 2019
particularly stoats (Mustela erminea), which die from secondary 
poisoning (Murphy et al. 1999). To balance this positive 
impact against non-target 1080 poisoning risk, and to find risk 
mitigation pathways, requires improved quantification of risk 
Mt Arthur
and an understanding of its spatial and temporal variability.
Wangapeka ●
●● Nelson Lakes
Scrounging by kea as a potential influence on 1080 
poisoning risk
An outcome of the kea’s adaptability that could possibly affect 

Arthur's Pass
1080 poisoning risk is the exploitation of humans for food 

Glacier Country

by some individuals. Human food is obtained by stealing, 

soliciting, and scavenging food items, directly from people 
Abbey Rocks
● Mt Cook
and indirectly from rubbish bins and open landfills. We use 
the term ‘scrounging’ to describe this phenomenon. Young kea 
are initially attracted to scrounging sites by the presence of 

Treble Cone
other kea and then learn to scrounge by emulation, particularly 
●●● ● Milford Road
of adult males (Diamond & Bond 1991; Gajdon et al. 2004). 
Scrounging by kea is patchy in the landscape and can only 
be observed daily in three main areas of intensive tourist 
activity: (1) Milford Road, (2) West Coast ‘Glacier Country’ 
(Fox and Franz Josef Glaciers), and (3) Arthur’s Pass; and at 
some ski areas and high-use back-country huts, e.g. on the 
Milford Track (Fig. 1). Scrounging was previously common 
Figure 1.
at Mt Cook Village, but now is rare following changes in 
 Map of South Island, New Zealand, showing forests and 
scrub inhabited by kea (light green), locations of our study sites 
rubbish collection and disposal practices. In contrast, efforts 
(blue labels, blue shading shows aerial 1080 treatment boundaries) 
to prevent feeding of kea by visitors to the three places listed 
and sites where kea scrounge food from humans (pink).
above have been unconcerted and, at present, unsuccessful; 
kea continue to scrounge daily at these sites. In the remote 
back-country, scrounging at tramping huts was common prior 
to 1984, when open rubbish pits were provided by the forest 
service. The closure of these pits and a new ‘pack it in, pack 
The kea, a unique bird with positive and negative responses 
it out’ ethic has reduced kea visitation rates at huts to very 
to 1080
low levels. At scrounging sites kea are constantly exposed 
A species for which a negative short term impact of aerial 
to rewarding novel foods and become familiar with a wider 
1080 has been detected is the kea (Nestor notabilis) (Veltman 
range of food types than is normal in the wild. We hypothesise 
& Westbrooke 2011), a large endangered parrot endemic to 
that these experiences could lead to suppression of innate 
the South Island. Kea are sparsely distributed across about 
neophobic behaviours that might otherwise protect kea from 
three million hectares of indigenous South Island forests and 
1080 poisoning risk.
about 1.5 million hectares of adjoining sub-alpine shrublands, 
We evaluated the effect of proximity to scrounging sites, 
grasslands  and  herbfields  (Fig.  1)  (Robertson  et  al.  2007). 
plus four other potentially important variables, on kea survival 
Nearly all of the kea’s range is legally-protected montane 
through aerial 1080 operations. Specifically, we investigated 
wilderness, administered by the New Zealand Department of 
whether 1080 poisoning risk was:
Conservation (DOC). Roughly one-third (c. 27%) of the kea 
(1)  higher close to scrounging sites
species range had aerial 1080 applied between 2014 and 2016 
(2)  lower after repeated aerial 1080 treatments (i.e. at sites 
(DOC unpubl. data).
where aerial 1080 had been previously applied)
The kea is a distinctive member of the New Zealand 
(3)  higher with larger bait size
avifauna (de Kloet & de Kloet 2005), possessing several 
(4)  lower for adult kea than juvenile kea
traits that potentially elevate 1080 poisoning risk. Unlike kākā 
(5)  different for males and females.
(Nestor meridionalis), kea’s sympatric congener, which nest 
We could not evaluate whether prefeeding increased risk to kea 
in tree cavities and forage predominantly in the forest canopy 
because all aerial 1080 operations in the study used prefeed.
(Moorhouse 1997), kea nest underground (Jackson 1963) 
and frequently forage on the forest floor (Young et al. 2012; 
Greer et al. 2015). Kea are highly intelligent, omnivorous 
and adaptable, the latter particularly during the juvenile stage 
(Diamond & Bond 1991). Juvenile adaptability is a trait shared 
by many omnivorous generalists (Rozin 1976) and also by the 
We utilised planned aerial 1080 operations in the South Island 
congeneric kākā (Wilson et al. 1998; Loepelt et al. 2016). As 
to evaluate the research questions listed above. Between 2008 
such, kea are more likely than arboreal birds to encounter 1080 
and 2016, we monitored the survival of individual kea through 
pellets and may be more likely than specialist feeders to eat 
19 operations (n = 222 monitoring occasions) at 12 sites (Table 
them, thus giving rise to a potentially high negative impact 
1; Fig. 1). Seven of these sites were utilised once, three sites 
from non target mortality. However, positive consequences 
were utilised twice, and two sites were utilised three times. The 
of aerial 1080 for kea have also been reported (Kemp et al. 
number of kea monitored per operation (and per site) ranged 
2018), attributable to the control of invasive exotic predators, 
from 2 to 37 (Table 1; Fig. 2). We monitored 15 individuals 

Kemp et al.: Kea survival during aerial 1080 drops
Table 1. Details of the aerial 1080 operations in the South Island, New Zealand, at which 1080 poisoning risk to kea was 
measured. Number of previous aerial 1080 treatments and years since prior aerial 1080 are the basis for assignment of 
1080 history category scores. Counts of monitored kea are given by age and sex with numbers of deaths attributed to 1080 
poisoning in brackets. AF=adult female, AM=adult male, FYF=first year female, FYM=first year male, JF=second year 
female, JM=second year male.
Number of   
Years since  1080 
aerial 1080  prior aerial  history 
Month  size (g)  Scrounge  treatments 
category  AF 
Arawhata 2008 

2 (0) 
7 (0) 
0 (0)  0 (0)  0 (0)  1 (0)  10 (0)
Fox-Franz 2008 


3 (1) 
7 (4) 
0 (0)  0 (0)  3 (1)  4 (1)  17 (7)
Mt Arthur 2009 


2 (0) 
9 (0) 
0 (0)  0 (0)  0 (0)  2 (0)  13 (0)
Hawdon 2009 


3 (0) 
3 (0) 
1 (0)  1 (0)  1 (0)  1 (0)  10 (0)
Okarito 2011 


13 (2)  13 (2)  2 (1)  3 (1)  3 (2)  3 (0)  37 (8)
Wangapeka 2011 

4 (0) 
7 (0) 
0 (0)  0 (0)  1 (0)  1 (0)  13 (0)
Abbey Rocks 2011  Oct 

3 (0) 
4 (0) 
0 (0)  0 (0)  1 (0)  0 (0)  8 (0)
Copland 2012 

2 (0) 
0 (0) 
0 (0)  0 (0)  0 (0)  0 (0)  2 (0)
Hawdon 2012 


6 (0) 
0 (0) 
0 (0)  0 (0)  0 (0)  0 (0)  6 (0)
Otira 2013 

11 (2)  15 (3)  1 (0)  2 (0)  3 (0)  2 (0)  34 (5)
Mt Arthur 2014 

1 (0) 
3 (0) 
0 (0)  1 (0)  0 (0)  2 (0)  7 (0)
Anatoki 2014 


1 (0) 
1 (0) 
0 (0)  0 (0)  0 (0)  0 (0)  2 (0)
Wangapeka 2014 


4 (0) 
2 (0) 
1 (0)  1 (0)  0 (0)  0 (0)  8 (0)
Abbey Rocks 2014  Nov 


8 (0) 
10 (1)  2 (0)  0 (0)  0 (0)  1 (0)  21 (1)
Oparara 2014 


2 (0) 
1 (0) 
1 (1)  1 (1)  0 (0)  0 (0)  5 (2)
Hawdon 2014 


4 (0) 
0 (0) 
0 (0)  0 (0)  0 (0)  0 (0)  4 (0)
Rotoiti 2014 



1 (1) 
1 (0) 
0 (0)  0 (0)  0 (0)  0 (0)  2 (1)
Oparara 2016 


3 (0) 
2 (0) 
0 (0)  0 (0)  0 (0)  0 (0)  5 (0)
Wangapeka 2016 


8 (0) 
8 (0) 
2 (0)  0 (0)  0 (0)  0 (0)  18 (0)
81 (6)  93 (10)  10 (2)  9 (2)  12 (3)  17 (1)  222 (24)
Figure 2. Sample sizes and distribution of 
kea deaths with respect to sites at which kea 
survival through aerial 1080 operations was 
monitored in the South Island, New Zealand. 
All of the recorded kea deaths are attributed to 
direct consumption of 1080 baits. Blue = kea 
not poisoned, pink = kea poisoned.
Kea poisoned
Kea not poisoned
Number of k
y Rocks
Mt Ar

New Zealand Journal of Ecology, Vol. 43, No. 1, 2019
through a second aerial 1080 operation and then two of these 
distance from there to the nearest scrounging site (shown in 
were monitored through a third operation. Thus, the 222 
Fig. 1). By this method the monitored kea clearly comprised 
monitoring occasions were obtained from 205 different kea. 
two distinct groups, (1) kea living adjacent to (<20 km) and 
Monitoring was achieved using radio tags (n = 209 occasions) 
(2) kea living remote from (>40 km) scrounging sites (Fig. 3), 
or by observation of unique leg bands at active nest sites (n = 
with 110 and 112 cases in each group, respectively (Fig. 4a). 
13 occasions). Post aerial 1080 monitoring continued for at 
No monitored kea had average locations between 20–40 km 
least 3 weeks, by which time baits become non toxic due to 
from scrounging sites and none were recorded moving between 
rain leaching poison from the bait matrix (Bowen et al. 1995). 
groups during monitoring. We assume that this grouping 
Radio tags emitting a ‘mortality’ signal, indicating that the 
correlates with the likelihood of kea with cryptic scrounging 
transmitter was stationary, were retrieved as soon as possible, 
histories having scrounged as juveniles because movements 
usually within 24 hours of receiving this signal, to determine 
of 5–10 km are common for kea, movements of 20 km are 
whether the kea had died and the cause of death.
regular, but movements of >20 km are rare (DOC unpubl. 
data, Wilson & Brejaart 1992). Some kea in the ‘adjacent’ 
Factors affecting 1080 poisoning risk
group likely had no history of scrounging, and some long-
distance dispersers in the remote group probably scrounged as 
We surmised that a history of scrounging might increase 1080 
juveniles, but these exceptions make our results conservative. 
poisoning risk due to learned neophilia and/or suppressed 
The qualitative nature of this variable necessitates cautious 
neophobia. Scrounging histories could not be quantified for 
use of any quantitative predictions derived from it.
individual kea due to variability in the ages and locations at 
Assignment to a scrounging group of two kea monitored 
which birds were enrolled in the study. The scrounging histories 
at the Oparara 2014 aerial 1080 operation was complicated 
of kea first caught as adults are cryptic at >5 km from scrounging 
by the erection of a research hut in the study area partway 
sites because only adult male kea living <5 km from scrounging 
through the study. These two first-year kea, one male and one 
sites display an ongoing daily scrounging habit (scrounging 
female, were initially enrolled (i.e. captured and radio tagged) 
by adult females is rare) (pers. obs., Jackson 1960; Wilson & 
as ‘remote’ study birds in April 2014, at c. 4 months of age. 
Brejaart 1992). Adult females at all distances and adult males 
During spring 2014, the research hut was erected in the alpine 
at >5 km from a scrounging site were never directly observed 
zone about 800 metres from where these kea were caught. In 
scrounging but may have scrounged as juveniles.
the months leading up to this first-time aerial 1080 operation, 
We addressed the problem of cryptic individual scrounging 
these two kea were frequently observed at the hut, investigating 
histories by dividing our study kea into two groups, ‘adjacent’ 
the building and associated objects, materials and people. The 
and ‘remote’ based on distance to the nearest scrounging site. 
occupants of the hut intentionally avoided feeding the kea, but 
To assign each kea to a group we calculated an average location 
unguarded food was possibly stolen. In the analysis presented 
(latitude and longitude) as the mean of all positions recorded 
here, these two kea were scored as ‘remote’. We explored the 
for that kea during its monitoring history and measured the 
consequences of alternative treatment of these two kea with 
Figure 3. Histogram of distances between 
mean kea positions (obtained for each kea 
during monitoring history) and the nearest 
scrounging site where kea obtain food from 
humans (Fig. 1), showing a clear division 
of the sample into two scrounging groups 
‘adjacent’ and ‘remote’. All of the recorded 
kea deaths are attributed to direct consumption 
of 1080 baits. Blue = kea not poisoned, pink 
= kea poisoned.
Number of k
Distance to nearest scrounging site (km)

Kemp et al.: Kea survival during aerial 1080 drops
a. Scrounging
b. 1080 History
Figure 4. Numbers of kea monitored and 
the number that died with respect to the five 
modelled variables Scrounging, 1080 History, 
ea 90
ea 75
Toxic Bait Size, Age Class and Sex. Blue = 
kea not poisoned, pink = kea poisoned. Kea 
were assigned to a scrounging group based on 
distance to nearest scrounging site (‘adjacent’ 
Number of k
Number of k
≤20  km,  ‘remote’  ≥40  km,  Fig.  3).  1080 
History is a three-level categorical variable 
for which 1 = first time aerial 1080 treatment 
Scrounge group
1080 History
at site, 3 = repeated treatment 2–3 years after 
c. Toxic Bait Size
d. Age Class
previous treatment, and 2 = intermediate 
history including partial block treatments and 
longer intervals between treatments.
Number of k
Number of k
> 2 years
1−2 years
<1 years
Bait size (g)
e. Sex
Number of k
a. Study kea
Figure 5. Kernel densities (smoothed 
histograms) of distance to scrounging 
sites for the 222 kea monitoring cases 
in this study (a), the 4.5 million hectare 
kea species range (b), and the 27% of the 
kea species range treated with aerial 1080 
in the years 2014–2016 (c). Histograms 
were constructed using GIS generated 
distances between (1) mean locations 
for the individual study kea and the 
Distance (km) from scrounging sites
nearest scrounging site, and between (2) 
b. Entire species range
500 000 and (3) 100 000 random points, 
generated within the relevant polygons, 
and the nearest scrounging site.
Distance (km) from scrounging sites
c. Recent aerial 1080 programme
Distance (km) from scrounging sites

New Zealand Journal of Ecology, Vol. 43, No. 1, 2019
alternative analyses, in which the birds were (1) excluded 
Kea can be accurately assigned to yearly age classes until 2 
from the dataset, or (2) classified as ‘adjacent’.
years of age based on cere and eyelid colouration and plumage 
We assessed the representativeness of our sample by 
condition. Ceres and eyelids are bright yellow at hatching and 
visual comparison of kernel density (smoothed histograms) 
start fading to grey after 2 years. The rate of fading is variable 
of distance to the nearest scrounging site for (1) our sample 
after 2 years of age and some adult females retain faint yellow 
of monitored kea, (2) the kea species range, and (3) recent 
markings, making age cryptic beyond 2 years. We modelled 
aerial 1080 use (2014–2016) in kea habitat (Fig. 5a–c). Plots 
age as a three-level categorical variable comprising ‘first-year’ 
for (2) and (3) were constructed using distances from the 
(0–1 years), ‘juvenile’ (1–2 years), and ‘adult’ (3+ years) (Table 
nearest scrounging site to 500 000 and 100 000 random points, 
1). More adult than young individuals were monitored, with 
respectively, generated within the relevant polygons. Random 
174 adults : 29 juveniles : 19 first-years (Fig. 4d).
points and distances to scrounging sites were generated in the 
ESRI ArcMap geographic information platform. Comparing 
plots showed that our study kea poorly represented the South 
Sex was determined using body measurements (Bond et al. 
Island kea distribution (Fig. 5a, b). A representative sample 
1991) and/or by observing breeding behaviour (only female 
would have 14% of kea within 20 km of scrounging sites, 
kea incubate eggs). We expected that risk might vary between 
19%  20–40  km  from  scrounging  sites,  and  67%  >40  km 
the sexes due to dimorphic body size (females 750–800 grams, 
away. Hence, kea adjacent to scrounging sites were greatly 
males 900–1050 grams), and/or to the breeding biology of kea 
overrepresented in our study, and kea living 20–40 km were 
in which males forage more than females (adult males provide 
not represented. Similarly, our study kea do not reflect the 
food to females) (Table 1). The sex ratio of monitored birds 
full coverage of recent aerial 1080 use (Fig. 5a, c). During 
was approximately even, at 119 females : 103 males (Fig. 4e).
the period 2014–2016, 83% of aerial 1080 was applied >40 
km from scrounging sites (Fig. 5c).
Statistical analysis
Aerial 1080 history
After 3 weeks from the day of poison application, each 
monitored kea was scored as having either survived (0) or 
We expected that 1080 poisoning risk at a given site might 
died (1). We modelled this binary response using mixed-effects 
decrease with repeated aerial 1080 operations through selection 
models on the logit scale, with the predictor variables described 
for innately cautious phenotypes and/or the creation of bait 
above as fixed effects (Scrounging, 1080 History, Bait Size, 
shy individuals via sublethal poisoning (conditioned aversion). 
Age and Sex). Balance within the dataset was visually assessed 
We devised a three-tier classification system based primarily 
using pairwise plots (see Appendix S1 in Supplementary 
on the time since previous aerial 1080 treatment. Kea at sites 
Material). Non-independence of observations within sites was 
treated with aerial 1080 for the first time were assigned to 
category 1. Kea at sites that had been comprehensively treated 
addressed by specifying site as a random effect. We did not 
(entire block) within the previous 3 years were assigned to 
address the additional non-independence of the 17 monitoring 
category 3. Six operations did not clearly fall into either of 
occasions involving birds that we had previously monitored 
these categories due to extended inter-treatment intervals and/
because their 1080 history was modelled as a fixed effect. Social 
or partial treatments of blocks. Kea in these operations were 
interactions among kea may also cause non-independence, but 
assigned to category 2. Four of the 19 operations in the study 
we believe this effect will be limited because most of our birds 
were first time treatments. Eight of the operations were at 
were adult kea, which are largely solitary and unlikely to be 
sites treated within the previous 3 years. The remaining seven 
influenced by the behaviour of other kea. We could not test 
operations were at sites previously partially treated and/or 
for overdispersion in our data because established statistical 
treated 4 or more years prior (Table 1). Our sample comprised 
methods applicable to binary data are lacking.
37, 105 and 80 kea in categories 1–3, respectively (Fig. 4b).
Our model set comprised all possible additive combinations 
of the predictor variables, with no interactions due to limited 
Bait size
sample size. Modelling was implemented in R (R Core Team 
2016) using package MuMIn version 1.40.4. (Barton 2018). We 
The aerial 1080 operations used either 6 g or 12 g poison 
used Information Theoretic methods (AIC
pellets (Table 1) with equal toxin concentration (1.5 mg 
c) to rank models and 
assess the relative importance of the five modelled variables 
g−1). Because a 12 g bait contains twice as much toxin we 
(Akaike 1973; Burnham & Anderson 2002). For supported 
expected that these larger pellets might elevate risk to kea. 
variables we use the odds ratio to express the magnitude of 
Our sample comprised 76 and 146 records for 6 g and 12 g 
the effect, calculated as exponentiated model coefficients. For 
pellets, respectively (Fig. 4c).
example, the odds ratio showing the effect of variable y in a 
two-factor additive model is:
Kea have an extended juvenile phase, in which independence is 
attained at 1–6 months of age, followed by sexual maturity at 
𝑦𝑦 = 𝑒𝑒𝛽𝛽𝑖𝑖+𝛽𝛽𝑥𝑥+𝛽𝛽𝑦𝑦
3–4 years (JRK unpubl. data). Pre-adult kea tend to congregate 
where  β
into dynamic mixed-age flocks and can range over tens of square 
i,  βx  and  βy  are  the  model  coefficients  for  the 
intercept, variable x, and variable y, respectively. We used 
kilometres (Jackson 1960; Wilson & Brejaart 1992). Once 
the or_glm function in the oddsratio package version 1.0.2. 
mature, kea mate for life. Adult kea are relatively sedentary, 
for R (Schratz 2017) to calculate odds ratios and associated 
with core ranges spanning only a few square kilometres and 
95% confidence intervals.
social interactions are primarily with the mate and any recent 
offspring (JRK pers. obs.; Jackson 1960; Wilson & Brejaart 
1992). We expected that younger kea might be more susceptible 
to 1080 poisoning because they are in a phase of intensive 
learning and heightened behavioural flexibility.

Kemp et al.: Kea survival during aerial 1080 drops
Table 2. Ranking of the top ten alternative models by AICc of factors affecting kea survival through aerial 1080 operations. 
Scrounging = 2-level categorical variable (adjacent versus remote). 1080 History = 3-level categorical variable, scored per 
aerial 1080 operation based on time since previous aerial 1080 treatment (Table 1). k = number of parameters in the model.

Delta AICc Weight
Scrounging + 1080 History 

Scrounging + 1080 History + Age Class 

Scrounging + 1080 History + Sex 

Scrounging + 1080 History + Bait Size 

Scrounging + 1080 History + Age Class + Sex 

Scrounging + 1080 History + Age Class + Bait Size 

Scrounging + 1080 History + Sex + Bait Size 

Scrounging + 1080 History + Age Class + Sex + Bait Size 

1080 History 

9  6.663 0.013
Figure 6. Survival rates for kea exposed 
to aerial 1080, estimated using the 
best-supported model, with respect to 

Scrounging  (‘remote’  ≥40  km  from 

scrounging sites, ‘adjacent’ ≤20 km) and 

1080 History (1 = first time aerial 1080 
treatment at site, 3 = repeated treatment 

2–3 years after previous treatment and 2 
= intermediate history including partial 
block treatments and longer intervals 
between treatments). Error bars are 95% 
binomial confidence intervals.

al (%) 50
Sur 40
1080 History
strongly supported by AICc as a negative influence on the 
chance of a kea surviving an aerial 1080 operation (Table 2; 
We recorded 24 kea deaths, all within six operations (n = 222 
Fig. 6). The odds of survival was 6.9 times higher for kea in 
kea monitoring cases; Table 1). Poison-pellet cereal matter 
the ‘remote’ group than for the ‘adjacent’ group (95% CI = 
present in the digestive tracts of all necropsied corpses (n = 
2.1–31.6). An effect of 1080 History was also supported (Table 
17) indicated that primary poisoning was the cause of death. 
2). The odds of survival was 21.3 times higher at sites with 
Necropsies were not conducted in seven cases owing to 
a recent aerial 1080 treatment than at first time treatments 
advanced decomposition that resulted from logistical delays 
(95% CI = 3.6–406.9). The best-supported model contained 
retrieving  corpses  from  the  field.  The  seven  deaths  were 
both Scrounging and 1080 History. The Age Class variable 
considered for the study as due to 1080 poisoning.
also had moderate support. However, because this result may 
Membership of the ‘adjacent’ scrounging group was 
be an artefact of small sample sizes of young kea, we present 

New Zealand Journal of Ecology, Vol. 43, No. 1, 2019
further results for the best-supported model only. There was 
Therefore, we recommend further monitoring of young kea 
little to no support for Bait Size and Sex as predictors. The 
through aerial 1080 operations to improve our understanding 
random effect of Site was non-zero for all models that did not 
of how kea age influences 1080 poisoning risk.
include both Scrounging and 1080 History, but was near-zero 
for all models that included both.
Utility of estimates and extrapolation to other sites
Support for Age Class was absent in two alternative 
While primarily aiming to evaluate for a qualitative relationship 
analyses  in  which  two  first-year  kea  poisoned  during  the 
between scrounging and 1080 poisoning risk, our study has 
Oparara 2014 operation were either excluded or scored as 
generated a useful estimate of survival of remote kea through 
‘adjacent’ for scrounging (see Methods). The survival rates 
a repeated aerial 1080 operation. With only one death out of 
predicted by the best-supported model (Scrounging + 1080 
60 remote monitoring occasions in repeated operations, the 
History) for remote first time treatments improved considerably 
lower 95% confidence limit for kea survival in this context 
in the alternative analyses (Appendix S2), but neither alternative 
is 95.6% (Fig. 6). Given the high magnitude of benefits from 
suggested different conclusions or recommendations.
predator control measured for kea and other species with 
similar nesting ecology, such as kākā and whio (Hymenolaimus 
malacorhynchos) (Moorhouse et al. 2003; Whitehead et al. 
2008; Kemp et al. 2018), we expect to see strong kea population 
growth resulting from long term aerial 1080 programmes in 
Mechanisms behind 1080 poisoning risk
remote areas.
Our finding of lower poisoning risk from aerial 1080 operations 
The estimated survival rates derived from our analysis for 
for remote kea than for kea living adjacent to scrounging sites 
kea adjacent to scrounging sites should be treated with caution 
can be explained if scrounging causes suppression of innate 
due to the qualitative nature of our method for assigning kea to 
neophobic behaviours that protect against poisoning from novel 
the ‘adjacent’ and ‘remote’ groups (see Methods). In addition, 
foods (Rozin 1976; Galef et al. 1998; Galef & Whiskin 2001). 
aerial 1080 operations near scrounging sites vary in size, shape 
Suppression of neophobia could potentially be learned through 
and location relative to scrounging sites, and the intensity 
frequent exposure to rewarding novel foods, and/or a direct 
of scrounging activity varies from place to place. Further, 
result of sublethal lead poisoning. Lead poisoning is prevalent 
our samples possibly contained disproportionate numbers of 
among kea near scrounging sites (McLelland et al. 2010; Reid 
scrounging kea – these being presumably easier to catch for the 
et al. 2012) and this could potentially contribute to suppressed 
same reason that they are more prone to ingesting 1080 pellets 
neophobia through effects on brain function, thermoregulation, 
– hence possibly leading to over-estimated risk. Interpolation 
immune system function and body condition (Newth et al. 
of survival rates to kea living 20–40 km from scrounging sites 
2016). We could not assay our study birds for blood lead levels 
is especially to be avoided, for the reasons given above and 
for logistical reasons, so we could not distinguish the relative 
because the shape of the relationship between kea risk and 
importance of learned neophilia versus lead poisoning. Thus, 
distance is unknown. Notwithstanding these cautions, we 
in remote sites, initial interactions with 1080 pellets may 
conclude that 1080 poisoning risk near scrounging sites is cause 
comprise a mixture of neophilia and neophobia, in which kea 
for concern. Benefits from predator control would need to be 
may approach novel objects (neophilia), but initial interactions 
high and of extended duration (i.e. more than two breeding 
with the novel item are cautious (neophobia). The short field 
seasons) to achieve a positive net outcome under our lowest 
life of the 1080 pellets used in kea habitat appears to preclude 
mean survival estimate of 56.8% survival at first time aerial 
their acceptance as a familiar rewarding food in the remote 
1080 treatments adjacent to scrounging sites (Fig. 6). Should 
context. Thus, the potential for high 1080 poisoning risk to kea 
survival fall between the point estimate and the lower 95% 
does not necessarily manifest under current baiting practices.
confidence limit of 35.9%, achieving a positive net outcome 
Our finding of lessened risk where aerial 1080 had been 
in these circumstances is beyond reasonable expectations. 
used in recent years could be explained by (1) learned aversion 
Only 33% of the kea species range lies within 40 km of 
(bait shyness) induced by sublethal 1080 doses, and (2) selection 
scrounging sites and only 17% of recent (2014–2016) aerial 
for neophobic individuals due to mortality of neophilic ones 
1080 applications have been within this range (Fig. 5). Thus, 
from 1080 poisoning. However, bait shyness can be precluded 
costs to the kea population from aerial 1080 use adjacent to 
by using prefeed, at least in some animals (Ross et al. 2000), 
scrounging sites might have been offset by benefits accrued 
and all the operations in this study were prefed. Therefore, 
in remote areas. However, the limitations of our study and the 
we favour the latter explanation that strong selection pressure 
potential to rapidly reduce the prevalence of scrounging by kea 
applied by aerial 1080 results in an increasing prevalence of 
are such that we recommend immediate mitigation action for 
innate neophobic behaviours. Possibly the effect of 1080 history 
the benefit of kea population recovery and to address concern 
differs between ‘remote’ and ‘adjacent’ groups, but we had 
about kea welfare.
insufficient data to test for such an interaction. To enable this 
idea to be tested, we recommend further radio-tag monitoring 
Managing aerial 1080 poisoning risk
of kea at remote first-time treatments, if DOC’s and/or TbFree 
Approaches to managing 1080 poisoning risk to kea near 
New Zealand’s South Island aerial 1080 programme expands.
scrounging sites potentially include (1) reducing scrounging 
Our study found moderate support for an effect of Age 
by removing the rewards that kea obtain from anthropogenic 
Class on 1080 poisoning risk, which may reflect the enhanced 
sources, (2) not prefeeding aerial 1080 operations near 
behavioural plasticity observed for juveniles within the genus 
scrounging sites, (3) using low-strength 1080 pellets (0.8 mg 
Nestor (Wilson et al. 1998; Loepelt et al. 2016). We caution 
g−1 pellets are available for use in New Zealand), (4) adding 
that our data set contained relatively few young kea, and that 
bird repellents to baits, (5) aversion training of individual 
two poisoned first-year kea were scored as ‘remote’ although 
kea at scrounging sites, and (6) spatial solutions such as 
they are known to have scrounged around a newly built hut. 
attracting kea out of treatment blocks during aerial 1080 

Kemp et al.: Kea survival during aerial 1080 drops
operations, and capture and holding of kea during aerial 
1080 operations. However, not all these options are currently 
In summary, we are confident that aerial 1080 application is not 
viable. Not prefeeding will almost certainly compromise pest 
cause for concern for kea in repeated aerial 1080 operations at 
control effectiveness (Coleman et al. 2007; Nugent et al. 
remote sites, particularly when the operations are designed to 
2011), reducing benefits to native species other than kea. Bird 
maximise predator control benefits to kea, as risk to kea is easily 
repellents and aversion training are potentially viable solutions, 
offset by these benefits. Proximity to scrounging sites is strongly 
but both require research and development (Orr-Walker et al. 
associated with elevated 1080 poisoning risk, but this risk is 
2012; Cowan et al. 2016; Crowell et al. 2016). Spatial solutions 
not well quantified. We recommend (1) the implementation of 
involving catching and holding kea would be stressful for 
a scrounge-management programme, (2) research to improve 
the birds and some will always avoid capture. Attracting kea 
our understanding of the links between scrounging, neophobia 
away from 1080 blocks may not succeed, especially for adult 
and 1080 poisoning risk, (3) further measurement of risk to 
females due to their more sedentary nature. The remaining 
young kea, and (4) further measurement of risk at first treatments 
two options – reducing scrounging and using lower strength 
in remote sites. If the South Island aerial 1080 programme 
1080 pellets – are potentially viable and could be immediately 
comprises mostly remote repeated operations, then a positive 
implemented. Managers should focus first on preventing the 
long term net outcome for kea is likely.
establishment of new scrounging sites, which is a real risk 
in the context of rapidly expanding tourism in New Zealand. 
Second, the attractiveness of established scrounging sites should 
be actively reduced through careful management of waste and 
This research was carried out by DOC with co-funding from 
of human behaviour. The effectiveness of any such programme 
TBfree NZ, a subsidiary of OSPRI New Zealand. The Kea 
should be tested by re-evaluating aerial 1080 poisoning risk near 
Conservation  Trust  contributed  significantly,  through  the 
managed scrounging sites after 5 to 10 years of management. 
funding of ongoing field studies in the Hawdon and Wangapeka 
Scrounging may never be eliminated, but the current level of 
Valleys. Radio tagging and banding of kea was conducted in 
effort to prevent scrounging is very low (JRK pers. obs.) and 
accordance with Department of Conservation Animal Ethics 
the recent reduction of scrounging rates at back-country huts 
Guidelines. For their significant contributions to the field effort 
suggests similar reductions are achievable at Arthur’s Pass, 
that went into this study we thank Francesca Cunninghame, 
Glacier Country, and Milford Road. A successful scrounge-
Joseph Fraser, Troy Makan, Dan Palmer, Brent Barrett, Josh 
reduction programme would have the additional benefit of 
Adam, Jacinda Amey, Kirsty Moran, Robyn Blyth, Tom 
reducing mortality arising from other hazards such as lead 
Goodman, Lucy Rossiter and Liam Bolitho. We thank the 
and cars. Using lower strength 1080 pellets is not current best 
staff of the various DOC Offices that facilitated our field work. 
practice, but the recent change from 0.8 mg g−1 to 1.5 mg g−1 
Comments from Susan Timmins, Clayson Howell, Ximena 
was based on theory (Frampton et al. 1999) and benefits to 
Nelson and Alex Taylor greatly improved the manuscript, for 
pest control efficacy of the higher concentration have not been 
which we are grateful. We also thank Luis Ortiz-Catedral and 
experimentally demonstrated in the field.
Deb Wilson (NZJE referee and associate editor), plus one other 
anonymous referee for their valuable input to the manuscript.
Potential for prefeed to affect 1080 poisoning risk
We were unable to quantify the effect of prefeeding on 1080 
poisoning risk to kea because all of the operations in the study 
were prefed. However, risk to remote living kea is clearly not 
elevated to concerning levels by a single application of prefeed 
Akaike H 1973. Information theory and an extension of the 
at the sowing rates used in this study. Changes to 1080 baiting 
maximum likelihood principle. In: Petrov BN, Csaki F 
practices, such as using two rounds of prefeeding (e.g. Nugent 
eds. Budapest: Akademiai Kiado, Proceedings of the 
et al. 2011) or longer life baits such as Wanganui Number 
2nd International Symposium on Information Theory. 
7 pellets (Bowen et al. 1995), could elevate risk to kea and 
Pp. 267–281.
therefore should be accompanied by kea monitoring. The 
Barton K 2018. MuMIn: Multi-Model Inference; R 
high survival of remote adult kea is consistent with studies 
package version 1.40.4.
of kākā, in which adults exhibited a disinclination to explore 
package=MuMIn (accessed May 2018).
potential new food sources (Wilson et al. 1998; Loepelt et al. 
Bond AB, Wilson KJ, Diamond J 1991. Sexual dimorphism 
2016).  Juvenile  kākā,  conversely,  learned  to  exploit  novel 
in the kea Nestor notabilis. Emu 91: 12–19.
foods more readily. For kea, which learn socially (Diamond 
Bowen LH, Morgan DR, Eason CT 1995. Persistence of 
& Bond 1991; Gajdon et al. 2004), 1080 poisoning risk to 
sodium monofluoroacetate (1080) in baits under simulated 
juveniles may be naturally mitigated through these learning 
rainfall. New Zealand Journal of Agricultural Research 
processes. However, young kea sometimes forage and explore 
38: 529–531.
independently of adults, at which time they may investigate 
Burnham KP, Anderson DR 2002. Model selection and 
baits. If so, their investigation of baits may be coupled with a 
multimodel inference: a practical information-theoretic 
cautious testing strategy (Rozin 1976) within which a single 
approach. 2nd edn. New York, Springer-Verlag. 488 p.
prefeed  application  may  provide  sufficient  familiarisation 
Coleman JD, Caley P 2000. Possums as a reservoir of bovine 
time to elevate 1080 poisoning risk to a concerning level. 
TB. In: Montague TL ed. The brushtail possum. Biology, 
We recommend behavioural studies of the process by which 
impact and management of an introduced marsupial. 
foods become accepted by kea to improve understanding of 
Lincoln, Manaaki Whenua Press. Pp. 92–104.
the role of prefeeding in determining 1080 poisoning risk, and 
Coleman JD, Fraser KW, Nugent G 2007. Costs and benefits 
the potential implications of multiple prefeed applications.
of pre-feeding for possum control. New Zealand Journal 
of Zoology 34: 185–193.

New Zealand Journal of Ecology, Vol. 43, No. 1, 2019
Cowan P, Booth L, Crowell M 2016. Repellents with potential 
Fraser I, August A, August C 2003. Control of introduced 
to protect kea and other native birds from aerial poisoning 
mammalian predators improves kaka Nestor meridionalis 
for possum and rat control. New Zealand Journal of 
breeding success: reversing the decline of a threatened 
Ecology 40: 29–41.
New Zealand parrot. Biological Conservation 110: 33–44.
Crowell M, Martini M, Moltchanova E 2016. Effect of the 
Murphy  EC,  Robbins  L,  Young  JB,  Dowding  JE  1999. 
addition of bird repellents to aerially applied 1080 baits 
Secondary poisoning of stoats after an aerial 1080 poison 
on rat and possum abundance. New Zealand Journal of 
operation in Pureora forest, New Zealand. New Zealand 
Ecology 40: 49–59.
Journal of Ecology 23: 175–182.
Diamond J, Bond AB 1991. Social behavior and the ontogeny 
Newth JL, Rees EC, Cromie RL, McDonald RA, Bearhop 
of foraging in the kea (Nestor notabilis). Ethology 88: 
S, Pain DJ, Norton GJ, Deacon C, Hilton GM 2016. 
Widespread exposure to lead affects the body condition 
Elliott G, Kemp J 2016. Large-scale pest control in 
of free-living whooper swans Cygnus cygnus wintering 
New Zealand beech forests. Ecological Management & 
in Britain. Environmental Pollution 209: 60–67.
Restoration 17: 200–209.
Nugent G, Warburton B, Thomson C, Sweetapple P, Ruscoe WA 
Frampton CM, Warburton B, Henderson RJ, Morgan DR 1999. 
2011. Effect of prefeeding, sowing rate and sowing pattern 
Optimising bait size and 1080 concentration (sodium 
on efficacy of aerial 1080 poisoning of small-mammal 
monofluoroacetate) for the control of brushtail possums 
pests in New Zealand. Wildlife Research 38: 249–259.
(Trichosurus vulpecula). Wildlife Research 26: 53–59.
Orr-Walker T, Adams NJ, Roberts LG, Kemp JR, Spurr EB 
Gajdon GK, Fijn N, Huber L 2004. Testing social learning in a 
2012. Effectiveness of the bird repellents anthraquinone 
wild mountain parrot, the kea (Nestor notabilis). Learning 
and D-pulegone on an endemic New Zealand parrot, the 
& Behavior 32: 62–71.
kea (Nestor notabilis). Applied Animal Behaviour Science 
Galef BGJ, Rudolf B, Whiskin EE, Choleris E, Mainardi M, 
137: 80–85.
Valsecchi P 1998. Familiarity and relatedness: effects on 
O’Donnell C, Hoare J 2012. Quantifying the benefits of long-
social learning about foods by Norway rats and Mongolian 
term integrated pest control for forest bird populations in 
gerbils. Animal Learning and Behavior 26: 448–454.
a New Zealand temperate rainforest. New Zealand Journal 
Galef BGJ, Whiskin EE 2001. Interaction of social and 
of Ecology 36: 131–140.
individual learning in food preferences of Norway rats. 
Powlesland RG, Knegtmans JW, Marshall I 1999. Costs and 
Animal Behaviour 62: 41–46.
benefits of aerial 1080 possum control operations using 
Green W, Rohan M 2012. Opposition to aerial 1080 poisoning 
carrot baits to North Island robins (Petroica  australis 
for control of invasive mammals in New Zealand: risk 
longipes), Pureora Forest Park. New Zealand Journal of 
perceptions and agency responses. Journal of the Royal 
Ecology 23: 145–159.
Society of New Zealand 42: 185–213.
Powlesland RG, Wills DE, August ACL, August C 2003. 
Greene TC, Dilks PJ, Westbrooke IM, Pryde MA 2013. 
Effects of a 1080 operation on kaka and kereru survival 
Monitoring selected forest bird species through aerial 
and nesting success, Whirinaki Forest Park. New Zealand 
application of 1080 baits, Waitutu, New Zealand. 
Journal of Ecology 27: 125–137.
New Zealand Journal of Ecology 37: 41–50.
R Core Team 2016. R: a language and environment for statistical 
Greer A, Gajdon GK, Nelson XJ 2015. Intraspecific variation 
computing. R Foundation for Statistical Computing. 
in the foraging ecology of kea, the world’s only mountain- 
Vienna, Austria.
and rainforest-dwelling parrot. New Zealand Journal of 
Reid C, McInnes K, McLelland JM, Gartrell BD 2012. 
Ecology 39: 254–261.
Anthropogenic lead (Pb) exposure in populations of a 
Jackson JR 1960. Keas at Arthurs Pass. Notornis 9: 39–58.
wild parrot (kea Nestor notabilis). New Zealand Journal 
Jackson JR 1963. The nesting of keas. Notornis 10: 319–326.
of Ecology 36: 56–63.
Kemp JR, Mosen C, Elliott GP, Hunter CM 2018. Effects of 
Robertson CJR, Hyvonen P, Fraser M, Pickard CR 2007. Atlas 
the aerial application of 1080 to control pest mammals 
of bird distribution in New Zealand. Wellington, The 
on kea reproductive success. New Zealand Journal of 
Ornithological Society of New Zealand. 296 p.
Ecology 42: 158–168.
Ross JG, Hickling GJ, Morgan DR, Eason C 2000. The role 
de Kloet RS, de Kloet SR 2005. The evolution of the spindlin 
of non-toxic prefeed and postfeed in the development and 
gene in birds: sequence analysis of an intron of the 
maintenance of 1080 bait shyness in captive brushtail 
spindlin W and Z gene reveals four major divisions of the 
possums. Wildlife Research 27: 69–74.
Psittaciformes. Molecular Phylogenetics and Evolution 
Rozin P 1976. The selection of food by rats, humans, and other 
36: 706–721.
animals. Advances in the Study of Behaviour 6: 21–76.
Loepelt J, Shaw RC, Burns KC 2016. Can you teach an old 
Schratz P 2017. R package ’oddsratio’: Odds ratio calculation 
parrot new tricks? Cognitive development in wild kaka 
for GAM(M)s & GLM(M)S. R package version 1.0.2. 
(Nestor meridionalis). Proceedings of the Royal Society (accessed 
B 283: 20153056.
May 2018).
McLelland JM, Reid C, McInnes K, Roe WD, Gartrell BD 2010. 
Van Vianen J, Burge OR, MacFarlane AT, Kelly D 2018. The 
Evidence of lead exposure in a free-ranging population 
effects of single aerial 1080 possum-control operations on 
of kea (Nestor notabilis). Journal of Wildlife Diseases 
common forest birds in the South Island, New Zealand. 
46: 532–540.
New Zealand Journal of Ecology 42: 169–178.
Moorhouse R 1997. The diet of the North Island kaka 
Veltman CJ, Westbrooke IM 2011. Forest bird mortality and 
(Nestor meridionalis septentrionalis) on Kapiti Island. 
baiting practices in New Zealand aerial 1080 operations 
New Zealand Journal of Ecology 21: 141–152.
from 1986 to 2009. New Zealand Journal of Ecology 
Moorhouse RJ, Greene TC, Dilks PJ, Powlesland RG, Moran 
35: 21–29.
L, Taylor GE, Jones A, Knegtmans JW, Wills D, Pryde M, 
Whitehead AL, Edge K-A, Smart A, Hill G, Willans M 2008. 

Kemp et al.: Kea survival during aerial 1080 drops
Large scale predator control improves the productivity of a 
rare New Zealand riverine duck. Biological Conservation 
141: 2784–2794.
Wilson K-J, Brejaart R 1992. The kea: a brief research review. 
In: Joseph L ed. Issues in the conservation of parrots 
in Australasia and Oceania: challenges to conservation 
biology. Royal Australian Ornithologists Union Report 
no. 83. Victoria, Royal Australian Ornithologists Union. 
Pp. 24–28.
Wilson PR, Karl BJ, Toft RJ, Beggs JR, Taylor RH 1998. The 
role of introduced predators and competitors in the decline 
of kaka (Nestor meridionalis) populations in New Zealand. 
Biological Conservation 83: 175–185.
Wright J 2011. Evaluating the use of 1080: predators, poisons 
and silent forests. Wellington, Office of the Parliamentary 
Commissioner for the Environment. 85p.
Young  LM,  Kelly  D,  Nelson  XJ  2012.  Alpine  flora  may 
depend on declining frugivorous parrot for seed dispersal. 
Biological Conservation 147: 133–142.
Received 19 December 2017; accepted 27 June 2018
Editorial board member: Deb Wilson
Supplementary Material
Additional supporting information may be found in the 
supplementary material file for this article:
Appendix S1. Pairwise plots of modelled predictor variables 
of kea survival of aerial 1080 operations.
Appendix S2. Survival rates for kea exposed to aerial 1080, 
estimated using the top model, with respect to Scrounging and 
1080 History, with respect to three alternative categorisations of 
two juvenile kea monitored through the Oparara 2014 operation.
The New Zealand Journal of Ecology provides supporting 
information supplied by the authors where this may assist 
readers. Such materials are peer-reviewed and copy-edited 
but any issues relating to this information (other than missing 
files) should be addressed to the authors.