Content uploaded by Kerry Weston
Author content
All content in this area was uploaded by Kerry Weston on Oct 07, 2017
Content may be subject to copyright.
Capturing the cryptic: a comparison of detection methods
for stoats (Mustela erminea) in alpine habitats
Des H. V. Smith
A
and Kerry A. Weston
B,C
A
Wildland Consultants Ltd, PO Box 33499, Barrington, Christchurch 8244, New Zealand.
B
Science and Policy Group, Department of Conservation, Private Bag 4715, Christchurch Mail Centre,
Christchurch 8140, New Zealand.
C
Corresponding author. Email: kweston@doc.govt.nz
Abstract
Context. The ability to monitor the spatial distribution and abundance of species is essential for detecting population
changes, and assessing the progress of conservation management programs. Stoats (Mustela erminea)areaserious
conservation pest in New Zealand, but current monitoring methods are not sensitive enough to detect stoats in all
situations.
Aims. We compare the effectiveness of the most commonly employed method for monitoring mustelids in New Zealand,
footprint-tracking tunnels, with two alternative detection methods, camera traps and artificial nests. We were interested
in determining whether alternative detection methods were more sensitive in detecting stoats than tracking tunnels.
Methods. We established a network of tracking tunnels, artificial nests and camera traps within alpine habitat. Devices
were checked for stoat detections weekly across two seasons, in spring–early summer and autumn. Differences in detection
rates and cost effectiveness among methods were analysed among seasons.
Key results. In spring–early summer, the time to first stoat detection using footprint-tracking tunnels was 61 days,
compared with 7 days for camera traps and 8 days for artificial nests. The rate of stoat detection using artificial nests was
significantly higher than it was using tracking tunnels (coef = 3.05 1.29, P= 0.02), and moderately higher using camera
traps (coef = 1.34 1.09, P= 0.22). In autumn, when overall detectability of stoats was higher, there was no significant
difference in detection rates among the three methods, although camera traps again recorded the earliest detection.
Artificial nests were the most cost effective detection method in both seasons.
Conclusions. Artificial nests and camera traps were more efficient at detecting stoats during their spring breeding season
(when they are known to be difficult to detect), compared with the more established footprint-tracking tunnel method.
Artificial nests have potential to be developed into a monitoring index for small mammals, although further research is
required. Both methods provide an important alternative to footprint tracking indices for monitoring stoats.
Implications. Our study demonstrated the importance of calibration among different monitoring methods, particularly
when the target species is difficult to detect. We hypothesise that detection methods that do not rely on conspicuous,
artificially constructed devices, may be more effective for monitoring small, cryptic mammals.
Additional keywords: mustelid, relative abundance indices, trail camera, track surveys.
Received 25 August 2016, accepted 16 June 2017, published online 6 October 2017
Introduction
Monitoring the spatial distribution and abundance of species
is essential for detecting population changes over time, and
assessing progress against the objectives of conservation
management. However, for cryptic species occupying remote
habitats, complete population counts are difficult. For this
reason, incomplete counts or indices of relative abundance are
often used (Thompson et al.1998). Indices of relative abundance
are effective only if they track change in population size over
time (Engeman 2005). An important first step is that they are
able to detect a species given it is present, and that there is no
spatial or temporal variation in this (Thompson et al.1998). If
an index is unable to reliably detect target species at certain
times of year, or in particular habitats, conservation management
decisions based on the index will be unreliable.
Globally, footprint-tracking indices are a common method
of monitoring small mammals such as American mink (Mustela
vison) in the United Kingdom (Bonesi and Macdonald 2004;
Thompson 2006), striped skunks (Mephitis mephitis), racoons
(Procyon lotor) and stoats (Mustela erminea) in Canada (Martin
and Fahrig 2012); and feral cats (Felis catus), dingoes (Canis
dingo) and red foxes (Vulpes vulpes) in Australia (Edwards et al.
2000). In New Zealand, the footprint-tracking tunnel index,
which utilises baited ink cards inside plastic tracking tunnels,
CSIRO PUBLISHING
Wildlife Research
http://dx.doi.org/10.1071/WR16159
Journal compilation CSIRO 2017 www.publish.csiro.au/journals/wr
has been developed and standardised in forested habitat for
monitoring stoats and other small mammals (Brown and Miller
1998; Gillies and Williams 2013).
Stoats are naturally distributed throughout the Holarctic
region of the northern hemisphere (King and Powell 2006).
Since their introduction to New Zealand in the 1880s, in an
attempt to control rabbits, stoats have become major predators
of indigenous fauna across a full suite of available habitat
types (Elliott 1996; Elliott et al.1996; McLennan et al.1996;
Wilson et al.1998; Ratz and Murphy 1999; Cuthbert et al.2000;
Murphy et al.2004; Smith et al.2008b;O’Donnell et al.2015,
2016). Stoat populations erupt periodically in southern beech
(Nothofagaceae) forests in response to mast seedfall events (King
1983; Dilks et al.2003). However, in most years and habitat
types, stoats in New Zealand are at low densities and are sparsely
distributed, having large home-range sizes and dispersing
extensive distances (Murphy and Dowding 1994; Murphy and
Dowding 1995; Alterio et al.1998; Smith and Jamieson 2005;
Smith et al.2007).
There is growing concern that tracking tunnels are not
always sensitive enough to confirm presence of stoats, especially
in situations where stoat density and probability of detection
are low (Choquenot and Ruscoe 2000; Gillies and Williams
2013; Pickerell et al.2014). In particular, stoats are known
to be difficult to detect during their spring breeding season
(September to November; Murphy and Dowding 1994;
Murphy and Dowding 1995). Rock wren (Xenicus gilviventris)
is a threatened alpine passerine that also breeds during spring
(Michelsen-Heath 1989). Recent video monitoring at rock
wren nests showed high rates of nest predation by stoats,
despite stoats never being detected using tracking tunnels
(Kerry Weston, Department of Conservation, unpubl. data).
This raises important questions regarding whether it is possible
to monitor stoats accurately, within different habitats and
seasons, such as in alpine habitat during the spring rock wren
breeding season.
However, other monitoring techniques are available. Remote-
camera trapping using digital infrared technology (Cutler
and Swann 1999; Swann et al.2004; McCallum 2013) has
demonstrated potential to detect and monitor a variety of
medium- to large-sized mammals across a diversity of habitats
(Rowcliffe et al.2008; Tobler et al.2008; McCallum 2013);
however, the relative effectiveness of this method for detecting
and monitoring small, fast-moving mammal species needs to
be further evaluated (Cutler and Swann 1999; Kelly and
Holub 2008; De Bondi et al.2010; Glen et al.2014; Anile
and Devillard 2016). Artificial nests, which typically use a real
egg paired with a fake egg, made of some medium that allows
predator-teeth marks to be identified, are traditionally used as a
proxy to study nest survival, and measure variation in predation
rates associated with different nest, habitat, spatial and temporal
variables (Major and Kendal 1996). The use of artificial nests
to study nest predation has been widely criticised as being a
poor predictor of reproductive success in real bird populations
(Faaborg 2004; Thompson and Burhans 2004; Robinson et al.
2005). However, little consideration has been given to the
possibility that they may have value as a tool for detecting and
indexing small mammalian predators (Møller 1989; Major and
Kendal 1996).
Stoats, like other mustelids, are thought to have a neophobic
response towards foreign objects (Brown and Miller 1998;
Zuberogoitia et al.2006; King et al.2009). It is, therefore,
possible that tracking tunnels, which require the stoat to enter
them, may evoke neophobic responses and this may be more
pronounced in some habitats or during certain times of the
year. Camera traps and artificial nests may be less likely to
cause a neophobic response.
So as to address concerns around monitoring stoats in New
Zealand, particularly in alpine areas during the spring breeding
season, we set up a trial to compare the effectiveness of tracking
tunnels, camera traps and artificial nests in detecting stoats. Our
primary objective was to determine how long tracking tunnels
take to detect stoats at an alpine site during spring–early summer,
and to see how this compares with camera-traps and artificial
nests. We also aimed to determine whether the relative efficiency
of these devices in detecting stoats changes in autumn, when
stoats are easier to detect. If camera traps or artificial nests are
more effective at detecting stoats, then stoat monitoring in New
Zealand may need to be reviewed, particularly in alpine habitat,
which may also have important implications for small mammal
monitoring elsewhere.
Materials and methods
Study site
We undertook this research at Bealey Spur and Cass-Lagoon
Saddle, located on public conservation land near Arthur’s Pass,
South Island, New Zealand (43020S, 171330E). The study
area ranged in altitude from 1000 m to 1600 m asl, and largely
comprised alpine grassland (Chionochloa spp.), with mixed
alpine scrub and cushion vegetation. Below the study area,
mountain beech (Fuscospora cliffortioides) forest predominates,
which has an altitudinal limit resulting in a treeline at 900–1000 m
asl. Interspersed within the mountain beech forest are areas of
regenerating manuka (Leptospermum scoparium) and at one
location, a small exotic Douglas fir(Pseudotsuga menziesii)
plantation (Fig. 1).
Detection methods
Tracking tunnels
Tracking tunnels consisted of a black polypropylene plastic
cover (350 mm (W) 900 mm (L) 1.5 mm thick) stapled onto
a heavy wooden base (100 mm (W) 535 mm (L) 25 mm
thick) with a wire bracket fixed at each end (Gillies and
Williams 2013). This tunnel design was selected to minimise
disturbance from kea (Nestor notabilis) and prevent tunnels
from collapsing beneath snow or blowing away in the exposed
alpine conditions. For all tracking tunnels, Black Trakka Ink
Cards (Gotcha Traps, Warkworth, New Zealand) were used to
record footprints. Tunnels were baited with a 30 mm 30 mm
cube of fresh rabbit (Oryctolagus cuniculus) meat. Bait was
placed in the middle of the card, and the tracking card was
secured inside the tunnel by a stopper nail at either end of the
wooden base. Eight tracking tunnel lines were established at
Bealey Spur and Cass-Lagoon Saddle (Fig. 1). Tunnels were
established 3 weeks before commencement of the study to
allow resident stoats to become familiar with the tunnels in
their environment. The spatial layout of tracking-tunnel lines
BWildlife Research D. H. V. Smith and K. A. Weston
followed the standardised method for monitoring stoats in New
Zealand (Gillies and Williams 2013), with five tracking tunnels
spaced 100 m apart per line. Each line was at least 1 km apart to
achieve spatial independence following the recommendation of
Brown and Miller (1998). The standard protocol is to run tunnels
for three nights, with a minimum of four surveys per year between
November and February, the period of peak stoat abundance
(Gillies and Williams 2013). However, to determine time until
first detection of stoat footprints, we ran tracking tunnels
continuously throughout two field seasons. Season 1 was from
20 October 2014 to 6 January 2015, and Season 2 was from 7
March 2015 to 10 May 2015. Initial checks were made after the
first three nights, as per the standard tracking-tunnel protocol
(Gillies and Williams 2013), followed by weekly checks until
the end of each field season. The bait was replaced during each
check and ink cards were replaced if they had been tracked.
Artificial nests
The design of artificial nests followed the method of Smith
et al.(2008a) where a nest bowl was created at the base of
tussocks, and a domestic chicken (Gallus gallus) egg was paired
with a similar-sized and -shaped white plastaline egg (Van
Aken plastaline modelling clay, Takapuna Art Supplies, New
Zealand). The plastaline egg had a wire loop through it that was
pegged to the ground, preventing it from being easily removed
by a predator. Stoats are able to remove the hen egg, but leave
distinguishable incisor and/or canine bite marks (Fig. 2) in the
plastaline egg when they try to remove it. Artificial nests were
deployed in clusters containing three artificial nests 2–3 m apart.
Artificial-nest clusters were paired with tracking-tunnel stations,
by placing them at a perpendicular distance of 300 m from the
tunnel in the centre of each line (Fig. 1). Artificial-nest clusters
were checked weekly, at the same time as the tracking-tunnel
inspections. However, if a nest had been preyed on, it was not
replaced. Identification of teeth imprints on artificial nests used
the methods described in Smith et al.(2008a), with additional
reference examples collected from the captive stoat population
at the Johnston Memorial Laboratory, at Lincoln University,
New Zealand.
Camera traps
Fourteen Keepguard 12MP 720P infrared digital camera units
(Keepway Industrial (Asia) Co.) were placed throughout the
study area (Fig. 1). These units use a passive infrared motion
sensor and LED infrared flash to detect and record passing
animals. Camera units were not paired with the tracking-tunnel
lines, but were placed at the mid-point of 1-km
2
grid cells
used for the Topo50 NZTM series maps (Fig. 1). For some
Legend
Artificial nest
Camera trap
Tracking tunnel
0 0.75 1.50 km
N
Fig. 1. Layout of stoat-monitoring devices within alpine habitat near Arthurs Pass, South Island, New Zealand.
A comparison of detection methods for stoats Wildlife Research C
cameras, this centring was offset to ensure that the cameras fell
within alpine habitat if they were otherwise too close to the
ecotonal boundary with beech forest. Camera units were secured
to a metal stake ~30 cm above the ground and were baited with
a large chunk of fresh rabbit meat (~200 g), which was pegged
down on the ground ~1.5 m in front of the camera. To minimise
false triggering and to gain a clear field of view for video
footage, dense vegetation within the field of view of the
camera was avoided and tussocks were tied back or trimmed
where necessary (Kelly and Holub 2008). Camera units were
programmed to record a 30-s segment of video each time
they were triggered, with a 30-s time delay between triggers.
This time delay was to avoid excessive number of ‘events’
triggered by individuals that remained at the bait for a
prolonged period (De Bondi et al.2010). As with the tracking
tunnels and artificial nests, cameras were checked weekly, at
which time they were re-baited, and AA batteries and SD
cards were replaced.
Cost comparison
To compare cost effectiveness among the three detection
methods, we calculated and partitioned costs into initial set-up
costs and ongoing costs per monitoring occasion. We then
compared the cost for each method, by using the number of
monitoring occasions required for a monitoring technique to
detect a stoat in each season.
Statistical analysis
The timing of stoat detections by camera traps was identified
using the time stamp on the video footage. For tracked tracking
tunnels and preyed-on artificial nests, timing of detections was
identified as the date of the weekly checks and, consequently,
estimates of time to detection for these methods may be biased
upward. Given the large size of stoat home ranges and the scale
of their movements (Murphy and Dowding 1994,1995; Smith
and Jamieson 2005; Smith et al.2007), the presence of stoat
tracks in at least one tunnel was considered a detection for that
line. For artificial nests, the imprints of stoat teeth in one nest
comprised a detection for that cluster.
Right-censored Kaplan–Meier survival curves (Kaplan and
Meier 1958) were used to estimate the median time to first
detection for each method. A separate Kaplan–Meier curve
was used for each season. Given that the number of detections
was low, particularly in Season 1 when tracking tunnels recorded
only two detections, confidence intervals were not able to be
estimated for all methods. Cox’s proportional-hazard models
(Crawley 2007) were used to assess the relationship among
detection rates for each method and season. This was undertaken
by comparing two Cox’s proportional-hazard models using
Akaike’s information criterion (AIC; Burnham and Anderson
2002); the first model included an interaction effect between
method and season, and the second considered only differences
between the two seasons. Analyses were undertaken within the
package ‘survival’(Therneau 2014) in R version 3.2.3 (R Core
Team 2013).
Tracking-tunnel lines, artificial nest clusters, and camera-trap
sites were not independent between the two seasons. However,
there were insufficient data for mixed-effects modelling.
Consequently, most of the analyses described above, with the
exception of the Cox’s proportional-hazard models, are within
season. Residuals from the Cox’s proportional-hazard models
were plotted and checked for correlation between repeated
measures to ensure the model validity.
(a)
(b)
(c)
Fig. 2. Predator tooth imprints left on artificial plastaline eggs within
alpine habitat near Arthurs Pass, South Island, New Zealand. (a) Stoats
made small holes with their canines and distinctive combed trenches with
incisors, (b) mice left characteristic fan-fluted marks and (c) possums often
chewed eggs and left substantial gouging.
DWildlife Research D. H. V. Smith and K. A. Weston
Results
Kaplan Meir analysis showed that the rate at which the three
different monitoring methods detected stoats was highly variable
in spring–early summer (Fig. 3a). The first detection by a camera
trap was on Day 7 (median 43.5 days), and on Day 8 for artificial
nests (median 27.5 days); in comparison, tracking-tunnel lines
did not detect a stoat until early summer, i.e. Day 61 (median
69 days). In autumn, detection rates were more rapid for all
three methods (Fig. 3b). Stoats were detected by all three
methods at, or before, the first weekly check, with cameras
detecting stoats first, after 2 days (median =9 days; 95%
CI = 4–19). Both tracking tunnels and artificial nests had
detected stoats at the first weekly check, with a median
detection time of 11 days. Overall, median time to detection
was significantly longer for spring–early summer (39 days;
95% CI = 30–72 days) than for autumn (9 days; 95%
CI = 5–16 days).
The Cox’s proportional-hazard model including an interaction
effect between season and detection method better described
stoat detection rates than did the model including the effect of
season alone (DAIC = 25.94). In spring–early summer, the rate
of stoat detection using artificial nests was significantly higher
(coef = 3.05 1.29, P= 0.02) than that using tracking tunnels,
and moderately higher than that using camera traps (coef =
1.34 1.09, P= 0.22). In autumn, the differences among
detection methods were not significant; detection rates using
artificial nests (coef = 0.35 1.91, P= 0.46) and cameras
(coef = 0.48 1.06, P= 0.33) were not significantly different
from those using tracking tunnels.
Cost comparison
The initial set-up costs of the three detection methods also
varied considerably, ranging from NZ$590 for artificial nests
to NZ$5316 for trail cameras (Table 1). The most expensive
component of expenditure was the purchase of the digital-camera
units. Labour comprised a substantial proportion of the cost for
tracking-tunnel monitoring, owing to the construction of heavy-
duty wooden base tunnels suitable for longer-term use in alpine
conditions and their transport by field staff into the study area.
The subsequent costs associated with each monitoring occasion
were less variable (NZ$340–518) and largely comprised labour,
travel, consumables (bait and ink cards) and analysis (Table 2).
Time taken to check the devices was similar, as approximately
the same distance on the ground was covered by field staff
travelling among the different devices. However, the hours
associated with analysis of camera footage increased the ongoing
costs for this detection method (Table 2). The quantity of camera
footage collected (and, hence, requiring analysis) depended
largely on the number of recording events triggered, although
we estimated this to average one 8-h day per monitoring occasion.
In spring, a stoat was detected at the first monitoring occasion
for both artificial nests and trail cameras, producing a total
estimated cost to detect at least one stoat of NZ$930 and
NZ$5834 respectively. For tracking tunnels, nine monitoring
occasions were required to achieve one stoat detection. The
total estimated cost for a tracking-tunnel stoat detection in
spring–early summer was NZ$5406. In autumn, given that all
three detection methods had detected a stoat at the first monitoring
occasion, costs were much more similar among the three
methods, namely NZ$930 for artificial nests, NZ$2174 for
tracking tunnels and NZ$5834 for trail cameras. This, of
course, is assuming the hypothetical that set-up costs were the
same and the required equipment had not already been purchased.
Discussion
Stoats are notoriously difficult to detect or trap; and have been
shown to actively avoid traps and baited tunnels within their
home ranges for considerable periods, especially during their
spring breeding season and when natural prey are abundant
1.0
0.8
0.6
0.4
0.2
0.0
020406080
Day
02010 4030 50 60
Da
y
Proportion detected
1.0
0.8
0.6
0.4
0.2
0.0
Proportion detected
Artifical nests Cameras Tracking tunnels
(a)
(b)
Fig. 3. Rates at which three different monitoring methods, namely,artificial-
nest clusters, camera traps and tracking-tunnel lines, detected stoats (Mustela
erminea) during two seasons, (a) spring–early summer and (b) autumn, within
alpine habitat near Arthurs Pass, South Island, New Zealand.
A comparison of detection methods for stoats Wildlife Research E
(Murphy and Dowding 1994; Murphy and Dowding 1995;
Alterio et al.1999; King et al.2009; Veale et al.2013). In
the present study, we were unable to detect alpine stoats during
spring using conventional tracking-tunnel methods. Despite
their low detectability, stoats are active predators during
spring, which is the time when threatened indigenous alpine
bird species are highly vulnerable during nesting; thus, this
is an important time to detect and monitor stoats (Michelsen-
Heath 1989; Maxwell 2001;O’Donnell et al.2016). Our research
suggested that alternative detection methods such as artificial
nests and camera traps are more efficient and cost effective at
detecting stoats during spring than are tracking tunnels.
In autumn, overall detectability was higher, and the relative
efficiency and costs of the three detection methods were more
similar. Seasonal variability in the relative efficiency of different
monitoring methods has also been shown in previous studies
(Bonesi and Macdonald 2004; Zuberogoitia et al.2006; Vine
et al.2009; Pickerell et al.2014). For example, American
mink sign surveys in England exhibited peaks during the mink
breeding season, although the number of mink trapped during
these months only weakly related to the proportion of sign
detected (Bonesi and Macdonald 2004). In Australia, the
detection rates of red foxes using spotlighting peaked in
autumn and was lowest winter, although there was little
seasonality in detection rates of the foxes using camera traps
(Vine et al.2009).
Camera traps were the first method to detect stoats in both
seasons, indicating that they are an effective early detection
method. During spring, the median detection time for camera
traps was 25.5 days earlier than the median detection time for
tracking tunnels. Camera traps are, therefore, likely to be the most
time-efficient detection device in situations such as predator-free
wildlife sanctuaries, where wildlife managers require efficient
and accurate detection tools to both evaluate the success of
any initial predator control and rapidly detect any subsequent
reinvasion. Another distinct advantage of camera traps, as
highlighted in the present study, is the time stamp, which
creates an accurate record of when the animal was first detected.
However, the initial expenditure associated with establishing
camera trapping is high, because of the outlay of electronic
equipment required. Consequently, they are likely to be of most
advantage over alternative methods when detectability is low.
Artificial nests were the most cost-effective detection method,
and yielded the highest overall detection rates in spring, when
detectability was low, suggesting that this technique has potential
to be developed into a monitoring index for small mammals.
Despite the widespread use, and criticism of artificial nests as
a surrogate in nest-survival studies, they have rarely been
explored as a method of detecting small mammals (Major and
Kendal 1996; Getzlaff et al.2013). Although, the target species
of the present study was stoats, our artificial nests also recorded
imprints of rodents and brushtail possums (Trichosurus
vulpecula; Fig. 2), suggesting that artificial nests may be
useful as a method of detecting and indexing other small
mammals. Given the wide range of predators we recorded
on plastaline eggs, it is possible that some stoat imprints
were either misidentified or masked by other predator species.
Before artificial nests are employed more widely as a
monitoring method, a calibration study using camera traps
should be undertaken to estimate the rate at which teeth
Table 1. A comparison of set-up costs for three stoat (Mustela erminea)-monitoring methods within alpine habitat near Arthurs Pass, South Island, New
Zealand. All costs are calculated in New Zealand dollars for the Year 2014 and are exclusive of GST
Tracking tunnels Artificial nests Trail cameras
Item Quantity Unit cost Total cost Item Quantity Unit cost Total cost Item Quantity Unit cost Total cost
Wood base 40 $2 $80 Plastaline 4 $8.50 $34 Keepguard 12MP digital camera 14 $209 $2926
Plastic covers 40 $ 6.50 $260 4-mm wire 1 $10 $10 Metal security box 14 $28 $392
4-mm wire 1 $40 $40 Hen eggs 2 cartons $3 $6 AA rechargable batteries 112 $6 $672
Labour 52 h $25 $1300 Labour 18 h $25 $450 4-mm wire 1 $10 $10
Travel 300 km $0.30 $90 Travel 300 km $0.30 $90 8 GB SD cards 28 $16 $448
Warratah 14 $15 $210
Metal hose clip 28 $6 $168
Labour 16 h $25 $400
Travel 300 km $0.30 $90
Total cost $1770 $590 $5316
Table 2. A comparison of costs per monitoring occasion for three stoat (Mustela erminea)-monitoring methods within alpine habitat near Arthurs Pass,
South Island, New Zealand. All costs are calculated in New Zealand dollars for the Year 2014 and are exclusive of GST
Tracking tunnels Artificial nests Trail cameras
Item Quantity Unit cost Total cost Item Quantity Unit cost Total cost Item Quantity Unit cost Total cost
Rabbit bait 40 $0.20 $8 Labour 8 $25 $200 Rabbit bait 14 $2 $28
Ink cards 40 $1.40 $56 Travel 300km $0.30 $90 Labour 8 $25 $200
Labour 8 $25 $200 Analysis 2 h $25 $50 Travel 300 km $0.30 $90
Travel 300 km $0.30 $90 Analysis 8 h $25 $200
Analysis 2 h $25 $50
Total $404 $340 $518
FWildlife Research D. H. V. Smith and K. A. Weston
imprints are either misidentified or lost as a result of secondary
visits by other species.
Characteristics of a detection technique that increase
sensitivity by maximising the number of interactions are clearly
desirable (Engeman 2005). It is assumed that artificial nests
give the appearance of an abandoned or unattended nest, and,
therefore, may be less likely to evoke neophobia. For cameras
traps, the approach is also passive, with the lure placed on open
ground and representing carrion. The camera is also placed
some distance from the lure and animals are often detected
before they interact with the lure. In contrast, tracking tunnels
and traps placed in the landscape are highly conspicuous and
require the target animal to enter, eliciting a cautious response.
It would now be useful to test whether different devices cause
different neophobic responses, with trials investigating stoat
behaviour around passive baits compared with tunnels and
traps.
The higher detection rates of artificial nests over the other
methods during spring could be linked with the foraging
behaviour of the stoats present during the spring season. Given
that in New Zealand, young stoats do not start venturing from
their dens until early December, stoats detected before this
time must have been adults (at least 1-year old) that have
survived over winter (King and Powell 2006). These adults
could be more experienced at avoiding tunnels, and rather,
focussed on searching for nests, which are common in the
alpine zone during spring. By contrast, in autumn, the annual
cohort of stoat young will boost the numbers available to be
detected, with inexperienced animals also likely to be more
naïve to artificial devices.
Artificial nests may have been negatively biased by our
protocol of removing nests without replacement once they had
been preyed on. This is in contrast with tracking tunnels and
camera traps, for which we replaced tracking cards and SD
cards each time. If artificial nests had been replaced once they
had been preyed on and monitoring of eggs continued, then
the overall number of stoat detections for artificial nests may
have also been higher, particularly in autumn when detection
rates were higher.
Carefully considered trail-camera placement is vital for
obtaining good-quality footage, as has been previously
acknowledged in numerous camera-trapping field trials
(Swann et al.2004; Kelly and Holub 2008; Vine et al.2009;
De Bondi et al.2010; Glen et al.2014). We were able to
minimise false triggering of the cameras associated with
moving vegetation; however, they were still triggered by high
winds, sun glare and heavy precipitation, which often filled up
SD cards and depleted battery life. Maintaining a wide field of
vision is beneficial when attempting to detect fast-moving,
cryptic animals such as mustelids; however, this comes with
a trade off against increased processing time of the footage
attributed to false triggers. The development of effective
image-recognition software for the automated identification of
target predators will assist in increasing the efficiency of camera
traps as a monitoring technique.
Whether, or to what extent, footprint tracking indices are
appropriate for monitoring small mammals may depend on the
species and the environment, and whether the animal is required
to enter an artificially constructed device. Silveira et al.(2003)
compared the use of camera traps, track surveys and transect
counts for wildlife in grassland habitat on the central–western
Brazilian plateau and found that track surveys were best for
estimating species richness. However, the sampling approach
in the study of Silveira et al.(2003) was passive, with the
sampling unit for tracks being long sections of dirt road;
which did not require animals to enter artificially constructed
devices. Edwards et al.(2000) also found passive track surveys
along roads to be a time-efficient method for indexing feral cat
and dingo abundance in Australia. However, they noted that
spotlight surveys offered higher precision. These studies, and
ours, have demonstrated the importance of calibration among
different monitoring methods.
Conclusions
So as to conserve threatened wildlife, conservation managers
need to be able to monitor threats when and where they occur.
However, in New Zealand, wildlife managers found that a well
established footprint tracking-tunnel index was unable to detect
stoats at alpine sites during the Austral spring, even though
frequent predation of rock wren nests by stoats at these sites
was recorded on remote trail cameras. Our research has shown
that artificial nests and camera traps are more efficient and
cost effective than are tracking tunnels at detecting stoats
during spring; however, in autumn, tracking tunnels are just as
effective at detecting stoats. The length of time and seasons
through which a stoat-monitoring program needs to run will
depend on the specific objectives of that program. For example,
if the objective is to detect stoats in spring, then camera traps
or artificial nests will be required. However, if detection of
stoats in spring is not required, then tracking tunnels alone
may suffice. Artificial nests have been widely criticised as a
proxy for studying nest predation, but have been overlooked
as a method of detecting and indexing small mammals. We
recommend that further work into developing artificial nests
as a small mammal index is undertaken. However, we also
stress that camera traps exhibit enormous potential for the
early detection of small and fast-moving, cryptic mammals.
Conflicts of interest
The authors declare no conflicts of interest.
Acknowledgements
We thank Colin O’Donnell and Jo Monks, Department of Conservation,
who helped with project design and provided useful discussions
around monitoring objectives. We also thank Paul van Dam-Bates, Ian
Westbrooke and Georgina Pickerell for contributing advice on survival
analysis, and the staff at the Johnston Memorial Laboratory, at Lincoln
University for collecting reference plastaline eggs from the enclosures of
captive predators on our behalf. A huge thanks will go to all of those that
assisted with field work: Kathrin Affeld, Fraser Maddigan, Tess Carney,
James Holborow, Jamie McAulay, Ian Westbrooke, Aaron Hamilton and
Derek Brown. Lastly, we thank Jamie Sanderlin, Carolyn King and an
anonymous referee for constructive comments on earlier drafts of this
manuscript.
References
Alterio, N., Moller, H., and Ratz, H. (1998). Movements and habitat use of
feral house cats Felis catus, stoats Mustela erminea and ferrets Mustela
A comparison of detection methods for stoats Wildlife Research G
furo, in grassland surrounding yellow-eyed penguin Megadyptes
antipodes breeding areas in spring. Biological Conservation 83,
187–194. doi:10.1016/S0006-3207(97)00052-9
Alterio, N., Moller, H., and Brown, K. (1999). Trappability and densities of
stoats (Mustela erminea) and ship rats (Rattus rattus) in a South Island
Nothofagus forest, New Zealand. New Zealand Journal of Ecology 23,
95–100.
Anile, S., and Devillard, S. (2016). Study design and body mass influence
RAIs from camera trap studies: evidence from the Felidae. Animal
Conservation 19,35–45. doi:10.1111/acv.12214
Bonesi, L., and Macdonald, D. W. (2004). Evaluation of sign surveys as
a way to estimate the relative abundance of American mink (Mustela
vison). Journal of Zoology 262,65–72. doi:10.1017/S0952836903
004448
Brown, J. A., and Miller, C. J. (1998). ‘Monitoring Stoat (Mustela erminea)
Control Operations: Power Analysis and Design.’https://ir.canterbury.ac.
nz/handle/10092/11221 [Accessed 11 April 2014]
Burnham, K., and Anderson, D. (2002). Information and likelihood theory:
a basis for model selection and inference. In ‘Model Selection and
Multimodel Inference: a Practical Information-Theoretic Approach’.
(Eds K. Burnham, D. Anderson.) pp. 49–97. (Springer: New York.)
Choquenot, D., and Ruscoe, W. A. (2000). Mouse population eruptions in
New Zealand forests: the role of population density and seedfall. Journal
of Animal Ecology 69, 1058–1070. doi:10.1046/j.1365-2656.2000.
00462.x
Crawley, M. J. (2007) ‘The R Book.’(John Wiley and Sons: Chichester,
UK.)
Cuthbert, R. J., Sommer, E., and Davis, L. S. (2000). Seasonal variation
in the diet of stoats in a breeding colony of Hutton’s shearwaters. New
Zealand Journal of Zoology 27, 367–373. doi:10.1080/03014223.
2000.9518246
Cutler, T. L., and Swann, D. E. (1999). Using remote photography in
wildlife ecology: a review. Wildlife Society Bulletin 27, 571–581.
De Bondi, N., White, J. G., Stevens, M., and Cooke, R. (2010). A comparison
of the effectiveness of camera trapping and live trapping for sampling
terrestrial small-mammal communities. Wildlife Research 37, 456–465.
doi:10.1071/WR10046
Dilks, P., Willans, M., Pryde, M., and Fraser, I. (2003). Large scale stoat
control to protect mohua (Mohoua ochrocephala) and kaka (Nestor
meridionalis) in the Eglinton Valley, Fiordland, New Zealand. New
Zealand Journal of Ecology 27,1–9.
Edwards, G., De Preu, N., Shakeshaft, B., and Crealy, I. (2000). An evaluation
of two methods of assessing feral cat and dingo abundance in central
Australia. Wildlife Research 27, 143–149. doi:10.1071/WR98067
Elliott, G. P. (1996). Productivity and mortality of mohua (Mohoua
ochrocephala). New Zealand Journal of Zoology 23, 229–237.
doi:10.1080/03014223.1996.9518082
Elliott, G. P., Dilks, P. J., and O’Donnell, C. F. (1996). The ecology of
yellow-crowned parakeets (Cyanoramphus auriceps)inNothofagus
forest in Fiordland, New Zealand. New Zealand Journal of Zoology
23, 249–265. doi:10.1080/03014223.1996.9518084
Engeman, R. M. (2005). Indexing principles and a widely applicable
paradigm for indexing animal populations. Wildlife Research 32,
203–210. doi:10.1071/WR03120
Faaborg, J. (2004). Truly artificial nest studies. Conservation Biology 18,
369–370. doi:10.1111/j.1523-1739.2004.00486.x
Getzlaff, C. L., Sievwright, K. A., Hickey-Elliott, A. B., and Armstrong, D. P.
(2013). Predator indices from artificial nests and tracking tunnels:
do they tell the same story? New Zealand Journal of Ecology 37,
232–239.
Gillies, C. A., and Williams, D. (2013). ‘DOC Tracking Tunnel Guide
v2.5.2: Using Tracking Tunnels to Monitor Rodents and Mustelids.’
(Department of Conservation: Hamilton, New Zealand.)
Glen, A., Warburton, B., Cruz, J., and Coleman, M. (2014). Comparison
of camera traps and kill traps for detecting mammalian predators: a field
trial. New Zealand Journal of Zoology 41, 155–160. doi:10.1080/
03014223.2014.898667
Kaplan, E. L., and Meier, P. (1958). Nonparametric estimation from
incomplete observations. Journal of the American Statistical Association
53,457–481. doi:10.1080/01621459.1958.10501452
Kelly, M. J., and Holub, E. L. (2008). Camera trapping of carnivores:
trap success among camera types and across species, and habitat
selection by species, on Salt Pond Mountain, Giles County, Virginia.
Northeastern Naturalist 15, 249–262. doi:10.1656/1092-6194(2008)15
[249:CTOCTS]2.0.CO;2
King, C. M. (1983). The relationship between beech Nothofagus sp.
seedfall and populations of mice Mus musculus, and the demographic
and dietary responses of stoats (Mustela erminea); in three New Zealand
forests. Journal of Animal Ecology 52, 141–166. doi:10.2307/4593
King, C. M., and Powell, R. A. (2006) ‘The Natural History of Weasels
and Stoats: Ecology, Behavior, and Management.’2nd edn. (Oxford
University Press: New York.)
King, C. M., McDonald, R. M., Martin, R. D., and Dennis, T. (2009). Why is
eradication of invasive mustelids so difficult? Biological Conservation
142, 806–816. doi:10.1016/j.biocon.2008.12.010
Major, R. E., and Kendal, C. E. (1996). The contribution of artificial nest
experiments to understanding avian reproductive success: a review of
methods and conclusions. The Ibis 138, 298–307. doi:10.1111/j.1474-
919X.1996.tb04342.x
Martin, A. E., and Fahrig, L. (2012). Measuring and selecting scales of
effect for landscape predictors in species–habitat models. Ecological
Applications 22, 2277–2292. doi:10.1890/11-2224.1
Maxwell, J. M. (2001). Fiordland takahe: population trends, dynamics and
problems. In ‘The Takahe: Fifty Years of Conservation Management
and Research’. (Eds W. Lee and I. Jamieson.) pp. 61–79. (University of
Otago Press: Dunedin, New Zealand.)
McCallum, J. (2013). Changing use of camera traps in mammalian field
research: habitats, taxa and study types. Mammal Review 43, 196–206.
doi:10.1111/j.1365-2907.2012.00216.x
McLennan, J., Potter, M., Robertson, H., Wake, G., Colbourne, R., Dew, L.,
Joyce, L., McCann, A., Miles, J., and Miller, P. (1996). Role of predation
in the decline of kiwi, Apteryx spp., in New Zealand. New Zealand
Journal of Ecology 20,27–35.
Michelsen-Heath, S. (1989) The breeding biology of the rock wren,
Xenicus gilviventris, in the Murchison Mountains, Fiordland National
Park, New Zealand. M.Sc. Thesis, University of Otago, Dunedin, New
Zealand.
Møller, A. P. (1989). Nest site selection across field–woodland ecotones:
the effect of nest predation. Oikos 56, 240–246. doi:10.2307/3565342
Murphy, E. C., and Dowding, J. E. (1994). Range and diet of stoats (Mustela
erminea) in a New Zealand beech forest. New Zealand Journal of Ecology
18,11–27.
Murphy, E. C., and Dowding, J. E. (1995). Ecology of the stoat in Nothofagus
forest: home range, habitat use and diet at different stages of the beech
mast cycle. New Zealand Journal of Ecology 19,97–109.
Murphy, E. C., Keedwell, R. J., Brown, K. P., and Westbrooke, I. (2004).
Diet of mammalian predators in braided river beds in the central South
Island, New Zealand. Wildlife Research 31, 631–638. doi:10.1071/
WR03033
O’Donnell, C. F. J., Clapperton, B. K., and Monks, J. M. (2015).
Impacts of introduced mammalian predators on indigenous birds of
freshwater wetlands in New Zealand. New Zealand Journal of Ecology
39,19–33.
O’Donnell, C. F. J., Weston, K. A., and Monks, J. M. (2016). Impacts of
introduced mammalian predators on New Zealand’s alpine fauna. New
Zealand Journal of Ecology 41,1–22.
HWildlife Research D. H. V. Smith and K. A. Weston
Pickerell, G. A., O’Donnell, C. F. J., Wilson, D. J., and Seddon, P. J. (2014).
How can we detect introduced mammalian predators in non-forest
habitats? A comparison of techniques. New Zealand Journal of
Ecology 38,86–102.
R Core Team (2013). ‘R: a Language and Environment for Statistical
Computing.’(R Foundation for Statistical Computing: Vienna, Austria.)
Ratz, H., and Murphy, B. (1999). Effects of habitat and introduced
mammalian predators on the breeding success of yellow-eyed penguins
Megadyptes antipodes, South Island, New Zealand. Pacific Conservation
Biology 5,16–27. doi:10.1071/PC990016
Robinson, W. D., Styrsky, J. N., Brawn, J. D., and Stouffer, P. (2005). Are
artificial bird nests effective surrogates for estimating predation on real
bird nests? A test with tropical birds. The Auk 122, 843–852. doi:10.1642/
0004-8038(2005)122[0843:AABNES]2.0.CO;2
Rowcliffe, J. M., Field, J., Turvey, S. T., and Carbone, C. (2008). Estimating
animal density using camera traps without the need for individual
recognition. Journal of Applied Ecology 45, 1228–1236. doi:10.1111/
j.1365-2664.2008.01473.x
Silveira, L., Jácomo, A. T. A., and Diniz-Filho, J. A. F. (2003). Camera
trap, line transect census and track surveys: a comparative evaluation.
Biological Conservation 114, 351–355. doi:10.1016/S0006-3207(03)
00063-6
Smith, D. H. V., and Jamieson, I. (2005). Lack of movement of stoats ( Mustela
erminea) between Nothofagus valley floors and alpine grasslands, with
implications for the conservation of New Zealand’s endangered fauna.
New Zealand Journal of Ecology 29,45–52.
Smith, D. H. V., Wilson, D. J., Moller, H., Murphy, E. C., and van Heezik, Y.
(2007). Selection of alpine grasslands over beech forest by stoats
(Mustela erminea) in montane southern New Zealand. New Zealand
Journal of Ecology 31,88–97.
Smith, D. H. V., Wilson, D. J., Moller, H., and Murphy, E. C. (2008a). Using
artificial nests to explore predation by introduced predators inhabiting
alpine areas in New Zealand. New Zealand Journal of Zoology 35,
119–128. doi:10.1080/03014220809510109
Smith, D. H. V., Wilson, D. J., Moller, H., Murphy, E. C., and Pickerell, G.
(2008b). Stoat density, diet and survival compared between alpine
grassland and beech forest habitats. New Zealand Journal of Ecology
32, 166–176.
Swann, D. E., Hass, C. C., Dalton, D. C., and Wolf, S. A. (2004). Infrared-
triggered cameras for detecting wildlife: an evaluation and review.
Wildlife Society Bulletin 32, 357–365. doi:10.2193/0091-7648(2004)32
[357:ICFDWA]2.0.CO;2
Therneau, T. (2014). ‘A Package for Survival Analysis in S. R Package
Version 2.37.’https://CRAN.R-project.org/package=survival [Accessed
11 August 2014]
Thompson, H. (2006). The use of floating rafts to detect and trap American
mink Mustela vison for the conservation of water voles Arvicola terrestris
along the River Wensum in Norfolk, England. Conservation Evidence 3,
114–116.
Thompson, F. R., and Burhans, D. E. (2004). Differences in predators
of artificial and real songbird nests: evidence of bias in artificial nest
studies. Conservation Biology 18, 373–380. doi:10.1111/j.1523-1739.
2004.00167.x
Thompson, W. L., White, G. C., and Gowan, C. (1998) ‘Monitoring
Vertebrate Populations.’(Academic Press: San Diego.)
Tobler, M., Carrillo-Percastegui, S., Leite Pitman, R., Mares, R., and Powell,
G. (2008). An evaluation of camera traps for inventorying large- and
medium-sized terrestrial rainforest mammals. Animal Conservation 11,
169–178. doi:10.1111/j.1469-1795.2008.00169.x
Veale, A., Edge, K. A., McMurtrie, P., Fewster, R., Clout, M., and Gleeson, D.
(2013). Using genetic techniques to quantify reinvasion, survival and
in situ breeding rates during control operations. Molecular Ecology 22,
5071–5083. doi:10.1111/mec.12453
Vine, S., Crowther, M., Lapidge, S., Dickman, C. R., Mooney, N., Piggott, M.,
and English, A. (2009). Comparison of methods to detect rare and
cryptic species: a case study using the red fox (Vulpes vulpes). Wildlife
Research 36, 436–446. doi:10.1071/WR08069
Wilson, P., Karl, B., Toft, R., Beggs, J., and Taylor, R. (1998). The role of
introduced predators and competitors in the decline of kaka (Nestor
meridionalis) populations in New Zealand. Biological Conservation
83, 175–185. doi:10.1016/S0006-3207(97)00055-4
Zuberogoitia, I., Zabala, J., and Martínez, J. A. (2006). Evaluation of sign
surveys and trappability of American mink: management consequences.
Folia Zoologica 55, 257–263.
A comparison of detection methods for stoats Wildlife Research I
www.publish.csiro.au/journals/wr