Content uploaded by Rachael L. Sagar
Author content
All content in this area was uploaded by Rachael L. Sagar on Nov 28, 2022
Content may be subject to copyright.
1Sagar et al.: Mast effects on subantarctic mice
Population trends of house mice during tussock mast seeding on Auckland Island
Rachael L. Sagar1* , Finlay S. Cox1, Stephen R. Horn1 and James C. Russell2
1Department of Conservation, PO Box 743, Invercargill 9840, New Zealand
2University of Auckland, Private Bag 92019, Auckland 1142, New Zealand
*Author for correspondence (Email: rsagar@doc.govt.nz)
Published online: 28 November 2022
New Zealand Journal of Ecology (2022) 46(3): 3497 © 2022 New Zealand Ecological Society.
DOI: https://doi.org/10.20417/nzjecol.46.3497
RESEARCH
Abstract: House mice (Mus musculus) are an invasive species on Auckland Island in the New Zealand
subantarctic and planning for their eradication is underway. Mast seeding events cause rodent populations to
irrupt, though little is known about this phenomenon in snow tussock grass (Chionochloa spp.) systems on
on Auckland Island for the 2 years following a mast event, and with which tools to monitor abundance, to
inform planning of bait application for eradication. Mouse populations were studied using kill and live trapping
at two sites on Auckland Island, and mouse density was estimated using spatially explicit capture-recapture
models. Mouse population density was highest during summer mast seeding of Chionochloa antarctica and
then declined the following winter and subsequently remained low for the following year. Breeding remained
seasonal, with a pulse in early summer and a very low level continuing through winter in both years, regardless
of mast conditions. These results are similar to those from other cool temperate Southern Ocean islands where
seasonal resource availability appears to drive breeding. Throughout the study the capture probability of mice
was generally higher when population density was lower, which highlights that conclusions about population
trends could be misleading if abundance indices are not calibrated to measures of population density. Mouse
eradication should preferentially take place outside of a mast event but would likely still succeed during and
for Southern Ocean islands, which is particularly important where eradications are planned.
Keywords: Chionochloa antarctica, density, spatially explicit capture-recapture, eradication, Mus musculus,
rodent, Southern Ocean
Introduction
The phenomenon of irruptive rodent population responses to
periods of high seed production (mast seeding or masting)
by grasses or forest trees is well documented (Bogdziewicz
et al. 2016). In New Zealand, these dynamics are best known
in southern beech (Nothofagus spp.) forest, where rodent
populations, including house mice (Mus musculus), irrupt
in response to periodic beech masting (King 1983). Beech
forest is not the only mast seeding system that drives mouse
population dynamics in New Zealand. Alpine grasses including
snow tussocks (Chionochloa spp.) have periodic mast seeding,
which has a similar impact on mouse population dynamics
on the main islands of New Zealand to beech mast events,
though evidence is limited to a single study (Wilson & Lee
2010). Mouse populations rise in the autumn following a mast
summer, and may remain high through the following winter,
spring and summer, then decline prior to the next winter,
likely owing to resource limitations and predation (King 1983;
Wilson & Lee 2010). Studies of mouse population trends in
response to cereal crop availability in Australia show that the
length of breeding seasons, population density and rate of
multiple factors including the population density at the start
of breeding, disease prevalence and the level of the preceding
winter’s rainfall (Singleton et al. 2005). Such a variable
other systems.
House mice are a highly successful invasive species,
present on all continents except Antarctica (Boursot et al.
1996). Their introduction has been almost entirely accidental,
assisted by human movement, or their range expanded through
Mice are present on many Southern Ocean islands where they
have negative impacts on native biodiversity (Courchamp et al.
2003; Angel et al. 2009). There is widespread documentation
of the consumption of plant matter and macroinvertebrates
by mice on Southern Ocean islands (e.g. Angel et al. 2009;
Houghton et al. 2019; Russell et al. 2020). Over the last decade
species, including predation on large live seabirds, has emerged
from Southern Ocean islands (Wanless et al. 2012; Dilley et al.
2015). Unsurprisingly, eradicating mice and other mammalian
pests from such islands is becoming an increasingly common
2 New Zealand Journal of Ecology, Vol. 46, No. 3, 2022
Holmes et al. 2019; Spatz et al. 2022).
Mice have been documented since 1840 on Auckland
Island, the main island in the Auckland Island group in the
New Zealand Subantarctic Islands Area (NZSIA; Fig. 1), and
and cultural values (World Heritage Convention 1998). The
nuclear DNA of Auckland Island mice reveals they are the
subspecies M. musculus domesticus, are distinct from other
mouse populations in New Zealand, and likely arrived with
sealers from North America (Veale et al. 2018). Mice are
found at all altitudes and across all habitats on Auckland Island
on Auckland Island are suspected to be driven by the pulsed
availability of seed from tussock mast seeding events, with
other plants and invertebrates sustaining smaller populations
between masting events (Harper 2010b; Russell et al. 2018,
2019).
The eradication of feral pigs (Sus scrofa), mice and feral
cats (Felis catus) from Auckland Island is considered feasible
(Horn et al. 2022) and planning for its implementation is
underway. There is a history of pioneering eradications on
Southern Ocean islands, and in particular within the NZSIA
(Russell et al. 2022 and references therein). To date the largest
successful mouse eradication in the world was on subantarctic
Macquarie Island (12 785 ha), which was cleared of rabbits
(Oryctolagus cuniculus), ship rats (Rattus rattus) and mice in
2014 (Springer 2016). Subantarctic Antipodes Island (2012
ha) in the NZSIA was cleared of mice in 2016 (Horn et al.
2019). If successful, the eradication of mice from Auckland
Island (45 891 ha) would be a 350% increase in the global land
area cleared of mice, a huge jump in the scale of operations.
The logistical challenges of completing bait spread at the
scale of Auckland Island requires a deviation from current best
practice (Horn et al. 2022; Livingstone et al. 2022; Oyston et
al. 2022). A lower bait sowing rate (two applications, each of
4 kg ha brodifacoum) and timing operations during summer
have been proposed, c.f. two applications, each of 8 kg ha
of brodifacoum during winter (best practice: Broome et al.
2017). A bait uptake trial simulating these adjusted methods
was successfully carried out on Auckland Island during
summer (early February) 2019 (Russell et al. 2019), leading
bait application strategy encompassed two baiting applications
a few weeks apart. However, during summer 2018/2019 a
mast-seeding of tussock occurred on Auckland Island. Tussock
mast events occur every few years and are suspected to cause
Island (Harper 2010b; Russell et al. 2018). Uncertainty over
the population dynamics of mice (e.g. density, home range size
and breeding activity) following a mast event on Auckland
Island translates to uncertainty in eradication success under
such circumstances. Potentially, mice will not consume toxic
bait when other food sources are plentiful, and/or they might
have fewer opportunities to encounter bait if population
densities are higher and home ranges smaller. Understanding the
legacy of mast events on mouse populations across Auckland
Island is important to inform the risk to success should an
eradication follow a mast event. The objective of this study
Island in relation to seeding cycles of the dominant tussock
grass Chionochloa antarctica. In addition, we compared two
methods of measuring mouse abundance.
Methods
Site description
This study was undertaken at two sites across three habitat types
on Auckland Island (45 891 ha), the main island in the Auckland
Island Motu Maha group (56 816 ha; 50.69°S, 166.08°E) in
the New Zealand subantarctic area, 465 km south of mainland
New Zealand (Fig. 1). The climate is characterized by strong,
prevailing westerly winds and frequent rain and cloud cover
and cool temperatures (2–12°C) all year round (Fraser 2020).
Three major distinct vegetation types cover the island. A thin
band of coastal forest (approx. 5000 ha) extends up to 50 m
inland along the more sheltered eastern and southern side of
Metrosideros
umbellata). A thick scrub band (approx. 20 000 ha) extends
from the coast in places, to approx. 250 m a.s.l.. Scrub is
dominated by dracophyllum (Dracophyllum longifolium),
myrsine (Myrsine divaricata
a.s.l. and in exposed coastal areas, tussock grasses dominate
(approx. 20 000 ha). The predominant species is snow tussock
(Chionochloa antarctica). Chionochloa antarctica tussock
seeding event.
Live and kill trapping were undertaken around the Smith
Harbour area (Fig. 1), and only kill trapping was undertaken
in the Deas Head area (Fig. 1). Trap grids and transects were
established in the three main habitat types: coastal forest, scrub
and tussock. Trapping grids and transects were established
in coastal forest close to sea level (<20 m a.s.l.), in scrub at
100–200 m a.s.l., and in tussock at 200–350 m a.s.l. (Fig.
Phocarctos
hookeri) were present in the study area and very occasionally
interfered with traps. Field work at each site was done when it
could take place alongside other work programs on feral pigs
(Cox et al. 2022) and feral cats (Glen et al. 2022; Rodriguez-
Recio et al. 2022), meaning sampling periods across sites
were not always aligned.
Mouse population density
Deas Head
Sampling was undertaken in early March 2019, late August
2019, late November 2019 and mid-March 2020. During each
sampling event, nine kill trap transects were established in
the Deas Head area, three in each of the three habitat types
following Harper (2010b) (Fig. 2; Table 1). The exception
was in March 2020 when logistical constraints meant only
two transects were set in each habitat type (Table 1).
Transects were 625 m long and 500 m apart, each with
25 Victor® snap kill traps (Woodstream Coporation Inc.,
Lancaster, USA) spaced 25 m apart. Traps were baited with
peanut butter and secured under small tunnels to reduce risks
to non-target species and to prevent precipitation setting traps
cleared daily for 3 days. All killed mice were necropsied and
their sex, reproductive status and weight (nearest 1 g) were
recorded. Females with perforate vaginas and males with scrotal
testes were considered to be reproductively active. Typically,
considered adult; lighter mice were considered juvenile. To
investigate patterns of cohort recruitment, the distribution of
weights within trips was examined for patterns of bimodality.
3Sagar et al.: Mast effects on subantarctic mice
Figure 1. Subantarctic Auckland Island
relative to mainland New Zealand showing
study sites at Deas Head and Falla Peninsula
during house mouse population studies in
2019–2020.
Table 1. Number of live or kill trapping grids (G) or transects (T) in each habitat by site during each sampling event in
2019–2020 at Auckland Island. Shading indicates when traps were not operated.
__________________________________________________________________________________________________________________________________________________________________
Site Habitat Feb 2019 Aug 2019 Nov 2019 Jan 2020 Mar 2020 Aug 2020
__________________________________________________________________________________________________________________________________________________________________
Live Kill Live Kill Live Kill Live Kill Live Kill Live Kill
__________________________________________________________________________________________________________________________________________________________________
Deas Head Forest 3T 3T 3T 2T
Scrub 3T 3T 3T 2T
Tussock 3T 3T 3T 2T
__________________________________________________________________________________________________________________________________________________________________
Smith Harbour Forest 1G 5G 1G 1G 1G 2T
Scrub 1G 2G 1G 1G 1G 2T
Tussock 2G 2G 1G 1G 2G 2T
__________________________________________________________________________________________________________________________________________________________________
4 New Zealand Journal of Ecology, Vol. 46, No. 3, 2022
Figure 2. Study sites at Auckland Island: (A) Deas Head and (B) Smith Harbour, including Falla Peninsula, showing positions of live
and kill trapping grids and transects set for house mice. n = 25 traps per transect; n = 49 traps per grid (7 × 7; only perimeters shown).
(A)
(B)
5Sagar et al.: Mast effects on subantarctic mice
Smith Harbour
In February 2019, 13 grids of 49 traps, 10 m apart in seven
rows and seven columns (following Russell 2012), were
established around the Smith Harbour area, including Falla
Peninsula, spread across the three habitat types (Fig. 2; Table
1). Four of these grids had Longworth traps (Penlon Ltd,
Oxfordshire, UK), baited with carrot and peanut butter and
provided with dry polyester wool bedding (Table 1). The four
(west of Falla Peninsula) caught no mice. These traps were
operated for 7 nights, checked daily, with mice euthanised by
nine grids of Victor® snap kill traps were baited with peanut
butter and secured under small tunnels as above. These traps
were checked and cleared daily for 7 days. In August 2019
and January 2020 three live capture grids were repeated (all
on Falla Peninsula; Table 1).
In August 2020 all four live capture grids were repeated.
Additionally, in August 2020 six kill trap transects were
operated, two in each of the three habitat types (Fig. 1; Table
1). Transects were set 500 m away from other transects or
grids. Each transect was 625 m long, with 25 Victor® snap kill
traps spaced 25 m apart. Traps were baited with peanut butter
and secured under small tunnels as above. Traps were opened
when deployed and checked and cleared daily for 3 days.
All killed mice were necropsied and measurements
recorded for sex, reproductive status and weight (nearest 1 g).
Tussock monitoring
of seed production (Kelly et al. 2008). Sixteen permanent
transects (20–25 m) were established in tussock habitat at
each of Falla Peninsula and Deas Head in February 2019 (Fig.
2). Transect locations and bearings were randomized within
areas dominated by C. antarctica.
tussocks where the centre of the tussock canopy was located
within 1 m either side of the transect line were sampled and
tagged for repeat measures. Following Kelly et al. (2008),
along with the basal diameter and a visual estimation of the
percent of the basal area carrying live tillers. Tussocks were
re-sampled following the same protocol at Falla Peninsula
in January 2020 (n = 16 transects) and Deas Head in March
2020 (n = 6 transects).
Density estimation
The density of mice was estimated using spatially explicit
capture–recapture models with half-normal detection curves,
Known deaths (in kill traps, at the end of live trap sessions,
and accidental deaths in live traps) were incorporated into
these models. Previous modelling of mouse density at Smith
Harbour using only the January 2019 live and kill-trap grid
data showed that habitat, trap type (Longworth or Victor®),
et al. 2019). Therefore, full likelihood models were used
to determine how capture probability (g0) and the scale of
movement (σ) varied with temporal covariates only, while
density was always allowed to fully vary by grid and transect
Because no recaptures could occur in kill traps, only data
from the live trapping grids at Smith Harbour could be used
to estimate σ
estimated σ, and how it varied with year (2019 or 2020) and
season (warm or cold) either individually or additively, or with
month (3 levels: Jan, Feb, Aug), trip (4 levels: Feb 2019, Aug
2019, Jan 2020, Aug 2020) or session (12 levels: the individual
grids), all while g0 was held constant (7 models). This process
determined the most important temporal covariate for σ.
The second model set investigated how g0 varied with the
same temporal covariates in the live trapping grids at Smith
Harbour, while σ was held constant (7 models). This process
determined the most important temporal covariates for g0, on
the same dataset used to do so for σ.
Because capture probability can be estimated from both live
and kill traps, data from both Smith Harbour and Deas Head
could be used to estimate g0 and its covariates. Therefore, the
third model set investigated how g0 varied with the temporal
covariates for the combined live trapping grids and kill trapping
grids and transects at both Smith Harbour and Deas Head,
while σ was held constant (7 models). This process determined
the most important temporal covariates for g0, on the entire
trapping dataset, to contrast with the most important temporal
A model combining the most supported covariates of
model set comparisons was then constructed. An additional
fourth model set based on this most supported temporal
model was then used to investigate if there were additional
capture probability and/or density (4 models), with density
still fully varying with grid or transect. All models in each
model set (see Appendix S1) were compared using Akaike’s
information criterion (AIC) and model weights (Burnham &
Anderson 2002).
used to provide estimates of mouse density at all trapping
grids and transects on Auckland Island. However, due to the
inferential limitations from extrapolating live to kill trap data
across sites, the density estimates should be interpreted only
as indicating patterns in space and time rather than as absolute
estimates at each grid.
The trapping rate (mice per 100 corrected trap nights
(CTN)) was used as an index of mouse abundance. CTN was
calculated by removing half a trap night for every night a
trap was unavailable due to a mouse capture or to non-target
interference (Nelson & Clark 1973). CTN was calculated for
each grid and transect set in each habitat and averaged for
each habitat at each sampling event.
Results
In total, 1116 mice were caught over 2 years. At Smith Harbour,
201 mice were caught in the live-trap grids (Feb 2019, Aug
2019, Jan 2020, Aug 2020), 232 in the kill-trap grids (Feb 2019)
and 98 in the kill-trap transects (Aug 2020). At Deas Head
585 mice were caught in the kill-trap transects (Feb 2019, Aug
2019, Nov 2019, Mar 2020). The bimodal distribution of body
weights in November 2019 indicated mice were beginning to
breed with the recruitment of a new cohort (Fig. 3E), followed
by a second cohort in Jan 2020 (Fig. 3F), which by March
2020 (and also by Feb 2019 the previous year) was mostly
fully incorporated into the adult population (Figs. 3G, A, B).
In August, breeding was largely absent with only a few large
juveniles in the population (Figs. 3C, D, H).
For the 201 mice caught in the Smiths Harbour live-trap
6 New Zealand Journal of Ecology, Vol. 46, No. 3, 2022
Figure 3. Frequency distribution of body weights (g) of adult (dark grey) and juvenile (light grey) house mice captured on Auckland
Island during Feb 2019–Aug 2020. (A) Deas Head Feb 2019; (B) Smith Harbour Feb 2019; (C) Deas Head Aug 2019; (D) Smith Harbour
Aug 2019; (E) Deas Head Nov 2019; (F) Smith Harbour Jan 2020; (G) Deas Head Mar 2020; (H) Smith Harbour Aug 2020.
7Sagar et al.: Mast effects on subantarctic mice
grids, there were an additional 264 recaptures that allowed
estimation of mouse scale of movement (σ
Mouse scale of movement was best estimated by a model where
it varied with session i.e. unique to every grid (model weight
σ models cannot be
extrapolated to kill-traps where no recaptures occurred, we
removed this model from further consideration. The next best
σ (model weight
= 0.52), although there was some evidence that σ varied with
trip (model weight = 0.24) or season (model weight = 0.21).
The same data from live-trap grids were used to determine
covariates of mouse capture probability (second model set).
Mouse capture probability was best estimated by a model
where it varied with session i.e. unique to every grid (model
weight = 0.56) or trip (model weight = 0.44). Repeating this
analysis on the entire dataset of live trapping grids and kill
trapping grids and transects at both Smith Harbour and Deas
Head (third model set), mouse capture probability was best
estimated by a model where it varied with trip (model weight
= 0.96).
the entire mouse trapping data set was one where capture
probability varied with trip and scale of movement was
on capture probability (model weight = 0.33) and density
(model weight = 0.19).
2019 Smith Harbour: 28–104 mice ha and Deas Head: 76–104
mice ha) and declined thereafter at both sites (Fig. 4; Aug
2019 Deas Head: 12–31 mice ha and Smith Harbour: 12–24
mice ha). Density remained low the following summer (Nov
2019 Deas Head: 13–19 mice ha; Mar 2020 Deas Head:
6–45 mice ha; Jan 2020 Smith Harbour: 10–32 mice ha)
and remained at a similar level through the subsequent winter
(Fig. 4; Aug 2020 Smith Harbour 4–31 mice ha). Capture
Figure 4.
(light grey) and tussock (mid-grey) and capture probability (g0; dashed line) over time at two sites, Deas Head (A) and Smith Harbour
8 New Zealand Journal of Ecology, Vol. 46, No. 3, 2022
Figure 5. An index of abundance (mean captures per 100 corrected trap nights ± SE) of house mice at each site by broad habitat
0 ± SE; dashed line) over
time at two sites, Deas Head (A) and Smith Harbour (B).
probability (g0
index (mice 100 per CTN) followed a similar pattern to
capture probability (Fig. 5). Overall, mouse density, capture
probability and abundance indices trends were consistent
between sites, though all measures were generally higher at
Deas Head than Smith Harbour (Figs. 4 & 5). Trends were also
trip at Smith Harbour, where mouse density in kill-trap grids
varied across habitats as much as 28–104 mice ha (Fig. 4).
Very little bycatch or trap interference occurred at either
site. At Deas Head, three tomtits (Petroica macrocephala
marrineri) were caught in traps, while pigs interfered with
traps on 27 occasions and sealions did so on three occasions.
At Smith Harbour, pigs interfered with traps on only three
occasions (pigs were being locally eradicated at the start of
our study and allowed to repopulate thereafter; Cox et al.
2022) and three bellbirds (Anthornis melanura) were caught
in snap traps.
Snow tussock C. antarctica
in summer 2019 with >92% of plants having at least one
plant). This result contrasted with summer 2020, when <1%
Discussion
The population trends of mice on Auckland Island broadly
follow the irruptive patterns of house mice in relation to
masting in other temperate New Zealand systems (Ruscoe
et al. 2001; Wilson & Lee 2010). However, both population
densities and the timing of the increase on cool-temperate
9Sagar et al.: Mast effects on subantarctic mice
systems. Following an autumn seed fall in warm-temperate
New Zealand systems, mouse population peaks are generally
seen during the following winter-spring and sometimes as late
as summer (Wilson & Lee 2010). On Auckland Island the
highest mouse population densities were observed immediately
months of high food availability. The seeding of tussock on
cool-temperate subantarctic Auckland Island (Nov–Feb) is
earlier and briefer than New Zealand South Island tussock
(Edgar & Connor 2000) and beech (Fuscospora spp.) forests
(Jan–Apr; Wilson & Lee 2010). This timing likely accounts
for the earlier population increase on Auckland Island than in
warm temperate systems. The mouse population density on
Auckland Island had declined by the winter following the mast
event and remained at a similar level during the subsequent
summer and winter. This result is further supported by evidence
from Russell et al. (2018), where very low mouse population
density was measured at the same Deas Head site outside of
a mast year. Although the highest density recorded was in
periods it is possible the true population peak was not sampled.
Furthermore, we only studied two sites, and found variation in
density even among sites of seemingly identical habitat, and
outside of mast seeding (Russell et al. 2018). Accordingly,
we only have a coarse view of mouse population dynamics
on a very large island.
Estimated mouse densities on Auckland Island around
the population peak were lower than reported from summer
studies on subantarctic Gough Island (R. Cuthbert unpubl. data
in Rexer-Huber et al. 2013) and Marion Island (Matthewson
et al. 1994), though comparable to Antipodes Island (Russell
2012). Winter density on Auckland Island was comparable to
winter densities on other Southern Ocean islands, including
Steeple Jason Island (Rexer-Huber et al. 2013) and Marion
Island (Matthewson et al. 1994; Ferreira et al. 2006), which
are all at much higher latitudes than Gough Island (R. Cuthbert
unpubl. data in Rexer-Huber et al. 2013). However, comparison
of densities among studies, and even of densities within our
capture devices, and assumptions made when estimating
density. Mice were or are the sole extant introduced mammal
on Steeple Jason, Antipodes, Gough and Marion Islands, in
contrast to Auckland Island where they co-exist with pigs
and cats. Evidence from Marion Island before and after the
eradication of cats shows that abiotic factors, not predation
pressure, drive mouse density there (Aarde & Jackson 2007).
While cats undoubtedly prey on mice on Auckland Island
(Harper 2010a), the apparent large demographic responses
of mice to masting mean it is unlikely cat predation has any
the bottom-up role of resource availability drives population
responses.
As in warm-temperate New Zealand beech (King 1983;
Fitzgerald et al. 2004) and tussock (Wilson & Lee 2010)
systems, the mouse population increase on Auckland Island
is likely the result of high juvenile survival and subsequent
recruitment as a result of increased food supply. However,
our study appears not to have captured the initial breeding
occurred toward the end of seeding and the wide distribution of
weights shows juveniles were recruiting to the population and
adults were still breeding at this time. The similar distribution
of weights from both winters studied suggests that breeding
largely ceases through this season, regardless of if a mast event
occurred the previous summer. The bi-modal distribution of
weights from summers shows that multiple seasonal breeding
pulses occurs during this time, and outside of a mast event
breeding slows again by late summer. Very few mice breed over
winter on Southern Ocean islands, including Auckland Island
(Matthewson et al. 1994; Avenant & Smith 2004; Harper 2010b;
Rexer-Huber et al. 2013; Elliott et al. 2015). Studies on other
Southern Ocean islands have occurred in the absence of mast
breeding is not elevated in a mouse population on a Southern
Ocean island following a mast event the previous summer. In
contrast elevated winter breeding following tussock or tree
masting the previous summer is typical in warm temperate
New Zealand island systems (King 1983; Wilson & Lee 2010).
how likely individuals are to interact with traps. Throughout
the study the capture probability of mice (g0) was generally
the resource limitation that has constrained density at these
times probably also motivates mice to interact with traps more.
Furthermore, when population density is high mouse home
ranges are smaller, meaning they will have fewer opportunities
Abundance indices such as corrected captures are often
used as a proxy for population trends (Fitzgerald et al. 2004
and references therein). In this study, capture probability and
an abundance index (captures per 100 CTN) followed similar
Abundance index values for the winter following masting
are considerably higher than Harper (2010b) reported for
the same trapping regime on Auckland Island, also assumed
to be following a mast event. Harper’s (2010b) study was
undertaken earlier in winter (June–July c.f. late August here)
and it is possible tussock seeds and invertebrates remained more
plentiful during this period, leading to lower catch rates. Data
from Antipodes Island show that although density was similar
between trapping periods in summer 2011 (60–147 mice ha)
and winter 2013 (74–104 mice ha), abundance indices were
lower in winter 2013 (21.7–22.9 mice 100CTN) than summer
2011 (28.7–33.1 mice 100CTN (Russell 2012; Elliott et al.
2015). Likewise, mice abundance indices declined during
winter (Jun–Aug 2005) on Macquarie Island, though mouse
densities for this period are lacking (K. Springer unpubl. data).
Reduced captures in winter is surprising, because resource
availability presumably declines in winter on Southern Ocean
islands (fewer seabirds present, invertebrate activity and
vegetation growth reduced) compared to summer, and as a result
capture probability should be higher. Together these results
highlight that index of abundance measures, such as CTN, do
not reliably indicate actual population density, and at least on
and index measures need to be established and studies that
assume a linear relationship between indices and population
density may be misleading (Ruscoe et al. 2001).
This study shows the bait uptake trial undertaken on
Auckland Island in summer 2019 (Russell et al. 2019)
coincided with the highest estimates of mouse density and
lowest capture probability during a mast seeding event. Thus
the successful bait uptake by 99% of mice in that trial remains
particularly encouraging for the feasibility of eradicating
mice from Auckland Island at any stage of a tussock mast
10 New Zealand Journal of Ecology, Vol. 46, No. 3, 2022
seeding cycle (Russell et al. 2019). This outcome is critical
because the timing of mouse baiting on Auckland Island will
require determinations months if not years in advance of any
eradication.
gap in knowledge for Southern Ocean islands, particularly
where tussock mast seeding occurs and is a consideration in
eradication planning. Increasingly, larger or more complex
eradications are being proposed that will require adaptation
of best practice (e.g. Horn et al. 2022; Livingstone et al. 2022;
Oyston et al. 2022). We recognize the interaction of pests, food
resources and the environment mean there is likely no ‘one size
planning to reduce uncertainty and risk where departure from
best practice is required.
Author contributions
RLS, FSC, SRH and JCR designed the study and undertook
wrote the manuscript with input from FSC and SRH.
Acknowledgements
Thanks to Paul Jacques, Rose Hanley-Nickolls, Lindsay Chan,
Kirsten Rogers, Veronika Frank, Mark Le Lievre, Micaela
Graham Parker, Kalinka Rexer-Huber, Brent Barret, Jenny
Rickett, Penny Pascoe, Iain Graham, Caitlyn Thomas, Rose
Collen, Rowan Hindmarsh-Walls and James Ware for assistance
of MY Evohe, MV Awesome, MV Searanger and HMNZS
Canterbury for safe transport to and from the island. Thanks
to the Murihiku district team and Kaitiaki Roopu for their
support of the work. Performed under University of Auckland
animal ethics R2095. Thanks to a reviewer for feedback on
the manuscript.
References
Aarde R, Jackson T 2007. Food, reproduction and survival
of mice on sub-Antarctic Marion Island. Polar Biology
30: 503–511.
Angel A, Wanless RM, Cooper J 2009. Review of impacts of
the introduced house mouse on islands in the Southern
Ocean: are mice equivalent to rats? Biological Invasions
11: 1743–1754.
house mouse progression in Eurasia: a palaeontological
and archaeozoological approach. Biological Journal of
the Linnean Society 41: 13–25.
Avenant NL, Smith VR 2004. Seasonal changes in age class
structure and reproductive status of house mice on Marion
Island (sub-Antarctic). Polar Biology 27: 99–111.
Bogdziewicz M, Zwolak R, Crone EE 2016. How do vertebrates
respond to mast seeding? Oikos 125: 300–307.
likelihood methods for capture-recapture studies.
Biometrics 64: 377–385.
Boursot P, Din W, Anand R, Darviche D, Dod B, Von Deimling
F, Talwar GP, Bonhomme F 1996. Origin and radiation of
the house mouse: mitochondrial DNA phylogeny. Journal
of Evolutionary Biology 9: 391–415.
Broome KG, Golding C, Brown K, Horn SR, Corson P, Bell
P 2017. Mouse eradication using aerial baiting: current
agreed best practice used in New Zealand (Version 1.0).
New Zealand Department of Conservation internal
document DOC-3034281. Wellington, New Zealand,
Department of Conservation Te Papa Atawhai. 23 p.
Burnham KP, Anderson DR 2002. Model selection and
inference a practical information-theoretic approach. 2nd
edn. Springer-Verlag. 488 p.
Courchamp F, Chauis J-L, Pascal M 2003. Mammal invaders
on islands: impact, control and control impact. Biological
Reviews 78: 347–383.
Cox FS, Horn SR, Bannister WM, Macdonald NL 2022. A
local eradication pilot study of methods for feral pig
eradication on Auckland Island. New Zealand Journal of
Ecology 46: 3490.
Dilley BJ, Schoombie S, Schoombie J, Ryan PG 2015.
spreads rapidly at Marion Island. Antarctic Science 28:
73–80.
Edgar E, Connor HE 2000. Flora of New Zealand Volume V:
Gramineae (Grasses): Vol. V. Manaaki Whenua Press.
Oikos 106: 598–610.
dependent home-range size revealed by spatially explicit
capture–recapture. Ecography 39: 676–688.
Elliott GP, Greene TC, Nathan HW, Russell JC 2015. Winter bait
preparation for mouse (Mus musculus) eradication. DOC
Research and Development Series 345. Wellington, New
Zealand, Department of Conservation. 34 p.
Ferreira SM, Van Aarde RJ, Wassenaar TD 2006. Demographic
responses of house mice to density and temperature on
sub-Antarctic Marion Island. Polar Biology 30: 83–94.
ad mast seeding of southern beeches in the Orongorongo
Valley, New Zealand. New Zealand Journal of Ecology
31: 167–184.
Fraser A 2020. The climate and weather of the Auckland Islands.
Unpublished Report - prepared for the Department of
Conservation Te Papa Atawhai. Invercargill, New Zealand,
Glen AS, Sagar RL, Brav-Cubitt T, Jacques PM 2022.
Monitoring and detection of feral cats on Auckland Island.
New Zealand Journal of Ecology 46: 3494.
Harper GA 2010a. Diet of feral cats on subantarctic Auckland
Island. New Zealand Journal of Ecology 34: 259–261.
Harper GA 2010b. Habitat use by mice during winter on
subantarctic Auckland Island. New Zealand Journal of
Ecology 34: 262–264.
Holmes ND, Spatz DR, Oppel S, Tershy B, Croll DA, Keitt
A, Aguiree-Muñoz A, Raine AF, Knapp CR, Hung CH,
Wingate D, Hagen E, Méndez-Sánchez F, Rocamora G,
Yuan HW, Fric J, Millett J, Russell JC, Liske-Clark J,
Vidal E, Jourdan H, Campbell K, Springer K, Swinnerton
K, Gibbons-Decherong L, Langrand O, de Brooke ML,
McMinn M, Bunbury N, Oliveira N, Sposimo P, Geraldes
11Sagar et al.: Mast effects on subantarctic mice
P, McClelland P, Hodum P, Ryan PG, Borroto-Páez R,
SA, Cranwell S, Micol T, Butchart SHM 2019. Globally
important islands where eradicating invasive mammals
e0212128.
Horn SR, Greene T, Elliott GP 2019. Eradication of mice
from Antipodes Island, New Zealand. In: Veitch CR,
Clout MN, Martin AR, Russell JC, West CJ eds. Island
invasives: scaling up to meet the challenge. Proceedings
of the International Conference on Island Invasives
2017. Occasional Paper of the IUCN Species Survival
Commission 62. Gland, Switzerland, IUCN. Pp. 131–137.
Horn SR, Sagar RL, Frank VK, Cox FS, Jacques PM, Ware
J, Hanley-Nickolls R, Leask EP, MacDonald NL, Kirby-
Crowe MS, Le Lievre ME, Broome KG 2022. The next
frontier: assessing the feasibility of eradicating mammalian
pests from Auckland Island. New Zealand Journal of
Ecology 46: 3500.
Houghton M, Terauds A, Merritt D, Driessen M, Shaw J 2019.
The impacts of non-native species on the invertebrates of
Southern Ocean Islands. Journal of Insect Conservation
23: 435–452.
Kelly D, Turnbull MH, Pharis RP, Sarfati MS 2008. Mast
seeding, predator satiation, and temperature cues in
Chionochloa (Poaceae). Population Ecology 50: 343–355.
King CM 1983. The relationships 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. The Journal of
Animal Ecology 52: 141.
Livingstone J, Horn SR, Broome KG, Sagar RL 2022. How low
can you sow? House mouse eradication on Motuareronui/
Adele Island. New Zealand Journal of Ecology 46: 3498.
Matthewson DC, Van Aarde RJ, Skinner JD 1994. Population
biology of house mice (Mus musculus) on sub-Antarctic
Marion Island. South African Journal of Zoology 29:
99–106.
Nelson L, Clark FW 1973. Correction for sprung traps in
Mammalogy 54: 295–298.
Maud Island for reducing aerial baiting sow-rates for
the eradication of house mice. New Zealand Journal of
Ecology 46: 3499.
Rodriguez-Recio M, Sagar RL, Chan L, Cox FS, Jacques PM
2022. Spatial ecology meets eradication of feral cats on
Auckland Island. New Zealand Journal of Ecology 46:
3493.
Rexer-Huber K, Parker GC, Reeves M, Stanworth AJ, Cuthbert
RJ 2013. Winter ecology of house mice and the prospects
for their eradication from Steeple Jason (Falkland Islands).
Polar Biology 36: 1791–1797.
Ruscoe WA, Goldsmith R, Choquenot D 2001. A comparison
of population estimates and abundance indices for house
mice inhabiting beech forests in New Zealand. Wildlife
Research 28: 173–178.
Russell JC 2012. Spatio-temporal patterns of introduced mice
and invertebrates on Antipodes Island. Polar Biology 35:
1187–1195.
Russell JC, Broome KG 2016. Fifty years of rodent eradications
in New Zealand: another decade of advances. New Zealand
Journal of Ecology 40: 197–204.
Russell JC, Horn SR, Harper GA, McClelland P 2018. Survey of
introduced mammals and invertebrates on Auckland Island,
March–April 2015. DOC Research and Development
Series 352. Wellington, New Zealand, Department of
Conservation. 17 p.
FS, Horn SR 2019. Mouse bait uptake and availability
trials on Falla Peninsula, Auckland Island. DOC Research
and Development Series 363. Wellington, New Zealand,
Department of Conservation. 11 p.
Russell JC, Peace JE, Houghton MJ, Bury SJ, Bodey TW
2020. Systematic prey preference by introduced mice
exhausts the ecosystem on Antipodes Island. Biological
Invasions 22: 1265–1278.
Russell JC, Horn SR, Broome KG 2022. Restoration of New
Zealand subantarctic islands. New Zealand Journal of
Ecology 46: 3482.
Singleton GR, Brown PR, Pech RP, Jacob J, Mutze GJ, Krebs
CJ 2005. One hundred years of eruptions of house mice in
Australia - a natural biological curio. Biological Journal
of the Linnean Society 84: 617–627.
Spatz DR, Holmes ND, Will DJ, Hein S, Carter ZT, Fewster
RM, Keitt B, Genovesi P, Samaniego A, Croll DA, Tershy
BR, Russell JC 2022. The global contribution of invasive
vertebrate eradication as a key island restoration tool.
Springer K 2016. Methodology and challenges of a complex
multi-species eradication in the sub- Antarctic and
Zealand Journal of Ecology 40: 273–278.
Veale AJ, Russell JC, King CM 2018. The genomic ancestry,
landscape genetics and invasion history of introduced mice
in New Zealand. Royal Society Open Science 5: 170879.
GM, Kritzinger P, Ryan PG, Slabber M 2012. Predation
of Atlantic Petrel chicks by house mice on Gough Island.
Animal Conservation 15: 472–479.
Wilson DJ, Lee WG 2010. Primary and secondary resource
pulses in an alpine ecosystem: snow tussock grass
(Chionochloa Mus
musculus) populations in New Zealand. Wildlife Research
37: 89–103.
World Heritage Convention 1998. New Zealand sub-antarctic
Cultural Organization. http://whc.unesco.org/en/list/877/
Editorial board member: George Perry
Received 21 April 2021; accepted 10 August 2022
Supplementary Material
Additional supporting information may be found in the online
version of this article:
Appendix S1. Model sets for estimating density, capture
probability (g
explicit capture-recapture.
The New Zealand Journal of Ecology provides online
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
