CSIRO PUBLISHING

www.publish.csiro.au/journals/wr Wildlife Research, 2004, 31, 327–337

Home range, movement and habitat utilisation of the Carpentarian

rock-rat (Zyzomys palatalis) in an isolated habitat patch

Helen PuckeyA, Milton LewisB, David HooperB and Carrie MichellB

A
Parks and Wildlife Service, Department of Infrastructure, Planning and Environment,
PO Box 2130, Alice Springs, NT 0871, Australia.
B
Parks and Wildlife Service, Department of Infrastructure, Planning and Environment,
PO Box 334, Katherine, NT 0851, Australia.

Abstract. Radio-telemetry was used to examine the home range, movement and habitat utilisation of the critically
endangered Carpentarian rock-rat (Zyzomys palatalis) in an isolated habitat patch in the Gulf of Carpentaria
hinterland over a 13-month period. Two home-range estimators were used in the study, (i) minimum convex polygon
(MCP) and (ii) fixed Kernel (KL), the latter also being used to estimate core areas of activity. Based on a total
sample size of 21 individuals, the mean MCP home range was 11165 m2, similar to the mean KL home range of
10687 m2. Core areas were, on average, 11.9% of the KL home-range estimate. There was no significant difference
in the size of home range or core area of males and females. Juveniles had a significantly smaller home range than
adults. Home ranges and, to a lesser degree, core areas were non-exclusive, with multiple areas of overlap (averaging
41% and 38% respectively) within and between all age and gender categories, but especially between males and
between juveniles. Movement frequencies showed that animals made many short forays in a central area close to
the arithmetic home-range mean and far fewer long forays of distances greater than 100 m from the central area.
The spatial and temporal activity of Z. palatalis was concentrated in, but not confined to, the ‘valley’ and ‘slope’
habitats, with fewer movements of rats onto the surrounding ‘plateau’. Resource selection analyses showed that Z.
palatalis tended to prefer valley and slope habitats over the plateau and that the proportion of point locations was
significantly higher for adults in the slope habitat and for juveniles in the valley habitat. Most home ranges were
centred on the ecotone between these two habitat types. Although isolated and spatially limited, these habitat
patches provide high-quality resources for dense populations of Z. palatalis. This study exemplifies a species’
attempt to make efficient use of a limited resource in an otherwise hostile environment. Even small declines in
habitat area or quality due to their vulnerability to fire would impact upon many animals.
WR03025
eHta.olPm.ueckreayngeofteh Carpnetarianrock-rat

Introduction
The Carpentarian rock-rat (Zyzomys palatalis), is a small
(120 g) conilurine rodent that occurs in isolated habitat
patches in the Gulf of Carpentarian hinterland, Northern
Territory. Despite extensive survey of the Gulf region
(Menkhorst and Woinarski 1992; Churchill 1996; Trainor
1996; Puckey 2003) it appears to be confined to five
isolated patches of rainforest associated with rocky
sandstone gorges and permanent water on Wollogorang
Station. Like its close relatives, the large rock-rat
(Zyzomys woodwardi) and the Kakadu rock-rat (Zyzomys
maini), Z. palatalis feeds on fleshy fruits and large seeds
from plants within monsoon vine thicket (Begg and
Dunlop 1985). This habitat type is extremely limited in
extent and vulnerable to fires in the Gulf region. The
species has been classified as critically endangered under
the IUCN Red List criteria due to its extremely limited
total range, small number of populations (which are
fragmented), specific habitat requirements, and the
likelihood that the extent or quality of its habitat is
declining (Baillie and Groombridge 1996).
The persistence of Z. palatalis is threatened by altered fire
regimes, grazing by feral animals and stock, weed intrusion,
and the stochastic hazards associated with small, fragmented
populations (Brook et al. 2002). Population and habitat-
simulation modelling predicted the population viability of Z.
palatalis to be highly sensitive to hot, late dry season fires
(Brook et al. 2002). As controlled fuel-reduction burns
around the gorges have been suggested to reduce this threat
(Brook et al. 2002) it is important to understand how Z.
palatalis is using the rainforest patches and the surrounding
broadleaf woodlands.

© CSIRO 2004 10.1071/WR03025 1035-3712/04/030327

328 Wildlife Research
Fig. 1. Location map of study area at Wollogorang Station, Northern Territory.
The objective of this study was to determine how
Z. palatalis uses the isolated habitat patches in which it
occurs, with a view to improving conservation management
of the existing populations and the sites that support them.
Estimation of home range provides an indication of the
amount of space the animal needs to survive, while
determination of habitat utilisation helps to characterise
features of that space that are important to its survival.
Specifically, we sought to estimate and describe home range
and core areas, examine for seasonal, gender and age
differences in these, and to examine patterns of habitat use in
relation to availability and characteristics.
Methods
Study site
The study was conducted on Wollogorang Station in the Northern
Territory, ~450 km north-east of Tennant Creek (Fig. 1). The site is
within the Gulf Fall Uplands bioregion, which occupies a total land area
of 120000 km2 across Northern Territory and Queensland (Thackway
and Cresswell 1996). Rolling to steep, rugged sandstone hills with
eucalypt woodlands and spinifex understorey characterise this
bioregion.
The site is part of a small gorge system with permanent springs and
monsoon vine thickets known locally as Banyan Gorge. The southern
branch of the gorge where the study was conducted is ~1 km in length
and characterised by boulder slopes and rock walls up to 80 m in
elevation. Within the more protected parts of the gorge are small
patches of wet rainforest dominated by Syzygium angopheroides.
Pockets of dry monsoon vine thicket occur on the rocky slopes,
particularly beneath cliffs and rocky outcrops. The main valley floor
has a canopy of Corymbia ptychocarpa with ground cover of Mnesithea
rottboellioides. The surrounding vegetation is mixed eucalypt
woodlands with a spinifex understorey.
The climate at the study site is typical of the monsoonal Gulf
Savanna Region with an average annual rainfall of 850 mm, ~80% of
which falls from December to March. Wet season (October–April)
temperatures range from a daily minimum of 23°C to a maximum of
39°C, while in the dry season (May–September) temperatures typically
range from a minimum of 10°C to a maximum of 33°C (Northern
Territory Bureau of Meteorology).
Wollogorang Station has a long history of cattle (Bos taurus)
grazing. Although few stock currently frequent the study site due to its
topography, there are signs of cattle and feral pigs (Sus scrofa) in the
H. Puckey et al.
gorge. Feral cats (Felis catus) occur throughout the study area. Less
than 6 km from the study site an excision in the pastoral lease is
occupied by an active copper mine.
Radio-tracking
The movements of Z. palatalis were studied using radio-telemetry over
the period June 2001 to July 2002. Animals were captured in Elliott
traps (23 × 8 × 9 cm) baited with peanut butter, rolled oats and honey.
Lines of 20–25 traps spaced 15 m apart were run opportunistically in
each of the habitat types within the study area. Captured rats were
preferentially selected for tracking in order to obtain equal gender and
age representation as well as a representative sample of rats from across
the entire study area. Individual rats were measured, weighed and fitted
with 1.9-g radio-collars, which were ~1.7% of mean bodyweight.
Collars were made with a button cell battery and transmitter, epoxy
adhesive and double-stranded whip antennae (7 cm long) and fitted
with surgical tubing and Vetbond™ glue to secure the knot (model
LTM-392, Titley Electronics, Ballina, New South Wales). Collar fitting
took less than 15 min per animal, after which animals were released at
their point of capture. Thirty-six Z. palatalis were collared during the
study, with a maximum of six during any individual tracking session. Of
these, 30 retained their collars, continued to emit a radio signal and
were successfully tracked for at least four nights. Animals were then
recaptured and the collar removed. As animals were not all permanently
marked it was not known whether individuals were tracked on more
than one occasion. Traps and bait were removed from the study area
before tracking commenced so that movement was not biased towards
trap sites. Animals were tracked in June, September and November of
2001 and February, April and July of 2002.
In June 2001 active tracking was carried out on foot over four nights
using a collapsible 3-element hand-held directional antenna and
portable receiver (Titley Electronics). An attempt was made to locate
each animal hourly. Generally, animals were not sighted but accurate
locations were achieved by considering signal direction while circling
around the estimated location until their location could be refined to
within 10 m. Once located, the position of the animal was recorded
using a Global Positioning System (GPS). The time and some notes on
habitat were also recorded.
During all other tracking periods, fixed receivers were used to
triangulate positions on Z. palatalis at prescribed intervals. Five fixed
receivers were situated at least 60 m apart on high points around the top
of the gorge. Each receiver consisted of a 4-element Yagi directional
antenna horizontally mounted on 2.5-m wooden poles, the bases of
which were mounted in a compass rose marked in 1° divisions.
Synchronised location fixes were taken by researchers every 15 min for
each animal from nightfall (1900–2000 hours) until dawn (0500–0530
Home range of the Carpentarian rock-rat
hours) for four nights during each tracking session. The constant
sampling intervals that were used should be sufficient to overcome
problems associated with autocorrelation of the data (Rooney et al.
1998; de Solla et al. 1999). Systematic error (consistent difference in
direction between the true bearing and the bearing determined by
radio-location from each station) was corrected using the mean of a
series of bearings taken on transmitters at known locations throughout
the area typically used by radio-collared Z. palatalis.
Triangulation was carried out using LOCATE II (Nams 2000), which
uses the Lenth maximum-likelihood method to estimate transmitter
location. Bearings of individual stations were manually rejected if they
clearly did not fit the triangulation combination of all other stations. The
resultant UTM coordinates for each position were then mapped and
analysed using ArcView 3.1 software (Environmental Systems Research
Institute, Redlands, California, USA) and the Animal Movement
extension for ArcView (Hooge and Eichenlaub 2000). All triangulation
calculations with an error ellipse of >20 m2 were excluded from further
analysis. All positions resulting from fewer than three receivers were also
excluded. MCP area bootstrapping (see below) was used to eliminate a
further 9 rats whose coefficient of variation did not reach a minimum
of 25%. All home-range, animal-movement and habitat-utilisation
analyses were thus carried out on a dataset of 21 animals with 20–86
positions recorded for each animal.
Home-range estimation
Home-range area was estimated using both the Minimum Convex
Polygon (MCP) method and the fixed Kernel (KL) method. MCPs were
constructed by connecting the outermost location records of an animal
to form a convex polygon with a quantifiable area. The MCP method
has been used widely due to its simplicity and comparability between
studies (Harris et al. 1990). However, the number of location records,
gaps in the range area and outliers can create bias in the size and shape
of home range using this method (White and Garrott 1990). The KL
method of estimating home range is a nonparametric statistical method
that estimates probability densities and is less influenced by such biases
(Seaman and Powell 1996). The KL estimator was chosen to (1)
calculate the utilisation distribution encompasing 95% of points
(defined here as the KL home range) and (2) calculate the areas of high
usage encompassing 50% of points within the home range (defined as
the KL core area). ArcView uses least-squares cross-validation to select
the optimal smoothing factor (h) in calculating the KL utilisation
distribution. Realising that home-range estimates are affected by the
length of time spent collecting data, we acknowledge that areas
presented here are ‘4-day’ home-range estimates and probably
underestimate the lifetime home-range area of Z. palatalis.
A home-range estimate was initially calculated using pooled data
for all animals, after which home-range size was compared between
season, gender and ages using ANOVA univariate tests of significance
(Statistica 5.0). Age categories (adult and juvenile) were assessed by a
combination of weight and reproductive condition. Records from June,
September and July were considered to be dry season and those from
November, April and February as wet season. These groupings were
broadly based on rainfall and plant phenology.
The overlap in non-exclusive MCP home-range areas and the
overlap in non-exclusive KL core areas was quantified as a percentage
using the equation proposed by Horsup (1994):
100((X∩Y)/X + (X∩Y)/Y)/2,
where (X∩Y) = area of home range overlap between individual X and
individual Y.
Where variances lacked homogeneity in comparing gender
overlaps, a non-parametric Kruskal–Wallis test was used to determine
the significance of differences in overlap.
Wildlife Research 329
Habitat utilisation
The study area was divided into three broad habitat types delimited on
the basis of floristic and topographic differences and mapped to an
accuracy of ~15 m using digital aerial photography, topographic maps
and ground-truthing with a base station and differential GPS. A brief
description of each habitat type follows:
(1) Valley areas were characterised by tall, dense stands of Syzygium
angophoroides, Corymbia ptychocarpa, Grevillea pteridifolia and
Melaleuca leucadendra with a sometimes dense understorey of
Melastoma affine, Scleria lingulata and Mnesithea rottboellioides.
This habitat type was generally on flat to gently sloping ground with
variable rock cover, a well developed leaf litter layer in drier months
and the presence of, or close proximity to permanent water.
(2) Slope areas were steep boulder slopes with associated rock walls.
The vegetation was particularly speciose, characterised by low open
woodland of Corymbia aspera, Erythrophleum chlorostachys,
Owenia vernicosa, Livistona inermis and patches of monsoon vine
thicket with Ficus spp., Celtis philippensis, Diospyros humilis and
Antidesma parvifolium.
(3) Plateau areas were elevated and rocky, dominated by open
woodlands of Eucalyptus miniata, E. tetrodonta and Terminalia
carpentariae with a Triodia bitextura understorey. Acacia
asperulacea, Tephrosia spechtii, perennial Sorghum sp. and
Eriachne ciliata were also particularly common at these sites.
For the purpose of analysis of habitat selection, the total area of
habitat available was calculated as the total present in all individual
MCP areas, with a buffer of 39 m (the average distance moved between
successive point locations). As the study area is centred on a steep gorge
system, the area of slope habitat would be underestimated using
conventional two-dimensional mapping techniques. Thus the surface
area of the sloping habitat was calculated on the basis of the average
angle of the slope and has been applied to all area calculations.
Two aspects of habitat utilisation were investigated: (i) the
proportion of point locations in each of three habitat types; and (ii) the
area of each of the habitat types in MCP home ranges in proportion to
their availability in the project area.
The percentages of fixes in each of the three habitat types were
normalised using arcsin-transformation for each individual and
analysed using a GLIM model to determine the effects of season,
gender and age on habitat selection.
Using Resource Selection Analysis Software for Windows (1999 ©
Fred Leban), Friedman Tests of MCP home-range areas were calculated
to rank the difference in selection and availability across the three
habitat types.
Habitat ordination
Detailed habitat assessments were carried out at 47 quadrats
(20 × 20 m) throughout the tracking area in July 2002. These quadrats
were selected to cover a cross-section of the habitat types and to broadly
represent the spatial range of the study area. Cover values for each plant
species present in the quadrats were recorded. Environmental attributes
recorded at each quadrat included aspect, slope, distance to water, fire
impact (on an index of 1–5), rock cover in seven different rock size
classes, soil texture and depth, ground cover (using 100-m
point-intercept transects), canopy height and percentage cover of
vegetation at five different height classes.
To identify and characterise species and environmental gradients
and to test the validity of the three chosen habitat divisions, ordinations
were carried out on the plant species cover values of all quadrats using
PATN (Belbin 1994). Ordination was by the semi-strong-hybrid
multi-dimensional scaling methodology (SSH) based on a Bray–Curtis
dissimilarity coefficient. Nonparametric Kruskal–Wallis ANOVAs
330 Wildlife Research H. Puckey et al.

were used to determine the significance of differences in environmental

18000
attributes across the three habitat types.
16000
Results
Radio-tracking studies assume that data collected are 14000
representative of the population and that the transmitter had
no effect on the behaviour or movement pattern of the 12000

Area (m2)
animals. Preliminary studies of collared Z. palatalis in
10000
captivity indicated no sign of distress, discomfort or
noticeable change in behaviour within the first few hours of 8000
collaring (Helen Puckey, personal observations). In the field,
a lack of effect on body mass of studied animals has been 6000
used to indicate normal mobility during the study period
(Swihart and Slade 1985; Moro and Morris 2000). In this 4000
study, rock-rats that were recaptured after radio-tracking 2000
showed no significant loss in body mass (Wilcoxon matched JM JF AM AF
pairs test: Z = 0.2, n = 25, P = 0.8), nor were there any Class
obvious signs of tissue damage around the collars. This
Fig. 2. Mean (± s.e.) MCP home-range estimates (m2) of Z.
suggests that the radio-collars had little effect on the palatalis studied at Wollogorang Station, Northern Territory. JM =
condition and movements of the rock-rats. juvenile male (n = 4), JF = juvenile female (n = 3), AM = adult male
(n = 7) and AF = adult female (n = 7).
Home range
Active tracking on foot and remote tracking from fixed Table 1). Core areas of activity (50% KL utilisation
receivers gave similar estimates of home-range size. Thus distributions) on average represented 11.9% (mean =
the active tracking data from June were included with those 1272 m2, s.d. = 921) of the KL home-range estimates.
of the other periods in all analyses. Home-range estimates The mean adult MCP home-range area of 12912 m2
were similar for MCP (mean = 11165 m2, s.d. = 5824) and (s.d. = 5951) was significantly larger than the mean juvenile
KL methods (mean = 10687 m2, s.d. = 8710) (Fig. 2, MCP area of 7673 m2 (s.d. = 3869) (F1,19 = 4.42, P = 0.049).

Table 1. Home-range estimates and tracking effort for Z. palatalis at Wollogorang Station,
Northern Territory
AM, adult male; AF, adult female; JM, juvenile male; JF, juvenile female

Animal Age/sex Dates No. of MCP 95% KL 50% KL

No. tracked fixes (m2) (m2) (m2)
01 AM Jun. 2001 48 8325 6204 789
02 AF Jun. 2001 48 4246 2012 225
03 AM Jun. 2001 30 8668 13318 1026
04 JM Sep. 2001 22 910 4702 687
05 JF Nov. 2001 42 7122 7724 920
06 JF Nov. 2001 64 13071 8754 1219
07 AM Nov. 2001 55 14187 14748 2604
08 AM Nov. 2001 58 13395 13056 2065
09 AM Nov. 2001 43 17255 11504 1197
10 JM Feb. 2002 66 5468 4589 421
11 JF Feb. 2002 86 7388 3965 459
12 JM Feb. 2002 69 9935 6975 970
13 JM Feb. 2002 43 9813 8241 1091
14 AF Apr. 2002 74 12855 12059 1238
15 AM Apr. 2002 48 17053 26789 3390
16 AF Apr. 2002 80 14125 7577 605
17 AF Apr. 2002 20 25248 40965 3666
18 AF Jul. 2002 74 19805 6390 950
19 AF Jul. 2002 35 9917 10595 1504
20 AF Jul. 2002 50 12683 9418 850
21 AM Jul. 2002 25 3001 4838 837
Home range of the Carpentarian rock-rat Wildlife Research 331

Table 2. GLIM models based on the percentage of point locations of Z. palatalis in

slope habitats and valley habitats at Wollogorang Station, Northern Territory

Habitat Deviance Effect d.f Wald P

explained statistic
Slope 26% Intercept 1 41.9 0.000
Age 1 6.3 0.011
Gender 1 0.0 0.985
Season 1 0.5 0.469
Valley 46% Intercept 1 151.2 0.000
Age 1 13.9 0.000
Gender 1 2.0 0.147
Season 1 1.9 0.162

100% Visual comparison of the fixes within spatially

Mean proportion point locations

90% overlapping core areas indicated that little temporal

80%
70%
60%
50%
40%
30%
20%
10%
0%
dry wet male female juv.
(n = 8) (n = 13) (n = 11) (n = 10) (n = 7) (n = 14) (n = 21)
Fig. 3. Mean percentage of point locations of Z. palatalis occurring
in each of the three habitat types studied at Wollogorang Station with
effects of season, gender and age. The total column represents the
percentage of each habitat type available in the total study area.
There was no significant difference in MCP home-range
areas of females (mean = 12646 m2, s.d. = 6224) and males
(mean = 9819 m2, s.d. = 5365) when age classes were pooled
(F1,19 = 1.25, P = 0.28). Wet-season MCPs were not
significantly different to dry-season MCPs (F1,19 = 3.12,
P = 0.09). There was no significant difference in the size of
KL 50% core areas between males and females (F1,19 = 0.25,
P = 0.62).
The home ranges of Z. palatalis tracked during the same
period overlapped with those of up to 3 other individuals and
had a mean area of intercept of 4254 m2 MCP (s.d. = 2873)
and a mean percentage overlap of 41% (s.d. = 27).
Home-range overlap was observed within and between sex
and age categories and any differences between these
categories were not statistically significant. Z. palatalis were
less likely to have overlapping core areas (n = 11 compared
with n = 18 for overlapping total home ranges), with the
mean area of intercept being small (MCP = 343 m2,
s.d. = 402) and a mean percentage overlap of 38%. A
Kruskal–Wallis ANOVA showed there was no significant
difference in the proportion of overlap between male:male,
male:female or female:female core-area overlap
(H2, 11 = 2.4, P = 0.307). Nor was there a significant
difference in the overlaps of core areas between age
categories (H2, 11 = 5.6, P = 0.06).
separation occurred. This is supported by direct observations
during the active tracking period when two collared animals
plateau (1 female and 1 male) occupied the same boulder pile for
slope
valley
extended and repeated periods. Uncollared conspecifics
were observed within the home ranges of collared rats and
probably ~50% of Z. palatalis in the study area were
adult total collared, thus areas of overlap expressed are underestimates.
Animal movements
The mean distance moved between successive point
locations (≥15 min apart) for all Z. palatalis was 38.8 m
(s.d. = 18.4) and the mean total distance moved over a
4-night period was 1488 m (s.d. = 679.8). The maximum
distance moved between successive points for any animal
was 269 m. These straight-line distances are conservative
estimates as they do not represent the animal’s true course.
Often animals did not move at all between location estimates.
The frequency of movements is difficult to interpret due to
variable gaps in the data where a successful fix could not be
made on an animal for some time. Generally, animals moved
many short distances from the centre of their home range
(88% of locations being within 60 m of the arithmetic mean)
and only occasionally moved larger distances (3% of
positions being further than 100 m from the arithmetic
mean).
Habitat utilisation
When the percentage of fixes in each of the three habitat
types were analysed using a GLIM model to determine the
effects of season, gender and age on habitat selection,
successful models were developed for slope and valley areas
but not for plateau areas (Table 2).
The mean proportion of point locations was highest for all
season, age and gender categories in the slope habitat
(Fig. 3). However, there were significantly more point
locations in slope areas for adults than juveniles and in valley
areas for juveniles than adults (Table 2, Fig. 3). The effects
of season and gender were not statistically significant.
332 Wildlife Research
1.5
1.0
0.5
Axis 2
0.0
–0.5
–1.0
–1.5
–2 –1 0 1 2 Cox et al. 2000) and the house mouse (Mus domesticus)
Axis 1
Fig. 4. Ordination of 47 quadrats based on plant species
abundances. 䊉 = valley sites, 䊐 = slope sites, 䉱 = plateau sites
(stress value 0.15).
Friedman Tests using MCP home-range areas indicated
that rock-rats showed a significant non-random preference
for slope and valley areas over plateau areas (F2,21 = 23.1,
P < 0.000). Indeed, most home ranges were centred on the
ecotone between these two habitat types.
Habitat ordination
Ordination based on the abundance of all 100 plant species
present in at least 5 quadrats differentiated slope, valley and
plateau sites within a broader continuum of vegetation (Fig.
4). An assessment of the most common plant species and
their locations within the three habitat types conforms well
with the ordination (Table 3). Kruskal–Wallis ANOVAs
determined 24 habitat characteristics that differed
significantly between valley, slope and plateau sites
(Table 4).
Radio-tracking of Z. palatalis required considerable
effort for a relatively small amount of useful data.
Difficulties with signal strength and signal bounce are
inherent in the study of animals in a rocky gorge
environment with dense vegetation cover. Given these
limitations, and even with a configuration of five receiving
stations, most position fixes resulted from a combination of
only three successful bearings. In order to minimise the mass
of the collars for a species that has unusually fragile skin, the
whip antennae was relatively short and fine, which limited
the strength of the signal and led to many of the transmitters
emiting a poor signal by the end of four nights of tracking.
Discussion
Home range
The area required for a home range is influenced by the
interaction of a species’ metabolic requirements and the
H. Puckey et al.
distribution of its food resources (McNab 1963). A small
rodent such as Z. palatalis requires a high-energy diet, which
may be obtained from fruits and seeds of a relatively small
but resource-rich home range (mean MCP = 1.1 ha). MCP
home-range estimates from radio-tracking studies of other
rodents in Australia are mostly restricted to smaller species
of the arid zones such as the Lakeland Downs mouse
(Leggadina lakedownensis) and Forest’s mouse (Leggadina
forresti) (4.8–5.3 ha, 22 g and 0.6–3.6 ha, 20 g respectively:
Moro and Morris 2000), the plains mouse (Pseudomys
australis) (0.4 ha, 45 g: Brandle and Moseby 1999) and the
sandy inland mouse (Pseudomys hermannsburgensis)
(3.5 ha, 18 g: Moro and Morris 2000), or introduced rodents
such as the black rat (Rattus rattus) (0.34–0.76 ha, 320 g:
(2.8 ha, 20 g: Moro and Morris 2000). Although some
caution is necessary in comparing MCP home-range areas
for species of different studies, the home range for
Z. palatalis in this study does fall within the estimates
obtained for the arid-zone and exotic rodents.
Generally, in other small mammal species pressure on
males to increase access to mates results in larger home
ranges for males than females (Bubela et al. 1991). This was
not the case for Z. palatalis, in which there was no
significant difference between females and males. The effect
of reproductive condition in females was not tested in the
current study. Although Trainor (1996) found distinct
seasonality in breeding of Z. palatalis, the current study
found pregnant females and juveniles in the population
throughout the year. There were no discernible differences in
the home-range areas between the wet and dry season.
Juveniles had significantly smaller home ranges than adults.
The home ranges of Z. palatalis exhibited a high degree
of spatial overlap both within and between gender and age
categories. This extensive overlap may be the result of a high
population density within the site. These factors, coupled
with lack of paternal care for young, may facilitate a
promiscuous mating system with opportunistic matings
between males and females (Emlen and Oring 1977). Core
areas showed a much lower degree of overlap, with most
examples occurring between a male and a female. In only
one instance was there overlap between the core areas of two
adjacent females (both adults) and the degree of overlap was
extremely low. Females in this population may be defending
resources such as food or nesting sites within the core area.
Females in captivity tend to show much greater aggression
than males and frequently injure other individuals in the
same enclosure (Helen Puckey, personal observation).
Habitat utilisation
Habitat types at Banyan Gorge occur as a gradation from
plateau through slope to valley. These broadly defined
macrohabitats differed substantially in terms of plant species
composition, vegetation structure, topography and rock
Home range of the Carpentarian rock-rat Wildlife Research 333

Table 3. Representative plant species and their percentage occurrence in quadrats in each of the
three habitat types of the study area at Wollogorang Station, Northern Territory
Only plants occurring in >50% of quadrats of at least one habitat type are included. n = total number of
sites surveyed in each habitat type

Species Plateau Slope Valley

(n = 9) (n = 26) (n = 12)
Triodia bitextura 100 23 0
Acacia asperulacea 89 4 0
Eriachne ciliata 89 19 0
Sorghum spp. 89 15 0
Stemodia lythrifolia 89 85 8
Cheilanthes brownii 78 58 0
Tephrosia spechtii 78 15 0
Acacia humifusa 67 50 8
Corymbia capricornia 67 8 0
Terminalia carpentariae 67 58 25
Cajanus aromaticus 56 38 0
Calytrix exstipulata 56 27 0
Cheilanthes sp. 56 50 8
Cymbopogon procerus 56 27 0
Eucalyptus miniata 56 12 0
Gardenia fucata 56 46 0
Petalostigma quadriloculare 56 27 0
Triumfetta sp. 56 50 8
Boronia lanceolata 78 96 50
Alphitonia excelsa 56 85 42
Aristida exserta 56 85 8
Corymbia aspera 33 85 25
Erythrophleum chlorostachys 56 85 58
Acacia hemsleyi 67 81 8
Digitaria sp. 11 81 8
Antidesma parvifolium 11 77 8
Owenia vernicosa 56 77 8
Distichostemon hispidulus 56 73 8
Exocarpos latifolius 11 73 33
Livistona inermis 56 73 42
Sida macropoda 11 73 17
Tephrosia conspicua 56 69 50
Acacia lamprocarpa 0 62 58
Croton habrophyllus 0 50 8
Jasminum molle 0 50 0
Lindsaea ensifolia 0 12 92
Scleria lingulata 0 4 83
Grevillea pteridifolia 0 12 75
Syzygium angophoroides 0 12 75
Buchanania obovata 44 42 67
Corymbia ptychocarpa 0 38 67
Hibiscus sp. 0 58 67
Melastoma affine 0 23 67
Blechnum orientale 0 0 58
Cyperus haspan 0 8 58
Dicranopteris linearis 0 8 58
Mnesithea rottboellioides 0 19 58
Acacia neurocarpa 0 31 50
Cajanus acutifolius 0 19 50
Fimbristylis pauciflora 0 8 50
Glochidion sp. 0 19 50
Melaleuca leucadendra 0 8 50
Total number of species present 89 160 95
334 Wildlife Research H. Puckey et al.

Table 4. Median values for selected environmental attributes that characterise each of the three habitat types in the
study area (measured at 47 quadrats), Wollogorang Station, Northern Territory
Codes for significance of H statistic resulting from Kruskal–Wallis ANOVA tests to determine significant differences in habitat
variables are as follows: * = P < 0.05, ** = P < 0.01, ** = P < 0.001

Habitat attribute Valley Slope Plateau K–W H Significance

(n = 12) (n = 26) (n = 9) level
Patch size (ha) <1 <1 1–5 15.55 ***
Slope (°) 4.5 45 30 21.36 ***
Distance to water (m) 0 <50 50–500 33.03 ***
Fire Imp (index 0–5) 2 1 1 18.16 ***
Canopy height (m) 12.5 8 7 27.69 ***
Canopy cover (%) 25 5.5 5 20.63 ***
Soil depth (cm) 10–40 <10 <10 21.56 ***
Rock cover (%): Pebbles 0 <2 <2 12.47 **
Small stones 0 <2 <2 14.09 **
Stones <2 <2 2–10 20.89 ***
Small rocks <2 2–10 10–20 22.07 ***
Rocks <1 10–20 10–20 20.96 ***
Big rocks 0 10–20 2–10 12.31 **
Outcrop 0 10–20 2–10 10.80 **
Ground cover (%): Rocks 10 50 50 20.10 ***
Litter 50 35 20 24.28 ***
Hummock grass 0 0 5 20.79 ***
Sedges 2 1 0 21.44 ***
Forbes 2 1 0 18.51 ***
Structural cover (%): 10 m 10–25 <5 0 18.94 ***
5–10 m 10–25 5–10 <5 9.27 *
3–5 m 5–10 5–10 <5 6.67 *
1–3 m 10–25 5–10 <5 12.87 **
0.5–1 m 10–25 5–10 5–10 11.30 **

cover. Characteristics that varied significantly between
habitats and probably contribute to the dominance of Z.
palatalis activity in the slope habitat are rockiness
(especially that of larger rocks and outcrops) and medium to
high structural complexity in the vegetation cover that comes
with high species richness. This conforms with Trainor et al.
(2000), who found that the occurrence of Z. palatalis was
associated strongly with rockiness and plant species
richness. Studies on other mammals have indicated that
caves, crevices and rockpiles are important topographic
features providing (a) shelter from predators and fire, (b)
lower levels of vegetation damage from grazing, (c) a
reduction in the extent and uniformity of fire, (d)
topographic complexity for nesting and foraging, and (e)
support for greater floristic diversity (Freeland et al. 1988;
Woinarski et al. 1992; Smith and Quin 1996;Trainor et al.
2000). Other studies have also found small mammal species
to be strongly associated with vegetation of high structural
complexity (Barnett et al. 1978; Braithwaite and Gullen
1978; Friend et al. 1988; Moro 1991; Cox et al. 2000).
Preference for slope habitats by Z. palatalis might be
attributed to stability of food resources in this extremely
seasonal environment. The slope habitat offers the most
structural complexity as well as the highest diversity of plant
species: 160 plant species were recorded on the slope habitat,
compared with 89 species on the plateau and 95 in the valley.
It has been suggested that, in the seasonally dry tropics,
greater water availability in rocky gorge and escarpment
habitats drives higher species richness and plant productivity
compared with surrounding woodland habitats (Freeland
et al. 1988). Trainor et al. (2000) found that the distribution
of Z. palatalis was significantly associated with variables
typical of dry monsoon rainforest: high tree species richness,
woody plant species richness, and shrub cover. The higher
plant species richness of slope habitat potentially offers a
greater range of fruit and seed resources, many of which have
been identified in the diet of both Z. palatalis and Z. maini
(including Antidesma parvifolium, Owenia vernicosa,
Terminalia carpentariae, Exocarpos latifolius, Livistona
inermis and Jasminum molle) (Begg and Dunlop 1985;
Helen Puckey, unpublished data).
Preliminary faecal analyses indicated that 84% of the
Z. palatalis diet comprises fruits and dicotyledon seeds
(Helen Puckey, unpublished data). The diet of the central
rock-rat (Zyzomys pedunculatus) and Z. maini also contains
a large proportion of seed and fleshy fruits (85% and 72%
respectively: Watts 1977; Nano et al. 2003). Fleshy-fruited
plants in most tropical regions show distinct fruiting
Home range of the Carpentarian rock-rat
seasonality (Kimura et al. 2001). Some seasonal changes in
diet have been shown for the common rock-rat (Z. argurus)
and Z. maini (Begg and Dunlop 1985). Mass fruiting of
species like Syzigium angophoroides and Antidesma
parvifolium occurs from the beginning of the wet season
when other seed resources (particularly grass seeds) are lost
either to germination or rotting. Fruiting of most broadleaf
woodland and dry monsoon rainforest plants in Banyan
Gorge is greatest in the dry season (Trainor 1996). However,
some plants such as Ficus spp. and Terminalia carpentariae
show asynchronous fruiting within the population,
producing fruit consistently throughout the year. Other
species with hard seed testa such as Livistona inermis,
Owenia vernicosa and T. carpentariae can persist as
resources in the leaf litter or rock crevices for well over a year
when other fleshy fruits become scarce. Rocky substrates
have been found to retain fallen grass seeds longer than
gravel-free substrates, where seeds become buried or burnt
(Woinarski 1990). Seed caching or scatter-hoarding is
common among rodents where fruiting resources fluctuate
substantially, such that individuals forage for maintenance
requirements and also for food to store (Swihart et al. 1988;
Hoch and Adler 1997; Brewer and Rejmanek 1999; Yasuda
et al. 2000; Forget and Vander Wall 2001). Small seed caches
have been found on rock ledges at Banyan Gorge (Helen
Puckey, personal observation) and are thought to be secure
eating sites or stores for later consumption by Z. palatalis.
Seasonal shifts in habitat use have been demonstrated in
other rodent species (King and Moller 1997; Cox et al.
2000); however, Z. palatalis continued to show a preference
for slope habitats regardless of season.
Although Z. palatalis spends a greater proportion of time
within gorge habitats (i.e. valley and slope) many individuals
do make occasional forays into the surrounding broadleaf
woodlands of the plateau habitat. For the purpose of
fine-scale home-range and habitat-utilisation analysis this
paper considered only triangulation data with an error ellipse
of <20m2. If this is relaxed to 50 m2 to include a coarser scale
of data, some individuals showed infrequent but reliable
forays up to 1.5 km from the main gorge habitats. The
species may depend on plateau habitat for longer-distance
dispersal (for example, to other branches of the gorge
system) or foraging for supplementary foods, particularly
when preferred fruits within the gorge are scarce. The
abundance of T. carpentaria on the plateau sites provides
such a resource.
Prior to this study the only recorded movements for
Z. palatalis were from trapping data and ranged up to 120 m
(Trainor 1996). The distances moved by other small
mammals generally exceed those moved by Z. palatalis in
this study (Read 1984; Leung et al. 1993; Moro and Morris
2000), especially species in more arid environments, where
distances of up to 14 km have been recorded for a rodent
species (Dickman et al. 1995). Movements of small
Wildlife Research 335
mammals in tropical parts of Australia appear to be more
limited, with distances of >1 km seldom being recorded
(Dickman et al. 1995). For a frugivorous specialist,
long-range movements through suboptimal habitat to other
isolated resource patches may not be feasible and may be a
restricting factor in the current distribution of the species.
In the plateau habitat Z. palatalis is particularly
susceptible to both direct and indirect impacts of fire. The
high fuel levels in the spinifex understorey of the plateau
habitat enable hot fires to burn in a spatially homogeneous
rather than mosaic fashion to the boundaries of the slope and
valley habitats. Excessively frequent fire has been
established as a potential threat to Z. palatalis and rainforest
habitats in the Gulf region (Russell-Smith and Bowman
1992; Brook et al. 2002). Possible impacts of fire include
reduced population numbers (direct loss of animals and the
following indirect effects), reduced resource availability
(loss of plant species, seeds, fruits, nest places and refuge),
increased predation (loss of vegetation cover) and reduced
individual fitness (lowered breeding success associated with
fewer resources) (Sutherland and Dickman 1999). Studies of
other rock rat species (Z. argurus and Z. woodwardi) have
shown fewer lactating and pregnant females in the year after
a fire and lower juvenile recruitment than in pre-fire years as
well as shifts in habitat use (Begg et al. 1981). Gradual
elimination of fire-sensitive plant species typical of the slope
and valley is of concern because these plants characterise the
preferred habitat of Z. palatalis and constitute the greatest
proportion of its diet.
Management implications
Identification and characterisation of the habitat use of
Z. palatalis are essential in conservation management aimed
at the protection of essential habitats. Previous research on
the spatial organisation of the species has been restricted to
mark–recapture studies, which are limited by seasonal
trappability and few data points (Trainor 1996). Assessing
home-range size and overlap using radio-tracking techniques
has enabled a detailed measure of the species’ use of space
and its sociality. Assessment of habitat characteristics in
relation to use of space by this species helps in our
understanding of its ecology and also assists managers to
predict the effects of disturbances such as fire on population
dynamics.
Maintenance of habitats that provide monsoon vine
thickets and scree slopes associated with permanent water in
rocky gorges appears essential to the conservation of Z.
palatalis. Sites must be managed not only to protect these
primary habitats, but also the surrounding broadleaf
woodlands. Low-intensity fuel-reduction burns in the
surrounding habitat during the early dry season are
necessary to exclude late dry-season high-intensity wildfires
and protect the habitat and food plants of Z. palatalis.
336 Wildlife Research
Acknowledgments
This study was funded by the Parks and Wildlife
Commission of the Northern Territory and a grant from the
Natural Heritage Trust. We thank John Burke, Jodie Gogler,
Nerida Holznagel, Andrew O’Connor, John Papple, Nick
Shaw and Dave Wurst and all the volunteers for assistance in
the field. Paul Zlotkowski of Wollogorang Station and Val
and Tony Inwood at Redbank Mine provided logistical
support. Tony Griffiths and Andrew Gibbons assisted with
the data analysis. We are grateful to John Woinarski, Alaric
Fisher, Glen Edwards and two anonymous referees for
discussion and comments on the manuscript.
References
Baillie, J., and Groombridge, B. (1996). ‘IUCN Red List of Threatened
Animals.’ (IUCN: Gland, Switzerland.)
Barnett, J. L., How, R. A., and Humphreys, W. F. (1978). The use of
habitat components by small mammals in eastern Australia.
Australian Journal of Ecology 3, 277–285.
Begg, R. J., and Dunlop, C. J. (1985). The diets of the large rock-rat,
Zyzomys woodwardi, and the common rock-rat, Zyzomys argurus
(Rodentia: Muridae). Australian Wildlife Research 12, 19–24.
Begg, R. J., Martin, K. C., and Price, N. F. (1981). The small mammals
of Little Nourlangie Rock, NT. V. The effects of fire. Australian
Wildlife Research 8, 515–527.
Belbin, L. (1994). ‘PATN: Pattern Analysis Package Users Guide.’
(CSIRO Division of Wildlife and Ecology: Canberra.)
Brandle, R., and Moseby, K. E. (1999). Comparative ecology of two
populations of Pseudomys australis in northern South Australia.
Wildlife Research 26, 541–564.
Braithwaite, R. W., and Gullen, P. K. (1978). Habitat selection by small
mammals in a Victorian heathland. Australian Journal of Ecology 3,
109–127.
Brewer, S. W., and Rejmanek, M. (1999). Small rodents as significant
dispersers of tree seeds in a Neotropical forest. Journal of
Vegetation Science 10, 165–174.
Brook, B. W., Griffiths, A. D., and Puckey, H. L. (2002). Modelling
strategies for the management of the critically endangered
Carpentarian rock-rat (Zyzomys palatalis) of northern Australia.
Journal of Environmental Management 65, 355–368.
doi:10.1016/S0301-4797(02)90561-X
Bubela, T. M., Happold, D. C. D., and Broome, L. S. (1991). Home
range and activity of the broad-toothed rat, Mastacomys fuscus, in
subalpine heathland. Wildlife Research 18, 39–48.
Churchill, S. K. (1996). Distribution, habitat and status of the
Carpentarian rock-rat, Zyzomys palatalis. Wildlife Research 23,
77–91.
Cox, M. P. G., Dickman, C. R., and Cox, W. G. (2000). Use of habitat
by the black rat (Rattus rattus) at North Head, New South Wales: an
observational and experimental study. Austral Ecology 25,
375–385. doi:10.1046/J.1442-9993.2000.01050.X
De Solla, S. R., Bonduriansky, R., and Brooks, R. J. (1999). Eliminating
autocorrelation reduces biological relevance of home range
estimates. Journal of Animal Ecology 68, 221–234. doi:10.1046/
J.1365-2656.1999.00279.X
Dickman, C. R., Predavic, M., and Downey, F. J. (1995). Long-range
movements of small mammals in arid Australia: implications for
land management. Journal of Arid Environments 31, 441–452.
Emlen, S., and Oring, L. (1977). Ecology, sexual selection and the
evolution of mating systems. Science 197, 215–223.
H. Puckey et al.
Freeland, W. J., Winter, J. W., and Raskin, S. (1988). Australian rock
mammals: a phenomenon of the seasonally dry tropics. Biotropica
20, 70–79.
Friend, G. R., Dudzinski, M. L., and Cellier, K. M. (1988). Rattus colleti
(Rodentia: Muridae) in the Australian wet-dry tropics: seasonal
habitat preferences, population dynamics and the effects of buffalo
(Bubalus bubalis). Australian Journal of Ecology 13, 51–66.
Forget, P. M., and Vander Wall, S. B. (2001). Scatter-hoarding rodents
and marsupials: convergent evolution on diverging continents.
Trends in Ecology and Evolution 16, 65–67. doi:10.1016/
S0169-5347(00)02072-3
Harris, S., Cresswell, W. J., Forde, P. G., Trewhella, W. J., Woollard, T.,
and Wray, S. (1990). Home-range analysis using radio-tracking data
– a review of problems and techniques particularly as applied to the
study of mammals. Mammal Review 20, 97–123.
Hoch, G. A., and Adler, G. H. (1997). Removal of black palm
(Astrocaryanum standleyanum) seeds by spiny rats (Proechimys
semispinosus). Journal of Tropical Ecology 13, 51–58.
Hooge, P. N., and Eichenlaub, B. (2000). Animal movement extension
to ArcView. Version 2.0. US Geological Survey, Alaska Biological
Science Center, Anchorage, AK.
Horsup, A. (1994). Home range of the allied rock-wallaby, Petrogale
assimilus. Wildlife Research 21, 65–84.
Kimura, K., Yumoto, T., and Kikuzawa, K. (2001). Fruiting phenology
of fleshy-fruited plants and seasonal dynamics of frugivorous birds
in four vegetation zones on Kinabalu, Borneo. Journal of Tropical
Ecology 17, 833–858.
King, C. M., and Moller, H. (1997). Distribution and response of rats
Rattus rattus, R. exulans to seedfall in New Zealand beech forests.
Pacific Conservation Biology 3, 143–155.
Leung, L. K. P., Dickman, C. R., and Moore, L. A. (1993). Genetic
variation in fragmented populations of an Australian rainforest
rodent, Melomys cervinipes. Pacific Conservation Biology 1, 58–65.
McNab, B. K. (1963). Bioenergetics and the determination of home
range size. American Naturalist 97, 133–141. doi:10.1086/282264
Menkhorst, K. A., and Woinarski, J. C. Z. (1992). Distribution of
mammals in monsoon rainforests of the Northern Territory. Wildlife
Research 19, 295–316.
Moro, D. (1991). The distribution of small mammal species in relation
to heath vegetation near Cape Otway, Victoria. Wildlife Research 18,
605–618.
Moro, D., and Morris, K. (2000). Movements and refugia of Lakeland
Downs short-tailed mice, Leggadina lakedownensis, and house mice,
Mus domesticus, on Thevenard Island, Western Australia. Wildlife
Research 27, 11–20. doi:10.1071/WR99016
Nams, V. O. (2000). ‘Locate II User’s Guide.’ (Pacer: Nova Scotia,
Canada.)
Nano, T. J., Smith, C. M., and Jefferys, E. (2003). Investigation into the
diet of the central rock-rat (Zyzomys pedunculatus). Wildlife
Research 30, 513–518. doi:10.1071/WR01084
Puckey, H. (2003). Additional records of the Carpentarian rock-rat
(Zyzomys palatalis) at Redbank, close to the type locality. Northern
Territory Naturalist 17, 43–45.
Read, D. G. (1984). Movements and home ranges of three sympatric
dasyurids, Sminthopsis crassicaudata, Planigale gilesi and P.
tenuirostris (Marsupialia), in semiarid western New South Wales.
Australian Wildlife Research 11, 223–234.
Rooney, S. M., Wolfe, A., and Hayden, T. J. (1998). Autocorrelation data
in telemetry studies: time to independence and the problem of
behavioural effects. Mammal Review 28, 89–98. doi:10.1046/
J.1365-2907.1998.00028.X
Russell-Smith, J., and Bowman, D. M. J. S. (1992). Conservation of
monsoon rainforest isolates in the Northern Territory. Biological
Conservation 59, 51–63. doi:10.1016/0006-3207(92)90713-W
Home range of the Carpentarian rock-rat
Seaman, D. E., and Powell, R. A. (1996). An evaluation of the accuracy
of Kernel density estimators for home range analysis. Ecology 77,
2075–2085.
Smith, A. P., and Quin, D. G. (1996). Patterns and causes of extinction
and decline in Australian conilurine rodents. Biological
Conservation 77, 243–267. doi:10.1016/0006-3207(96)00002-X
Sutherland, E. F., and Dickman, C. R. (1999). Mechanisms of recovery
after fire by rodents in the Australian environment: a review.
Wildlife Research 26, 405–419.
Swihart, R. K., and Slade, N. A. (1985). Testing for independence of
observations in animal movements. Ecology 66, 1176–1184.
Swihart, R. K., Slade, N. A., and Bergstrom, B. J. (1988). Relating body
size to the rate of home range use in mammals. Ecology 69,
393–399.
Thackway, R., and Cresswell, I. D. (1996). ‘An Interim Biogeographic
Regionalisation for Australia.’ (Australian Nature Conservation
Agency: Canberra.)
Trainor, C. R. (1996). Habitat use and demographic characteristics of
the endangered Carpentarian rock-rat Zyzomys palatalis. M.Sc.
Thesis, Northern Territory University, Darwin.
Trainor, C., Fisher, A., Woinarski, J., and Churchill, S. (2000).
Multiscale patterns of habitat use by the Carpentarian rock-rat
(Zyzomys palatalis) and the common rock-rat (Z. argurus). Wildlife
Research 27, 319–332. doi:10.1071/WR97040
http://www.publish.csiro.au/journals/wr
Wildlife Research 337
Watts, C. H. S. (1977). The foods eaten by some Australian rodents
(Muridae). Australian Wildlife Research 4, 151–157.
White, G. C., and Garrott, R. A. (1990). ‘Analysis of Wildlife
Radio-tracking Data.’ (Academic Press Inc.: California.)
Woinarski, J. C. Z. (1990). Effects of fire on the bird communities of
tropical woodlands and open forests in northern Australia.
Australian Journal of Ecology 15, 1–22.
Woinarski, J. C. Z., Braithwaite, R. W., Menkhorst, K. A., Griffin, S.,
Fisher, A., and Preece, N. (1992). Gradient analysis of the
distribution of mammals in Stage III of Kakadu National Park, with
a review of the distribution patterns of mammals across
north-western Australia. Wildlife Research 19, 233–262.
Yasuda, M., Miura, S., and Hussein, N. A. (2000). Evidence for food
hoarding behaviour in terrestrial rodents in Pasoh Forest Reserve, a
Malaysian lowland rainforest. Journal of Tropical Forest Science
12, 164–173.
Manuscript received 2 April 2003; accepted 23 December 2003