Upload
brasilia
View
0
Download
0
Embed Size (px)
Citation preview
B I O L O G I C A L C O N S E R V A T I O N 1 4 1 ( 2 0 0 8 ) 3 2 0 – 3 3 1
. sc iencedi rec t .com
ava i lab le at wwwjournal homepage: www.elsevier .com/ locate /b iocon
Multi-scale patterns of habitat use by re-introducedmammals: A case study using medium-sized marsupials
Graeme R. Finlaysona,*, Emerson M. Vieirab, David Priddelc, Robert Wheelerc,Joss Bentleyd, Chris R. Dickmana
aInstitute of Wildlife Research, School of Biological Sciences, University of Sydney, Sydney, NSW 2006, AustraliabLaboratorio do Ecologia de Mamiferos, Centro 2, Universidade do Vale do Rio dos Sinos, UNISINOS,
CP 275 Sao Leopoldo, RS 93022-000, BrazilcDepartment of Environment and Climate Change (NSW), P.O. Box 1967, Hurstville, NSW 2220, AustraliadAustralian Wildlife Conservancy, Scotia Sanctuary, via Wentworth, NSW 2648, Australia
A R T I C L E I N F O
Article history:
Received 14 May 2007
Received in revised form
10 October 2007
Accepted 14 October 2007
Available online 26 November 2007
Keywords:
Habitat use
Re-introduction
Conservation
Endangered
Marsupial
0006-3207/$ - see front matter � 2007 Elsevidoi:10.1016/j.biocon.2007.10.008
* Corresponding author: Tel.: +61 (0)2 9351 86E-mail address: [email protected]
A B S T R A C T
Knowledge of the habitat requirements of threatened species at both local and landscape
scales is crucial for maintaining viable populations and for making conservation and man-
agement decisions. Here, we use live trapping and radio-tracking to investigate habitat use
by four species of threatened marsupials – burrowing bettongs (Bettongia lesueur), brush-
tailed bettongs (B. penicillata), greater bilbies (Macrotis lagotis), and bridled nailtail wallabies
(Onychogalea fraenata). The study populations had been re-introduced to Scotia Sanctuary in
western New South Wales, Australia, within a predator-proof area. All showed preferences
for particular macrohabitats while resting by day, with M. lagotis and B. penicillata selecting
Eucalyptus woodland with Triodia understorey and B. lesueur and O. fraenata selecting Euca-
lyptus woodland with shrubs. However, they showed no such partiality at night. Bettongia
penicillata used areas with Triodia and litter but few herbs for shelter, while burrows of M.
lagotis avoided shrubs. Habitat components that influenced trap captures were: crust cover
and herb layer cover (negative) for B. penicillata, trees <5 m in height and number of shrubs
(both negative) for B. lesueur, crust cover for M. lagotis, and crust cover and trees <5 m high
for O. fraenata (both negative). There was also a negative association at this scale between B.
penicillata and both B. lesueur and M. lagotis, suggesting the possibility of competition. Our
results support the idea that studies at multiple spatial scales are crucial to understand
the habitat use and requirements of threatened fauna, and should therefore be incorpo-
rated into future re-introduction programs.
� 2007 Elsevier Ltd. All rights reserved.
1. Introduction
Habitat can be defined as the physical space required by an
organism to gain essential resources for survival and repro-
duction (Partridge, 1978). Patterns of habitat use are not ran-
dom, with animals generally using certain habitats in
er Ltd. All rights reserved
83; fax: +61 (0)2 9351 411d.edu.au (G.R. Finlayson
preference to others. Such preferences might be influenced
by both abiotic and biotic factors, with the latter including
availability of feeding resources, risk of predation, and intra-
and interspecific competition (Falkenberg and Clarke, 1998).
The scale on which animals respond to habitat differences
can also vary, with some showing distinct responses to differ-
.
9.).
B I O L O G I C A L C O N S E R V A T I O N 1 4 1 ( 2 0 0 8 ) 3 2 0 – 3 3 1 321
ences both between habitats and between patches within
habitats (i.e. microhabitats) (Torres-Contreras et al., 1997; Kelt
et al., 1999; Moura et al., 2005).
Understanding habitat preferences and patterns of habitat
utilization by threatened fauna can be critical for informing
management and conservation decisions. For example, key
microhabitats that protect remaining breeding populations
of the midwife toad (Alytes muletensis) from the impacts of
introduced species are now used to determine possible
re-introduction sites for this species (Moore et al., 2004).
Knowing species’ habitat requirements is important also for
conservation planning in relation to reserve size and location
(Noss et al., 2002) and for incorporation into models that pre-
dict locations of suitable habitat patches (Metzger et al., in
press; Smith et al., in press). Moreover, assessment of the
habitats used by threatened species can be at the forefront
of conservation efforts for those species (Cahill and Matthy-
sen, in press).
Since European settlement of Australia, the decline in the
mammalian fauna has been severe (Burbidge and McKenzie,
1989; Dickman, 1994a, b; Short and Smith, 1994; Short, 1998;
Burbidge and Manly, 2002). This decline has been attributed
to many factors including predation by the introduced red
fox (Vulpes vulpes) and cat (Felis catus), and competition with
introduced herbivores such as the European rabbit (Oryctola-
gus cuniculus) (Morton, 1990). In an attempt to restore ele-
ments of Australia’s biodiversity, re-introductions into large
fenced areas, where the aforementioned threats have been
removed, provide opportunities for creating large, secure
self-sustaining populations of threatened species and also
for studying aspects of their biology to assist future conserva-
tion and management. As part of the management of these
populations, post-release studies into aspects of the species’
behaviour and ecology are essential (IUCN, 2006). Information
on the local abundance and success of re-introduced popula-
tions is also of utmost importance (Lindenmayer, 1994).
Furthermore, multiple-species re-introduction programs pro-
vide unique opportunities to gain insights about former inter-
species interactions which no longer occur elsewhere.
One such re-introduction program is at Scotia Sanctuary in
western New South Wales, where 4000 ha of relatively intact
old-growth low-growing, multi-stemmed eucalypt woodland
have been fenced and all introduced large and medium-sized
mammals (foxes, cats, goats (Capra hircus) and rabbits) eradi-
cated. Since November 2004 a series of re-introductions has
been carried out, with burrowing bettongs (Bettongia lesueur),
brush-tailed bettongs (B. penicillata), greater bilbies (Macrotis
lagotis) and bridled nailtail wallabies (Onychogalea fraenata)
being experimentally released into the fenced area. At the time
of European settlement, all four species were widespread
across much of arid and semi-arid Australia (Dickman et al.,
1993; Strahan, 1995). In particular, bettongs were reported to
have some of the most extensive ranges of anyof the Australian
marsupials (Troughton, 1957; Finlayson, 1958) and were likely
to have occupied a wide range of vegetation communities
across their range. Bilbies also ranged over much of mainland
Australia, but are now restricted to just 20 per cent of their for-
mer distribution in parts of the Tanami Desert (Northern Terri-
tory), Pilbara and southern Kimberley (Western Australia), and
as an isolated population in south-western Queensland
(Southgate, 1990). The bridled nailtail wallaby suffered such a
severe range reduction and population decline in the early
20th century that it was presumed extinct (Maxwell et al.,
1996). A single population was rediscovered in central Queens-
land in 1973 (Gordon and Lawrie, 1980).
These species, along with many other native species that
once dominated the semi-arid regions of Australia, probably
played a pivotal role in landscape functioning. For example,
bettongs disperse fungal spores (Claridge and May, 1994), in-
crease productivity of herbs (Noble et al., 2007), disperse seeds
(Murphy et al., 2005) and create nutrient patchiness by their
extensive digging (Garkaklis et al., 2003; James and Eldridge,
2007). There is also evidence that selective browsing by these
species assisted with the regulation of fire-promoting popula-
tions of native shrubs (Noble and Grice, 2002). Insights into
patterns of habitat use should help wildlife managers to
understand and address the loss of such ecological processes.
The future success of re-introductions of species back into
parts of their original ranges also relies on the understanding
of the interaction between these species and their
environment.
The re-introduction of the four species at Scotia Sanctuary
was part of a large-scale multi-species biodiversity restora-
tion project undertaken by the Australian Wildlife Conser-
vancy in collaboration with the New South Wales
Department of Environment and Climate Change and the
University of Sydney. This initiative was the first re-introduc-
tion involving two species of bettongs, and the first in New
South Wales involving all four species. It provided a unique
opportunity to study patterns of habitat use by these species.
Neither the broad vegetation formations (macrohabitats) nor
the finer-scale habitat components used by these species
are known for the habitats studied here, although they cover
much of the former ranges of each species. We aimed to eval-
uate habitat use at these different scales and to quantify pat-
terns of habitat selection and partitioning during diurnal
(resting) and nocturnal (active) periods. Partly because of
the species’ high population densities, we also investigated
whether competition for resources might potentially play a
role in shaping patterns of space use among species. Specifi-
cally, we predicted that any habitat partitioning would be
most pronounced between the two bettong species, as would
be expected from competitive avoidance. Both species histor-
ically co-occurred within the region and, although their diets
probably differ, increased dietary overlap could be expected
within an enclosed area (Sampson, 1971; Christensen, 1980;
Robley et al., 2001; Murphy et al., 2005). We expected our re-
sults to provide insight into the habitat requirements of all
species, hence informing management decisions in both this
and other multi-species re-introductions.
2. Study site
The study was carried out at Scotia Sanctuary (64,000 ha;
141�10 0E, 33�10 0S), 150 km south of Broken Hill, on the bound-
ary of the arid and semi-arid climatic zones (Fig. 1). Rainfall
averages 257 mm a year and is highly irregular. The region en-
dures hot summers with mean daily temperatures of 17–33 �Cand cool winters with mean daily temperatures of 5–17 �C.
Evaporation rates are high, with evaporation rates that are
Fig. 1 – Scotia Sanctuary, New South Wales, showing the location of the sanctuary and surrounding reserves. The top inset
shows the area in relation to Australia and the bottom inset shows the release site in relation to the entire sanctuary.
322 B I O L O G I C A L C O N S E R V A T I O N 1 4 1 ( 2 0 0 8 ) 3 2 0 – 3 3 1
six times higher than the average annual rainfall (Westbrooke
et al., 1998). The sanctuary lies within a region of sedimentary
rocks of Cainozoic age, with red earthy sands and sandy sol-
onized brown soils overlying sandy clays, generally of aeolian
origin (Westbrooke et al., 1998). The dominant landforms in-
clude east–west parallel sand dunes with narrow sandy
swales and open calcareous swales of varying width.
Four dominant vegetation communities, or macrohabitats,
occur within the 4000-ha fenced site used for animal releases:
Eucalyptus woodland with an understorey of Triodia scariosa;
Eucalyptus woodland over shrubs; Casuarina pauper woodland;
and shrubland (previously cleared woodland undergoing
regeneration) (Fig. 2). Dominant Eucalyptus species are E. ole-
osa, E. costata, E. dumosa, and E. socialis. Frequently occurring
grasses and herbs in these communities include Austrostipa
spp., Vittadinia cuneata complex, Dissocarpus paradoxus, Cheno-
podium cristatum and Podolepis capillaries (Westbrooke et al.,
1998; J. Bentley, pers. obs). Despite previous disturbance, Sco-
tia Sanctuary supports some of the most intact Eucalyptus and
Casuarina woodland in the region. The sanctuary has a short
grazing history, and only small areas of woodland have burnt
in the release site since 1975 (Fig. 2).
3. Methods
3.1. Re-introduction background
Releases occurred within the fenced site between November
2004 and September 2005, with 172 brush-tailed bettongs,
118 burrowing bettongs, 40 greater bilbies, and 161 bridled
nailtail wallabies released there from on-site captive breeding
colonies. During the period of the current study there were
two releases, in June 2005 and September 2005. Numbats
(Myrmecobius fasciatus) had been released into the fenced site
Fig. 2 – The 4000 ha release site within Scotia Sanctuary showing the main vegetation communities within the reserve;
Eucalyptus with a Triodia scariosa understorey ( ), Eucalyptus with an understorey of mixed shrubland (h), Casuarina pauper
woodland ( ), and mixed shrubland ( ). Some areas of Eucalyptus woodland have been burnt since 1975 (j). Trap sites and
tracks are also indicated.
B I O L O G I C A L C O N S E R V A T I O N 1 4 1 ( 2 0 0 8 ) 3 2 0 – 3 3 1 323
in 1999 and greater stick-nest rats (Leporillus conditor) were re-
leased in 2006.
3.2. Post-release radio-tracking
As part of the re-introduction program, ten individuals of
each species were radio-tracked for periods of two weeks (1)
immediately following release and (2) six weeks after release.
Bettongs and wallabies were fitted with collar-mounted VHF
transmitters (25 g and 31 g respectively) and bilbies with a
9 g tail-mounted VHF transmitter (SirTrack Wildlife Tracking
Solutions, New Zealand). All transmitters were less than 2%
of body weight. Animals were radio-tracked daily during each
two-week period to their diurnal resting place (burrow, den, or
shelter; hereafter referred to as shelter site). A description of
each shelter site was recorded and the location determined
using a GPS unit (Garmin� 12X). Only those shelter sites lo-
cated 6–8 weeks after release were used to describe habitat
use because sites used during the first two weeks were con-
sidered likely to be temporary (i.e. prior to the establishment
of a stable home range).
3.3. Animal trapping
Trapping sessions were conducted every three months be-
tween June 2005 and June 2006. A trapping network of 114 trap
sites was established along an extensive system of tracks
which covered the entire study site (Fig. 2). Each trap site con-
tained three wire cage traps (small: 57 · 23 · 23 cm; medium:
62 · 25 · 25 cm; and large: 76 · 33 · 32 cm), and was about
500 m from its nearest neighbour. Traps were set at dusk, cov-
ered in hessian sacks to protect animals from the elements,
324 B I O L O G I C A L C O N S E R V A T I O N 1 4 1 ( 2 0 0 8 ) 3 2 0 – 3 3 1
and baited with a mixture of rolled oats, peanut butter, honey
and vanilla essence. Trap sites were grouped into two series,
or runs, of approximately equal size. Each series was trapped
for three nights, with all trapping taking place over six consec-
utive nights. In an attempt to limit possible sampling bias, the
direction of travel along a particular series was reversed on
alternate nights, that is, for example, the starting point on
night 1 was the finishing point on night 2. We started to check
and close traps 2–3 h after sunset, and finished before sunrise.
Bettongs and wallabies were identified using individually num-
bered passive integrated transponder tags (AVID�, supplied by
Austock Rural Services Pty Ltd.), with new animals being tagged
at first capture. To identify bilbies recaptured during each trip
we marked the ear of each bilby with ink. We recorded the date,
site of capture, passive integrated transponder tag number,
species and sex of each animal trapped.
3.4. Habitat variables
Since the trapping grid covered the entire study site (Fig. 2),
macrohabitat availability was calculated as the sum of trap sta-
tions within each of the four major vegetation communities
present. We also measured 10 fine-scale habitat variables that
could potentially influence the distribution of the four species
within the study site. We considered each trap station to be a
sampling point, and measured variables along three 20-m tran-
sect lines, the starting point of each being selected randomly
within a 20-m radius of the trap or shelter site. We estimated
the following variables: crust cover (% cover of hardened soil
surface); number of Triodia hummocks; trees at two different
heights (<5 m and >5 m); herb cover (plants <0.5 m in height);
litter cover; amount of bare ground; the number of shrubs;
the area covered by logs (>10 cm diameter); and the area cov-
ered by vegetative debris (uprooted trees or shrubs, usually
associated with roads). Crust cover, litter cover, bare ground
cover and herb cover were calculated using a step-point meth-
od (Evans and Love, 1957; Etchberger and Krausman, 1997) as
the proportion of total footsteps along each line that fell within
each of these categories. Proportions were transformed using
arcsine prior to analysis (Zar, 1996). We recorded all other hab-
itat variables 2-m either side of the transect line. Trees with
trunks outside the transect, but with overhanging branches,
were counted as being within the transect. All variables were
measured once around each trap site during March 2006. We
considered the arithmetic mean of the total of each habitat var-
iable recorded on the three transects as the value for that par-
ticular variable at each trap site.
The same 10 habitat variables were also measured at 30
shelter sites (10 each for B. lesueur, B. penicillata and M. lagotis)
that had been found by radio-telemetry. We did not include
O. fraenata as this species is substantially larger than the other
three and individuals generally rest above ground in the
shade of trees or shrubs. We did not wish to disturb and scat-
ter resting animals at their shelters by our sampling protocol.
3.5. Data analysis
3.5.1. Diurnal habitat useTo detect patterns of diurnal habitat use at the macrohabitat
scale we used chi-squared contingency tests to determine
whether the distributions of shelter sites of the four species
showed any association with the distributions of available
macrohabitats within the study site.
To analyze patterns of diurnal habitat use at a finer scale
and detect differences among species we used a multi-re-
sponse permutation procedure within the program PC-ORD
4.20 (McCune and Mefford, 1999). The multi-response permu-
tation test statistic is based on the within-group average of
pairwise distance measures between object response values
in Euclidian data space (Zimmerman et al., 1985). It is a
non-parametric procedure for testing the hypothesis of no
difference between two or more groups of entities and is anal-
ogous to a t-test or a one-way F-test, but is not constrained by
the assumptions required by these tests (Biondini et al., 1988).
We compared differences among four groups: three spe-
cies (B. lesueur, B. penicillata and M. lagotis) and the available
habitat (i.e. measured at all trap sites). Although adequate
for comparing groups with different sample sizes, as in this
study, the multi-response permutation analysis only indi-
cates differences among groups; it does not identify which
variables drive the differences. Consequently, when this anal-
ysis indicated a significant difference among the four groups,
we compared each of the 10 original variables among these
groups using a Kruskal–Wallis test to indicate which variables
contributed to the observed inter-group differences. For these
analyses we used SigmaStat (version 2.03; SPSS Inc., Rich-
mond, California).
3.5.2. Nocturnal habitat useTo detect patterns of nocturnal habitat use at the macrohab-
itat scale we again used chi-squared tests to determine
whether the distributions of trapping locations were associ-
ated with those of the macrohabitats within the study site.
For each individual, we used only the first capture at least
six months after release and the first capture in each subse-
quent trapping session. Multiple trap encounters of the same
individual within a single trapping session were not used as
they were considered to be non-independent.
To analyze habitat utilization at a finer-scale during noc-
turnal periods we used data on species occurrences at trap
sites and habitat variables recorded at each trap site to run
a canonical correspondence analysis. Data were again re-
stricted to the first capture at least six months after release
and the first capture in any subsequent trapping session.
The canonical correspondence analysis, performed using
PC-ORD 4.20, provides an explanatory analysis of potential
relationships between two or more sets of variables (McGari-
gal et al., 2000), in this case, the association between species
occurrences (captures at trap sites) and habitat variables. We
also used the Monte Carlo permutation procedure within PC-
ORD to test the significance of relationships between habitat
variables and species occurrences. This results in a species-
environment correlation coefficient (ter Braak, 1995).
We used multiple regression analysis to determine the ef-
fect of the finer-scale habitat variables, as well as the presence
of the other species, on the occurrence of each re-introduced
species during periods of nocturnal habitat use. This analysis
yields a more specific prediction of the interaction between
each species and the measured habitat variables and can be ap-
plied after the use of an exploratory analysis, such as canonical
B I O L O G I C A L C O N S E R V A T I O N 1 4 1 ( 2 0 0 8 ) 3 2 0 – 3 3 1 325
correspondence analysis, to accurately confirm its results (ter
Braak, 1995; Dalmagro and Vieira, 2005). The technique is also
useful as it provides insight into the potential for competition
between species and describes community patterns of multi-
ple species associated with particular habitat variables (Fox
and Luo, 1996; Luo et al., 1998). Following Dalmagro and Vieira
(2005), we considered the number of captures of each species as
the dependent variable and the finer-scale habitat variables as
the independent variables. We standardized the species vari-
ables by dividing each value by the standard deviation, thus
equalling thevariance to 1 to provide a more accurate represen-
tation of possible competitive effects (Fox and Luo, 1996; Luo
et al., 1998). We ran a backward stepwise multiple regression
analysis (a-to-enter = 0.15, a-to-remove = 0.15) with each spe-
cies as the dependent variable and the fine-scale habitat vari-
ables and occurrences of the other three species as the
independent variables. Variables were checked for inter-corre-
lation and three (Triodia cover, % bare soil and % litter cover)
were removed from the final regression analysis to reduce
any autocorrelation problems (Philippi, 1993). Analyses were
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Casuarina pauper Eucalyptus & Triod
Prop
ortio
n
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Casuarina pauper Eucalyptus & Triodia
Prop
ortio
n
Fig. 3 – (a) The proportion of available habitat types (j) and the p
each habitat type for four re-introduced species: B. lesueur ( ), B
Sanctuary. (b) The proportion of available habitat types (j) and
habitat type for four re-introduced species: B. lesueur ( ), B. penic
period at Scotia Sanctuary.
performed using Systat Software Inc. v. 11.0 (Richmond, Cali-
fornia, USA).
4. Results
4.1. Animal trapping
In total, we captured the four study species 2564 times in 3420
trap nights between June 2005 and June 2006. This comprised
836 captures of 162 individual B. lesueur, 1485 captures of 195
B. penicillata, 107 captures of 94 M. lagotis and 136 captures of
83 O. fraenata.
4.2. Habitat use
4.2.1. Diurnal habitat useIn all, 1023 diurnal locations were obtained for all species.
Pair-wise comparisons between macrohabitats that were
available and those used for diurnal resting indicated habitat
selection by each of the four species (Fig. 3a). The frequency of
ia Eucalyptus & shrubs Shrubland
Eucalyptus & shrubs Shrubland
roportion of daytime shelter or burrow sites recorded within
. penicillata (h), M. lagotis ( ) and O. fraenata ( ) at Scotia
the proportion of nocturnal captures recorded within each
illata (h), M. lagotis ( ) and O. fraenata ( ) within a 12-month
326 B I O L O G I C A L C O N S E R V A T I O N 1 4 1 ( 2 0 0 8 ) 3 2 0 – 3 3 1
burrow sites used by B. lesueur differed among the available
macrohabitats (Fig. 3a; v2 = 190.74, df = 3, p < 0.001), with more
burrows than expected located within Eucalyptus woodland
with a shrubby understorey. The frequency of shelter sites
used by B. penicillata also differed from that available
(Fig. 3a; v2 = 297.27, df = 3, p < 0.001), with disproportionately
more shelters located in Eucalyptus woodland over Triodia. A
similar pattern was found for burrows of M. lagotis (Fig. 3a;
v2 = 264.38, df = 3, p < 0.001). The frequency of diurnal shelter
Table 1 – Median values (25–75% percentiles) of the ten habitare-introduced to Scotia Sanctuary
Available Bettongia penicilla
Trees < 5 m (80 m2) 0.3 (0–1.00)a 3.8 (1.3–4.7)b
Trees > 5 m (80 m2) 0.3 (0–1.00)a 1.8 (1.50–3.33)b
Shrubs > 0.5 m (80 m2) 7.3 (3.7–10.3)a 3.8 (1.00–5.67)a,b
Log length (m) 0 (0–1.0) 0 (0–0)
Debris area (m2) 0 (0–2.7) 0 (0–0)
Crust cover (%) 17.4 (10.3–29.3) 13.5 (12.3–19.0)
Herb layer cover (%) 17.7 (5.7–34.8)a 3.8 (0.0–4.3)b
Triodia (80 m2) 0 (0–5.0)a 26.3 (24.0–26.9)b
Bare soil (%) 30.1 (16.7–42.9) 23.9 (17.5–28.0)
Litter (%) 22.2 (6.2–39.4)a 57.1 (53.8–67.2)b
Data for available habitat were collected at trapping sites (see text). P va
variable and were less than 0.05 where indicated. Pairwise a posteriori
lowercase letters.
M.lagotis
B.penicillata
Debris area
Loglength
Shrubs
Trees>5m
Trees<5m
Crust cover
Fig. 4 – Ordination diagram from canonical correspondence analy
trapping locations at least six months after release at Scotia Sanc
arrows.
sites used by O. fraenata indicated selection for Eucalyptus
woodland with a shrubby understorey and shrubland, and
avoidance of Eucalyptus woodland with Triodia understorey
(Fig. 3a; v2 = 80.89, df = 3, p < 0.001).
When comparing the association of shelter sites and fine-
scale habitat variables, we found a marked difference between
the four groups (three species and available habitat) based on
the multi-response permutation procedure (p < 0.0001).
Kruskal–Wallis tests revealed significant differences for six of
t variables measured at shelter sites of three species
ta Bettongia lesueur Macrotis lagotis P
1.2 (0.3–2.0)a,b 0.8 (0.2–1.8)a,b <0.001
0.8 (0.33–1.67)a,b 1.3 (0.7–2.0)a,b <0.001
5.5 (3.7–0.7)a,b 3.2 (1.3–5.0)b 0.008
0 (0–3.7) 0 (0–1.2) 0.25
0 (0–0.3) 1.0 (0–2.7) 0.20
10.6 (5.3–24.3) 23.5 (15.7–28.0) 0.36
4.0 (1.7–12.9)a,b 8.4 (5.9–13.5)a,b <0.001
3.5 (0–10.9)a 11.33 (0.17–33.9)a,b <0.0001
41.6 (29.2–46.6) 29.8 (9.6–41.6) 0.21
36.4 (30.6–46.6)a,b 37.9 (31.1–43.2)a,b <0.001
lues are for Kruskal–Wallis tests comparing the four groups for each
significant differences (Dunn–Sidak test; P < 0.05) are indicated with
B.lesueur
O.fraenata
Herb layercover
sis of the distribution of four re-introduced species based on
tuary. Species are indicated by a (j) and habitat variables by
Table 2 – Habitat and species associations for trapping captures of four mammal species re-introduced to Scotia Sanctuary
Environmental variables Target species
Bettongia lesueur Bettongia penicillata Macrotis lagotis Onychogalea fraenata
Habitat
Crust 2.214b 1.216a �1.298a
<5 m trees �0.175a �0.253a
Herb �0.626a
Shrubs �0.019a
Logs
Debris
>5 m trees
Species
B. lesueur �0.196a
B. penicillata �0.264 �0.254a
M. lagotis �0.205a
O. fraenata
Data are based on regression coefficients of the fine scale habitat variables using backwards stepwise analysis. Only the significant partial
regression coefficients of variables that remained in the final model are shown.
a (p < 0.05).
b (p < 0.01).
B I O L O G I C A L C O N S E R V A T I O N 1 4 1 ( 2 0 0 8 ) 3 2 0 – 3 3 1 327
the habitat variables (Table 1). Pair-wise comparisons indi-
cated a difference between the location of B. penicillata shelter
sites and trees less than 5 m in height, trees greater than 5 m in
height, % herb layer cover, Triodia cover and % litter cover. The
shelter sites of B. lesueur differed from those of B. penicillata
with respect to spinifex cover, but were not significantly differ-
ent to the available habitat. The shelter sites of M. lagotis were
associated with fewer shrubs than in the available habitat
(Table 1).
4.2.2. Nocturnal habitat useA total of 991 captures of the four study species was used to
determine whether the nocturnal use of macrohabitats was
random compared to the distribution of habitats within the
study area. No difference was detected among any of the four
species (Fig. 3b; v2 < 3.1; df = 3, p > 0.20, for all species).
At the finer scale, the Monte Carlo permutation showed a
significant species-environment interaction (correlation coef-
ficient = 0.494, p = 0.005), with canonical correspondence
analysis revealing that different habitat variables were associ-
ated with each species (Fig. 4). The occurrence of B. penicillata
was related to crust cover and trees less than 5 m high, whilst
that of B. lesueur was associated with an increased cover of
herbs. There was also a distinct separation between these
two species along the first axis of the ordination, suggesting
a marked negative association in their occurrence (Fig. 4).
Associations were less obvious for either M. lagotis or
O. fraenata.
The multiple regression revealed similar patterns to those
of the canonical correspondence analysis, with significant
associations between species and particular habitat variables
for the four species (Table 2). There was a negative correlation
between the occurrence of B. lesueur and trees less than 5 m, a
strong positive correlation between the occurrence of B. peni-
cillata and crust cover, and negative correlations between the
occurrence of O. fraenata with crust and trees less than 5 m in
height (Table 2). The values obtained for the multiple regres-
sion were: M. lagotis, R2 = 0.117, p = 0.027; B. penicillata,
R2 = 0.287, p < 0.001; B. lesueur, R2 = 0.155, p = 0.002; and for O.
fraenata, R2 = 0.134, p = 0.007. Competition coefficients were
significant between both species of Bettongia and between B.
penicillata and M. lagotis (Table 2).
5. Discussion
The four study species showed some similarities and differ-
ences in their patterns of habitat use at the two scales evalu-
ated, and also between day and night. All showed selection
for diurnal resting locations but appeared to be less selective,
or were randomly distributed, during nocturnal foraging peri-
ods. Our results highlight the need for evaluating patterns of
habitat use not only for activity periods but also during rest-
ing. The resting period is often neglected in habitat-use stud-
ies as data collection is often based primarily on captures
(Vernes, 2003). However, as each of the species in this study
probably faces higher risks of predation by day, it is also dur-
ing this period that they might be more selective in terms of
habitat use.
A lack of preference for any particular macrohabitat dur-
ing foraging was also observed in the same study area for
numbats (Vieira et al., 2007). Although there is little informa-
tion on population densities for numbats or for any of our
study species in western New South Wales prior to European
settlement, our results may indicate that sufficient food and
other resources were available across all habitats and that
selection for any particular habitat conferred no net benefit.
Also, these species are highly mobile; for example, both
B. penicillata and B. lesueur often move more than a kilometre
in a single night (Christensen, 1980; Christensen and Burrows,
1995). Consequently, there is a chance that traps separated by
500 m caught animals passing on their way to preferred forag-
ing sites. Alternatively, within the fenced area where long-
distance dispersal is limited, foraging in unsuitable areas
may be forced upon individuals once the carrying capacity
within each favoured habitat is reached and territories are
established. In the latter situation, assessment of the impact
328 B I O L O G I C A L C O N S E R V A T I O N 1 4 1 ( 2 0 0 8 ) 3 2 0 – 3 3 1
of these animals on the overall health of the environment
should be a priority. Trends in body condition over a longer
period of time should assist in determining which hypothesis
holds true.
At a finer scale we recorded differences among species in
habitat utilization for both diurnal resting and nocturnal forag-
ing. However, all specieswere more selective for diurnal resting
sites than for nocturnal foraging habitats, as indicated by the
generally lower values of the regressions. Differences in fine-
scale habitat selection were particularly evident for the two
species of Bettongia, leading to a negative relationship between
the numbers of captures of each of these two species.
Since release, the population of B. penicillata has increased
and then declined whereas the population of B. lesueur has in-
creased steadily (G. Finlayson, unpublished data). The differ-
ence in population trends between the two species may be
due to differences in habitat use during a prolonged drought
that extended through the current study. By burrowing, B.
lesueur may be better able to cope with drought than its con-
gener, which does not generally use burrows but prefers to
construct elaborate nests above ground (Christensen, 1995).
A burrowing lifestyle can be expected to be advantageous in
this semi-arid environment, especially during extended dry
periods. Burrows provide shelter from environmental ex-
tremes (Ancel et al., 1998) and so reduce water loss and
expenditure of energy (Darden, 1972; Hansell, 1993; Shimmin
et al., 2002). There is evidence that throughout the more arid
expanses of their historical range, populations of B. lesueur
vastly outnumbered those of B. penicillata, a trend that was re-
versed in the higher rainfall areas of south-western Australia
(Finlayson, 1958). Despite some fine-scale habitat partition-
ing, the negative association between the two bettong species
revealed by the regression analysis may be suggestive of com-
petition. The continued monitoring of these bettong popula-
tions is thus of utmost importance, particularly as both
translocated and naturally occurring populations of B. penicil-
lata elsewhere have declined significantly in recent years
(A. Wayne, pers. comm.). Regression analysis also indicated
potential competitive effects between B. penicillata and M. lag-
otis. More detailed studies of the fine-scale movements, diets
and interactions between B. lesueur, B. penicillata and M. lagotis
may be useful to balance the long term management of these
species at Scotia Sanctuary.
The species most selective for diurnal shelter was B. peni-
cillata. At the macrohabitat scale this species favoured areas
of Eucalyptus woodland with a Triodia understorey, with a high
proportion of shelters occurring at the base of Triodia hum-
mocks. This pattern was confirmed at a finer scale using
habitat variables measured at shelter sites. Interestingly, in
contrast to previous observations of this species (Christensen,
1980; 1995; Christensen and Leftwich, 1980), only a few B. pen-
icillata used burrows or piles of debris for shelter, and this
generally occurred within the first two weeks after release,
prior to the establishment of stable home ranges.
The results of the canonical correspondence analysis and
the higher regression coefficient for B. penicillata suggest fur-
ther that this species is more highly associated with certain
habitat variables during nocturnal foraging than are the other
species. The association between this species and crust cover
can be attributed to the species’ association with spinifex, as
crust cover and spinifex cover were strongly correlated. It
seems that B. penicillata favours areas of increased spinifex,
which provide cover for both diurnal shelter and protection
during nocturnal foraging. In a population of B. penicillata in
Western Australia, scrub density and bare ground were iden-
tified as important characteristics of the preferred habitat for
this species, with animals absent from open areas and areas
with extremely dense ground cover (Christensen, 1980). The
Kruskal–Wallis tests revealed that the shelter sites of B. peni-
cillata were also associated with sites with more trees less
than 5 m high.
Cover can have strong effects on the use of habitat by
small mammals (Dickman, 1992; Hughes and Ward, 1993). In
a concurrent study examining the use of microhabitats by
the re-introduced bettongs at Scotia, Pizzuto et al. (2007)
found that nocturnal foraging movements and activity points
of both B. penicillata and B. lesueur were associated with high
levels of ground cover. In our study, we found B. lesueur to
be positively, but not significantly, correlated with herb layer
and negatively correlated with trees less than 5 m high. This
may be a product of them preferring disturbed habitat where
trees have been removed and where drainage lines and soil
disturbance favour the construction of large warrens and
the growth of species in the herb layer. It is also possible that
favoured food resources such as chenopod forbs, which dom-
inate the herb layer and are more abundant in these open
areas, are the driving force behind some of the observed dis-
tribution patterns observed. After re-introduction on the
north-west coast of Western Australia, B. lesueur favoured
feeding sites in regenerating vegetation that had been re-
cently burnt (Christensen and Burrows, 1995).
We found no evidence of habitat partitioning between the
two semi-fossorial species, B. lesueur and M. lagotis. Both spe-
cies were trapped in similar ratios across all macrohabitats.
They both constructed their burrows in three of the four mac-
rohabitats, but overlapped only in two. Macrotis lagotis fa-
voured areas of Eucalyptus woodland with a Triodia
understorey whereas B. lesueur tended to construct burrows
in Eucalyptus woodland with a shrubby understorey. At the
Arid Recovery Reserve in South Australia, where both species
have been successfully re-introduced, also into a fenced area,
both species tended to construct their burrows in dune habi-
tat (most similar to the Eucalyptus woodland with a Triodia
understorey on Scotia Sanctuary) (Moseby and O’Donnell,
2003; Finlayson and Moseby, 2004). In the current study, con-
trasting rates of capture between species make it difficult to
draw more definitive conclusions.
At two sites in Queensland, O. fraenata has been shown to se-
lect shrubs, hollow logs, fallen timber and dense grass tussocks
for diurnal shelter, and is generally solitary (Fisher, 2000). At
Scotia Sanctuary, this species generally rested at the base of
shrubs in shallow depressions termed hip-holes (Eldridge and
Rath, 2002) within areas of Eucalyptus woodland with a shrubby
understorey. Animals also used fallen logs, piles of debris and
large burrows for shelter. Debris is common on Scotia Sanctu-
ary where trees have been felled for the construction of fences
and roads. This characteristic may explain the negative corre-
lation we observed between captures of O. fraenata and small
trees. After the first release, in December 2004, O. fraenata
favoured the shelter of shrubs at the interface of Eucalyptus
B I O L O G I C A L C O N S E R V A T I O N 1 4 1 ( 2 0 0 8 ) 3 2 0 – 3 3 1 329
woodland with a shrubby understorey and more open and pre-
viously cleared shrubland, which supports favourable pasture
for foraging with a dense herb layer of grasses and forbs. Patchy
clearing of regrowth to create edges near pasture has been pre-
viously noted as an appropriate management strategy for this
species (Fisher, 2000). For many other Australian vertebrates,
particularly macropods, the creation of habitat mosaics is re-
garded as an appropriate management objective (Law and
Dickman, 1998).
In conclusion, our results indicate that, at a macrohabitat
scale, the four species do not select habitat types for foraging
but select specific habitats for diurnal resting, with M. lagotis
and B. penicillata preferring areas of Eucalyptus woodland with
Triodia understorey and B. lesueur and O. fraenata seeming to
select areas of Eucalyptus woodland with a shrubby understo-
rey. At a finer scale, both species of Bettongia showed distinct
patterns of habitat use, with B. penicillata being the most
selective in terms of the habitat variables we measured. As
this is also the species whose population has been declining
in the study area, this may imply a competitive effect of
B. lesueur on B. penicillata; this interpretation is supported by
the negative relationship between both species that we
detected at a fine habitat scale. However, these results must
be approached with caution, as competition may be a func-
tion of the state of the habitat at the time of the study, which
is likely to vary in association with climatic conditions or
wildfire in the region. Seasonally fluctuating exploitative
competition has been hypothesised between B. lesueur and
O. cuniculus in semi-arid coastal Western Australia where die-
tary overlap was higher during the summer months when for-
age was less available (Robley et al., 2001). There could also be
a change in the behaviour of all species once second and third
generation individuals establish in their populations. Previ-
ous experimental studies in other taxa have demonstrated
that captive-bred individuals can behave differently in re-
introduced populations (Kelley et al., 2006), and therefore
potentially drive ‘unnatural’ behaviour patterns. We also
found that distinct differences between species in their habi-
tat use emerged at different spatial scales. The apparent
superiority of B. lesueur is possibly more related to habitat
use at the micro-habitat scale or may vary in relation to cli-
matic conditions which fluctuate dramatically in arid sys-
tems. Our results reinforce the view that multi-species
studies at multiple spatial scales are crucial to understand
the habitat use of threatened fauna and should be incorpo-
rated into species recovery efforts by conservation agencies.
This is particularly important for re-introductions of multi-
species populations into parts of their former range in Austra-
lia, where land-use practices have dramatically altered the
landscape, and should remain a priority in recovery planning.
Acknowledgements
This project was completed in accordance with approval from
the University of Sydney’s Animal Ethics Committee. Project
title: ‘Scotia endangered mammal recovery project’, project
approval number L04/12-2004/2/4010. All field work was con-
ducted under scientific permit number S10614 Department
of Environment and Climate Change (New South Wales). We
thank the Australian Wildlife Conservancy for providing field
equipment, on-site accommodation and other facilities. We
also thank numerous volunteers who provided assistance
with fieldwork. Funding was provided by an Australian Wild-
life Conservancy postgraduate award (to Graeme Finlayson)
and by a Postdoctoral Grant to Emerson Vieira from the Brazil-
ian Government (CAPES – Fundacao Coordenacao de Aper-
feicoamento de Pessoal de Nıvel Superior). We are grateful
to the editor and two anonymous referees for helpful reviews
and comments.
R E F E R E N C E S
Ancel, A., Fetter, L., Groscolas, R., 1998. Changes in egg and bodytemperature indicate triggering of egg desertion at a bodymass threshold in fasting blue petrels (Halobaena caerulea).Journal of Comparative Physiology B: Biochemical, Systemic,and Environmental Physiology 168, 533–539.
Biondini, M.E., Mielke Jnr, P.W., Berry, K.J., 1988. Data-dependentpermutation techniques for the analysis of ecological data.Plant Ecology 75, 161–168.
Burbidge, A.A., Manly, B.F.J., 2002. Mammal extinctions onAustralian islands: Causes and conservation implications.Journal of Biogeography 29, 465–473.
Burbidge, A.A., McKenzie, N.L., 1989. Patterns in the moderndecline of Western Australia’s vertebrate fauna: causesand conservation implications. Biological Conservation 50,143–198.
Cahill, J.R.A., & Matthysen, E., in press. Habitat use by twospecialist birds in high-Andean Polylepis forests. BiologicalConservation, doi:10.1016/j.biocon.2007.07.022.
Christensen, P., 1995. Brush-tailed Bettong. In: Strahan, R. (Ed.),The Mammals of Australia. Reed New Holland, Sydney, pp.292–293.
Christensen, P., Burrows, N., 1995. Project desert dreaming:Experimental reintroduction of mammals to the GibsonDesert, Western Australia. In: Serena, M. (Ed.), ReintroductionBiology of Australian and New Zealand Fauna. Surrey Beatty &Sons, Chipping Norton, pp. 199–207.
Christensen, P., Leftwich, T., 1980. Observations of the nest-building habits of the Brush-tailed Rat-kangaroo or Woylie(Bettongia penicillata). Journal of the Royal Society of WesternAustralia 63, 33–38.
Christensen, P.E.S., 1980. The biology of Bettongia penicillata (Gray,1837) and Macropus eugenii (Desmarest, 1817), in relation to fire.Forests Department of Western Australia Bulletin 91, 1–90.
Claridge, A.W., May, T.W., 1994. Mycophagy among Australianmammals. Australian Journal of Ecology 19, 251–275.
Dalmagro, A.D., Vieira, E.M., 2005. Patterns of habitat utilization ofsmall rodents in an area of Araucaria forest in Southern Brazil.Austral Ecology 30, 353–362.
Darden, T.R., 1972. Respiratory adaptations of a fossorialmammal, the pocket gopher (Thomomys bottae). Journal ofComparative Physiology 78, 121–137.
Dickman, C.R., 1992. Predation and habitat shift in the housemouse, Mus domesticus. Ecology 73, 313–322.
Dickman, C.R., 1994a. Mammals of New South Wales: Past,present and future. Australian Zoologist 29, 158–165.
Dickman, C.R., 1994b. Native mammals of western New SouthWales: Past neglect, future rehabilitation? In: Lunney, D.,Hand, S., Reed, P., Butcher, D. (Eds.), Future of the Fauna ofWestern New South Wales. Royal Zoological Society of NewSouth Wales, Mosman, pp. 81–91.
330 B I O L O G I C A L C O N S E R V A T I O N 1 4 1 ( 2 0 0 8 ) 3 2 0 – 3 3 1
Dickman, C.R., Pressey, R.L., Lim, L., Parnaby, H.E., 1993.Mammals of particular conservation concern in the WesternDivision of New South Wales. Biological Conservation 65,219–248.
Eldridge, D.J., Rath, D., 2002. Hip-holes: kangaroo (Macropus spp.)resting sites modify the physical and chemical environment ofwoodland soils. Austral Ecology 27, 527.
Etchberger, R.C., Krausman, P.R., 1997. Evaluation of fiver methodsfor measuring desert vegetation. Wildlife Society Bulletin 25,604–609.
Evans, R.A., Love, R.M., 1957. The step-point method of sampling –practical tool in range management. Journal of RangeManagement 10, 208–212.
Falkenberg, J.C., Clarke, J.A., 1998. Microhabitat use of deer mice:effects of interspecific interaction risks. Journal ofMammalogy 79, 558–568.
Finlayson, G.R., Moseby, K.E., 2004. Managing confinedpopulations: the influence of density on the home range andhabitat use of reintroduced burrowing bettongs (Bettongialesueur). Wildlife Research 31, 457–463.
Finlayson, H.H., 1958. On central Australian mammals. Part III.The Potoroinae. Records of the South Australian Museum 13,235–307.
Fisher, D.O., 2000. Effects of vegetation structure, food and shelteron the home range and habitat use of an endangered wallaby.Journal of Applied Ecology 37, 600–671.
Fox, B.J., Luo, J., 1996. Estimating competition coefficients fromcensus data: a re-examination of the regression technique.Oikos 77, 291–300.
Garkaklis, M.J., Bradley, J.S., Wooller, R.D., 2003. The relationshipbetween animal foraging and nutrient rich patchiness insouth west Australian woodland soils. Australian Journal ofSoil Research 41, 665–673.
Gordon, G., Lawrie, G.C., 1980. The rediscovery of the bridlednail-tailed wallaby, Onychogalea faenata (Gould) (Marsupialia:Macropodidae) in Queensland. Australian Wildlife Research 7,339–345.
Hansell, M.H., 1993. The ecological impact of animal nests andburrows. Functional Ecology 7, 5–12.
Hughes, J.J., Ward, D., 1993. Predation risk and disturbance tocover affect foraging behaviour in Namib Desert gerbils.Animal Behaviour 46, 1243–1245.
IUCN, 2006. The IUCN position statement on translocation ofliving organisms: introductions, re-introductions andre-stocking. IUCN, Gland.
James, A.I., Eldridge, D.J., 2007. Reintroduction of fossorial nativemammals and potential impacts on ecosystem processes inan Australian desert landscape. Biological Conservation 138,351–359.
Kelley, J.L., Magurran, A.E., Macias Garcia, C., 2006. Captivebreeding promotes aggression in an endangered Mexican fish.Biological Conservation 133, 169–177.
Kelt, D.A., Meserve, P.L., Patterson, B.D., Lang, B.K., 1999. Scaledependence and scale independence in habitat associations ofsmall mammals in southern temperate rainforest. Oikos 85,320–334.
Law, B.S., Dickman, C.R., 1998. The use of habitat mosaics byterrestrial vertebrate fauna: implications for conservation andmanagement. Biodiversity and Conservation 7, 323–333.
Lindenmayer, D.B., 1994. Some ecological considerations andcomputer-based approaches for the identification ofpotentially suitable release sites for reintroductionprogrammes. In: Serena, M. (Ed.), Reintroduction Biology ofAustralian and New Zealand Fauna. Surrey Beatty & Sons,Chipping Norton, pp. 1–5.
Luo, J., Monamy, V., Fox, B.J., 1998. Competition between twoAustralian rodent species: a regression analysis. Journal ofMammalogy 79, 962–971.
Maxwell, S., Burbidge, A.A., Morris, K. (Eds.), 1996. The 1996 ActionPlan for Australian Marsupials and Monotremes. WildlifeAustralia, Canberra.
McCune, B., Mefford, M.J., 1999. Multivariate Analysis of EcologicalData, Version 4.20. Mjm Software, Glenedon Beach, Oregon.
McGarigal, K., Cushman, S.A., Stafford, S.G., 2000. MultivariateStatistics for Wildlife and Ecology Research. Springer-Verlag,New York.
Metzger, K.L., Sinclair, A.R.E., Campbell, K.L.I., Hilborn, R.,Hopcraft, J.G.C., Mduma, S.A.R., & Reich, R.M., in press. Usinghistorical data to establish baselines for conservation: Theblack rhinoceros (Diceros bicornis) of the Serengeti as a casestudy. Biological Conservation, doi:10.1016/j.biocon.2007.06.026.
Moore, R.D., Griffiths, R.A., Roman, A., 2004. Distribution of theMallorcan midwife toad (Alytes muletensis) in relation tolandscape topography and introduced predators. BiologicalConservation 116, 327–332.
Morton, S.R., 1990. The impact of European settlement on thevertebrate animals of arid Australia: a conceptual model.Proceedings of the Ecological Society of Australia 16,201–213.
Moseby, K.E., O’Donnell, E., 2003. Reintroduction of the greaterbilby, Macrotis lagotis (Reid) (Marsupialia: Thylacomyidae), tonorthern South Australia: Survival, ecology and notes onreintroduction protocols. Wildlife Research 30, 15–27.
Moura, M.C., Caparelli, A.C., Freita, S.R., Vieira, M.V., 2005. Scale-dependent habitat selection in three didelphid marsupialsusing the spool-and-line technique in the Atlantic forest ofBrazil. Journal of Tropical Ecology 21, 337–342.
Murphy, M.T., Garkaklis St., M.J., Hardy, G.E., 2005. Seed caching bywoylies Bettongia penicillata can increase sandalwood Santalumspicatum regeneration in Western Australia. Austral Ecology30, 747–755.
Noble, J.C., Grice, A.C., 2002. Fire regimes in semi-arid and tropicalpastoral lands: managing biological diversity and ecosystemfunction. In: Bradstock, R.A., Williams, J.E., Gill, A.M. (Eds.),Flammable Australia: Fire Regimes and the Biodiversity of aContinent. Cambridge University Press, Cambridge, pp.373–400.
Noble, J.C., Muller, W.J., Detling, J.K., Pfitzner, G.H., 2007.Landscape ecology of the burrowing bettong: Warrendistribution and patch dynamics in semiarid easternAustralia. Austral Ecology 32, 326–337.
Noss, R.F., Carroll, C., Vance-Borland, K., Wuerthner, G., 2002. Amulticriteria assessment of the irreplaceability andvulnerability of sites in the greater Yellowstone ecosystem.Conservation Biology 16, 895–908.
Partridge, L., 1978. Habitat selection. In: Krebs, J.R., Davies, N.B.(Eds.), Behavioural Ecology: An Evolutionary Approach.Blackwell Scientific Publications, Oxford, pp. 351–376.
Philippi, T.E., 1993. Multiple regression: herbivory. In:Scheiner, S.M., Gurevitch, J. (Eds.), Design and Analysis ofEcological Experiments. Chapman and Hall, New York, pp.183–210.
Pizzuto, T.A., Finlayson, G.R., Crowther, M.S., Dickman, C.R., 2007.Microhabitat use of the brush-tailed bettong (Bettongiapenicillata) and burrowing bettong (B. lesueur) in semi-aridNew South Wales: implications for reintroduction programs.Wildlife Research 34, 271–279.
Robley, A.J., Short, J., Bradley, S., 2001. Dietary overlap between theburrowing bettong (Bettongia lesueur) and the European rabbit(Oryctolagus cuniculus) in semi-arid coastal Western Australia.Wildlife Research 28, 341–349.
Sampson, J.C., 1971. The biology of Bettongia penicillata Gray, 1837.PhD thesis, University of Western Australia.
Shimmin, G.A., Skinner, J.D., Baudinette, R.V., 2002. The warrenarchitecture and environment of the southern hairy-nosed
B I O L O G I C A L C O N S E R V A T I O N 1 4 1 ( 2 0 0 8 ) 3 2 0 – 3 3 1 331
wombat (Lasiorhinus latifrons). Journal of Zoology (London) 258,469–477.
Short, J., 1998. The extinction of rat-kangaroos (Marsupialia:Potoroidae) in New South Wales, Australia. BiologicalConservation 86, 365–377.
Short, J., Smith, A., 1994. Mammal decline and recovery inAustralia. Journal of Mammalogy 75, 288–297.
Smith, C.S., Howes, A.L., Price, B., & McAlpine, C.A., in press.Using a Bayesian belief network to predict suitable habitat ofan endangered mammal - The Julia Creek dunnart (Sminthopsisdouglasi). Biological Conservation, doi:10.1016/j.biocon.2007.06.025.
Southgate, R.I., 1990. Distribution and abundance of the GreaterBilby, Macrotis lagotis Reid (Marsupialia: Paramelidae). In:Seebeck, J.H., Brown, P.R., Wallis, R.I., Kemper, C.M. (Eds.),Bandicoots and Bilbies. Surrey Beatty and Sons, Sydney, pp.293–302.
Strahan, R. (Ed.), 1995. Mammals of Australia. Reed Books, Sydney.Ter Braak, C.J.F., 1995. Ordination. In: Jongman, R.H.G., Braak,
C.J.F.t., Tongeren, O.F.R.v. (Eds.), Data Analysis in Communityand Landscape Ecology. Cambridge University Press,Cambridge, pp. 91–173.
Torres-Contreras, H., Silca-Aranguiz, E., Marquet, P.A., Camus,P.A., Jaksic, F.M., 1997. Spatiotemporal variability of rodentsubpopulations at a semiarid neotropical locality. Journal ofMammalogy 78, 505–513.
Troughton, E., 1957. Furred Animals of Australia. Angus andRobertson, Sydney.
Vernes, K., 2003. Fine-scale habitat preferences and habitatpartitioning by three mycophagous mammals in tropical wetsclerophyll forest, north-eastern Australia. Austral Ecology 28,471–479.
Vieira, E.M., Finlayson, G.R., Dickman, C.R., 2007. Habitat use anddensity of numbats (Myrmecobius fasciatus) reintroduced in anarea of mallee vegetation, New South Wales. AustralianMammalogy 29, 17–24.
Westbrooke, M.E., Miller, J.D., Kerr, M.K.C., 1998. The vegetation ofthe Scotia 1: 100 000 map sheet, western New South Wales.Cunninghamia 5, 665–684.
Zar, J.H., 1996. Biostatistical Analysis, 4rd ed. Prentice Hall,Englewood Cliffs, New Jersey.
Zimmerman, G.M., Goetz, H., Mielke, P.W., 1985. Use of animproved statistical method for group comparisons to studyeffects of prairie fire. Ecology 66, 606–611.