Download AND Gehyra variegata) IN REMNANT HABITAT

Biological Conservation 71 (1995) 25-33
0 1994 Elsevier Science Limited
Printed in Great Britain. All rights reserved
0006-3207/95/$9.50
ELSEVIER
0006-3207(94)00017-4
PERSISTENCE OF TWO SPECIES OF GECKO (Oedura reticulata
AND Gehyra variegata) IN REMNANT HABITAT
Stephen
Molecular
Evolution and Systematics,
Sarre
Research School of Biological Sciences, Australian National University, PO Box 475,
Canberra, ACT, 2601, Australia
Graeme
T. Smith
CSIRO, Division of Wildlife and Ecology, LMB 4, Midland,
WA 6056, Australia
&
Jacqueline
A. Meyers
CSIRO, Division of Wildlife and Ecology, PO Box 84. Lyneham,
(Received
7 January
generalists,
dispersal,
2602, Australia
1994; revised version received and accepted 28 February
1994)
INTRODUCTION
Abstract
Identifving
the characteristics
that make a species vulnerable to extinction jtillowing habitat fragmentation
is
one of the most pressing problems in conservation biology. One suggestion is that habitat specialists are less
able to cope with rapid changes to their habitat or to
move through the modljied landscape than habitat generalists and so will be more vulnerable to extinction.
We
examine this hypothesis by comparing the distributior of
two species of gecko, Gehyra
variegata
and Oedura
reticulata,
in patches of remnant woodland. The former
species is a habitat generalist relative to the latter and
shows a markedly higher level of persistence (97% remnant occupancy vs 72%). Logistic regression modelling
of the presence or absence of 0. reticulata revealed a sig@cant
correlation
between
the number of smoothbarked eucalypts (both species preferred by 0. reticulata)
in the remnant and the presence of 0. reticulata.
This
suggests that the probability
of extinction for a given
population is related to the amount of suitable habitat in
the remnant and is a function of processes operating at
the population level rather than on a regional basis. PitfalI trapping in this study and evidence from another
long-term study suggest that the movement of 0. reticulata between remnants is negligible. As a consequence,
this species has been unable to form a metapopulation
at
equilibrium. In contrast, it is likely that G. variegata is
maintaining
its widespread
distribution
through
a
metapopulation
structure. These results demonstrate
the
importance of the ability to form a metapopulation for a
species to maintain persistence
in recently and highly
fragmented
ecosystems.
Keywords: persistence,
habitat
cialists, lizards, fragmentation,
tion, extinction.
ACT,
Clearing
natural
vegetation
for agriculture
or other
development
has occurred at an unprecedented
rate in
recent centuries. The survival of species within a landscape of fragmented
natural
habitat
will depend on
their ability to exploit new habitats such as agricultural
lands and to cope with the changes in their habitat.
These changes include a reduction
in the area of available habitat,
changed spatial relationships
of habitat,
increased fluxes of solar radiation,
wind and water and
the disruption
of ecological processes such as nutrient
cycling,
energy transfer,
competition
and predation
(Saunders et al., 1991).
One major consequence
of clearing native vegetation
is that continuous
or near continuous
populations
of
species are broken into smaller and more insular local
populations.
Small populations
are likely to be more
vulnerable
to extinction
through
stochastic
demographic, genetic and environmental
processes (Shaffer,
1987) although
what constitutes
a minimum
viable
population
size is contentious
(Shaffer,
1981; Soul&
1987; Nunney
& Campbell,
1993) and will vary between species and ecosystems. The relative importance
of extinction
pressures related to population
size, as
opposed to the ‘driven’ causes of extinction
(Caughley,
1994), is difficult to assess. In many cases of humaninduced
habitat
fragmentation,
the minimum
viable
population
size may be irrelevant,
because, regardless
of the population
size of the species, it is unable to
cope with the array of changed conditions
and becomes
locally or even regionally extinct.
The probability
of the regional extinction of a species
in a fragmented
landscape
will also be affected by the
interaction
between sub-populations
of that species. If
a set of populations
interacts on a regional basis (i.e.
habitat spemetapopula25
Stephen
26
Sarre, Graeme
T. Smith,
form a metapopulation:
Levins,
1970; Gilpin,
1987;
Hanski & Gilpin,
1991) then the extent and nature
of the extinction
rate will be affected (Opdam,
1990;
Harrison,
1991; Verboom et al., 1991). Harrison (1991)
has described
four basic models of equilibrium
and
non-equilibrium
metapopulation
structures,
but as
yet few studies
of metapopulation
dynamics
have
been applied to recently fragmented
ecosystems.
Thus
the importance
of metapopulations
to the conservation of species in fragmented
ecosystems
remains
unclear.
The interaction
of subpopulations
and the probability of their extinction
is likely to vary between species.
Consequently,
the ability to predict the vulnerability
of
a given species to fragmentation
is one of the most
pressing problems
facing conservation
biologists.
The
breadth
of habitat
that a species can occupy may
be one characteristic
useful for predicting
the vulnerability of a species to fragmentation,
the general expectation
being that habitat
generalists
will exhibit
low vulnerability
to habitat
fragmentation
(Opdam,
1990; Harris & Silva-Lopez,
1992; Hobbs, 1992). The
basis for this expectation
is that species with narrow
niches are less able to exploit new habitats or disperse
through the modified landscape (Harris & Silva-Lopez,
1992) than species with more general habitat requirements.
In this paper we explore the relative abilities of two
species of gecko (Oedura reticulata and Gehyra variegata) to persist in a fragmented
environment.
0. reticulata is a habitat
specialist
relative
to G. variegata
(Kitchener
& How, 1982; Kitchener
et al., 1988). The
hypothesis tested is that the habitat generalist G. varie
gata will exhibit a higher degree of persistence in habitat remnants
than the habitat
specialist.
A second,
related prediction
derived from island biogeographic
theory (MacArthur
& Wilson, 1967) is that the rate of
extinction
of both species will be related to the size of
the habitat remnant,
the distance to the nearest source
population
and the time since isolation. Specifically, the
extinction
rate is expected to increase with decreasing
remnant size, increasing distance between remnants and
increasing time since isolation.
Jacqueline
MATERIALS
A. Meyers
AND METHODS
Life history characteristics of the two species of gecko
The reticulated velvet gecko 0. reticulata (Bustard 1969)
is restricted to the southwest of Western Australia
although its distribution
within the wheatbelt
region is
extensive (Chapman
& Dell, 1985). It is also limited in
its range of habitat,
being arboreal
and mainly confined to mature smooth-barked
Eucalyptus species with
dead wood (How & Kitchener,
1983). In contrast, the
tree dtella G. variegata (Dumeril
& Bibron 1836) is
widely distributed
over the southern
half of Australia
(Cogger, 1992) although its current systematic status is
uncertain
primarily because of chromosomal
variation
(King,
1979; Moritz,
1986; Cogger,
1992). Moritz
(1986) identified G. variegata as the chromosomal
form
2n = 40B, which is restricted to the southwest of Western Australia.
Chromosomes
from the Gehyra populations used in this study were not examined,
but 2n =
40B is the only chromosome
race found in the region
(C. Moritz, pers. comm.) and so is probably
the form
of G. variegata studied here. Voucher specimens of the
taxa studied will be lodged with the Western Australian
Museum.
G. variegata is a habitat generalist living in
logs and fallen timber, tree trunks or shrubs (Bustard,
1968; Kitchener
et al., 1988) and can also be found
under rocks or in highly disturbed habitats (Kitchener
et al., 1988; personal observations).
A summary of the
life history
characteristics
of both species obtained
from a long-term population
study (How & Kitchener,
1983; Kitchener et al., 1988) is provided in Table 1.
Study area
The study area is 1680 km* between Kellerberrin
and
Trayning
in the Western Australian
wheatbelt (Fig. 1).
The district
is gently
undulating,
with low relief
(100 m) and occasional granite outcrops (Beard, 1980).
Approximately
93% of the original vegetation has been
cleared for agriculture
since 1900, leaving over 450 vegetation remnants.
These range in size from less than
1 ha to 1190 ha (Arnold & Weeldenburg,
1991). The
dominant
agricultural
activity is wool and wheat production.
Table 1. Summary comparison of basic life history parameters for the Konnongorring populations of 0. reticuluta and G. variegata
(Modified from Kitchener et al., 1988)
Species characteristics
0. retieulata
G. variegata
Habitat breadth”
Adult sex ratio
Sexual dimorphism
Longevity (years)
Annual population
survival (x f SD) (%)
Estimated survival to 24 months (%I)
Adult females gravid each season (%)
Clutch size
Clutches/season
‘M adult size when sexually mature
Age of sexual maturity (years)
Mean SVL of reproductively
active female (mm)
1.276
-1: 1
female > male
-14
82.5 f 6.8
33350
c. 100
2
5161
-1: 1
female = male
-14
86.7 +_3.4
39-50
100
1
2
90.0
-2.8
49.9 k 3.2
‘Estimated
from the frequency
94.6
-4.8
64.4 + 4.1
of use of perch sites. SVL = snout vent length
Persistence
of lizards in remnant
habitat
21
Since
0. reticulata
is most commonly
found on
smooth-barked
eucalypts
such as E. salubris and E.
salmonophloia,
the survey was restricted to woodland
remnants
containing
these two species. All remnants
surveyed were unfenced and located within paddocks
used for sheep and wheat production.
The former distribution
of vegetation
associations
containing
these
species (as well as those containing
E. wandoo, which is
has been mapped
also favoured
by 0. reticulata)
(McArthur,
in press; M. G. Brooker, pers. comm.) and
is presented
in Fig. 2. This figure suggests that the
likely distribution
of 0. reticulata within the study area
before European
settlement was widespread but discontinuous. The current known distribution
of 0. reticulata in the study area (Fig. 2) supports the notion that
the species was widespread.
G. variegata also shows a
wide current distribution
(Fig. 2) and given its broad
habitat preferences
would have been distributed
over
most of the region prior to clearing.
Fig. 1. The location of the study area (dark shading) near
Kellerberrin
in Western Australia.
The lightly shaded area
indicates the extent of the Western Australian wheatbelt.
Sampling
The presence or absence of both species of gecko was
determined
by surveying 32 vegetation remnants (31 for
G. variegata because its absence could not be confirmed
in one remnant)
in October-November
1989, January
1991 and February
1992. These remnants represent the
total number
of remnants
containing
homogeneous
stands of E. salubrislE. salmonophloiu in the study area
that have not been cut (i.e. harvested
for timber) or
burnt
since isolation
and that were isolated
before
1937. Although some variation in smooth-barked
eucalypt size structure was observed between remnants,
an
@ G. variegata present
0 0. reticulata present
Fig. 2. The distribution
of Oedura reticulata and Gehyra variegata within the study area. The dark shaded areas represent the habitat remnants, and the light shaded areas the distribution
of habitat suitable for 0. reticulata before clearing (from McArthur,
in
press; M. G. Brooker, pers. comm.)
28
Stephen
Sarre, Graeme
T. Smith,
intensive
survey of habitat usage by 0. reticulata in
three of the remnants
(168, 170, 175, Sarre, in prep),
found no preference by 0. reticulata for any of the tree
characteristics
(height, height of bole, diameter, degree
of senescence,
crown depth,
number
of trees with
touching
crowns,
crown density
and foliage cover)
measured.
Consequently,
no assessment
of smoothbarked eucalypt
variation
was made in the surveys
described
here. Each remnant
was surveyed
by two
observers for a minimum
of 2 h (or until both species
were observed) from dusk onwards using head torches.
More intensive
surveys
in several of the remnants
(Sarre, unpublished
data) showed that even at very low
population
densities both species are usually observed
within 30 min of commencing
the survey. Intensive
hand searching was carried out during the day if one or
both of the species were not seen during the night surveys. If the species were still not found it was deemed
absent from the remnant.
Environmental variables and data analysis
Six variables thought to be important
to the persistence
of 0. reticulata (but not G. variegata, which was present
in all but one of the remnants)
were measured.
These
were owner, area (ha), total number
of trees, total
number of smooth-barked
eucalypts (E. salubris + E.
salmonophloia),
distance
from the nearest
woodland
(m) and years since isolation). The variable ‘owner’ was
included to account for possible variation
in management practices between owners that may affect the persistence of 0. reticulata. In the majority of cases, the
remnants had either been managed by the same family
or by the same pair of owners since clearing. The area
of each remnant
and the distance
between remnants
were calculated
from aerial photographs.
The total
Fig. 3. The vegetation
absent.
The shaded
Jacqueline
A. Meyers
number
of trees and the total number
of smoothbarked eucalypts were obtained by total counts within
each remnant
while the time since clearing was obtained from extensive surveys of all current and past
land holders (G. W. Arnold & J. R. Weeldenburg,
pers.
comm.). If an exact clearing date was not available,
then the midpoint of the estimated limits of the time of
clearing was used. A further three variables,
the proportion of the total number of trees that were smoothbarked
eucalypts,
the
density
of smooth-barked
eucalypts (tree/ha) and the density of total trees, were
also calculated
to determine
if habitat density was important to the persistence of 0. reticulata.
Because the response variable has a binomial
distribution, logistic regression
analysis was used to determine the relationship
between
the presence
of 0.
reticulata in vegetation remnants and the variables. The
continuous
variables
were fitted as both linear and
quadratic functions to test for possible curvature in the
response. Each independent
variable was added to the
null model and its level of significance
determined
using the change in deviance of the fitted model from
that of the null model (Nicholls, 1991a,b). The variable
that accounted
for the greatest change in deviance was
included in the model and then all other variables were
added in turn to determine
if they significantly
improved the model. This process was continued
until no
additional
variables significantly
improved
the model.
The statistical
package GLIM (McCullagh
& Nelder,
1989) was used for all modelling.
Movement between adjacent populations
To obtain a measure of the extent of movement
into
and out of the remnants,
53 pitfall traps (45 X 30 cm)
were installed at a spacing of 7 m, 15 m beyond the
remnants surveyed for the presence/absence
of Oedura reticulata. 0, 0. retiedata present; 0, 0. reticulata
area is the distribution
of habitat suitable for 0. reticulata before clearing (from McArthur,
in press; M. G.
Brooker, pers. comm.).
Persistence
of lizards in remnant habitat
edge of remnant
170 (Fig. 3). A fibreglass drift fence
(30 cm in height) was erected between the pitfalls so
that the remnant was surrounded.
Two in every three
pitfalls were fenced to ensure that only those animals
entering the remnant from the surrounding
wheat fields
could fall into them. The remaining pitfalls were fenced
so that only the animals leaving the remnant could be
captured.
Both species have been caught in identical
pitfalls used in other studies (G. Smith, personal observations).
RESULTS
Distribution of 0. reticulata
The persistence of the two
remnants
varied markedly.
from 28% of the remnants
variegata was absent from
2). Consequently,
a logistic
and G. variegata
species within the woodland
0. reticulata was absent
(Table 2; Fig. 3), whereas G.
only a single remnant (Table
regression model could only
29
be developed for 0. reticulata. Five variables - remnant size, total number
of trees, total number of E.
salubris, total number
of E. salmonophloia
and total
number
of smooth-barked
eucalypts
were each
significantly
correlated
with the presence of 0. reticufata, with the latter being the most significant
(2 =
8.03, p < 0.005, d.f. = 1; Table 3). Interestingly,
neither
time since isolation nor distance to the nearest woodland were significant explanatory
variables for the presence of 0. reticulata. The size of each remnant and the
distance to the nearest woodland
varied over several
orders of magnitude
(0.1-8.0 ha for the former and
O-l 100 m for the latter; Table 2), providing
a good
range with which to examine the effects of these variables on persistence. In contrast, the difference between
the ages of the remnants was small (54-78 years; mean
= 69, SD = 8.0), which may preclude the possibility of
obtaining a significant relationship
between age and the
presence of 0. reticulata if the relationship
is weak.
The 32 remnants
sampled were on properties owned
Table 2. The data collected for each vegetation remnant surveyed
Remnant”
46a
168
170
171
175
183
185
216
217
220
222
223
269
277
303
313
319
442
443
444
446
449
461
469
S84
S85
S128
S136
s143
s147
S167
s173
Ownerh
Dixon
Beresford
Anderson
Anderson
Gorfin
Barnes
Riley
Ryan
Laird
Reedy
Reedy
Reedy
Dixon
Leake
Moore
Leake
MacNeil
Reed
Ryan
Ryan
Neck
Rolston
Hocking
Neck
Barnes
Barnes
Anderson
Enright
Beresford
Beresford
Ryan
Nelson
Size
(Ha)
1.0
0.6
0.5
1.4
0.6
4.0
3.0
4.0
3.5
2.9
2.8
6.5
2.3
4.0
3.0
3.7
1.8
5.4
0.8
1.0
5.6
3.7
8.0
2.2
0.4
0.8
0.1
0.1
0.5
0.3
0.4
0.1
Distance
to nearest
woodland
(ml
Year(s)
of isolation
100
250
200
1920-10
1920
1912
1912
1930
1936
19:5
1909-20
1925
1920
1910-20
9 1O-20
910-20
92040
1930
9lCk-20
910-20
91620
909-25
909-25
910-20
93040
925-30
910-20
1936
1936
192040
1910-20
1926
1920
1909-25
1910-20
150
1000
350
400
600
0
300
250
300
0
50
1100
900
180
4.50
800
800
800
0
50
550
500
700
350
300
600
400
600
300
Years since
isolation
No.
E. salub.’
No.
E. salmd
Total
Trees
Presence
absence’
Oedura
60
70
78
78
60
54
75
75
65
70
75
75
75
60
60
75
75
75
73
73
75
55
63
75
54
54
60
75
65
70
73
75
177
138
89
140
76
443
304
570
951
279
190
270
291
314
206
307
120
350
68
49
200
520
497
191
6
60
14
8
15
17
13
22
12
0
28
0
10
0
0
0
13
4
7
45
9
159
4
0
88
21
0
0
124
1.50
50
5
8
0
0
0
0
0
3
0
190
138
119
160
108
449
320
570
970
297
219
315
300
473
270
313
208
405
69
52
334
800
606
196
33
60
14
8
40
21
24
22
1
1
1
1
1
1
1
1
1
0
0
1
1
1
0
1
1
1
0
1
1
1
1
1
1
1
0
0
0
1
0
0
Gehyra
1
1
1
1
1
1
1
1
1
0
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
“The number designated for each remnant by CSIRO (M. Brooker, pers. comm.), where prefixed by S there is no CSIRO number.
‘Current owner.
‘Eucalyptus salubris.
dE. salmonophloia.
‘1, present; 0, absent.
30
Stephen
Sarre, Graeme T. Smith, Jacqueline
A. Meyers
Table 3. Summary of the fitted statistical model to estimate the presence of Oedura reticulata in remnant patches of vegetation
The variables are Owner, current owner; Size, area of remnant (ha), Distance, distance from the nearest woodland (m); Age, years
since isolation; E. salubris, the number of E. salubris per remnant; E. salmonophloia, the number of E. salmonophloia per remnant;
Total trees, the total number of trees per remnant; Smooth-barked eucalypts, total number of E. salubris + E. salmonophloia. The
model assumes a binomial distribution for 0. reticulata. The levels of significance (ns, p > 0.05, *0.05 Zp < 0.01; **O.Ol >p > 0.005;
***p <0.005) refer to the change in deviance associated with the term added at each step.
Model
Null model
Owner
Size
Distance
Age
E. salubris
E. salmonophloia
Smooth-barked
Total trees
Smooth-barked
Smooth-barked
Total trees/size
Current model
Smooth-barked
+owner
+size
+distance
+age
+total trees
+total trees/size
+smooth-barked
+smooth-barked
Final model
Smooth-barked
eucalypts
eucalyptsitotal
eucalyptsisize
eucalypts
eucalyptsitotal
eucalypts/size
eucalypts
Deviance
d.f.
38.02
15.28
33.22
36.80
37.83
31.19
Change in
deviance
d.f
P
22.75
4.80
1.22
0.19
6.83
7.13
8.03
7.20
1.50
1.12
0.87
18
1
ns
*
1
30.89
29.99
30.80
36.52
36.90
37.15
31
13
30
30
30
30
30
30
30
30
30
30
1
1
1
1
1
ns
ns
**
**
***
**
1
1
1
ns
ns
ns
29.99
8.29
29.41
29.86
29.68
27.83
29.58
29.95
29.70
30
29
29
29
29
29
29
29
29
21.69
0.58
0.13
0.31
2.16
0.41
0.04
0.29
18
1
1
ns
ns
ns
ns
ns
ns
ns
ns
29.99
30
8.03
by 19 different farmers with no more than three remnants per owner. This kind of ownership
pattern minimises the possibility
that any patterns observed in the
regression
analysis were merely artefacts
of the land
management
practices of individual
owners. The lack
of significance of the variable ‘owner’ in the regression
model supports this view. The smooth-barked
eucalypt
species (E. salubris and E. salmonophloia)
preferred by
0. reticulata
were the most numerous
tree species
(> 64%) in all remnants
except S84 (42%) and S147
(37%) which contained
several mallee form and roughbarked eucalypts.
After the most significant
variable
‘total number of smooth-barked
eucalypts’ was fitted to
the data, no other variable
contributed
significantly
when added to the model. Thus the best explanatory
model took the form
p =
,-04492+0.008275 T/(1+,-04492+0008275T)
(where p = probability
that 0. reticulata is present, T =
total number of smooth-barked
eucalypts
in the remnant) which estimates the probability
that 0. reticulata
will occur in a given remnant (Fig. 4). Figure 4 suggests
that remnants with a total of 411 smooth-barked
eucalypts will have a 95% probability
of containing
0. reticulata after the mean time interval of 69 years, but it
also indicates that the probability
of the presence of the
species drops sharply below this number of trees. This
implies that there is a threshold
amount
of habitat,
1
1
1
1
1
beyond which the population
can be reasonably
certain
of being present. Below this threshold,
the probability
of persistence
drops to where a population
in a remnant containing
55 smooth-barked
eucalypts has only a
50% probability
of persisting for 69 years.
Movement
between
adjacent
populations
The pitfall
trapping
over the summer
of 1990/91
around remnant
170 demonstrated
that, even at a time
of peak activity,
neither
species exhibited
a high
propensity
for dispersal beyond the boundaries
of the
remnant.
In a total of 3,445 pitfall nights between 5
November
1990 and 20 January 1991, a single G. variegata was captured on the inside of the drift fence (i.e.
leaving the remnant).
Neither species was captured in
the pitfalls outside the drift fence although
remnant
170 was only 200 m from the nearest woodland
(a
roadside verge known to contain 0. reticulata) and 700
m from remnant
171 (known to contain both species).
Although this cannot be construed as evidence for zero
movement
between the remnants,
it certainly suggests
that rates of dispersal are low for both species. This
evidence
is supported
by the findings
of Kitchener
et al. (1988) who, over a 3-year period of capturemark-recapture
sampling
in a habitat
remnant,
captured only one unmarked
adult 0. reticulata and 10
unmarked
adult G. variegata that were possible immigrants.
Persistence
of lizards in remnant
habitat
31
0.8
j-
/
0.2
..
0.0
0
200
.
I
I
I
I
400
regression line
~~~~~~~~~
95% confidence
600
Number of smooth-barked
limits
I
I
800
1000
eucalypts
Fig. 4. Logistic regression of the probability that Oedura reticuluta is present in a vegetation remnant and the total number of
trees in the remnant. 0: remnants which contained (probability of presence = 1) or did not contain (probability of presence = 0)
0. reticulata.
DISCUSSION
The survey of woodland
remnants
in the Kellerberrin
region demonstrates
quite clearly that, in terms of persistence, the habitat generalist
G. variegata has coped
considerably
better with the changes wrought by habitat clearance
than the habitat
specialist 0. reticulata.
Although
G. variegata
was almost
certainly
more
widespread
than 0. reticulata before clearing, the remnants that did not contain
0. reticulata were all formerly part of larger regions of woodland that are likely
to have contained
both species. In most cases, remnants that did not contain
0. reticulata were in the
same paddock
or within a few hundred
metres of
woodland
that did contain
the species. Current
absences are thus likely to be due to extinction
rather
than absence at the time of remnant formation.
The significant
correlation
of total
number
of
smooth-barked
eucalypts with the presence of 0. reticulata suggests that habitat quantity
is important
in determining
the likelihood
that a population
of 0.
reticulata will have persisted to the present time. The
high proportion
of absences
of 0. reticulata in the
small remnants
(low tree numbers)
is consistent
with
island biogeographic
theory
(MacArthur
& Wilson,
1967), which predicts that populations
on small ‘islands’ will have a greater probability
of extinction
than
populations
on larger ‘islands’. The underlying
basis
for this prediction
is that small ‘islands’, by implication, will contain
small populations
which are more
vulnerable
to stochastic genetic, demographic
and environmental
extinction
processes (MacArthur
& Wilson,
1967; Shaffer, 1981; Goodman,
1987).
Although
the presence/absence
data for 0. reticulata
are consistent
with minimum
viable population
size
theory, it is important
to make the distinction
between
the effects of habitat per se and the population
level
phenomena
that may be causing the ultimate demise of
the species in local populations.
There are many possible causes of local extinction
besides those related to
stochastic processes but which may also be related to
remnant
area. These include insufficient
resources, increased predation
and greater influence of external factors such as edge effects, where smaller remnants
have
a proportionately
larger edge than large remnants.
Our
data do not discriminate
between these possible causes,
although it is interesting
that the index of habitat size
(total
number
of smooth-barked
eucalypts)
has a
higher statistical
significance
than the correlated
variable of remnant
size. This implies that the limit on
population
size provided by habitat availability
may be
more critical to persistence than actual remnant area.
In contrast
to 0. reticulata, there appears to be no
lower size limit to habitat remnants
in which G. variegata can persist. Two explanations
are possible for this
result.
First,
G. variegata
may have some innate
32
Stephen
Sarre, Graeme
T. Smith, Jacqueline
characteristic
that allows it to persist in small habitat
remnants. This means that the changes to the ecosystem
caused by fragmentation
have not been detrimental
to
G. variegata (e.g. food and shelter have been sufficiently
abundant
and changes in predation
have been insufficient to cause extinction),
but it would also mean that
this species has been able to maintain
populations
of
sufficient size in even very small remnants
to avoid
extinction
through
stochastic
processes. Demographic
extinction
models
developed
by Goodman
(1987),
Kinnaird
and O’Brien (1991) and others suggest that,
when environmental
variation
is included in extinction
models, quite large population
sizes are required for
even moderately
short persistence times. Evidence gathered from several of the remnants included in this study
(Sarre, unpublished
data) suggests that population
sizes
of G. variegata within the remnants are often very small
(sometimes
fewer than 20 individuals)
and well below
the thousands
required by the models to persist for 23
generations.
It is therefore unlikely that these populations are persisting for long in complete isolation.
The second explanation
for the higher persistence
levels of G. variegata is that it is able to move between
the remnants
and consequently
the district population
has a spatial structure that approximates
a metapopulation (a group of interacting
local populations:
Gilpin,
1987; Harrison,
1991) rather than a series of insular
populations.
Such a structure would promote high levels of remnant occupancy through the recolonisation
of
remnants
following
extinction
events or the ‘topping
up’ (Brown & Kodrik-Brown,
1977) of extant populations from adjacent
or nearby populations.
In either
case, the high occupancy
rate exhibited by G. variegata
suggests an equilibrium
rather than a non-equilibrium
metapopulation
structure.
Several characteristics
of G. variegata suggest that
some movement
between remnants
is likely and that it
could
form
a metapopulation
in the Kellerberrin
region. For example, it is a highly mobile species and
exhibits rapid rates of recolonisation
(Moritz,
1987)
and high levels of gene flow over several kilometres
(Moritz,
1992). It is also unlikely
that cleared land
will represent
a complete
barrier to movement
given
that they have been found in highly disturbed
areas
(Kitchener
et al., 1980, 1988; Kitchener,
1982; S. Sarre,
personal
observations).
However,
the evidence
from
pitfalling
(this study) and from the intensive
markrecapture study of Kitchener
et al. (1988) suggests that
rates of movement
between remnants
if they do occur
will be low. Consequently,
any metapopulation
structure is likely to be similar to the type ‘C’ structure
described
by Harrison
(1991) where extinction
and
dispersal rates are in approximate
equilibrium.
In this
case both rates are likely to be low.
In conclusion,
the contrast between the two species is
stark. The habitat generalist (G. variegata) has a high
level of remnant occupancy
and exhibits the characteristics of a metapopulation
at equilibrium.
It is not possible to say definitively
whether this equilibrium
has
been maintained
by high rates of dispersal or low rates
A. Meyers
of extinction, but it is likely that dispersal rates are low,
suggesting that extinction
rates are also low. In contrast, 0. reticulata exhibits comparatively
low rates of
remnant occupancy, suggesting a high rate of extinction
relative to dispersal. If this species forms a metapopulation, then over the 54-78 years since fragmentation,
extinction rates have exceeded recolonisation
rates. It is
likely (given the lack of statistical
significance
of the
variable ‘distance to the nearest woodland’,
anecdotal
evidence from Kitchener et al. (1988) and this study and
estimates of maternal gene flow (S. Sarre, unpublished
data)) that dispersal rates of 0. reticulata approach
zero. The implications
of this are that the proportion
of
remnants
containing
0. reticulata will continue
to decline until only populations
with a very low probability
of extinction will remain. An important
consequence
of
such a pattern of extinctions
is that even if dispersal is
occurring, it will decline with a decline in occupancy because there are fewer source populations
and greater
distances
between the remaining
populations.
Consequently, the contribution
made by dispersal to reducing
the extinction rate is likely to decline with time.
Recent debate surrounding
metapopulation
structure
(Hastings & Wolin, 1989; Bengtsson,
1991; Ebenhard,
199 1; Hanski & Gilpin, 1991; Hansson,
199 1; Harrison,
1991; Kindvall & Ahlen, 1992) has tended to concentrate on metapopulations
that are in equilibrium.
This
is understandable
because equilibrium
systems are particularly amenable
to study questions
relating population size and environmental
variation
to rates of
extinction
and colonisation.
Non-equilibrium
systems
have received scant attention
and yet in both a practical and theoretical
sense an understanding
of them is
clearly critical to the regional conservation
strategies
adopted for many species. This study shows the contrast in persistence between two species, both of which
were probably widespread
and numerous
before European settlement
but show contrasting
patterns in their
ability to cope with the subsequent
rapid and major environmental
changes. The ability to form an equilibrium metapopulation
may be critical to the persistence
of such species in habitat remnants.
ACKNOWLEDGEMENTS
We thank B. Brooker, G. Hanna, A. Heideman, A. Sarre,
G. Sarre, I. Sarre, D. Stevens and J. Weeldenburg
for
their assistance in the field and G. Arnold, M. Brooker,
R. How, D. Saunders
and D. Shaw for their logistic
support and access to unpublished
data. We also thank
A. Georges, C. Margules, A. Nicholls, A. Sarre and D.
Shaw for their comments
on drafts of this manuscript,
A. Nicholls for statistical advice and Tara Goodsell for
producing
the maps. S. Sarre thanks J. Short and C.
Hannan
for their very kind accommodations
during
field work. This project was supported by a CSIRO Institute of Natural
Resources
and Environment
Postgraduate
Project Award and an Australian
Museum
Post-graduate
Research Award. S. Sarre is the recipient
of a Commonwealth
Post-graduate
Research Award.
Persistence
of lizards
REFERENCES
Arnold, G. W. & Weeldenburg,
J. R. (1991). The distributions
and characteristics of remnant native vegetation in parts of
the Kellerberrin,
Tammin,
Tray&g
and Wyalkatchem
Shires of Western Australia. CSIRO, Lyneham.
Beard, J. S. (1980). The vegetation of the Kellerberrin area
Perth.
Western Australia. Vegmap Publications,
Bengtsson, J. (1991). Interspecific competition
increases local
extinction rate in a metapopulation
system. Nature, Lond.,
340,713-15.
Brown, J. H. & Kodric-Brown,
A. (1977). Turnover rates in
insular biogeography:
effect of immigration
on extinction.
Ecology, 58, 44559.
Bustard, H. R. (1968). The ecology of the Australian
gecko,
Gehyra variegata in northern New South Wales. J. Zoo/.,
Lond., 154, 113-38.
Bustard,
H. R. (1969). The population
biology
of the
gekkonid lizard Gehyra variegata (Dumeril & Bibron) in
exploited forests in northern
New South Wales. J. Anim.
Ecol., 38, 3555 1.
Caughley, G. (1994). Directions
in conservation
biology. J.
Anim. Ecol., 63, 2 1544.
Chapman, A. & Dell, J. (1985). Biology and zoogeography
of
the amphibians
and reptiles of the Western
Australian
wheatbelt. Rec. West. Aust. Mus., 12(l), 146.
Cogger, H. G. (1992). Reptiles and amphibians of Australia.
Reed International
Books, Chatswood.
Ebenhard,
T. (1991). Colonization
in metapopulations:
a
review of theory and observations.
Biol. J. Linn. Sot., 42,
105-21.
Gilpin, E. (1987). Spatial structure and population
vulnerability. In Viable populations for conservation, ed. M. E. Soule.
Cambridge University Press, Cambridge, pp. 12540.
Goodman,
D. (1987). The demography
of chance extinction.
In Viable populations for conservation ed. M. E. Soule.
Cambridge University Press, Cambridge, pp. 1 l-34.
Hanski, I. & Gilpin, M. (1991). Metapopulation
dynamics:
brief history and conceptual
domain. Biol. J. Linn. Ser.,
42, 3-16.
Hansson, L. (1991). Dispersal and connectivity
in metapopulations. Biol. J. Linn. Sot., 42, 899103.
Harris, L. D. & Silva-Lopez,
G. (1992). Forest fragmentation
and the conservation
of biological diversity. In Conservation biology. The theory and practice of nature preservation
and management, ed. P. L. Fiedler & S. K. Jain. Chapman
& Hall, New York, pp. 197-237.
Harrison,
S. (1991). Local extinction
in a metapopulation
context: an empirical evaluation.
Biol. J. Linn. Sot., 42,
73-88.
Hastings, A. & Wolin, C. L. (1989) Within-patch
dynamics in
a metapopulation.
Ecology, 70, 126 l-6.
Hobbs, R. J. (1992). The role of corridors
in conservation:
solution or bandwagon?
TREE, 7, 389-92.
How, R. A. & Kitchener,
D. J. (1983). The biology of the
gecko Oedura reticulata Bustard, in a small habitat isolate
in the Western Australian
wheatbelt. Aust. Wildl. Res., 10,
543-56.
Kindvall,
0. & Ahlen, I. (1992). Geometrical
factors and
metapopulation
dynamics of the bush cricket, Metrioptera
bicolor Phihppi (Orthoptera:
Tettigoniidae).
Conserv. Biol.,
6, 520-9.
King,
M.
(1979).
Karyotypic
evolution
in
Gehyra
(Gekkonidae:
Reptilia), I. The Gehyra variegata ~ punctata
complex. Aust. J. Zool., 27, 373-93.
Kinnaird, M. F. & O’Brien, T. G. (1991). Viable populations
for an endangered
forest primate, the Tana River crested
in remnant
habitat
33
mangabey
Cercocebus galeritus galeritus. Conserv. Biol., 5,
203-13.
Kitchener, D. J. (1982). Predictors of vertebrate species richness in nature reserves in the Western Australian
wheatbelt. Aust. Wildl. Res., 9, 1-7.
Kitchener, D. J., Chapman, A., Dell, J. & Muir, B. G. (1980).
Lizard assemblage
and reserve size and structure in the
Western
Australian
wheatbelt
~ some implications
for
conservation.
Biol. Conserv., 17, 25-62.
Kitchener, D. J. & How, R. A. (1982). Lizard species in small
mainland
habitat
isolates and islands off south-western
Western Australia. Aust. Wildl. Res., 9, 357-63.
Kitchener,
D. J., How, R. A. & Dell, J. (1988). Biology of
Oedura reticulata and Gehyra variegata (Gekkonidae)
in an
isolated woodland
of Western Australia. J. Herpetol., 22,
401-12.
Levins, R. (1970). Extinction. In Some mathematical problems
in biology, ed. M. Gerstenhaber.
American
Mathematical
Society, Providence, RI, pp. 777107.
MacArthur,
R. H. & Wilson, E. 0. (1967). The theory of
island biogeography. Princeton University Press, Princeton,
New Jersey.
McArthur, W. M. (in press). Soil landscape map of the Kellerberrin region. Land resource Map, No. 12. Department
of
Agriculture, Perth.
McCullagh, P. & Nelder, J. A. (1989). Generalised linear models. Chapman & Hall, London.
Moritz, C. M. (1986). The population
biology of Gehyra
(Gekkonidae):
chromosome
change and speciation.
Syst.
Zool., 35, 4667.
Moritz,
C. (1987). The population
biology
of Gehyra
(Gekkonidae:
Reptilia), II. Individual movements
and colonization by Gehyra variegata and G. nana. Aust. J. Zool.,
35, 587-96.
Moritz, C. M. (1992). The population
biology of Gehyra
(Gekkonidae),
III. Patterns of microgeographic
variation.
J. Evolut. Biol., 5, 661-76.
Nicholls, A. 0. (1991a). Examples of the use of Generalised
Linear Models in analysis of survey data for conservation
evaluation.
In Nature conservation: cost-effective biological
surveys and data analysis, ed. by C. R. Margules & M. P.
Austin. CSIRO, Canberra, pp. 5463.
Nicholls, A. 0. (19916). An introduction
to statistical modelling using GLIM. In Nature conservation: cost-effective
biological surveys and data analysis, ed. C. R. Margules &
M. P. Austin, CSIRO, Canberra, pp. 191-201.
Nunney, L. & Campbell,
K. A. (1993). Assessing minimum
viable population
size: Demography
meets population
genetics. TREE, 8, 2349.
Opdam, P. (1990). Dispersal in fragmented
populations:
the
key to survival. In Species dispersal in agricultural habitats,
ed. R. G. H. Bunce & D. C. Howard. Belhaven Press, London, pp. 3-17.
Saunders, D. A., Hobbs, R. J. & Margules, C. R. (1991). Biological consequences
of ecosystem fragmentation:
a review.
Conserv. Biol., S(l), 1-15.
Shaffer, M. (1981). Minimum
population
sizes for species
conservation.
Bioscience, 31, 13 14.
Shaffer, M. (1987). Minimum viable populations:
coping with
uncertainty.
In Viable populations for conservation, ed. M.
E. Soule. Cambridge
University
Press, Cambridge,
pp.
69-86.
Soul& M. E. (ed.) (1987). Viable populations for conservation.
Cambridge University Press, Cambridge.
Verboom, J., Lankester, K. & Metz, J. A. J. (1991). Linking
local and regional dynamics in stochastic metapopulation
models. Biol. J. Linn. Sot., 42, 39-55.