Fire and Mammalian Secondary Succession in an Australian Coastal Heath
Author(s): Barry J. Fox
Source: Ecology, Vol. 63, No. 5 (Oct., 1982), pp. 1332-1341
Published by: Ecological Society of America
Stable URL: http://www.jstor.org/stable/1938861
Accessed: 15/02/2009 07:00
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=esa.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the
scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that
promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org.
Ecological Society of America is collaborating with JSTOR to digitize, preserve and extend access to Ecology.
http://www.jstor.org
Ecology, 63(5), 1982, pp. 1332-1341
? 1982 by the Ecological Society of America
FIRE AND MAMMALIAN SECONDARY SUCCESSION IN AN
AUSTRALIAN COASTAL HEATH'
BARRY J. FoX2
School of Biological Sciences, Macquarie University, North Ryde, New South Wales 2113 Australia
Abstract. A species-rich small-mammal community on 7 ha of diverse coastal heath in Myall
Lakes National Park, New South Wales, Australia, was studied for 7 mo before the area burned in
a wildfire (August 1974) and has subsequently been monitored for 5 yr since the fire. Species regularly
trapped were the dasyurids Antechinus stuartii and Sminthopsis marina, the peramelid Isoodon
macrourus, and the native murids Pseudomys novaehollandiae and P. gracilicaudatus, together with
the introduced Mus musculus. Pseudomys species, M. musculus, and S. marina reached greater
abundance on early seral stages, while the abundance there of both Rattus fuscipes and R. lutreolus
was much reduced. Species reach peak abundance in an orderly replacement sequence exhibiting a
mammalian secondary succession. A habitat accommodation model is proposed with species entering
the succession and reaching peak abundance as externally controlled changes in the vegetation fulfil
the habitat requirements of each species. Species leave the succession or are greatly reduced in
abundance when these local physical conditions move out of the optimal range for the species.
This study shows that a major perturbation of a small-mammal community can produce a variety
of responses among species, suggesting that regeneration time may represent a resource axis subdivided
by community members in this secondary succession. The rapid recovery illustrated by the
total number of individuals, as a measure of community response, supports the suggestion that fire
and fire frequency have been major factors in producing a fire-adapted, species-rich small-mammal
community on this heathland.
Key words: Australia; community; fire; habitat accommodation; heath; resource utilization;
small mammal; succession.
INTRODUCTION
In this paper I analyze responses of mammalian
populations and communities to a major perturbation.
The data set was obtained from a community of small
mammals occupying a mosaic of habitats in Australian
coastal heathland that was perturbed by fire. The variety
of responses exhibited by species suggests that
the ability of a species to respond to fire, or more
correctly, to the postfire environment, is part of a response
continuum. Regeneration time then becomes
a resource axis with each species exhibiting a different
resource utilization curve. High response ability or
lack of it then forms part of a species "regeneration
niche," an extension of the concept envisaged by
Grubb (1977). The result is a species replacement sequence
or succession that may facilitate coexistence,
increase species richness, and determine species composition
and community structure. I compare a number
of alternative mechanisms by which fire may affect
community structure and assess these against models
of succession proposed by Connell and Slatyer (1977);
then describe a habitat accommodation model to explain
the observed mammalian secondary succession.
Gill (1975), in a comprehensive review of fire and
the Australian flora, puts forward the idea that plant
species are adapted to a particular fire regime, incorI
Manuscriptreceived 11December1978;revised1August
1981;accepted 10August 1981.
2 Presentaddress:School of Zoology, Universityof New
SouthWales, Kensington,New SouthWales2033Australia.
porating fire intensity, frequency, and season, rather
than the occurrence of fire itself. This seems likely for
small-mammal species as well, given that small-mammal
habitat selection relates to both structural and floristic
components of the vegetation (Fox and Fox
1981). I consider here the idea that changes in smallmammal
abundance following fire might be thought of
as a response to vegetation change, itself responding
to a specific fire regime, rather than a response to the
phenomenon of fire. A review dealing specifically with
the response of heathlands to fire has been completed
recently by Specht (1981). Fire, either as wildfires or
as repeated controlled burns, is an important factor on
ecological as well as evolutionary time scales, particularly
with respect to its ability to effect substantial
habitat change. Such changes can drastically alter the
composition of the mammal community (Beck and
Vogl 1972).
Some of the possible mechanisms by which fire may
alter the structure of mammalian communities and influence
species richness are:
I) Frequent fires can create new habitats equally
available to all species present in the community
and to which they can immigrate;
2) Fire creates a sequence of microhabitats that are
a function of time since fire (plant secondary
succession), and these are selected preferentially
by different mammalian species;
3) Frequent fires can create underutilized habitats to
be occupied by fugitive species that are then replaced
by later mammal species;
October 1982 FIRE AND MAMMALIAN SECONDARY SUCCESSION 1333
A B
M ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~M
all> Woet H8eathc
6gree zeath A
FIG. 1. A: topograpic map of the study area with a 0.5-rn contour interval (arbitrary datum). Trapsites are marked on
each plot. The hatched area on Ml remained unburned. B: the major habitat types in the study area (redrawn from Fox[1981]).
4) Frequent fires can lead to the evolution (or co-evolution)
of "fire specialists" suited to the exploitation
of early seral stages;
5) Later species can enter the succession by displacing
species already present;
6) Early successional species may be able to alter
plant pyric succession (e.g., by grazing pressure)
to delay habitat changes favoring later successional
species, thus prolonging the mammalian succession.
Mechanism 1 would lead to either the tolerance or
inhibition models of Connell and Slatyer (1977), but
would exclude their facilitation model. Mechanisms 2,
3, and 4 would all represent modifications of the facilitation
model. In these cases, the conditions necessary
for later mammalian species are facilitated not
by previous mammal species but by vegetation
changes concomitant with plant secondary succession.
Thus, the cause of the successional change is external
to the mammalian community (Whittaker 1975: 171).
I have called these "habitat facilitation" models.
Mechanism 5 represents a similar modification of the
tolerance model, with the mammalian succession responding
to the external stimulus of vegetation
change, and I refer to this as the "habitat tolerance"
model. Individuals of later successional species enter
the succession in spite of the continued presence of
established populations of early successional species.
Changes in the vegetational succession could make the
habitat less suitable for early species, but might have
little effect on later species. Mechanism 6, although
mediated by vegetational change, is initiated within
the community, so that it represents a less drastic
change to the inhibition model as stated by Connell
and Slatyer (1977). Still-present individuals of early
successional species are able to suppress or exclude
subsequent colonists of all species. Early occupants
modify the environment to be less suitable for both
early and late-succession species.
This study cannot hope to test these mechanisms
completely, given that it is a single descriptive study
dealing with the effects of fire after the event, without
the benefit of replication and experimental treatments.
However, I have erected this framework of possible
mechanisms in an effort to understand this mammalian
succession as a basis for comparison, and perhaps to
stimulate testing of these ideas against other mammalian
secondary successions. By introducing the device
of response to externally induced changes in
vegetation, I have tried to extend Connell and Slatyer's
models to cover this mammalian succession.
The mammalian succession observed in open forest
within 10 km of this study showed close agreement
with these models based on 16 study plots from five
different fire regeneration ages (Fox and McKay 1981),
and this observation encouraged me to attempt this
analysis.
The study area
The study site was located in coastal heathland in
Myall Lakes National Park (latitude 32028'S, longitude
152024'E) north of Sydney, New South Wales, Australia.
The heath supports nine species of small mammals,
and is diverse when compared to other habitats
in the park and to other heathlands on the eastern
Australian coast (cf. Posamentier 1976). The outer
1334 BARRY J. FOX Ecology, Vol. 63, No. 5
TABLE 1. Trappability data for each species calculated as (A) Population trappability (Krebs 1966) using (1) proportion of
marked individuals of each species captured during each 4-day trapping session and averaging over the total number of
sessions that the species was present; (2) average proportion of marked individuals of each species captured each day over
an intensively sampled trapping period. (B) Individual trappability (Hilborn et al. 1976) using the number of days that an
individual was captured divided by the number of days that the individual was exposed to capture (excluding first and last
captures) averaged over all individuals with three or more captures.
Pseu- PseuSmin-
domys domys
Antechinus thopsis Mus novaehol- gracili- Rattus Rattus
stuartii marina musculus landiae caudatus fuscipes lutreolus
(A) Population trappability
Sessions/session Mean 0.81 0.69 0.98 0.82 0.83 0.85 0.80
(Total) Variance 0.13 0.21 0.003 0.09 0.08 0.05 0.02
N 16 18 13 18 31 11 6
Days/day Mean 0.61 0.54 0.90 0.66 0.56 0.89 0.74
(Intensive sampling) Variance 0.15 0.11 0.04 0.08 0.22 0.06 0.08
N 21 17 23 25 24 24 29
(B) Individual trappability
Days/day Mean 0.52 0.43 0.81 0.56 0.67 0.72 0.66
(Total) Variance 0.12 0.07 0.06 0.07 0.05 0.10 0.11
N 7 8 36 38 39 11 15
barrier high dunes are forested, while the low relief
inner barrier dunes support heath. The 4-5 m of relief
provided by these low dunes permit gradation from
swamp to tall dry heath (Osborne and Robertson 1939,
Carolin 1970). Such a gradation occurs on the study
site. Part of a former lagoon, the study site has a low
transverse dune supporting a patch of eucalypt woodland
with a dense, tall, dry heath understorey, thus
including the full range of habitats available. These are
shown in Fig. lB and are described in Fox (1980).
A wildfire in August 1974 swept over the heathland
from the open forest in the north, burning >2000 ha.
The fire was very hot at ground level, leaving only
sand and ash, and burned the canopy of the malleeform
eucalypts. Much of the adjacent open forest on
the high dunes was burned as well, both understory
and canopy. The area has experienced a high fire frequency
for at least the last 100 yr. Over the last 10 yr,
some sections of the nearby open forest have burned
in each of 1979, 1978, 1977, 1976, 1975, 1974, 1971,
and 1968 (Fox and McKay 1981: fire map). The result
has been a mosaic of forest patches with different regeneration
ages and histories. The existence of a high
fire frequency was reinforced when the study site (MI
and part of M2) burned, as part of another extensive
wildfire, in August 1980.
METHODS
In May 1974, a trapping grid with 20-m trap spacing
was established and trapped. The main study plot (MI)
was 7 ha, with a second 4-ha plot (M2) adjacent to it
but separated by a gravel road. M2 was survey trapped
in May 1974 and functioned as an unburned reference
plot after MI burned in August 1974. The purpose of
M2 was to monitor seasonal changes and help distinguish
them from changes on M1 occurring as part of
the regeneration process. Both plots are shown on the
topographic map in Fig. IA.
MI was trapped at =2-mo intervals (1974-1978) with
additional trapping sessions, sometimes at 2-wk intervals,
during breeding seasons. For each trapping session
the minimum number of animals known to be
alive on the study plot during the session (MNA) was
taken to be the simplest measure of abundance (Krebs
1966). M2 was monitored at -3-mo intervals for 16
trapping sessions comprising 5590 trap nights, while
MI was monitored for 34 trapping sessions comprising
16 350 trap nights.
Collapsible aluminum small-mammal traps
(33 x 10 x 9 cm) were set, one at each trapping station,
and baited with a mixture of rolled oats and peanut
butter. Traps were checked in the early morning,
except during winter when traps were also checked at
night to minimize trap mortality. In each session, trapping
was carried out for four successive nights with
no prebaiting period. At the time of first capture, all
animals were identified and given an individual toe
clip, and a numbered metal ear-tag was attached. At
the first capture in a trapping session, body measurements
were taken (head and body, tail, and hind foot).
For every capture the mass of the animal was determined
to the nearest 0.5 g, and the animal was identified
to species and individual. The sex and reproductive
condition were noted and the trap station
recorded for all captures.
RESULTS AND INTERPRETATIONS
To justify the use of the measure '"minimumnumber
alive" (MNA), I have presented trappability data for
all species (Table 1), calculated on both population
and individual bases using the methods of Krebs (1966)
and Hilborn et al. (1976). The only entry in the table
October 1982 FIRE AND MAMMALIAN SECONDARY SUCCESSION 1335
LUJ lYRI 2 YRI 3 YR 4 YRI 5 YR
CZ 5 - c
0 2-c\
0.<1 \~~UNBURNEDk
3 *
2-
BUNED
J M S IJ M S IJ M S IJ M S IJ M S | M S
1974 1975 1976 1977 1978 1979
FIG. 2. Changes with time in the density of animals known to be alive (MNA) at each trapping session. 0 0 indicates
the burned plot (Ml); 0-- -0 the unburned plot (M2). January, May, and September are marked on the time scale.
below the 0.50 trappability recommended by Hilborn
et al. (1976) is for Sminthopsis marina when calculated
for individuals on a day by day basis. However,
when calculated on the same 4-d sessional basis used
to determine MNA values, the appropriate trappability
is 0.69.
A minimum estimate (MNA) for the total number of
animals on each study plot is used to calculate densities
on each study plot for comparison (Fig. 2). Of
74 animals marked in the 3-mo period before the fire,
only 4 were recorded in the first trapping session 1 mo
after the fire. Chew et al. (1959) reported finding large
numbers of carcasses immediately after a wildfire in
chaparral. Animals were probably killed in this fire
too, although no carcasses were found at the first postfire
trapping session.
Both plots undergo a seasonal variation in total
abundance (MNA), peaking in early to midwinter and
reaching a low point in early to midsummer. Seasonal
changes in abundance are most marked in Pseudomys
novaehollandiae, Sminthopsis marina, and Antechinus
stuartii, and apparent local extinctions occur for
parts of the year. The extent to which this may be a
behavioral response, producing reduced trappability,
should not be overlooked. Kemper (1977) reported a
decrease in the number of males in her trappable population
during early summer, but she trapped males
later in the summer that were old enough to have been
in the population earlier. Fletcher (1977) has shown
seasonal changes in trappability for A. stuartii in rain
forest, with up to 18%of the population avoiding traps
completely.
MI shows a marked depression in MNA until mid-
1976, which is superimposed on the seasonal variation
observed in M2 (Fig. 2). I have interpreted the overall
population increase on Ml from late 1974 as a part of
the regeneration process, numbers peaking at the end
of 1977, then decreasing again. The very marked decline
in MNA on M2 during September and October
1974 may have resulted from the influx of predators,
particularly accipiters, observed following the fire on
Ml. Temporal changes in abundance (MNA) of all
species on both plots are presented in Fig. 3.
Postfire recolonization
Four stages can be recognized in the process of recolonization
by a species. First, dispersal from other
areas results in the presence of transient individuals
on burned areas. Later, dispersers are able to establish
residence on burned areas, thus becoming colonists.
A third stage is reached when colonists successfully
reproduce, and the final stage is the increase in abundance
associated with the recruitment of resident juveniles
in addition to dispersers. Different species excel
in different stages, so that they exhibit different
and independent rates of recolonization.
Rattus lutreolus (Fig. 3D), a specialist herbivore,
showed the most dramatic decline in MNA. Of 18 individuals
caught on MI in the 3 mo before the fire,
only 1 was recaptured in the 1st mo postfire. No individuals
were caught in the next 3.5 yr, despite the
presence of potential colonists in nearby unburned
vegetation. Furthermore, there was no successful
breeding until almost 5 yr after the fire. This species,
then, exhibits both poor dispersal and poor colonization
ability.
Rattus fuscipes (Fig. 3C), an omnivore, differed
from R. lutreolus in that individuals were able to colonize
for short periods. Successful breeding on Ml did
not occur until 4 yr after the fire, even though breeding
on M2 was successful in each year. This species, then,
shows good dispersal and colonizing ability, but is limited
by its failure to reproduce successfully until later
seral stages.
Antechinus stuartii (Fig. 3B), an insectivorous marsupial,
although present within 6 mo of the fire, remained
at a low level, only reaching 57% of its prefire
abundance after 3 yr and 70% after 5 yr. Juvenile recruitment
was successful in 1977 and again in 1978.
This species achieves the first three stages of recolo-
1336 BARRY J. FOX Ecology, Vol. 63, No. 5
J]. -*-A- -
1
-_
- .
101C
>, 25
51
15E
cc
YR YR YR 15- YR YR1
O10-
5-z
z, 15- F
lo-~~~~~~~~~~~~~-
5-0A r M . < . . , . . M S I J M
15-G~~~~~~~~~~~~~~~~~-
10-
U
5J
M S J M S 'J M S J M S J M S J M
1974 1975 1976 1977 1978 1979
FIG.3. Changeswith time in numbers(MNA) for each
species on each plot at each trappingsession. As MI is 7 ha
and M2is 4 ha, comparisonsshouldbe for relativechanges
ratherthanabsolutenumbers.(A) Sminthopsismarina,(B)
Antechinus stuartii, (C) Rattus fuscipes, (D) R. lutreolus,
(E) Mus musculus, (F) Pseudomys novaehollandiae, (G) P.
gracilicaudatus.Differentsymbol shapes indicatedifferent
taxonomicgroups;open vs. closed symbolsas in Fig. 2.
nization but is now limited by a slow increase in abun-
dance.
Sminthopsis marina (Fig. 3A), another insectivorous
marsupial, showed a marked increase in late 1976
after an unsuccessful initial attempt at recolonization
on Ml in 1975. Breeding was successful in 1976. This
species exhibits successful recolonization at each
stage.
Mus musculus (Fig. 3E), an introduced omnivore,
colonized M1 and began successful breeding in less
than a year, increased sufficiently in abundance in
mid-1975 to cause dispersal to M2 but disappeared
from both MI and M2 by the end of the year. Further
recolonization occurred in late 1976, causing a population
explosion in late 1977 followed by a precipitous
decline in early 1978. This species has excellent dispersal
(high numbers of transient individuals) and
short-term colonizing ability combined with an equally
high rate of reproduction and ability to increase. However,
populations are not stable. A similar irruption in
Mus numbers was noted during 1977 on recently
+8 -
6/76
+6-
0
+4_
Lu
+2_
LLJ ~~~~~~~9/7
o 0 - -2- /76 \ -2
6/77
ZI
-4
I I i I I I I
-6 -4 -2 0 +2 +4 +6 +8 +10
RELATIVE NUMBERS (Ml -7/4 M2)
FIG.4. ForPseudomysnovaehollandiae,therelationbetween
net migrationinto M1 in each quarterandthe relative
populationnumbers(Ml - 1.75M2),with accounttakenof
the smallerareaof M2 (i.e., for equivalentpopulations,abscissa
value is zero). Dates are midpointsof quarterswhen
dispersalwas recorded.
burned plots in nearby open forest (Fox and McKay
1981). The very large fluctuations in numbers of Mus
perhaps indicate the behavior of a fugitive species responding
to change, rather than that of a fire specialist.
Laurie (1946) reports that Mus has the ability to breed
all year round.
Pseudomys novaehollandiae (Fig. 3F), a generalist
granivore, benefited from very high dispersal. M2 provided
colonizers for MI in early 1976, but in turn received
animals in mid-1977 when animals were abundant
on Ml but in low numbers on M2. This is
illustrated by Fig. 4, which shows that not only the
direction but also the magnitude of dispersal are functions
of the relative populations on the two plots. This
species exhibits good performance at all stages and
shows longer occupation periods for individuals than
does M. musculus.
Pseudomys gracilicaudatus (Fig. 3G), a generalist
herbivore, had high juvenile survival producing good
recruitment. Good colonizing ability made up for a
lower dispersal rate. Abundance reached prefire levels
in 1 yr and increased sixfold within 2 yr, largely due
to high recruitment on Ml. Numbers did not increase
beyond the mid-1976 peak and slowly decreased during
1978 and 1979, even though successful breeding
occurred. These factors support the characterization
of P. gracilicaudatus as a fire specialist.
Spatial niche parameters calculated for species in
this community show significant rejection of null hypotheses
of noninteraction between species in relation
to their use of space (Fox 1981). These results are
consistent with circumstantial evidence for interaction
between congeners and pairs of similar species in this
community (Fox 1980). The strongest evidence of interaction
was the relationship between native species
and the introduced Mtus. Newsome and Corbett (1975)
October 1982 FIRE AND MAMMALIAN SECONDARY SUCCESSION 1337
TABLE 2. Yearly population maxima for each species over a 5-yr period on the unburned plot tested against a null hypothesis
of no significant difference between years. Kolmogorov-Smirnov tests (Siegel 1956) were used, and values for the statistic
D and significance levels are included together with the total number of individuals for each species as the degrees of
freedom.
Yearly population maximum (unburned plot)
1974 1975 1976 1977 1978 E D P
Pseudomys gracilicaudatus 0 1 0 4 2 7 .46 .1 > P > .05
Mus musculus 6 4 0 3 8 21 .18 > .2
Pseudomys novaehollandiae 8 3 14 7 5 37 .10 > .2
Sminthopsis marina 1 1 2 4 1 9 .18 > .2
Antechinus stuartii 0 1 2 2 0 5 .20 > .2
Rattus fuscipes 1 11 5 7 1 25 .16 > .2
Rattus lutreolus 4 6 4 4 3 21 .08 > .2
report that for central Australia, Mus displaces native
species. However, the observed pattern here is consistent
with Mus reacting to changes in abundance of
native species. The two dasyurids A. stuartii and S.
marina also exhibit a marked ecological separation
(Fox 1982).
Significance of changes in abundance
I use the criterion that the peak abundance for each
species (MNA) during prefire trapping sessions represents
the best estimate of peak abundance for that
species in any year, had the fire not occurred. To test
the appropriateness of this as an expected value, I
tested the unburned plot (M2) against the null hypothesis
of no significant differences in abundance between
years (i.e., a uniform cumulative frequency distribution
with five classes: 0.2, 0.4, 0.6, 0.8, 1.0), using the
Kolmogorov-Smirnov test for goodness of fit (Siegel
1956:47, Sokal and Rohlf 1969:571). Yearly maxima
(1974-1978) for each species on M2 are shown in Table
2, together with the test statistic (D) and its probability
of occurrence by chance alone. The null hypotheses
were accepted (P > .05): a uniform distribution is an
appropriate method for determining expected values.
Yearly maxima for the burned plot (MI) as shown in
Table 3 were then tested over the same period (1974-
1978). The zero values for R. lutreolus meant that population
data for that species had to be tested with the
Kolmogorov-Smirnov test also, but the more sensitive
G-test (log likelihood ratio; Sokal and Rohlf 1969:559)
was used for all other species (Table 3). All species
show significant changes from their 1974 maxima, so
that differences represent real changes in abundance.
Response indices
To quantify these differences for each species, I calculated
a fire response index for each year as the logarithm
of the ratio of the maximum value of MNA in
each year since the fire, to the maximum MNA before
the fire. The behavior of this index for each species is
shown in Fig. 5. Positive values represent increases
on prefire levels, and negative values decreases. A
total index was calculated from the logarithm of the
ratio of the sum of the five postfire years to five times
the prefire value. This overall fire response index was
used to evaluate the effect of fire on each species.
The classification of species using a fire response
index is in agreement with much of what limited information
is available in the literature. Quantifying the
response allows comparisons between species in this
study, and, I hope, between studies. The community
subdivides into four groups: one badly affected by the
fire (<-0.4, both Rattus species); one less badly affected
(-0.4 to -0.2, A. stuartii); a third group able
to maintain populations in fire-perturbed areas (-0.2
to 0.2, P. novaehollandiae, S. marina, and M. musculus);
and the last showing a positive overall response
to the perturbation (>0.2, P. gracilicaudatus). The
TABLE 3. Yearly population maxima for each species over a 5-yr period on the burned plot with results of a log likelihood
ratio test (Sokal and Rohlf 1969) for departure from a uniform distribution equal to the prefire (1974) value. For R. lutreolus
a Kolmogorov-Smirnov test (see Table 2) was used.
Yearly population maximum (burned plot)
1974 1975 1976 1977 1978 G P
Pseudomys gracilicaudatus 3 4 17 14 18 +169 < .001
MuS musculus 10 9 1 12 25 +44 < .001
Pseudomys novaehollandiae 10 4 16 11 13 +16.6 < .001
Sminthopsis marina 4 2 3 6 7 +8.2 < .05
Antechinus stuartii 7 2 2 4 3 - 19.6 < .001
Rattus fuscipes 7 1 2 3 2 -19.0 < .001
Rattus lutreolus 11 2 0 0 1 D = 0.52 < .001
1338 BARRY J. FOX Ecology, Vol. 63, No. 5
.8
[ ~ ~ ~, , I | oaI
*0
1974 197U 197601977 1978 1979 (7549)
10- ~ ~ ~ "*
Cn,
T I M E
FIG. 5. Fire response index (logarithm of ratio of maximum value of MNA in each year to prefire [1974] maximum MNA)
for each species as a function of time. Solid symbols show dry-habitat species, and open symbols show wet-habitat species.
division is based on each species' performance (see
Fig. 5). Antechinus stuartii can also be separated from
the two Rattus species because of its steady increase,
as opposed to their delayed recovery.
Succession
A gradient in postfire responses of species is clearly
demonstrated in Fig. 5. This is also reflected in the
time required for each species to reach its peak abundance
(Fig. 6). The sequences of species in wet heath
and dry heath are shown separately. Mus musculus
occurred predominantly in dry heath in 1975, but predominantly
in wet heath in 1977-1978, and is shown
in both habitats. Pseudomys occupies the early seral
stages, and Rattus clearly occupies the later seral
stages, with the dasyurids occupying mid- and late seral
stages in this succession. Newsome et al. (1975)
found that A. stuartii survived, with fluctuations, on
open forest plots, but not on heath plots. Christensen
and Kimber (1975) report a much longer period of 20
yr for a similar species, A. flavipes, to reach prefire
levels in Western Australia. No information is available
for S. marina apart from a study in a nearby
forest by Fox and McKay (1981) which shows it to
achieve maximum densities in forests whose understory
was burnt 2-6 yr previously.
The order and timing of this replacement sequence
is very similar to that observed on simultaneously
trapped, replicated plots of different regeneration ages
in a nearby open forest (Fox and McKay 1981). Early
and mid-seral stage species appear to increase and
then decrease, peaking when their particular habitat
requirements are best met. A significant body of
information is now emerging that identifies four species
of Pseudomys as fire- or disturbance-enhanced
species: P. novaehollandiae (Posamentier and Recher
1974, Posamentier 1976, Kemper 1977, Braithwaite
and Gullan 1978, Fox and Fox 1978, Fox and McKay
1981, and this study), P. shortridgei (Cockburn 1978),
P. albocinereus (reported as a personal communication
from C. R. Crouch in Cockburn 1978), and P.
gracilicaudatus (this study).
The community composition changes drastically as
a function of regeneration time. Beck and Vogl (1972)
were able to convert an oak forest to brush-prairie
savanna by increasing the fire frequency, and reported
concomitant changes in small-mammal species composition.
Cook (1959), however, shows changes in the
abundance of species in brush and grassland habitats,
rather than species replacement. Wetzel (1958) reports
a mammalian succession following strip-mining on a
Midwestern floodplain (USA), where a species replacement
sequence was clearly demonstrated. Pearson
(1959) has shown a similar mammal species replacement
during old-field succession. Christian (1977)
suggests that the rapid return of small mammals to a
recently burned desert grassland is attributable to the
fact that the normal ranges of the species encompass
October 1982 FIRE AND MAMMALIAN SECONDARY SUCCESSION 1339
1.0 FJ *. A
.8 Wet ~g*
LU Ha b i ta t .
z .6 Sp e c ies
;/O.' ~ \-0~~~~~~~~~~~~~~~~~~~
z
m .2 AA
1.0 - B
D ry 0.,>
.8 Habita t
Spei / / s 6
< .6 dje~
.2
0 -~~~>~~
0 12 3 4 5 6
T I M E (Y R)
FIG. 6. The succession of mammal species responding to changes in vegetation regenerating after fire. A: wet-habitat
species, B: dry-habitat species. Values for each species are shown as a proportion of that species' maximum abundance.
Prefire abundances are shown to the left of the abscissa scale and again at 6 yr, the estimated age of the site since the last
fire (hence the question mark).
natural areas equally as harsh. This is probably true
for Mus and perhaps Pseudomys as well. Conversely,
it may also be a reason why the two Rattus spp. appeared
so late.
The ground cover may not have been thick enough
to support R. lutreolus recolonization; Posamentier
(1976) has shown that R. lutreolus requires a dense
ground cover. Layne (1974) found that the reappearance
of harvest mice and cotton rats after fire in a
Florida flatwoods was correlated with the redevelopment
of ground cover. For R. fuscipes, Christensen
and Kimber (1975) report similar results up to 3 yr
after fire on their shrub-understory plot. Although they
had better colonization by males on their sword-grass
plot, females were largely transient, thereby precluding
successful breeding.
Newsome et al. (1975) indicate similar results on
small grids, but on their larger grids in open forest, R.
fuscipes survived and then increased in the 2nd yr
after fire. They commented on the patchy nature of
the forest which probably accounted for the R. fuscipes
survival after the fire. Cowley et al. (1969) and
Leonard (1970) report high R. fuscipes survival after
control burns which left 30% and 50%, respectively,
of the area unburnt. Leonard (1970) shows R. fuscipes
to have overwhelming preference for dense unburnt
areas on his plot. Fox and McKay (1981) found no R.
fuscipes colonists until 3 yr postfire. A period of increase
in R. fuscipes biomass followed until 5 yr postfire
when it began to plateau.
Comparison of succession models
In comparing the mechanisms by which fire may
affect community composition, I draw upon the information
from this study, together with relevant published
work to which I have referred. Mechanism 1
does not describe the situation observed, which suggests
that the vegetation present early in the succession
is entirely unsuitable for the late seral stage
Rattus. The presence of Rattus in nearby unburnt
vegetation throughout the study supports rejection of
this mechanism. Rejection of mechanism 1implies that
the tolerance and inhibition models of Connell and
Slatyer (1977) cannot be considered as suitable descriptions.
Mechanism 2 does appear to operate in
general and is exemplified by the orderly way in which
species reach their peak abundance (Fig. 6). Mechanism
3 is more specific and would appear to describe
the response of the fugitive M. musculus that is then
displaced by the native P. novaehollandiae, whose
response is better described by mechanism 5. The response
of the fire specialist P. gracilicaudatus is best
described by mechanism 4. There is no evidence that
any species has sufficient effect on the vegetation to
1340 BARRY J. FOX Ecology, Vol. 63, No. 5
satisfy mechanism 6. For both Rattus species information
is not complete, but they would appear to be
best described by mechanism 5.
Mechanism 2 is consistent with the habitat facilitation
model. However, the appropriateness of mechanisms
3, 4, and 5 for different species contravenes part
of this model. A new model is needed, one able to
encompass all small-mammal species in this secondary
succession. The one I propose incorporates elements
of both the habitat facilitation and tolerance models,
and I have termed it the habitat accommodation mod-
el.
Habitat accommodation model
In this model, species do not modify local physical
conditions. These conditions are changing in response
to external factors. Species are able to enter the
succession when the changing local physical conditions
(habitat) first meet their specific requirements
(habitat facilitation model), but as these conditions
move out of the optimal range for the species, the
animals leave the succession or are reduced in numbers
(habitat tolerance model). This may result from
the species no longer being able to obtain an adequate
share of resources as their competitive ability is re-
duced.
The habitat accommodation model accounts for all
species in this succession and would also appear to
describe the postfire mammalian secondary succession
observed in physiognomically similar California chaparral
(Wirtz 1977, R. D. Quinn, personal communi-
cation).
An example of another system that may be described
by a habitat accommodation model is the
succession observed on a salt marsh (Randerson
1979). Accretion of sediment is the external causal factor,
and although plant presence may accelerate accretion,
Randerson (1979) has shown that the plants
do not themselves cause the environmental changes.
Plant species enter the succession and reach peak
abundance when conditions are optimal for them, then
are greatly reduced in abundance as marsh development
continues. A similar externally induced succession
is observed in the appearance of the reproductive
structures of coprophilous fungi growing on herbivore
dung (Hudson 1972).
CONCLUSIONS
This study has shown that a major perturbation of
a small-mammal community can produce a variety of
responses among the species comprising the community.
These responses can be represented by a response
index that suggests a long-term resource axis
(regeneration time) partitioned among species. The
rapid recovery illustrated by total number of individuals
indicates that fire and fire frequency have been
important factors in the development of the heath
small-mammal community. This has been supported
by the marked increase in importance of the native
murid Pseudomys gracilicaudatus in the postfire community,
where it increased sixfold during the 1st 3 yr.
The habitat accommodation model, or a new model of
another form, is necessary to describe adequately the
range of responses, which represent a mammalian secondary
succession or species replacement series that
responds to the vegetation changes occurring as part
of the plant pyric succession. The need to maintain a
mosaic of regeneration ages for management and
maintenance of a diverse small-mammal community
(Pickett and Thompson 1978) has been demonstrated.
ACKNOWLEDGMENTS
This workhas been supportedby a CommonwealthPostgraduateResearchAwardto
theauthorandinpartby a grant
from the Departmentof Environment,Housing, and CommunityAffairs(Number76/328)to
Dr.G. M. McKay.Ithank
Dr. McKayfor his supportof this projectandforreadingan
earlierdraft of this manuscript.MarilynFox providedinvaluableassistance
in the field and providedmuch critical
commentand advice duringthe courseof the study, particularlyduringpreparationof
this manuscript.Three anonymous
referees providedmuch useful criticismthat was instrumentalin
greatlyimprovingthe manuscript.I wouldalso
like to thank Dr. Paul Adam for bringingthe salt marsh
successionto my notice andthe Schoolof Zoology,University
of New SouthWalesfor allowingme to use theirSmith
LakeFieldStation.AnnLonergankindlytypedthe draftsof
the manuscript.
LITERATURE CITED
Beck, A. M., and R. J. Vogl. 1972. The effects of spring
burning on rodent populations in a brush-prairie savanna.
Journal of Mammalogy 53:336-346.
Braithwaite, R. W., and P. Gullan. 1978. Habitat selection
by small mammals in a Victorian heathland. Australian
Journal of Ecology 3:109-127.
Carolin, R. C. 1970. Myall Lakes-an ancient and modern
monument. Proceedings of the Ecological Society of Australia
5:123-129.
Chew, R. M., B. B. Butterworth, and R. Grechman. 1959.
Effects of fire on the small mammal populations of chapparal.
Journal of Mammalogy 40:253.
Christensen, P. E., and P. C. Kimber. 1975. Effect of prescribed
burning on the flora and fauna of S. W. Australian
forests. Proceedings of the Ecological Society of Australia
9:85-106.
Christian, D. P. 1977. Effects of fire on small mammal populations
in a desert grassland. Journal of Mammalogy
58:423-427.
Cockburn, A. 1978. The distribution of Pseudomys shortridgei
(Muridae: Rodentia) and its relevance to the other
heathland Pseudomys. Australian Wildlife Research 5:213-
220.
Connell, J. H., and R. 0. Slatyer. 1977. Mechanisms of
succession in natural communities and their role in community
stability and organization. American Naturalist
111:1119-1144.
Cook, S. R., Jr. 1959. The effects of fire on a population of
small rodents. Ecology 40:102-108.
Cowley, R. D., A. Heislers, and E. H. M. Ealey. 1969. Effects
of fire on wildlife. Victoria's Resources 11:18-22.
Fletcher, H. L. 1977. Habitat relationships among small
mammals at Petroi, north-eastern New South Wales. Thesis.
University of New England, Armidale, New South
Wales, Australia.
October 1982 FIRE AND MAMMALIAN SECONDARY SUCCESSION 1341
Fox, B. J. 1980. The ecology of a small mammal community:
secondary succession, niche dynamics, habitat partitioning,
community structure and species diversity. Dissertation.
Macquarie University, North Ryde, New South
Wales, Australia.
. 1981. Niche parameters and species richness. Ecology
62:1415-1425.
1982. Ecological separation and coexistence in
Sminthopsis marina and Antechinus stuartii : a regeneration
niche? Pages 187-197 in M. Archer, editor. Carnivorous
marsupials. Royal Zoological Society of New South
Wales, Sydney, Australia.
Fox, B. J., and M. D. Fox. 1978. Recolonisation of coastal
heath by Pseudomys novaehollandiae (Rodentia : Muridae)
following sand mining. Australian Journal of Ecology
3:447-465.
Fox, B. J., and M. D. Fox. 1981. A comparison of vegetation
classifications as descriptors of small mammal habitat preference.
Pages 166-180 in A. N. Gillison and D. J. Anderson,
editors. Vegetation classification in Australia. Commonwealth
Scientific and Industrial Research Organization
and Australian National University Press, Canberra, Aus-
tralia.
Fox, B. J., and G. M. McKay. 1981. Small mammal responses
to pyric successional change in a eucalypt forest.
Australian Journal of Ecology 6:29-42.
Gill, A. M. 1975. Fire and the Australian flora: a review.
Australian Forestry 38:4-25.
Grubb, P. J. 1977. The maintenance of species-richness in
plant communities: the importance of the regeneration
niche. Biological Reviews 52:107-145.
Hilborn, R., J. A. Redfield, and C. J. Krebs. 1976. On the
reliability of enumeration mark and recapture census of
voles. Canadian Journal of Zoology 54:1019-1024.
Hudson, H. J. 1972. Fungal saprophytism. Arnold, London,
England.
Kemper, C. M. 1977. Biology of the New Holland mouse.
Dissertation. Macquarie University, North Ryde, New
South Wales, Australia.
Krebs, C. J. 1966. Demographic changes in fluctuating populations
of Microtus californicus. Ecological Monographs
36:239-273.
Laurie, E. M. 0. 1946. The reproduction of the house mouse
(Mus musculus) living in different environments. Proceedings
of the Royal Society, Series B 133:248-282.
Layne, J. M. 1974. Ecology of small mammals in a flatwood
habitat in North Central Florida. American Museum Novitates
2544:1-48.
Leonard, B. 1970. Effects of control burning on the ecology
of small mammal population. Pages 1-14 in Symposium
Papers, Second Fire Ecology Symposium. Forest Commission,
Victoria, Australia.
Newsome, A. E., and L. K. Corbett. 1975. Outbreaks of
rodents in semi-arid and arid Australia: causes, prevention
and evolutionary considerations. Pages 117-153 in I. Prakash
and P. G. Ghosh, editors. Rodents in desert environments.
W. Junk, The Hague, The Netherlands.
Newsome, A. E., J. McIlroy, and P. Catling. 1975. The effects
of an extensive wildfire on populations of 20 ground
vertebrates in S.E. Australia. Proceedings of the Ecological
Society of Australia 9:107-123.
Osborne, T. G., and R. N. Robertson. 1939. A reconnaissance
survey of the vegetation of the Myall Lakes. Proceedings
of the Linnaean Society of New South Wales
64:279-296.
Pearson, P. G. 1959. Small mammals and old field succession
on piedmont of New Jersey. Ecology 40:249-255.
Pickett, S. T. A., and J. N. Thompson. 1978. Patch dynamics
and the design of nature reserves. Biological Conservation
13:27-37.
Posamentier, H. G. 1976. Habitat requirements of small
mammals in coastal heathlands of N.S.W. Thesis. University
of Sydney, Sydney, Australia.
Posamentier, H. G., and H. F. Recher. 1974. Status of Pseudomys
novaehollandiae. Australian Zoologist 18:66-71.
Randerson, P. F. 1979. A simulation model of salt-marsh
development and plant ecology. Pages 48-67 in B. Knights
and A. J. Phillips, editors. Estuarine and coastal land reclamation
and water storage. Saxon House, Farnborough,
England.
Siegel, S. 1956. Nonparametric statistics for the behavioural
sciences. McGraw-Hill, Tokyo, Japan.
Sokal, R. R., and F. J. Rohlf. 1969. Biometry. W. H. Freeman,
San Francisco, California, USA.
Specht, R. L. 1981. Responses of selected ecosystems,
heathlands and related shrublands. Pages 395-415 in A. M.
Gill, R. H. Groves, and I. Noble, editors. Fire and the
Australian biota. Australian Academy of Science, Canberra,
Australia.
Wetzel, R. 1958. Mammalian succession in midwestern
floodplains. Ecology 39:262-271.
Whittaker, R. H. 1975. Communities and ecosystems, Second
edition. MacMillan, New York, New York, USA.
Wirtz, O. W., II. 1977. Vertebrate post-fire succession.
Pages 46-57 in Symposium Papers, Second Fire Ecology
Symposium on the Environmental Consequences of Fire
and Fuel Management in Mediterranean Ecosystems. General
Technical Report WO-3, United States Forest Service,
Washington, DC, USA.