Ecological Applications, 18(8), 2008, pp. 1967–1983
? 2008 by the Ecological Society of America
TESTING HYPOTHESES ASSOCIATED WITH BIRD RESPONSES
DAVID B. LINDENMAYER,1,3JEFF T. WOOD,1,2ROSS B. CUNNINGHAM,1CHRISTOPHER MACGREGOR,1MASON CRANE,1
DAMIAN MICHAEL,1REBECCA MONTAGUE-DRAKE,1DARREN BROWN,1RACHEL MUNTZ,1AND A. MALCOLM GILL1
1Fenner School of Environment and Society, Australian National University, Canberra, ACT 0200 Australia
2Statistical Consulting Unit, Australian National University, Canberra, ACT 0200 Australia
abundance of many elements of the earth’s biota. Predicting the response of biota to
disturbance is therefore important, but it nevertheless remains difficult to make accurate
forecasts of response. We tested predictions from disturbance-related theories and concepts in
10 vegetation types at Booderee National Park (southeastern Australia) using a retrospective
study of bird responses to fire history (over 35 years) on 110 sites and a prospective study
following a single wildfire event in 2003 at 59 of these sites. Our data did not support
predictions from the intermediate-disturbance hypothesis; observed bird species richness at a
site was significantly (F1,99¼ 6.30, P ¼ 0.014) negatively related to the number of fires since
1972 and was 8.7% lower (95% CI, 1.8–15.1%) for each additional fire. In contrast to fire
history effects, we found that after the 2003 fire, the vast majority of individual species and the
bird assemblage per se in most vegetation types recovered within two years. Thus, recovery
after a single fire did not reflect long-term effects of multiple fires on overall bird species
richness at a site. We postulated that the recovery of bird species richness and bird assemblage
composition after the 2003 fire would be fastest in structurally simple vegetation types and
slowest in structurally complex vegetation, but observed the opposite. Although observed bird
species richness in vertically heterogeneous forest and woodland had returned to prefire levels
by 2006, bird species richness in structurally simple vegetation types (e.g., sedgeland) had not.
Postfire vegetation regeneration, together with a paucity of early-successional specialists,
would explain the speed of recovery of the bird assemblage and why it changed relatively little
during our investigation.
Disturbance is a key ecological process influencing the distribution and
sedgeland; southeastern Australia; vegetation management; wildfire event.
avifauna; biological legacies; bird community; disturbance; fire history; fire severity;
Fire is a key ecological process in ecosystems around
the world (Agee 1993, Bradstock et al. 2002, Andersen
et al. 2003, Burton et al. 2003). However, in many
landscapes, the sequence of fires or fire history (sensu
Gill 1975) has changed in the past few centuries
(Covington 2003, Andersen et al. 2005) as a result of
such factors as active fire suppression (Zackrisson 1977,
Spies et al. 2004), logging (Thompson et al. 2007), urban
encroachment on natural areas (Cary et al. 2003), and
possibly climate change (Lenihan et al. 2003, Pittock
2005). Such changes are believed to be having negative
impacts on biodiversity (Woinarski and Recher 1997,
Schurbon and Fauth 2003, Noss et al. 2006). For
example, altered fire history is thought to be threatening
biota in Australia: more than 50 species of birds are at
risk and only land clearing threatens a greater number of
species (Woinarski 1999, Garnett and Crowley 2000).
Given this, a better understanding of the effects of fire
on biodiversity is vital to designing management
strategies that best manage areas of native vegetation
(Richards et al. 1999, Bradstock et al. 2005).
Many authors (e.g., Keith et al. 2002, Parr and
Andersen 2006) have highlighted the lack of attempts to
monitor and predict the impacts of the sequence of fires
on biodiversity. However, this can be complex because,
for example, (1) different fires can burn at different
severities in the same vegetation type (Smucker et al.
2005, Kotliar et al. 2007), giving rise to considerable
local- and landscape-scale patchiness (DeLong and
Kessler 2000, Mackey et al. 2002, Turner et al. 2003,
Kulakowski and Veblen 2007); (2) fires can burn
differently in different vegetation types (Agee 1993,
Andersen et al. 2005); and (3) differences in the timing of
a fire or series of fires can produce markedly different
impacts on biota (Bradstock et al. 2002, Thompson et al.
2007). Therefore, a research and management challenge
remains to identify ecologically appropriate and socially
acceptable fire management practices for different
vegetation types and their associated biota (Gill et al.
1999, Parr and Chown 2003, Spies et al. 2004, Haynes
et al. 2006).
Manuscript received 27 November 2007; revised 17 March
2008; accepted 25 March 2008. Corresponding Editor: J. M.
In this paper, we describe a study of bird responses to
fire in 10 vegetation types at Booderee National Park
within the Jervis Bay Territory in southeastern Austral-
ia. We quantified fire effects on birds through a
retrospective study of a series of fires over time (‘‘fire
history’’) and a prospective investigation of birds
following a single conflagration (a ‘‘fire event’’) in
December 2003. We explored fire effects at the
population, bird assemblage (community), and species
(individual species and all species combined) levels. We
tested predictions from theories and concepts directly or
indirectly linked with the impacts of natural disturbanc-
es on biodiversity.
First, we quantified relationships between fire history
and bird species richness and compared our findings
with predictions from the intermediate-disturbance
hypothesis. Under this hypothesis, species richness is
predicted to be highest on sites subject to intermediate
values for the number of fires and time since fire rates.
This is because they will tend to have a mixture of early-
successional and late-successional species (Connell
1978, Shiel and Burslem 2003). Since 1972, the number
of fires in any given part of Booderee National Park has
varied from none to five and the time since the last fire
ranges from four to 35 years. Given this information
and based on the intermediate-disturbance hypothesis,
we tested the prediction that bird species richness
should be highest in areas that have experienced 2–3
past fires and areas where ;12–16 years have elapsed
since the last fire.
Second, we examined relationships between vegeta-
tion structure, species richness, and bird species recovery
after a single major wildfire in 2003. Fire burns
differently in different vegetation types (Agee 1993,
Whelan 1995) and the impacts of fire on bird biota may
be mediated through temporal recovery patterns of
vegetation structure and composition. The vertical
vegetation structure hypothesis (MacArthur and Mac-
Arthur 1961, Recher 1969) predicts that bird diversity
will be highest in vertically complex vegetation types
such as forest and woodland and lowest in structurally
simple environments (e.g., sedgeland). However, the
recovery of vegetation structure may take longer in
structurally complex vegetation than where vertical
vegetation structure is simple. We tested the prediction
that the recovery of bird species richness and bird
assemblage composition after the 2003 fire would be
fastest in structurally simple vegetation types and
slowest in structurally complex vegetation characterized
by high levels of vertical heterogeneity (sensu Brokaw
and Lent 1999).
Third, we examined relationships between biological
legacies and bird persistence and recovery following the
2003 fire. Biological legacies are defined by Franklin
et al. (2000) as: ‘‘...organisms, organically-derived
structures, and organically-produced patterns that survive
from the pre-disturbance system.’’ They can influence the
type and pace of recovery of biota after disturbance
(Foster et al. 1998, Turner et al. 2003). The more of the
pre-disturbance stand structure that persists after fire,
the faster plant and animal populations should return to
prefire levels (Franklin et al. 2000, Whelan et al. 2002).
We tested the prediction that more species of birds
would persist and/or recover fastest on sites with high
levels of biological legacies remaining after the 2003 fire.
Fourth, we quantified relationships between bird
response to the 2003 fire and bird life history attributes.
Several studies have identified predictable relationships
between disturbance and animal life history attributes or
functional groups (e.g., Hansen and Urban 1992, Loyn
1993). These include investigations of fire (e.g., Raphael
et al. 1987, Imbeau et al. 1999, Saab et al. 2007). We
sought to determine if there were subsets of species from
the bird assemblage that responded in a similar way to
the 2003 fire and that had sets of life history attributes in
common. Finally, we examined the effects of the 2003
fire at Booderee National Park on a number of
Our overarching aim in completing this study was to
improve understanding of biodiversity responses to both
fire history and a single fire event. Such new knowledge
is important, given that several recent studies have
predicted that fire size, fire frequency, and/or fire
severity will increase in the near future as a consequence
of climate change (Cary 2002, Lenihan et al. 2003,
Pittock 2005, Westerling et al. 2006). An understanding
of the response of biodiversity to individual fire events
and sequences of fires will be crucial to better inform
appropriately targeted management practices for bird
communities in a wide range of ecosystems where fire is
an important ecological process. For example, managers
will seek knowledge on which species, vegetation types,
and fire patterns to monitor. There is considerable
precedent for targeted management of fire and bird
communities in many coastal ecosystems in eastern
Australia, particularly those that support threatened
plant and animal species and that are close to human
infrastructure that requires protection from wildfire
(Woinarski 1999, Garnett and Crowley 2000). In the
particular case of Booderee National Park, a detailed
fire management plan has been in place for almost 10
years (Department of Environment and Water Resourc-
We conducted this study at Booderee National Park,
a ;7500-ha area co-managed by the Wreck Bay
Aboriginal Community and Parks Australia (a section
of the Australian Federal Government’s Department of
the Environment and Water Resources). Booderee
National Park is located 200 km south of Sydney and
20 km south of the city of Nowra on the southern coast
of New South Wales, southeastern Australia (approxi-
DAVID B. LINDENMAYER ET AL.1968
Vol. 18, No. 8
mate midpoint is 358100S latitude, 1508400E longitude).
The area has a temperate maritime climate with an
average rainfall of 1150 mm per annum spread relatively
evenly over the year. Average minimum and maximum
air temperatures are 18–248C for January (summer) and
9.5–158C for July (winter) (Bureau of Meteorology
We targeted Booderee National Park for study
because it has some important features. First, it supports
an extremely diverse bird assemblage (Braithwaite et al.
1995), including important populations of a number of
high-profile taxa of conservation concern. For example,
Booderee National Park supports the largest known
remaining population of the Eastern Bristlebird (Da-
syornis brachypterus), a species believed to be sensitive to
the impacts of wildfires (Baker 1997, 2000). A second
reason why we targeted Booderee National Park for this
investigation was that it supports extraordinary patch-
iness and heterogeneity in vegetation types, ranging
from dry heathland to warm temperate rain forest,
which are markedly different in floristics and structure
(Ingwersen 1977, Mills 1995, Williams 1995, Taws 1998).
For this study, we recognized 10 broad categories of
native vegetation cover in Booderee National Park
A third factor influencing our selection of Booderee
National Park as a study area was that fire is a not-
uncommon form of disturbance across the full range of
vegetation types at Booderee National Park (Whelan
1995, Whelan et al. 2002), with the exception of warm
temperate rain forest (Gill et al. 1999). Fire severity may
be variable, ranging from low-severity surface fires to
high-severity stand-replacing events such as those that
occur approximately every 8–20 years in heathland
(Recher et al. 1975).
Some aspects of fire history dating back several
decades have been reasonably well documented and
carefully mapped at Booderee National Park (Ingwersen
1977, Taws 1998); the number of fires recorded for any
given area in the park varies from zero to five. Ignition
for these fires varies from unknown to lightning strike to
deliberate human ignition. The severity of past fires at
Booderee National Park has not been quantified, but
anecdotal information from long-term ranger staff in the
region indicates there was considerable spatial variation
in the severity of each fire as well as marked variation in
severity between individual fires.
Unlike some areas of temperate and boreal North
America, which support early-successional specialist
bird taxa (e.g., Murphy and Lehnhausen 1998, Hutto
2006), there are few unique bird assemblages closely
associated with recently burned habitats in any of the
vegetation types at Booderee National Park or similar
places elsewhere in coastal eastern Australia. A possible
exception is the Ground Parrot (Pezoporus wallicus),
which can quickly colonize burned heathland and
sedgeland; however, it may then occupy these areas for
many years (Keith et al. 2002).
Survey design and permanent site establishment
Important spatial data on vegetation cover, fire
history, and other variables have been captured in a
Geographic Information System (GIS) that has been
developed for Booderee National Park (ArcView, ESRI,
Redlands, California, USA). Using these spatial data,
we established a protocol for site selection at the
commencement of this project by identifying important
stratifying variables. The three stratifying variables that
we selected were: (1) vegetation, classified into 10
vegetation types (see Appendix A); (2) fire history,
classified into four classes of time since the last fire (0–10
years, 11–20 years, 21–30 years, and .30 years); and (3)
future burning, assigned to two classes (areas designated
for prescribed burning between 2002 and 2006 under the
management plan for Booderee National Park and those
to remain unburned). We manipulated the GIS to
overlay vegetation and fire history maps to form
‘‘homogeneous’’ polygons characterized by each of the
three classifying factors. We mapped these polygons and
then calculated the area of each polygon. We created a
list of all polygons and constructed a table of counts
(classified by vegetation, fire history, future fire, and
polygon area [ha]). We selected a stratified random
sample of polygons for study after excluding polygons
that contained places sacred to the local Wreck Bay
Aboriginal Community and polygons measuring ,1.5
ha in size, which were too small to contain a valid (100
m long) straight-line survey site. The selection process
that we employed ensured that: (1) the full range of
vegetation type by fire history and by future burning
classes was represented; (2) there was replication of each
class with a focus on replication of the most common
classes; (3) there was a good geographic ‘‘spread’’ of
selected polygons throughout the national park to avoid
potential problems with geographic bias (see Fig. 1); and
(4) the number of samples was generally proportional to
the total area occupied by each class. As a result, the
rarest combinations were represented only twice (for
example, woodland that had not been burned for more
than 30 years), whereas the most common combination
(woodland last burned between 11 and 20 years
previously) was represented nine times. We were acutely
aware that this approach risked missing the detection of
the fire responses of rarer species.
Using the GIS, we produced a set of geo-location
points and hard-copy maps of selected polygons to
ensure that each polygon could be precisely located on
the ground. We then established one 100 m long
permanent straight-line site within each selected poly-
gon. Our choice of site length was influenced by the
substantial heterogeneity in vegetation cover at Booder-
ee National Park, where abrupt changes in vegetation
type often occur over a short distance.
We permanently established a total of 110 field sites,
each consisting of star picket markers set at 0-, 20-, 40-,
60-, 80-, and 100-m points. We estimated and recorded
December 20081969 BIRD RESPONSE TO WILDFIRE
the coordinates of the first and last point at each site
using a global positioning system.
Following establishment of all 110 sites in our study
and the completion of bird counts in September 2003
(see below), a wildfire in late December 2003 burned
;50% of Booderee National Park. The fire was lit by an
arsonist and started at a location that was the
approximate midpoint of Booderee National Park; 59
sites were burned at varying levels of severity (see below
and Appendix A). We repaired the infrastructure on all
our damaged sites and recommenced surveys of all
burned and unburned sites in early 2004. However, the
prescribed burning program planned for Booderee
National Park was modified and the objectives of the
study that we report here were altered to focus on the
impacts of fire history dating back to 1972 and the
effects of the 2003 fire on birds.
Bird counting protocols
We completed two visits of 5 minutes each at the 20-m
and 80-m permanent points placed along the 110 sites,
for a total of 440 point-interval counts (sensu Pyke and
Recher 1983) annually. We recorded all birds seen or
heard within a polygon and assigned observations to
different distance classes from a point: 0–25 m, 25–50 m,
50–100 m, and .100 m. The survey protocol that we
used was specifically designed to quantify site occupancy
and for our statistical analyses, we did not assume that
individual counts at the two points on the same site were
independent. Each site was surveyed on a different day
by a different observer to reduce day effects on detection
and to overcome potential observer heterogeneity
problems (Cunningham et al. 1999, Field et al. 2002).
We completed surveys in late September each year,
which is the breeding season for the majority of species
and when summer migrants have arrived.
Vegetation data and other covariates
In addition to the stratifying (design) variables just
outlined, we measured a number of covariates (see
Appendix B) and used them as potential explanatory
variables in statistical modeling. For example, using past
fire-mapping data, we calculated the number of fires in
the past 35 years and the time elapsed since the last fire
at each site. These data were used in modeling bird
response to fire history.
We established vegetation plots measuring 20 3 20 m
in size at the 20–40 m and 60–80 m points at each site to
gather covariates for use for modeling of the response of
birds to the fire in 2003 (see Appendix B). We measured
in Booderee National Park, Australia.
The general location of Booderee National Park in the Jervis Bay Territory and the location of field survey sites (circles)
DAVID B. LINDENMAYER ET AL.1970
Vol. 18, No. 8
site-based and other covariates in 2004 (immediately
after the December 2003 fire) and again in late 2005
(immediately after the 2005 bird counts).
We recognized five categories of fire severity for the
2003 fire. We based this classification on two criteria.
The first was the direct effects on vegetation cover: 0, no
fire; 1, understory burned; 2, midstory burned but not
killed; 3, midstory killed; 4, midstory killed and
overstory burned. The second was based on the extent
of damage to metal tags attached to the top of each of
seven star pickets established at 20-m intervals along our
100 m long sites (zero damage¼no fire; damage class 5¼
a melted marker tag). For areas characterized by a mix
of fire severities, we chose the one that was dominant.
We used aerial photography coupled with ground-
based surveys to gather landscape-context data in 200 m
and 500 m radius concentric circles around each of our
110 field sites (Appendix B). These data included an
estimate of the amount (ha) of burned vegetation in each
vegetation type in these areas around each of the 110
sites. Finally, we gathered a range of measures of
floristic composition and vegetation structure at each of
the 110 survey sites (see Appendix B).
Bird life history attributes
Several studies have shown that avian responses to
disturbance can be linked with life history attributes
(e.g., Hansen and Urban 1992, Woinarski 1999, Brawn
et al. 2001). Given this, we collated data from the
literature on life history and other attributes for each
bird species. We summarized data on body mass, group
type (solitary, pairs, or flock), social system (monoga-
mous, polygamous, and so on), type of nest (cavity, cup,
mud bowl, and so on), nest placement (horizontal fork,
ground, and so on), nesting height, number of eggs laid
in a clutch, broods per year, movement behavior
(resident vs. migrant, latitudinal or altitudinal migrant),
and foraging guild.
We omitted one site located in the Booderee Botanic
Gardens from the statistical analysis because the site is
highly modified and supports a number of plant taxa
that are not locally endemic.
To test species richness responses, in the first stage of
statistical analysis, we explored relationships between
fire history and observed bird species richness at a site.
For each combination of year and site, we calculated the
total number of species that were observed. We fitted
quasi-Poisson generalized linear models (McCullagh
and Nelder 1989) with a log-link function to observed
species richness and we used deviance ratio tests to
assess the statistical significance of the contributions of
site attributes such as the number of past fires and the
time since fire, as well as the effect of the fire itself.
We investigated the response of the bird assemblage
to the 2003 fire in the second phase of our analysis. We
used correspondence analysis (Greenacre 1984, Digby
and Kempton 1987) to explore our data sets and identify
which bird life history attributes were linked significant-
ly with bird responses.
In the third phase of data analyses, we quantified the
responses of individual species to the 2003 fire by
plotting longitudinal (temporal) profiles based on bird
observation frequencies. We then examined whether
detection frequencies for individual species differed
significantly between sites that were burned in 2003
and those that were not.
Finally, we constructed statistical models of the
factors influencing the detection of a subset of three
species recorded between 2003 and 2006. Each species
was relatively common and represented one of three
broad types of longitudinal response curves that we
identified (Appendix D). We used generalized linear
models (GLM) (McCullagh and Nelder 1989) assuming
a quasi-binomial distribution and a logit-link function,
to model the probability of observation and hence
determine the significance and nature of individual
species’ responses to temporal effects, vegetation type,
vegetation structure, fire, and other effects. The number
of variables was always an order of magnitude less than
the number of sites. We assessed models using a
modified version of the Schwarz information criterion
(Schwarz 1978), (deviance/f ) þ p logen, where deviance
is the residual deviance and f is the scale parameter
estimated from fitting the GLM, p is the number of
parameters estimated, and n is the number of observa-
tions. After a model had been selected it was refitted as
generalized linear mixed model (GLMM) (McCulloch
and Searle 2001) to allow for the spatial site to site
component of variance. We tested the statistical
significance of effects using the method of Kenward
and Roger (1997). We discarded terms that were found
to be nonsignificant at this stage.
Observed species richness and fire history
We recorded 104 bird species from 39 families over
the four years of repeated field surveys; Appendix C lists
their common and scientific names. Prior to the 2003
fire, we recorded 71 species across our 110 survey sites.
The observed species richness detected on an individual
site varied from one to 24 species.
To quantify relationships between observed bird
species richness and fire history, we first examined
observed species richness–vegetation type interrelation-
ships. We found that observed bird species richness
varied significantly (F9,99 ¼ 5.1, P , 0.001) among
vegetation types (Table 1). We found that Casuarina
woodland had the highest observed species richness
(19.8 6 2.8 species, mean 6 SE) and it was substantially
higher than other kinds of vegetation such eucalypt
forest (14.2 6 1.2 species) and eucalypt woodland (12.4
6 1.0 species) (Table 1). The most structurally simple
vegetation type (sedgeland) was characterized by mod-
December 20081971BIRD RESPONSE TO WILDFIRE
erate levels of observed species richness (12.1 6 1.7)
We found that observed species richness at a site in
2003 was significantly (F1,99¼6.30, P¼0.014) negatively
related to the number of fires since 1972; sites that had
been burned more frequently had the fewest species. We
found that species richness per site was 8.7% lower (95%
confidence interval, 1.8–15.1%) for each additional fire.
The relationship between the number of fires and
observed species richness was similar across vegetation
types, although some kinds of vegetation rarely burned
over the past 35 years (Fig. 2).
We explored our pre-2003 observed bird species
richness data for relationships with other potential
explanatory variables but none was identified. For
example, we found no significant relationship between
observed species richness and time since fire.
Observed species richness relationships
after the 2003 wildfire
After the 2003 fire, we found that the only significant
covariate effects for observed species richness, apart
from vegetation type, came from: (1) fire severity, which
had a significant negative effect on observed species
richness (F4,106¼7.70, P , 0.001; see Fig. 3), and (2) the
percentage of unburned vegetation within 500 m of a
site, which had a significant positive effect on observed
species richness (F1,82¼ 5.72, P ¼ 0.019).
Pre- and post-2003 observed bird species richness
in burned and unburned areas
Our pre-2003 fire data for observed species richness in
the 10 vegetation types provided the basis for subse-
quent comparisons between both (1) burned and
unburned areas and (2) pre-2003 fire and post-2003 fire
levels of observed species richness for each vegetation
type. We observed similar numbers of species across all
sites combined in the years after the fire: 76 species in
2004, 63 in 2005, and 69 in 2006. However, we identified
significant (see Table 2) differences in observed species
richness between burned and unburned sites for some
vegetation types. In 2004, these were dry shrubland, wet
shrubland, sedgeland, wet heathland, and woodland. In
2005 and 2006, the difference between burned and
unburned sites was significant only for sedgeland.
We found that wet heathland and wet shrubland sites
had not returned to prefire levels by 2006. However, the
differences in observed species richness between burned
and unburned wet heathland and wet shrubland sites
were statistically significant only for 2004. For the
sedgeland sites, we note that there was a significant
difference between burned and unburned sites for
sedgeland in 2003 prior to the fire, indicating that these
two groups of sites already differed from each other.
Overall in 2003, 25 species were observed on the five
unburned sites compared with 30 on the burned sites.
The corresponding figures were 16 and 31 species in
2004, 12 and 24 species in 2005, and 14 and 25 species in
2006. We could not explain the reduction in observed
species richness in terms of the loss of particular
individual species, with the absence of different species
on different sites in different years. As an example, the
Yellow-faced Honeyeater was observed on burned sites
in 2004 and 2006 but not in 2005.
Our data do not provide any convincing evidence that
the immediate effects of fire persisted beyond the year
after the fire. In the case of sedgeland, the differences
observed in 2005 and 2006 were similar to those
observed prior to the fire.
Bird assemblage responses
We explored patterns of response of the bird
assemblage by analyzing data gathered for birds that
were recorded on a reasonably common basis. Our
criterion for selection was detection on 20 or more
occasions (;1%) out of a maximum possible 1760
detections (110 sites32 plots per site32 observers/site/
year 3 4 years), and this resulted in the inclusion of
36 species in the bird assemblage. When we applied
correspondence analysis (Greenacre 1984) to the matrix
of bird detection counts for each year separately, we
found that the first component was strongly related to
vegetation type; in particular, it was consistently related
to the complexity of the vegetation, especially as
measured by the number of vegetation layers.
The scores for the first component for the species in
different years were highly correlated (greater than 0.7
for all pairs of years), but for the second component
only the scores for 2003 and 2006 were highly correlated
(r¼0.79). This suggested that on the one hand there was
considerable stability in community structure, and that
on the other hand, reductions following the fire in 2003
had largely disappeared by 2006. Scores for sites in
different years had quite low correlations. This indicated
that there was not a strong association between bird
assemblages and particular sites.
The first component of correspondence analysis for all
combinations of sites and years had the same interpre-
tation as the first component for individual years,
whereas the second component represented a gradient
from species which appeared to prefer unburned sites to
major wildfire in 2003 from a quasi-Poisson generalized
linear model with vegetation type as the only predictor.
Fitted values for bird species richness prior to a
for a single site SE
DAVID B. LINDENMAYER ET AL.1972
Vol. 18, No. 8
those which appeared to tolerate medium-severity
burning (Fig. 4).
In an additional set of analyses of the bird species
assemblage, we sought to determine if there were
relationships between bird life history attributes and
bird responses as derived from correspondence analysis.
We found significant relationships between the first
dimension scores and maximum nest height (F1,34¼16.5,
P , 0.001) and type of nest (F4,31¼ 5.5, P ¼ 0.002).
There was a clear distinction between dome nests with a
low score and bowl, hollow, and suspended purse nests
with a high score. Cup nests had an intermediate score.
Some other life history attributes, such as dispersal ratio,
exhibited a statistically significant relationship with the
first component, dispersal ratio (F1,34¼7.21, P¼0.011),
cube root of body mass (F1,34¼ 4.64, P ¼ 0.039), and
mean length of wing (F1,34¼ 5.30, P ¼ 0.28). We found
no significant relationship between the second compo-
nent of correspondence analysis and any bird life history
Longitudinal responses of individual bird species
after the 2003 wildfire
We plotted longitudinal responses to fire severity
(including no burning) for all individual bird species
after the 2003 wildfire, except those that were observed
on 15 occasions or fewer (Appendix D). We identified
four broad types of longitudinal responses. (1) For many
species, the response curves in less severely burned areas
were very similar to those in unburned areas (e.g.,
Striated Thornbill and Fan-tailed Cuckoo). Some
species exhibited a general trend for an overall increase
in detection over time (e.g., Eastern Spinebill and
Silvereye), whereas the opposite was apparent for others
(e.g., Rufous Whistler, Noisy Friarbird, and White-
browed Scrubwren) (Appendix D). (2) Few species
exhibited significant interactions between temporal
differences in detection and burn severities (e.g., Shining
Bronze-Cuckoo [F8,203¼ 3.1, P ¼ 0.003]). In this case,
detections in 2006 for burned sites were significantly
lower (F1,94¼ 32.4, P , 0.001) than on unburned sites.
reflects a fitted relationship to the data. Crosses correspond to individual sites.
Relationship of bird species richness in 2003 with the number of fires for different vegetation types. The solid line
December 20081973 BIRD RESPONSE TO WILDFIRE
They also had not returned to prefire (2003) levels by
2006 (Appendix D). (3) Detections for a number of taxa
declined steeply initially after the fire but then returned
rapidly to prefire levels, typically by 2005 (e.g., Eastern
Yellow Robin, White-throated Treecreeper, and Eastern
Bristlebird) (Appendix D). (4) Some species exhibited a
spike in postfire detections in unburned sites, mirroring
the reduction in burned sites (e.g., Golden Whistler and
New Holland Honeyeater). The Australian Raven and
Laughing Kookaburra exhibited the opposite response;
decreasing detections on the unburned sites but increas-
ing in the burned sites (Appendix D). We based these
results on F tests derived from fitting GLMMs to pairs
Individual bird species models
We modeled three species with responses broadly
typical of the broad types of longitudinal profiles
summarized in Appendix D: the Eastern Whipbird
(Psophodes olivaceus), the Eastern Yellow Robin (Eop-
saltria australis), and the Rufous Whistler (Pachycephala
We found that detections of the Eastern Whipbird on
burned sites fell on all sites in 2004, but by 2005,
detections on the burned sites had returned to 2003
levels only to fall again in 2006. Analysis of data from
2004, 2005, and 2006 indicated that a number of fire-
related variables had a significant influence on the
detection rate for the Eastern Whipbird. In particular,
we found that the species was significantly more likely to
be recorded on sites with higher amounts of unburned
vegetation in the surrounding 500 m (F1,120¼ 35.5, P ,
0.001), and sites that were closer to areas of unburned
vegetation (F1,185¼ 6.9, P ¼ 0.010) (Table 3). The
Eastern Whipbird also was more likely to be recorded
on sites with lower values of elevation (F1,114¼ 4.4, P ¼
0.037) and higher numbers of understory plant species
(F1,271¼ 9.1, P ¼ 0.003).
We omitted wet heathland sites from model-fitting for
the Eastern Yellow Robin because it was uncommon on
are: 0, no fire; 1, understory burned; 2, midstory burned but not killed; 3, midstory killed; 4, midstory killed and overstory burned.
The solid line reflects a fitted relationship to the data. Crosses correspond to individual sites.
Bird species richness in 2004 in relation to fire severity for different vegetation types. Fire severity categories (see text)
DAVID B. LINDENMAYER ET AL.1974
Vol. 18, No. 8
them. We found that observations of the species declined
significantly on burned sites between 2003 and 2004, but
not on unburned sites. The best model for this species
for observations in 2004, 2005, and 2006 included only
one explanatory variable, percentage of leaf litter (F1,180
¼ 14.0, P , 0.001). The fitted model (with mean 6 SE)
for probability of observation, p, was
LogitðpÞ ¼ ?2:29ð60:235Þ
þ 0:01682ð60:00450Þ%Leaf Litter:
The Rufous Whistler exhibited a considerable decline
in abundance immediately after the 2003 fire for all sites,
burned and unburned. Low abundance persisted on sites
with few midstory species but recovered where there
were more midstory plant species (Table 4). This
interaction was highly significant (F3,305¼ 30.7, P ,
0.001). There was also a significant reduction in
abundance as the number of vegetation types within
500 m increased (F1,89¼8.0, P¼0.006), the coefficient in
the linear predictor being ?0.45 6 0.161.
To quantify bird response to fire, we used both a
retrospective study encompassing many sites with well-
known fire history dating back 35 years and a
prospective natural experimental approach comprising
many burned sites surveyed before and after fire that
were contrasted with matched unburned sites. In
addition, we reported findings at the population, bird
assemblage (community), and species (individual and all
species combined) levels. We then related these results to
postulates based on widely applied theories and
concepts. On the basis of this approach, we identified
several unexpected findings. Two in particular were: (1)
the rapidity of the recovery of the bird assemblage and
most individual taxa after a single (2003) fire, but a
significant effect of fire history as reflected by lower
observed species richness on sites where there had
been a number of fires over the past 35 years; and (2)
vegetation type differences in the response of bird
species richness that were opposite to those anticipated
on the onset of our investigation. We discuss these and
other key results in further detail in the remainder of
Fire and bird species richness
We postulated at the start of this study that, based on
the intermediate-disturbance hypothesis (sensu Connell
1978), we would observe the highest levels of observed
(after the 2003 survey) compared to unburned sites.
Temporal changes in bird species richness for five vegetation types following fire in 2003
and survey year
Bird species richness
(P) of effect?
Unburned Burned Difference
Notes: These tests are based on estimated effects prior to back-transformation. The effects
divided by their estimated standard errors were assumed to be normally distributed. The
approximate of F tests of Kenward and Roger (1997) give almost identical results because the
effective denominator degrees of freedom were greater than 165 in every case.
? From GLMM fit.
December 20081975BIRD RESPONSE TO WILDFIRE
species richness at sites subject to intermediate numbers
of fires (2–3 fires since 1972) and an intermediate period
of time since fire (;12–16 years). We also anticipated
that the highest observed species richness might occur on
sites burned in 2003 that were subject to intermediate
levels of fire severity. A normal bell-shaped curve would
be apparent as a result of such relationships (Wilson
1994). We found no substantive evidence to support
such patterns for birds in any of the vegetation types
studied. Observed species richness declined significantly
at a site with an increasing number of fires over the past
35 years, irrespective of vegetation type (Fig. 2).
Similarly, we showed that observed species richness
declined with increasing fire severity at sites burned in
2003 (Fig. 3), although for the vast majority of
vegetation types, the return to prefire levels was rapid
and was almost complete by 2005. Hence, the rapidity of
return to pre-2003 fire levels prevented any opportunity
for a normal-curve-shaped pattern of observed species
richness response to have developed. The findings of our
investigation add weight to the work of others that
suggest empirical support for the intermediate-distur-
bance hypothesis is inconsistent (e.g., Collins 1992,
Schwilk et al. 1997, Bascompte and Rodriguez 2000,
Beckage and Stout 2000, Schurbon and Fauth 2003).
Appendix C for abbreviation codes for individual bird species. The first component strongly reflected the influence of vegetation
type on the bird species assemblage, in particular, a contrast between structurally complex vegetation types such as forest and
woodland vs. more simply structured sedgeland and wet heathland. The second component reflected a contrast between bird species
that preferred unburned sites and those that tolerated sites subject to medium-severity fire.
First two components for species from correspondence analysis of data for all combinations of sites and years. See
sion model for observations of the Eastern Whipbird
Coefficients and standard errors for logistic regres-
Percentage unburned vegetation
within 500 m after 2003 fire
Number of understory species
DAVID B. LINDENMAYER ET AL.1976
Vol. 18, No. 8
Observed bird species richness, vegetation complexity,
and postfire response
An additional area of work in this study was the
relationship between vegetation type, species richness,
and postfire observed species richness. Based on research
by MacArthur and MacArthur (1961) and Recher
(1969) demonstrating relationships between species
richness and the vertical complexity of the vegetation,
we postulated that prefire observed bird species richness
would be highest in the most vertically complex
vegetation types, but after the 2003 wildfire it would
take longer in these vegetation types to return to prefire
levels. Our prefire data showed broad relationships
between vertical complexity and species richness; wood-
land and forest supported a higher average number of
species than structurally more simple vegetation types
such as sedgeland, heathland, and shrubland (Table 1).
However, the relationship was not particularly strong,
as the most structurally simple vegetation type, sedge-
land, had intermediate levels of observed species
richness (Table 1).
Our postfire data produced a more unexpected result,
opposite to that postulated at the beginning of the
investigation. That is, although observed bird species
richness in structurally complex vegetation types had
returned to prefire levels by 2006, it had not in more
structurally simple types like sedgeland sites. A possible
explanation for our findings for birds is that the speed of
bird recovery might be more closely allied with
differences in plant recovery mechanisms (Noble and
Slatyer 1981, Gill et al. 1999) than with initial (prefire)
vegetation complexity. Vegetation response in areas
such as woodland and forest has largely been via
epicormic resprouting from the trunks and large
branches of trees, the vast majority of which survived
the 2003 fire and quickly began resprouting. Structurally
complex vegetation cover has characterized these
vegetation types and this may be a key factor
contributing to the rapidity of postfire bird response.
Conversely, many plant species in sedgeland have
regenerated from seed shed during and immediately
after the fire, and the structural recovery of the
vegetation has been slower than in other vegetation
types. This highlights the fact that bird species’
responses to fire must first be underpinned by a good
understanding of the relationships between birds and
Many previous studies of fire and vertebrate responses
have contrasted burned and unburned areas and
generally have overlooked fire severity in areas that
have been burned (but see Smucker et al. 2005, Kotliar
et al. 2007) and the patchiness of the vegetation within
boundaries of a burned area (Schmiegelow et al. 2006).
However, several recent investigations have highlighted
the importance of quantifying fire severity in studies of
disturbance effects on avifauna. For example, Smucker
et al. (2005) and Kotliar et al. (2007) described
interspecific differences in response to fire severity.
Kotliar et al. (2007) showed that many taxa exhibited
positive or neutral responses in density within 1–2 years
of fire, even in locations where fire severity was
moderate to high. Some of these outcomes are consistent
with those of our study, including sites burned at high
severity in 2003. We identified significant negative effects
of the severity of the 2003 fire on observed species
richness, but despite an incomplete postfire bird
assemblage in sedgeland, for most other vegetation
types observed bird species richness rapidly returned to
prefire levels after the 2003 wildfire. We identified similar
patterns for many individual species. Thus, in common
with the results of Kotliar et al. (2007), we found that
many species could tolerate the effects of one moderate-
Our findings for the rapid response of the bird
assemblage following the 2003 fire were in marked
contrast to the results of many studies of northern
temperate and boreal forest. In some of these northern
temperate and boreal ecosystems, particularly those
where high-intensity disturbances are stand-replacing
events, areas can take prolonged periods to regenerate
after perturbation (Burton et al. 2003, Shatford et al.
2007). These areas can be floristically and structurally
distinct from mid- and late-successional stands (Frank-
lin et al. 2002, Franklin and Agee 2003, Fraser et al.
2004). They also can support a suite of early-succes-
sional species (Hutto 1995, 2006, Brawn et al. 2001,
numbers of midstory plant species for the four years of survey (linear predictor 6 its standard
error in parentheses).
Fitted observation rates for the Rufous Whistler (Pachycephala rufiventris) for varying
Number of midstory plant species
0.22 (1.24 6 0.220)
0.02 (?4.14 6 0.394)
0.02 (?3.74 6 0.321)
0.01 (?4.53 6 0.430)
0.19 (?1.47 6 0.205)
0.03 (?3.49 6 0.283)
0.10 (?2.16 6 0.359)
0.07 (?2.62 6 0.395)
0.16 (?1.69 6 0.340)
0.06 (?2.84 6 0.409)
0.36 (?0.56 6 0.0.704)
0.33 (?0.71 6 0.819)
Notes: Values were derived from a quasi-Poisson generalized linear mixed model with site as a
random effect and year, number of midstory plant species, and number of vegetation types within
500 m as fixed effects. The relationship between the observation rate and the predictors was
assumed to be linear on the logistic scale.
December 20081977BIRD RESPONSE TO WILDFIRE
Imbeau et al. 2001, Hoyt and Hannon 2002, Smucker
et al. 2005).
By contrast, the majority of vegetation types that we
studied generally recovered quickly and this, in part,
may explain why they are characterized by few, if any,
specialist early-successional birds. A paucity of early-
successional specialists also would explain why the
composition of the bird assemblage was little changed
over the duration of our investigation. Postfire condi-
tions, particularly high levels of rainfall, appear to have
promoted vegetation growth and structural regeneration
and, in turn, aided bird species recovery in most
vegetation types (Fig. 5). It is notable that our field
measurements of the vegetation suggest that vegetation
between 2003 and 2006: (A) two days after high-severity fire (category 3) in December 2003 at the site and (B) three years after the
2003 fire (photo credits: D. Lindenmayer and C. MacGregor).
Location close to the 0 m point of a woodland site in Booderee National Park that was surveyed for birds repeatedly
DAVID B. LINDENMAYER ET AL.1978
Vol. 18, No. 8
layers in forest and woodland vegetation types have
returned to pre-2003 fire levels, although not in others
such as sedgeland, where bird responses have been
slower. We also note that the use of the term ‘‘recovery’’
is inappropriate for a number of individual bird species
because detection rates for them changed little both
before and after the 2003 fire and between burned and
unburned areas (Appendix D). This occurred even on
those sites subjected to high-severity fire in 2003.
Biological legacies and bird responses
Several studies have highlighted the importance of
biological legacies in shaping the post-disturbance
responses of plants and animals (Foster et al. 1998,
Turner et al. 2003). Our findings further underscore the
importance of biological legacies. The amount of
unburned vegetation in the area surrounding a site was
a significant explanatory variable in logistic regression
models for observed species richness in 2004 (immedi-
ately following the 2003 wildfire) and also in statistical
models developed for a number of individual species
(e.g., the Eastern Whipbird; Table 4). In addition, we
found the response curves for some species in less
severely burned areas were similar to those in unburned
areas (Appendix D). These findings suggest that the
presence of biological legacies either facilitates the
persistence of a given species on a site following a fire
(see Whelan et al. 2002) or promotes the return of some
species to pre-disturbance levels (Turner et al. 2003,
Lindenmayer et al. 2005). We note, however, that
species richness and density may be misleading indica-
tors of habitat quality (van Horne 1983) because of
factors such as nest predation or habitat patches being
inhabited by one sex only (Temple and Cary 1988).
Further detailed studies of breeding success and other
population parameters (e.g., see Zanette et al. 2000)
would be required to establish if these measures are
related to the presence of birds at a site. However, our
data suggest that at many sites, populations of some
species such as the Eastern Bristlebird have been
increasing in the past two years (D. B. Lindenmayer,
Assemblage patterns and life history attributes
We identified interesting patterns for the bird
assemblage that contained several findings consistent
with our results for observed species richness and for
some individual bird taxa. The first component of
correspondence analysis (Fig. 4) strongly reflected the
influence of vegetation type on the bird species
assemblage, in particular, a contrast between structur-
ally complex vegetation types such as forest and
woodland vs. more simply structured sedgeland and
wet heathland. Notably, scores for the first component
for the species in different years were highly correlated
(.0.7 for all pairs of years), indicating stability in bird
community composition. We were able to link particular
life history attributes (nesting height and nest type) to
these outcomes. We expected these life history responses
because more species of birds are likely to nest higher in
taller forest and woodland vegetation. In addition,
structural features such as tree hollows are also more
prevalent in these vegetation types (reviewed by
Gibbons and Lindenmayer 2002), which would account
for the significant relationship that we observed.
We found that the second dimension of correspon-
dence analysis reflected a contrast between birds that
preferred unburned sites to those that tolerated sites
subject to medium-severity fire. We also found that the
impacts on bird community structure of the 2003 fire
had largely disappeared by 2006. No life history
attributes were associated with the second component
of correspondence analysis. A number of studies have
highlighted strong relationships between plant life
history attributes and sequences of fires (e.g., Noble
and Slatyer 1980). However, attempts to identify clear
and readily predictable associations between animal
responses and fire have not been particularly fruitful in
Australia (Whelan et al. 2002). Outcomes are often site
specific, and it is possible there may not be a set of ‘‘vital
attributes’’ (sensu Noble and Slatyer 1980) that deter-
mines responses for animals as there has been hypoth-
esized for plants (Whelan et al. 2002, Bradstock et al.
2005). As in the case of work on the rapidity of postfire
response, our findings for relationships between life
history attributes and animal responses to fire are in
marked contrast to those from the temperate and boreal
forest ecosystems of the northern hemisphere. In those
environments, there are often predictable relationships
between postfire habitat, animals, and their life history
attributes (e.g., Raphael et al. 1987, Imbeau et al. 1999,
Saab et al. 2007). The reasons for such differences are
not immediately clear but they may be associated with
the rapidity of postfire vegetation recovery. This would,
in turn, limit the spatiotemporal availability of suitable
habitat for early-successional specialists in the vegeta-
tion types that we studied. Another possible reason for
differences might be that the findings we have reported
are based on a single major study. Additional studies
from areas similar to the one in our investigation would
be required to strengthen the inference of fire responses
we have drawn.
Integrating fire management and biodiversity conser-
vation is a challenging task because of the many factors
that must be considered, particularly when the protec-
tion of human infrastructure is one of the objectives of
management. These challenges are set to become harder
as many researchers suggest that climate change will
increase fire frequency (e.g., Cary 2002, Lenihan et al.
2003, Westerling et al. 2006). Nevertheless, we believe
that the results of our study have some important
implications for improved fire management practices.
First, because we found that observed species richness
at a site was significantly negatively related to the
December 20081979BIRD RESPONSE TO WILDFIRE
sequence of past fires (Fig. 2), we suggest that an
informed approach to strategic spatial and temporal fire
management may be to ensure that there are some areas
exempt from additional planned prescribed fires. In
particular, areas that have experienced several fires (e.g.,
three or more) over the past 35 years might be those
where prescribed fire might be excluded. We also believe
that there may be a need to target the protection of
particular vegetation types such as those where the
return of observed species richness (or particular
individual taxa) to prefire levels is slowest. For example,
in the case of Booderee National Park, efforts might be
best focused on sedgeland and wet heathland, where our
data suggest that the return to prefire bird species
richness levels has been slowest (Table 2). Of course, this
recommendation is not a generic one that can be applied
uncritically to all ecosystems. For example, it would not
apply in the many vegetation types around the world
where prolonged fire suppression has had negative
impacts on biodiversity and the proactive use of fire
management is important (Zackrisson 1977, Harrington
and Sanderson 1994, Covington 2003, Ehle and Baker
2003, Spies et al. 2004).
A second key finding from our work that has
management implications is the relationship between
the amount of unburned vegetation surrounding a site
and the observed species richness, as well as the
persistence of individual species on sites. Almost all
wildfires are inherently patchy, leaving unburned and
partially burned areas (Turner et al. 2003, Kulakowski
and Veblen 2007). Such landscape heterogeneity is
known to be important for the maintenance of
biodiversity (De Long and Kessler 2000), although it is
not always clear which spatial mosaic of patch types is
the most appropriate one (Bradstock et al. 2005, Parr
and Andersen 2006). To maintain landscape heteroge-
neity, we believe that one objective of prescribed burning
might be to maintain some unburned areas with the
perimeter of such controlled fires.
Our data on the importance of the amount of
unburned vegetation surrounding a site also may have
implications for the suppression of wildfires. In some
jurisdictions, fire suppression activities involve ‘‘black-
out’’ burning or burning out fuel between a fuel break
and a fire. We believe that such practices might need to
be carefully managed in places where the conservation
of biodiversity is one of the objectives of management.
This is because ‘‘blackout’’ burning may lead to a loss of
landscape heterogeneity (Backer et al. 2004) with the
potential to negatively influence postfire responses of
some elements of the biota.
This project is part of an integrated study of fire and native
vegetation in Booderee National Park. It is funded by the
Australian Research Council and the Australian Government
Department of the Environment, Water, Heritage, and the Arts
and Department of Defence. The strong support of the Wreck
Bay Aboriginal Community is also most gratefully acknowl-
edged, and we are privileged to work on land that is co-
managed by them. Australian Bush Heritage, the Thomas
Foundation, and the Daimler Chrysler Environmental Re-
search Prize have helped support Darren Brown. Scott
Surridge, Peter Cochrane, and Con Boekel have been important
supporters of this project since its inception. Counts of birds
were completed through the generous assistance of highly
experienced expert volunteers from the Canberra Ornithologists
Group. In particular, we thank Bruce Lindenmayer, Jenny
Bounds, Martin Moffat, Terry Munro, Peter Fullagar, and
Mike Doyle. Comments by two anonymous referees signifi-
cantly improved earlier versions of the paper.
Agee, J. K. 1993. Fire ecology of the Pacific Northwest forests.
Island Press, Washington, D.C., USA.
Andersen, A. N., G. D. Cook, L. K. Corbett, M. M. Douglas,
R. W. Eager, J. Russell-Smith, S. A. Setterfield, R. J.
Williams, and J. C. Z. Woinarski. 2005. Fire frequency and
biodiversity conservation in Australian tropical savannas:
implications from the Kapalga fire experiment. Austral
Andersen, A. N., G. D. Cook, and R. J. Williams, editors. 2003.
Fire in tropical savannas: the Kapalga experiment. Springer-
Verlag, New York, New York, USA.
Backer, D. M., S. E. Jensen, and G. R. McPherson. 2004.
Impacts of fire suppression activities on natural communities.
Conservation Biology 18:937–944.
Baker, J. 1997. The decline, response to fire, status and
management of the Eastern Bristlebird. Pacific Conservation
Baker, J. R. 2000. The Eastern Bristlebird: cover dependent and
fire sensitive. Emu 100:286–298.
Bascompte, J., and M. A. Rodriguez. 2000. Self-disturbance as
a source of spatiotemporal heterogeneity: the case of the
tallgrass prairie. Journal of Theoretical Biology 204:153–164.
Beckage, B., and J. I. Stout. 2000. Effects of repeated burning
on species richness in a Florida pine savanna: a test of the
intermediate disturbance hypothesis. Journal of Vegetation
Bradstock, R. A., M. Bedward, A. M. Gill, and J. S. Cohn.
2005. Which mosaic? A landscape ecological approach for
evaluating between fire regimes, habitat and animals. Wildlife
Bradstock, R. A., J. E. Williams, and A. M. Gill, editors. 2002.
Flammable Australia: the fire regimes and biodiversity of a
continent. Cambridge University Press, Melbourne, Australia.
Braithwaite, L. W., M. P. Austin, and P. C. Catling. 1995.
Forest and woodland communities. Pages 91–98 in G. Cho,
A. Georges, and R. Stoutjesdijk, editors. Jervis Bay: a place
of cultural, scientific and educational value. Australian
Nature Conservation Agency, Canberra, Australia.
Brawn, J. D., S. K. Robinson, and F. R. Thompson. 2001. The
role of disturbance in the ecology and conservation of birds.
Annual Reviews of Ecology and Systematics 32:251–276.
Brokaw, N. V., and R. A. Lent. 1999. Vertical structure. Pages
373–399 in M. Hunter, editor. Managing biodiversity in
forest ecosystems. Cambridge University Press, Cambridge,
Bureau of Meteorology. 2007. Climate statistics for Australian
locations: Jervis Bay Point Perpendicular Lighthouse. Bureau
of Meteorology, Canberra, Australia. hhttp://www.bom.gov.
Burton, P. J., C. Messier, D. W. Smith, and W. L. Adamowicz.
2003. Towards sustainable management of the boreal forest.
National Research Council of Canada, Ottawa, Canada.
Cary, G. 2002. Importance of a changing climate for fire
regimes in Australia. Pages 26–48 in R. A. Bradstock, J. E.
Williams, and A. M. Gill, editors. Flammable Australia.
Cambridge University Press, Melbourne, Australia.
DAVID B. LINDENMAYER ET AL. 1980
Vol. 18, No. 8
Cary, G., D. B. Lindenmayer, and S. Dovers, editors. 2003. Download full-text
Australia burning: fire ecology, policy and management
issues. CSIRO [Commonwealth Scientific and Industrial
Research Organization] Publishing, Melbourne, Australia.
Collins, S. L. 1992. Fire frequency and community heteroge-
neity in tallgrass prairie vegetation. Ecology 73:2001–2006.
Connell, J. H. 1978. Diversity in tropical forests and coral reefs.
Covington, W. W. 2003. The evolutionary and historical
context. Pages 26–47 in P. Friederici, editor. Ecological
restoration of Southwestern ponderosa pine forests. Island
Press, Washington, D.C., USA.
Cunningham, R. B., D. B. Lindenmayer, H. A. Nix, and B. D.
Lindenmayer. 1999. Quantifying observer heterogeneity in
bird counts. Australian Journal of Ecology 24:270–277.
DeLong, S. C., and W. B. Kessler. 2000. Ecological character-
istics of mature forest remnants left by wildfire. Forest
Ecology and Management 131:93–106.
Department of Environment and Water Resources. 2007.
Threatened species and threatened ecological communities.
Department of the Environment and Water Resources,
Canberra, Australia. hhttp://www.environment.gov.au/
Digby, P. G. N., and R. A. Kempton. 1987. Multivariate
analysis of ecological communities. Chapman and Hall,
Ehle, D. S., and W. L. Baker. 2003. Disturbance and stand
dynamics in ponderosa pine forests in Rocky Mountain
National Park, USA. Ecological Monographs 73:543–566.
Field, S. A., A. J. Tyre, and H. P. Possingham. 2002. Estimating
bird species richness: how should repeat surveys be organized
in time? Austral Ecology 27:624–629.
Foster, D. R., D. H. Knight, and J. F. Franklin. 1998.
Landscape patterns and legacies resulting from large,
infrequent forest disturbances. Ecosystems 1:497–510.
Franklin, J. F., and J. K. Agee. 2003. Forging a science-based
national forest fire policy. Issues in Science and Technology
Franklin, J. F., D. B. Lindenmayer, J. A. MacMahon, A.
McKee, J. Magnuson, D. A. Perry, R. Waide, and D. R.
Foster. 2000. Threads of continuity: ecosystem disturbances,
biological legacies and ecosystem recovery. Conservation
Biology in Practice 1:8–16.
Franklin, J. F., T. A. Spies, R. van Pelt, A. Carey, D.
Thornburgh, D. R. Berg, D. B. Lindenmayer, M. Harmon,
W. Keeton, and D. C. Shaw. 2002. Disturbances and the
structural development of natural forest ecosystems with
some implications for silviculture. Forest Ecology and
Fraser, E., S. Landhausser, and V. Lieffers. 2004. The effect of
fire severity and salvage logging traffic on regeneration and
early growth of aspen suckers in north-central Alberta.
Forestry Chronicle 80:251–256.
Garnett, S. T., and G. M. Crowley. 2000. The Action Plan for
Australian birds 2000. Natural Heritage Trust, Canberra,
Gibbons, P., and D. B. Lindenmayer. 2002. Tree hollows and
wildlife conservation in Australia. CSIRO Publishing,
Gill, A. M. 1975. Fire and the Australian flora: a review.
Australian Forestry 38:4–25.
Gill, A. M., J. C. Woinarski, and A. York, editors. 1999.
Australia’s biodiversity: responses to fire. Plants, birds and
invertebrates. Environment Australia, Canberra, Australia.
Greenacre, M. J. 1984. Theory and applications of correspon-
dence analysis. Academic Press, Orlando, Florida, USA.
Hansen, A. J., and D. L. Urban. 1992. Avian response to
landscape pattern: the role of species’ life histories. Land-
scape Ecology 7:163–180.
Harrington, G. N., and K. D. Sanderson. 1994. Recent
contraction of wet sclerophyll forest in the wet tropics of
Queensland due to invasion by rainforest. Pacific Conserva-
tion Biology 1:3319–3327.
Haynes, R. W., B. T. Bormann, D. C. Lee, and J. R. Martin.
2006. Northwest forest plan: the first 10 years (1994–2003):
synthesis of monitoring and research results. USDA Forest
Service, Pacific Northwest Research Station, Portland,
Hoyt, J. S., and S. J. Hannon. 2002. Habitat associations of
black-backed and three-toed woodpeckers in the boreal
forest of Alberta. Canadian Journal of Forest Research 32:
Hutto, R. 1995. Composition of bird communities following
stand-replacement fires in northern Rocky Mountain conifer
forests. Conservation Biology 9:1041–1058.
Hutto, R. L. 2006. Toward meaningful snag-management
guidelines for postfire salvage logging in North American
conifer forests. Conservation Biology 20:984–993.
Imbeau, L., M. Mo ¨ nkko ¨ nen, and A. Descrochers. 2001. Long-
term effects of forestry on birds of the eastern Canadian
boreal forests: a comparison with Fennoscandia. Conserva-
tion Biology 15:1151–1162.
Imbeau, L., J.-P. Savard, and R. Gagnon. 1999. Comparing
bird assemblages in successional black spruce stands origi-
nating from fire and logging. Canadian Journal of Zoology
Ingwersen, F. 1977. Regeneration of vegetation after fire at
Jervis Bay and its implications for management. Thesis.
Australian National University, Canberra, Australia.
Keith, D., J. Williams, and J. Woinarski. 2002. Fire manage-
ment and biodiversity conservation: key approaches and
principles. Pages 401–425 in R. Bradstock, J. Williams, and
A. M. Gill, editors. Flammable Australia. The fire regimes
and biodiversity of a continent. Cambridge University Press,
Kenward, M. G., and J. H. Roger. 1997. Small sample inference
for fixed effects from restricted maximum likelihood.
Kotliar, N. B., P. L. Kennedy, and K. Ferree. 2007. Avifaunal
responses to fire in southwestern montane forests along a
burn severity gradient. Ecological Applications 17:491–507.
Kulakowski, D., and T. T. Veblen. 2007. Effect of prior
disturbance on the extent and severity of wildfire in Colorado
subalpine forests. Ecology 88:759–769.
Lenihan, J. M., R. Drapek, D. Bachelet, and R. P. Neilson.
2003. Climate change effect on vegetation distribution,
carbon, and fire in California. Ecological Applications 13:
Lindenmayer, D. B., R. B. Cunningham, and R. Peakall. 2005.
The recovery of populations of bush rat Rattus fuscipes in
forest fragments following major population reduction.
Journal of Applied Ecology 42:649–658.
Loyn, R. H. 1993. Effects of an extensive wildfire on birds in far
eastern Victoria. Pacific Conservation Biology 3:221–234.
MacArthur, R. H., and J. W. MacArthur. 1961. On bird species
diversity. Ecology 42:594–598.
Mackey, B., D. B. Lindenmayer, A. M. Gill, M. A. McCarthy,
and J. A. Lindesay. 2002. Wildlife, fire and future climate: a
forest ecosystem analysis. CSIRO Publishing, Melbourne,
McCullagh, P., and J. A. Nelder. 1989. Generalized linear
models. Second edition. Chapman and Hall, New York, New
McCulloch, C. E., and S. R. Searle. 2001. Generalized, linear,
and mixed models. John Wiley, New York, New York, USA.
Mills, K. 1995. Natural vegetation. Pages 71–81 in G. Cho,
A. Georges, and R. Stoutjesdijk, editors. Jervis Bay: a place
of cultural, scientific and educational value. Australian
Nature Conservation Agency, Canberra, Australia.
Murphy, E. C., and W. A. Lehnhausen. 1998. Density and
foraging ecology of woodpeckers following a stand replace-
ment fire. Journal of Wildlife Management 62:1359–1372.
December 20081981BIRD RESPONSE TO WILDFIRE