The spatial ecology of fire refuges in the
Victorian Central Highlands
Laurence Berry
Submitted in fulfilment of the requirements
for the degree of Doctor of Philosophy
of the Australian National University
January 2016
ii
iii
Declaration
This thesis is my own work, except where otherwise acknowledged
(see Preface and Acknowledgements).
Laurence Berry
January 2016
iv
v
Preface
This thesis is structured as a series of connected papers that have been published,
accepted, or submitted for publication at the time of thesis submission. I have listed
these papers at the end of the preface. All papers were intended as stand-alone pieces of
work. Consequently, there is some overlap in content between chapters, for example in
the nature of the background material and the description of study areas.
The format of this thesis complies with the Australian National University’s College of
Medicine, Biology and Environment guidelines for a thesis by compilation. In
accordance with these guidelines I have included a Context Statement at the beginning
of this thesis. The Context Statement provides a framework for understanding the
relationships between all aspects of the research.
I performed the majority of the work for the papers that form this thesis, including
developing research questions and experimental designs, data collection and analysis,
and writing the manuscripts. My supervisors (David Lindenmayer, Don Driscoll and
Sam Banks) provided advice on conceptualisation and interpretation of the findings and
assisted with manuscript revisions. The addition of different co-authors to each paper
reflects contributions from collaborators, which are detailed below. The author
contribution statement below has been agreed to in writing by all authors listed. Specific
contributions of co-authors to each paper are outlined below. Other contributions made
to this thesis are recognized in the Acknowledgements section of each paper.
vi
vii
I. Berry, L. E., Driscoll, D. A., Stein, J. A., Blanchard, W., Banks, S. C.,
Bradstock, R. A., & Lindenmayer, D. B. (2015). Identifying the location
of fire refuges in wet forest ecosystems. Ecological Applications.
Conceptualization: LB, DL; Design; LB, DD, Data extraction: LB, JS; Data analysis:
LB (with advice from WB); Manuscript drafting: LB; Manuscript editing & preparation:
LB, DD, JS, WB, SB, RAB, DBL
II. Berry, L. E., Driscoll, D. A., Banks, S. C., & Lindenmayer, D. B. (2015).
The use of topographic fire refuges by the greater glider (Petauroides
volans) and the mountain brushtail possum (Trichosurus cunninghami)
following a landscape-scale fire. Australian Mammalogy, 37(1), 39-45.
Conceptualization: LB, SB, DL; Design; LB, Field work and data collection: LB; Data
analysis: LB; Manuscript drafting: LB; Manuscript editing & preparation: LB, DD, SB,
DL
III. Berry, L. E., Driscoll, D. A., Banks, S. C., & Lindenmayer, D. B. (2015)
Bird use of fire refuges is contingent on landscape context and the
spatial extent of mixed severity fire. Diversity and Distributions. UNDER
REVIEW
Conceptualization and design: LB, DD, DL; Field work and data collection: LB; Data
analysis: LB; Manuscript drafting: LB; Manuscript editing & preparation: LB, DD, SB,
DL
IV. Berry, L. E., Lindenmayer, D. B., Driscoll, D. A., T. Dennis & Banks, S.
C. (2015) Fire severity patterns alter spatial and temporal movement
viii
patterns of an arboreal marsupial, the Mountain Brushtail Possum
Trichosurus Cunninghamii. International Journal of Wildland Fire.
UNDER REVIEW
Conceptualization and design: LB, SB; Field work and data collection: LB; Data
analysis: LB; Manuscript drafting: LB; Manuscript editing & preparation: LB, SB, TD,
DD, DL
V. Berry, L. E., Driscoll, D. A., Banks, S. C., & Lindenmayer, D. B. (2015)
Bird use of fire refuges is contingent on landscape context and the
spatial extent of mixed severity fire. In preparation for submission to
Frontiers in Ecology and Environment.
Conceptualization and design: LB; Data collection: LB; Data analysis: LB; Manuscript
drafting: LB; Manuscript editing & preparation: LB, SB, DD, DL
ix
Acknowledgements
I gratefully thank my supervisory panel David Lindenmayer, Don Driscoll and Sam
Banks for their guidance throughout my PhD candidature. I particularly acknowledge
David Lindenmayer for providing a source of inspiration during my study of the
Mountain Ash forests. I thank Don Driscoll for his availability and for his detailed and
constructive criticisms of my work. His continued support and input into my research
have greatly contributed to my development as an ecologist. I also thank Sam Banks
who has provided enthusiasm and a fresh perspective to the work in this thesis.
I would like to acknowledge the contributions of several others to the successful
completion of this thesis. John Stein provided assistance with the provision and use of
the spatial data explored in these papers, I have enjoyed our conversations throughout
my candidacy. I thank Wade Blanchard and Jeff Wood for their statistical advice. Ross
Bradstock provided enthusiasm and feedback on the first chapter. I thank Claire
Shepherd for her administrative support throughout my candidacy. A special thanks to
Lachlan McBurney and David Blair, whose help and camaraderie made my time spent
in the field highly enjoyable. Their attitude to forest ecology and conservation provides
a continuing source of inspiration.
Throughout my PhD I have benefited from numerous conversations and discussions. I
thank the following people who contributed constructively in various ways; Will
Batson, Ben Scheele, Annabel Smith, Claire Foster, Claudia Benham, Geoff Kay, Geoff
x
Carey, Phil Gibbons, Luke Kemp, Juliana Lazzari, John Evans and Karen Ikin. I also
thank the numerous volunteers who assisted in the field.
Finally, I would like to acknowledge the contributions of my family and friends. I thank
Fiona Tew for her positive attitude throughout my candidature and for constantly
pushing me to realise my ambitions. I owe enormous gratitude to my parents Margaret
and Andrew and my sisters Sophie and Amelia for their unwavering support, wisdom
and encouragement. They have made this journey less arduous and more rewarding. I
also thank the Tew family for their enthusiasm, help and friendship during my
candidature.
xi
Abstract
The spatial and temporal pattern of fire occurrence within landscapes is a principal
factor influencing species distributions and a core driver of biodiversity. However,
climate change, land use change, invasive species and detrimental land management
practices are altering the distribution, frequency, scale and intensity of large wildfires
globally. This poses a major challenge to biodiversity management as ecosystems adapt
to novel patterns of fire occurrence. Within fire-affected landscapes, areas which
experience unique disturbance regimes may act as refuges for biota, reducing the
impacts of fire on species and increasing their likelihood of survival. However, very few
studies have attempted to quantify the desirable spatial attributes of such areas within
fire mosaics for faunal conservation. This thesis aimed to quantify the ecological role
of fire refuges by examining the factors responsible for refuge establishment, how the
spatial properties of refuges influence their use by fauna, and the mechanisms
underpinning faunal responses.
To investigate the factors responsible for the spatial distribution of fire refuges in
montane forests I tested the operational validity of a pre-constructed fire simulation
model with actual fire severity patterns produced following a large fire in the modelled
landscapes. I found that for fires which occurred in extreme fire conditions, severity
patterns were largely determined by stochastic factors, such as weather. When fire
conditions were moderate, physical landscape properties appeared to mediate fire
severity distribution. The study highlighted that fire refuges are a potentially
ecologically important outcome of large wildfires. I recommend that detrimental land
xii
management practices are minimized to enable the ecological processes relevant to the
establishment and subsequent use of fire refuges to be maintained.
In recently burnt Mountain Ash forests in south-eastern Australia, I examined how fire
severity, patch size and landscape context influenced the abundance of arboreal
marsupials. We aimed to determine if fire refuges are an important mechanism for
facilitating the survival within extensively burnt landscapes. I found the mountain
brushtail possum had a positive response to a particular kind of topographic refuge
(unburnt peninsulas connected to larger areas of unburnt forest), whereas the greater
glider had a negative response to fire in the landscape. The study highlighted the need
for a more developed understanding of how post-fire habitat patterns facilitate species
survival within burnt landscapes.
In a correlative landscape-scale study, I examined how bird use of potential refuges was
influenced by 1) the size and connectivity of each refuge, 2) the extent of fire severities
at different scales in the surrounding landscape, and 3) the interaction between severity
patterns, vegetation structure and environmental gradients. I found that unburnt mesic
gullies facilitated the retention of forest birds within extensively burnt montane forest
landscapes. The study presented a key advance, in that the effects of fire-induced
habitat patterns on the distribution of fauna varied between areas depending on their
spatial relationships with key biotic and abiotic landscape patterns. I demonstrated that
developing contingent theory by examining ecological interactions between fire induced
habitat patterns and biotic and abiotic gradients is essential to understanding complex
faunal responses to fire.
xiii
Using GPS telemetry within a replicated landscape scale study design, I examined how
the spatial patterns of fire severity created by a large wildfire influenced the spatio-
temporal movement patterns of an arboreal marsupial, the Mountain Brushtail Possum,
Trichosurus cunninghammi. I found a difference in temporal movement dynamics,
habitat selection and spatial movement patters between forested landscapes which were
burnt to differing extents. Forest systems recently burnt at high severity may provide
suitable habitat for some species, if protected from subsequent disturbance such as
salvage logging. However, spatial and temporal patterns of habitat selection and use
differed considerably between burnt and undisturbed landscapes. The spatial outcomes
of ecological disturbances such as wildfires have the potential to alter the behaviour and
functional roles of fauna across large areas.
Employing a qualitative research approach, I identified the barriers and enablers to
spatially managing fire for biodiversity. I then developed a conceptual framework and
set of key steps to achieve the integration of spatial approaches to fire into management.
I identified that spatial approaches to fire management must co-exist within a complex
system of social and ecological feedbacks between landscapes, academic research,
socio-political land management systems, and environmental pressures. I suggest that
the integration of spatial approaches to fire can be achieved by developing community
understanding of fire science, improving the relevance of fire research outputs to land
management, amending existing government policy approaches and refining
management tools, structures, scales and monitoring to meet biodiversity and fire risk
objectives
xiv
The insights into fire refuge ecology provided by the papers in this thesis are highly
relevant to faunal conservation. Collectively, this thesis constitutes an important
contribution to global forest fire ecology and management and has implications for both
understanding the impacts of ecosystem disturbances on faunal persistence and
distributions, and for developing effective future research and conservation strategies.
xv
Table of Contents
Declaration ...................................................................................................................... iii
Preface .............................................................................................................................. v
Acknowledgements ......................................................................................................... ix
Abstract ........................................................................................................................... xi
Table of Contents .......................................................................................................... xv
Chapter 1. Context Statement.................................................................................. 17 1.1 Introduction .................................................................................................... 17
1.2 Overview of aims, methodologies and key results ......................................... 22
1.3 Concluding remarks ........................................................................................ 28
1.4 References ...................................................................................................... 29
Chapter 2. Identifying the location of fire refuges in wet forest ecosystems ....... 33 2.1 Abstract ........................................................................................................... 35
2.2 Introduction .................................................................................................... 37
2.3 Methods .......................................................................................................... 41
2.4 Results ............................................................................................................ 48
2.5 Discussion ....................................................................................................... 55
2.6 Conclusions .................................................................................................... 58
2.7 References ...................................................................................................... 59
2.8 Appendix 1 ..................................................................................................... 63
2.9 Appendix 2 ..................................................................................................... 66
2.10 Appendix 3 ..................................................................................................... 70
Chapter 3. The use of topographic fire refuges by the greater glider (Petauroides
volans) and the mountain brushtail possum (Trichosurus cunninghami) following a
landscape-scale fire. ...................................................................................................... 73 3.1 Abstract ........................................................................................................... 75
3.2 Introduction .................................................................................................... 76
3.3 Methods .......................................................................................................... 78
3.4 Results ............................................................................................................ 82
3.5 Discussion ....................................................................................................... 85
3.6 Acknowledgements ........................................................................................ 89
3.7 References ...................................................................................................... 90
xvi
Chapter 4. Bird use of fire refuges is contingent on landscape context and the
spatial extent of mixed severity fire ............................................................................. 95 4.1 Abstract........................................................................................................... 97
4.2 Introduction .................................................................................................... 98
4.3 Methods ........................................................................................................ 103
4.4 Results .......................................................................................................... 109
4.5 Management Recommendations .................................................................. 124
4.6 Acknowledgements ...................................................................................... 126
4.7 References .................................................................................................... 127
4.8 Appendix 1 ................................................................................................... 132
4.9 Appendix 2 ................................................................................................... 133
Chapter 5. Fire severity patterns alter spatial and temporal movement patterns
of an arboreal marsupial, the Mountain Brushtail Possum Trichosurus
cunninghamii ............................................................................................................... 141 5.1 Abstract......................................................................................................... 143
5.2 Introduction .................................................................................................. 144
5.3 Methods ........................................................................................................ 148
5.4 Results .......................................................................................................... 154
5.5 Discussion..................................................................................................... 154
5.6 Acknowledgements ...................................................................................... 167
5.7 References .................................................................................................... 168
5.8 Appendix 1. .................................................................................................. 168
Chapter 6. Spatially managing fire in forests for biodiversity: concepts, current
practices and future challenges .................................................................................. 173 6.1 Abstract......................................................................................................... 175
6.2 Introduction .................................................................................................. 177
6.3 Methods ........................................................................................................ 181
6.4 Results .......................................................................................................... 183
6.5 Discussion..................................................................................................... 190
Chapter 7. Conclusions ........................................................................................... 203
Appendices ......................................................................... Error! Bookmark not defined.
17
Chapter 1. Context Statement
1.1 Introduction
Fire is a core earth system process and an integral part of carbon, nutrient and vegetation
cycling (Bowman et al. 2009). Fire-prone vegetation communities cover 40% of the earth’s
land surface (Chapin et al. 2011). Fire is a core process in the life cycles of many species,
including as an important driver of plant reproduction (Weir et al. 2000; Burton et al. 2008;
Smith et al. 2014). Fire prone systems are often adapted to particular fire regimes (Gill 1975).
The spatial and temporal patterns of fire regimes occurrence within landscapes are a principal
factor influencing species distributions and a core driver of biodiversity (Bradstock et al.
2002; Lindenmayer et al. 2014).
However, climate change, altered land uses and detrimental management practices are
changing the distribution, frequency, scale and intensity of large wildfires globally
(McKenzie et al. 2004; Scholze et al. 2006; Westerling et al. 2006). Climate change is
predicted to increase forest fire activity by 10-50% (Flannigan et al. 2000). This poses a
major challenge to biodiversity conservation and management as ecosystems adapt to novel
patterns of fire occurrence. Altered fire regimes are one of three core drivers of biodiversity
decline alongside habitat loss and the impacts of invasive species (Evans et al. 2011).
Inappropriate fire regimes can change the structure and composition of ecological
communities, increasing the risk of species extirpations and extinctions (Gill and Bradstock
1995; Fisher et al. 2009).
Landscape-scale wildfires are an important form of ecosystem disturbance globally due to
their effects on habitat structure and resource availability (Smucker et al. 2005; Bowman et
18
al. 2009). The variegated configurations of habitat often produced following large fires are
commonly held to be essential for the persistence of fauna, due to an increase in niche
availability (Bradstock et al. 2002; Turner et al. 2003; Bond et al. 2005). These mosaics
consist of areas with differing fire regimes (fire intensity, time since last fire, mean fire
interval, seasonality and type (Gill 1975) and spatial properties (size, shape, isolation,
landscape context). Complex habitat mosaics are potentially important for biodiversity
conservation as they may enable both fire tolerant and fire phobic species to persist within
disturbed landscapes (Parr and Andersen 2006; Kelly et al. 2012; Taylor et al. 2012; Nimmo
et al. 2013).
The occurrence of fire refuges
Within fire mosaics, areas which experience disturbance regimes unique from those
prevailing in the surrounding landscape may act as refuges, reducing the impacts of fire on
species and increasing their likelihood of survival (Mackey et al. 2002; Lindenmayer et al.
2009b; Robinson et al. 2013). The occurrence of refuges may contribute to ecosystem
resilience and their presence within landscapes is likely to be linked to post-fire successional
trajectories (Mackey et al. 2012; Banks et al. 2015). Fire refuges may occur at a range of
spatial scales, as a function of landscape-scale habitat patterns, patch-scale areas of intact
habitat or as unburnt individual habitat features within the extent of large wildfires (Watson
et al. 2012; Robinson et al. 2013; Leonard et al. 2014). At the landscape scale, refuges may
occur as large areas of unburnt or low severity burnt forest (Figure 1.A), which retain
sufficient resources to support large populations of flora and fauna following fire, for
example, as fire shadows and sheltered aspects of mountains (Ager et al. 2007; Thompson
and Spies 2010). At the patch scale, fire refuges may occur as unburnt or low severity burnt
patches, remnants skips or isolated of habitat (Figure 1. B) which retain sufficient resources
19
to allow individuals to persist (Stuart-Smith et al. 2002; Burton et al. 2008). Fine scale
refuges may include sheltered features such as hollow trees, rocks or burrows (Figure 1. C)
which enable species to survive the immediate impacts of fire (Brennan et al. 2011). These
types of refuges are likely to occur in parts of landscapes which experience low severity,
long-interval fire regimes (Mackey et al. 2002; Leonard et al. 2014). It is postulated that fire
refuges perform three functions; enabling species to survive the immediate impacts of fire,
facilitating long term species survival in-situ and acting as a sources for re-colonization of
burnt landscapes as regenerating habitat becomes suitable for species (Robinson et al. 2013).
The ability of differing types of refuges to conserve biodiversity within burnt landscapes is
likely to be dependent upon their physical and biotic attributes and their spatial
characteristics.
Figure 1. Examples of potential fire refuge structures at the landscape (A) patch (B) and
feature (C) scales.
The influnce of fire refuge spatial attributes on their use by fauna
Understanding faunal responses to fire refuge attributes is essential to determining their role
in facilitating individual and population persistence and survival within extensively burnt
A C B
20
landscapes (Clarke 2008). The ability of individuals to persist within refuge patches will be
influenced by the spatial properties of that patch (Berry et al. 2015c). For fire refuges to
enable faunal survival in-situ, unburnt remnant patches must be of adequate size to provide
sufficient resources opportunities and preserve ecological processes necessary for long-term
population persistence, such as mating opportunities and low competitive pressures
(MacArthur and Wilson 1967; Fahrig 2003). Refuge use may be dependent both on the
presence of essential resources and the availability of complementary resources within
appropriate dispersal distance (Driscoll 2005; Kelly et al. 2012; Sitters et al. 2015).
Faunal response mechanisms and the importance of the spatial attributes of fire refuges
The landscape context of unburnt refuges, such as their distance to other refuge areas or
surrounding disturbance severity may influence their use by fauna (Ricketts 2001;
Lindenmayer et al. 2002). Faunal responses to landscape context in recently burnt landscapes
are likely to be determined in the short term by functional traits such as dispersal ability and
in the long-term by demographic processes such as the ability of species to form meta-
populations and avoid competitive pressures (Lindenmayer et al. 2002; Bradstock et al. 2005;
Elliott et al. 2012; Lindenmayer et al. 2013b). However, very few studies have attempted to
quantify the desirable spatial attributes of fire mosaics for faunal conservation (Clarke 2008).
The extent of relevant literature on each topic is detailed in the introduction of each chapter.
21
Figure 2. Thesis structure and sequence of paper
This thesis aimed to quantify the ecological role of fire refuges in landscapes disturbed by
extensive large-scale wildfires (Figure. 2). To achieve this, I examined three key areas of
theory relating to fire refuges. First, I aimed to identify the factors responsible for the
distribution of refuges within extensively burnt landscapes, establish how predictive
landscape models can be used by land managers in fire planning and determine how the
distribution of refuges is likely to be influenced by changing fire regimes. Secondly, I aimed
to understand how particular types of refuges and their landscape contexts influenced the
distribution of arboreal marsupials and birds within the extent of a large wildfire. Thirdly, I
sought to understand the mechanisms underpinning faunal distributions by examining how
Context Statement, Aims and methodology and summary of results
Part 1
Fire refuge occurence
Paper 1
Identifying the location of fire refuges in wet forest ecosystems
Part 2
Influence of fire refuges on faunal distribution
Paper 2
The use of topographic fire refuges by the greater glider (Petauroides volans) and the mountain brushtail possum (Trichosurus cunninghami)
following a landscape-scale fire.
Paper 3
Bird use of fire refuges is contingent on landscape context and the spatial extent of mixed severity fire
Part 3
Faunal response mechanisms to fire refuges
Paper 4
Fire severity patterns alter spatial and temporal movement patterns of an arboreal marsupial, the Mountain Brushtail Possum Trichosurus
cunninghamii
Spatial fire managementPaper 5
Spatially managing fire in forests for biodiversity: concepts, current practices and future challenges
Synthesis and Conclusions
22
the spatial extent of high severity wildfire influenced temporal and spatial patterns of faunal
movement. In the final chapter, I investigated various mechanisms for incorporating spatial
fire ecology principles into fire management practices to establish whether it is possible to
spatially manage landscapes for ecologically beneficial fire patterns, and if this can be
achieved within current policy frameworks.
1.2 Overview of aims, methodologies and key results
Paper I: Identifying the location of fire refuges in wet forest ecosystems
In paper I, I investigated how landscape topography and vegetation mediate the patterns of
fire severity generated by wildfire under differing weather conditions and whether we can use
predictive landscape fire models to identify areas of the landscape where fire refuges are
likely to occur. Understanding how physical and biotic landscape factors determine the
distribution of potential refuges is a fundamental step in establishing their potential role as an
agent of faunal distribution and survival in extensively burnt landscapes (Robinson et al.
2013; Leonard et al. 2014). This is particularly relevant in the mountain ash, Eucalyptus
regnans forest of the Victorian Central Highlands (VCH) where large-scale high severity
crown-fires and industrial clear fell logging have contributed to the decline of many species
and threaten the future viability of the ecosystem (Lindenmayer et al. 2013a; Burns et al.
2015). I tested the operational validity of a fire simulation model constructed in 2002 which
predicted the occurrence of potential refuges within two water catchments in the VCH by
comparing the predicted patterns of refuge occurrence with fire severity patterns following
the 2009 Black Saturday fires (Mackey et al. 2002).
Using a novel statistical model validation approach, I identified that under extreme fire
conditions, the distribution of fire refuges was limited to only extremely sheltered, fire-
23
resistant regions of the landscape. During extreme fire conditions, fire severity patterns were
largely determined by stochastic factors in our crown-fire adapted system, such as weather,
that could not be predicted by the model. When fire conditions were moderate, physical
landscape properties appeared to mediate fire severity distribution.
The study highlighted that fire refuges are potentially ecologically important outcomes of
large wildfires. Given that montane forest ecosystems are likely to experience altered fire
regimes in the future, it is essential that the broader climatic and spatial domain within which
fire refuges are identified. I suggest that within these envelopes, detrimental land
management practices are minimised to enable the ecological processes relevant to the
establishment and subsequent use of fire refuges to be maintained.
Paper II: The use of topographic fire refuges by the greater glider (Petauroides volans) and
the mountain brushtail possum (Trichosurus cunninghami) following a landscape-scale fire.
This paper questioned the recently emerging theme that small unburnt patches of vegetation
embedded within the extent of large fire act as ‘fire refuges’, facilitating species presence and
survival. I used a replicated observation study to examine how fire severity, patch size and
landscape context influenced the abundance of arboreal marsupials at 48 sites distributed
throughout the Mountain Ash (Eucalyptus regnans) forests of the Victorian Central
Highlands in south-eastern Australia.
This study was the first to directly examine the potential of mesic gullies to act as fire refuges
for arboreal marsupials in crown-fire forest ecosystems. I found the mountain brushtail
possum had a positive response to a particular kind of topographic refuge (unburnt peninsulas
connected to larger areas of unburnt forest), whereas the greater glider had a negative
response to the overall effects of fire in the landscape. The sugar glider and Leadbeater’s
24
possum were present in small unburnt patches and absent from burnt forest. However, given
the complex movement patterns and social interactions of these species, it is unlikely that
small, isolated unburnt forest fragments would provide sufficient resources to facilitate their
long-term persistence in-situ. I highlight the need for a more developed understanding of how
post-fire habitat patterns facilitate species survival within the burnt landscape and subsequent
recolonisation. I suggest further research is required to determine if fire refuges support
viable populations of these species in-situ in the long-term until subsequent recolonisation of
the surrounding regenerated forest can occur.
Paper III: Bird use of fire refuges is contingent on landscape context and the spatial extent of
mixed severity fire
Fire refuges have been identified as an important mechanism influencing the distribution and
persistence of fauna within extensively burnt landscapes. However, very few studies have
examined this relationship or identified desirable refuge characteristics, particularly relating
to responses to the spatial patterns of habitat created by fire. I conducted a replicated
landscape-scale observation study to determine whether areas with fire regimes differing
from those prevailing in the landscape (mesic gullies) acted as refuges for forest birds
following a large fire. I used a detailed model selection approach to examine how bird use of
potential refuges was influenced by the size and connectivity of the gullies, the extent of fire
severities at different scales in the surrounding landscape and the interaction between severity
patterns, vegetation structure and environmental gradients.
I found that unburnt mesic gullies facilitated the retention of forest birds within extensively
burnt montane forest landscapes. I found that many species responded positively to the
occurrence of intact forest patches regardless of their size or connectivity to the unburnt edge.
However, the ability of unburnt mesic gullies to support many species within the landscape
25
was contingent on appropriate proportions of fire severities at different scales in the
surrounding landscape and their interactions with elevation, precipitation, topographic
position and the availability of particular vegetation structures.
This study presents a key advance, in that the effects of fire-induced habitat patterns on the
distribution of fauna varied between areas depending on their spatial relationships with key
biotic and abiotic landscape patterns. To produce ecologically beneficial fire patterns, land
managers must aim to produce mosaics of mixed severity fire which overlap with a range of
biotic and abiotic gradients. This study demonstrates that developing contingent theory by
examining ecological interactions between fire induced habitat patterns and biotic and abiotic
gradients is the key to unravelling complex faunal responses to fire.
Paper IV: Fire severity patterns alter spatial and temporal movement patterns of an arboreal
marsupial, the Mountain Brushtail Possum Trichosurus cunninghamii.
Identifying how severe wildfires influence core faunal processes such as movement is
essential for understanding how predicted future increases in the scale and frequency of such
disturbances will affect ecosystems. This study aimed to understand species’ responses to
landscape-scale wildfire by examining how fire severity influences the spatio-temporal
patterns of movement by an arboreal mammal, the Mountain Brushtail possum, Trichosurus
cunninghamii.
I used GPS telemetry with a replicated landscape scale study design to map the movement of
18 individuals in landscapes burnt to differing extents by a large wildfire. I analysed the
relationship between movement patterns, landscape patterns, and resource availability.
I identified a change in temporal movement patterns in response to fire. In unburnt
landscapes, individuals moved the longest distance early and late in the night and had less
26
overlap in the areas used for foraging and denning. I also found habitat selection was
dependent on the spatial context of fire in the surrounding landscape. My results suggested a
trend for smaller home ranges in high-severity burnt landscapes.
Forest systems recently burnt at high severity may provide suitable habitat for some species,
if protected from subsequent disturbance. However, spatial and temporal patterns of habitat
use and habitat selection differ considerably between burnt and undisturbed landscapes. The
spatial outcomes of ecological disturbances such as wildfires have the potential to alter the
ecosystem function of fauna across large areas.
Paper V: Spatially managing fire in forests for biodiversity: concepts, current practices and
future challenges
Within the fire ecology literature, it is becoming increasingly recognized that the spatial
patterns generated by wildfires have a significant influence on the conservation of
biodiversity. This is particularly relevant to the fire-prone tall forest systems of south-eastern
Australia and the Pacific Northwest of the United States of America. Many spatially-focused
ecological studies conclude with suggested fire management recommendations to maintain or
improve the ecological value of fire-affected landscapes. However, these research findings
are rarely integrated into decision-making processes within fire management organizations or
translated into applied outcomes.
I employed a qualitative research approach to identify the barriers and enablers to spatially
managing fire for biodiversity and developed a conceptual framework to achieve the
integration of spatial approaches to fire into management. I conducted structured interviews
with experts in fire and biodiversity management and research working in fire-prone forest
ecosystems in the Pacific Northwest United States and south-eastern Australia. The trans-
27
pacific nature of this study enabled me to access a broad perspective of views on the spatial
management of fire for biodiversity. I aimed to 1) identify the barriers to successfully
spatially managing fire for biodiversity, 2) develop a conceptual approach to incorporating
spatial fire concepts into current management and research frameworks and 3) identify a set
of key actions require to facilitate the integration of spatial fire management approaches into
current frameworks.
I identified that spatial approaches to fire management must co-exist within a complex
system of social and ecological feedbacks between landscapes, academic research, socio-
political land management systems and environmental pressures. My findings suggest that
spatially managing fire can be achieved through a number of refinements to existing
processes. These steps relate to developing community understanding of fire science,
improving the relevance of fire research outputs to land management, amending existing
government policy approaches and refining management tools, structures, scales and
monitoring to meet biodiversity and fire risk objectives.
28
1.3 Concluding remarks
The study of fire refuges as a mechanism governing faunal distributions and persistence
within burnt landscapes is an emerging concept in fire ecology (Mackey et al. 2002;
Robinson et al. 2013). Identifying the desirable spatial characteristics and outcomes of fire
for faunal conservation has been highlighted as a research priority and an essential
component in mitigating the detrimental impacts of altered fire regimes on biodiversity
(Bradstock et al. 2005; Clarke 2008; Driscoll et al. 2010a).
My PhD research demonstrates how the occurrence of fire refuges and fire severity
heterogeneity are an integral component in the post-fire ecology of forested systems. My
work identified the factors responsible for the establishment of potential refuges, identified
how the type, spatial attributes, landscape context and relationship to environmental gradients
influenced faunal distributions, examined movement as a mechanism governing faunal
response patterns and examined how understanding of the spatial consequences of fire on
biota can be incorporated into contemporary fire management. I summarise how my work
contributes to the theoretical and applied understanding of fire refuges and post-fire
landscape patterns and their influences of faunal distribution and persistence in fire-prone
forests.
29
1.4 References
Ager, A. A., M. A. Finney, B. K. Kerns, and H. Maffei. 2007. Modeling wildfire risk to
northern spotted owl (Strix occidentalis caurina) habitat in Central Oregon, USA.
Forest Ecology and Management 246:45-56.
Banks, S. C., T. Lorin, R. E. Shaw, L. McBurney, D. Blair, M. D. Blyton, A. L. Smith, J. C.
Pierson, and D. B. Lindenmayer. 2015. Fine‐scale refuges can buffer demographic
and genetic processes against short‐term climatic variation and disturbance: a 22‐year
case study of an arboreal marsupial. Molecular Ecology.
Berry, L. E., D. B. Lindenmayer, and D. A. Driscoll. 2015b. Large unburnt areas, not small
unburnt patches, are needed to conserve avian diversity in fire‐prone landscapes.
Journal of Applied Ecology 52:486-495.
Bond, W. J., F. I. Woodward, and G. F. Midgley. 2005. The global distribution of ecosystems
in a world without fire. New Phytologist 165:525-538.
Bowman, D., J. K. Balch, P. Artaxo, W. J. Bond, J. M. Carlson, M. A. Cochrane, C. M.
D'Antonio, R. S. DeFries, J. C. Doyle, S. P. Harrison, F. H. Johnston, J. E. Keeley, M.
A. Krawchuk, C. A. Kull, J. B. Marston, M. A. Moritz, I. C. Prentice, C. I. Roos, A.
C. Scott, T. W. Swetnam, G. R. van der Werf, and S. J. Pyne. 2009. Fire in the Earth
System. Science 324:481-484.
Bradstock, R. A., M. Bedward, A. M. Gill, and J. S. Cohn. 2005. Which mosaic? A landscape
ecological approach for evaluating interactions between fire regimes, habitat and
animals. Wildlife Research 32:409-423.
Bradstock, R. A., J. E. Williams, and A. M. Gill. 2002. Flammable Australia: the fire regimes
and biodiversity of a continent. Cambridge University Press.
Brennan, K. E. C., M. L. Moir, and R. S. Wittkuhn. 2011. Fire refugia: The mechanism
governing animal survivorship within a highly flammable plant. Austral Ecology
36:131-141.
Burns, E. L., D. B. Lindenmayer, J. Stein, W. Blanchard, L. McBurney, D. Blair, and S. C.
Banks. 2015. Ecosystem assessment of mountain ash forest in the Central Highlands
of Victoria, south‐eastern Australia. Austral Ecology.
Burton, P. J., M. Parisien, J. A. Hicke, R. J. Hall, and J. T. Freeburn. 2008. Large fires as
agents of ecological diversity in the North American boreal forest. International
Journal of Wildland Fire 17:754-767.
Chapin, F. S., P. A. Matson, and P. Vitousek. 2011. Principles of terrestrial ecosystem
ecology. Springer Science & Business Media.
Clarke, M. F. 2008. Catering for the needs of fauna in fire management: science or just
wishful thinking? Wildlife Research 35:385-394.
Driscoll, D. A. 2005. Is the matrix a sea? Habitat specificity in a naturally fragmented
landscape. Ecological Entomology 30:8-16.
Driscoll, D. A., D. B. Lindenmayer, A. F. Bennett, M. Bode, R. A. Bradstock, G. J. Cary, M.
F. Clarke, N. Dexter, R. Fensham, G. Friend, M. Gill, S. James, G. Kay, D. A. Keith,
C. MacGregor, H. P. Possingham, J. Russel-Smith, D. Salt, J. E. M. Watson, D.
Williams, and A. York. 2010. Resolving conflicts in fire management using decision
theory: asset-protection versus biodiversity conservation. Conservation Letters 3:215-
223.
Elliott, C., D. Lindenmayer, S. Cunningham, and A. Young. 2012. Landscape context affects
honeyeater communities and their foraging behaviour in Australia: implications for
plant pollination. Landscape Ecology 27:393-404.
30
Evans, M. C., J. E. Watson, R. A. Fuller, O. Venter, S. C. Bennett, P. R. Marsack, and H. P.
Possingham. 2011. The spatial distribution of threats to species in Australia.
BioScience 61:281-289.
Fahrig, L. 2003. Effects of habitat fragmentation on biodiversity. Annual Review of Ecology
Evolution and Systematics 34:487-515.
Fisher, J. L., W. A. Loneragan, K. Dixon, J. Delaney, and E. J. Veneklaas. 2009. Altered
vegetation structure and composition linked to fire frequency and plant invasion in a
biodiverse woodland. Biological Conservation 142:2270-2281.
Flannigan, M. D., B. J. Stocks, and B. Wotton. 2000. Climate change and forest fires. Science
of the total environment 262:221-229.
Gill, A., and R. Bradstock. 1995. Extinction of biota by fires. Conserving biodiversity: threats
and solutions:309-322.
Gill, A. M. 1975. Fire and the Australian flora: a review. Australian forestry 38:4-25.
Kelly, L. T., D. G. Nimmo, L. M. Spence-Bailey, R. S. Taylor, S. J. Watson, M. F. Clarke,
and A. F. Bennett. 2012. Managing fire mosaics for small mammal conservation: a
landscape perspective. Journal of Applied Ecology 49:412-421.
Leonard, S. W., A. F. Bennett, and M. F. Clarke. 2014. Determinants of the occurrence of
unburnt forest patches: Potential biotic refuges within a large, intense wildfire in
south-eastern Australia. Forest Ecology and Management 314:85-93.
Lindenmayer, D., D. Blair, L. McBurney, S. Banks, J. Stein, R. Hobbs, G. Likens, and J.
Franklin. 2013a. Principles and practices for biodiversity conservation and restoration
forestry: a 30 year case study on the Victorian montane ash forests and the critically
endangered Leadbeater's Possum. Australian Zoologist 36:441-460.
Lindenmayer, D. B., W. Blanchard, L. McBurney, D. Blair, S. C. Banks, D. Driscoll, A. L.
Smith, and A. M. Gill. 2013b. Fire severity and landscape context effects on arboreal
marsupials. Biological Conservation 167:137-148.
Lindenmayer, D. B., W. Blanchard, L. McBurney, D. Blair, S. C. Banks, D. A. Driscoll, A. L.
Smith, and A. M. Gill. 2014. Complex responses of birds to landscape-level fire
extent, fire severity and environmental drivers. Diversity and Distributions 20:467-
477.
Lindenmayer, D. B., R. B. Cunningham, C. F. Donnelly, H. Nix, and B. D. Lindenmayer.
2002. Effects of forest fragmentation on bird assemblages in a novel landscape
context. Ecological Monographs 72:1-18.
Lindenmayer, D. B., C. MacGregor, J. T. Wood, R. B. Cunningham, M. Crane, D. Michael,
R. Montague-Drake, D. Brown, M. Fortescue, and N. Dexter. 2009. What factors
influence rapid post-fire site re-occupancy? A case study of the endangered Eastern
Bristlebird in eastern Australia. International Journal of Wildland Fire 18:84-95.
MacArthur, R. H., and E. O. Wilson. 1967. The Theory of Island Biogeography. Princeton
University Press, Princeton.
Mackey, B., S. Berry, S. Hugh, S. Ferrier, T. D. Harwood, and K. J. Williams. 2012.
Ecosystem greenspots: identifying potential drought, fire, and climate-change micro-
refuges. Ecological Applications 22:1852-1864.
Mackey, B., D. Lindenmayer, M. Gill, M. McCarthy, and J. Lindesay. 2002. Wildlife, Fire
and Future Climate: A Forest Ecosystem Analysis. CSIRO publishing, Collingwood.
McKenzie, D., Z. E. Gedalof, D. L. Peterson, and P. Mote. 2004. Climatic change, wildfire,
and conservation. Conservation Biology 18:890-902.
Nimmo, D., L. Kelly, L. Spence-Bailey, S. Watson, R. Taylor, M. Clarke, and A. Bennett.
2013. Fire Mosaics and Reptile Conservation in a Fire‐Prone Region. Conservation
Biology 27:345-353.
31
Parr, C. L., and A. N. Andersen. 2006. Patch mosaic burning for biodiversity conservation: a
critique of the pyrodiversity paradigm. Conservation Biology 20:1610-1619.
Ricketts, T. H. 2001. The matrix matters: Effective isolation in fragmented landscapes.
American Naturalist 158:87-99.
Robinson, N. M., S. W. Leonard, E. G. Ritchie, M. Bassett, E. K. Chia, S. Buckingham, H.
Gibb, A. F. Bennett, and M. F. Clarke. 2013. REVIEW: Refuges for fauna in fire‐prone landscapes: their ecological function and importance. Journal of Applied
Ecology 50:1321-1329.
Scholze, M., W. Knorr, N. W. Arnell, and I. C. Prentice. 2006. A climate-change risk
analysis for world ecosystems. Proceedings of the National Academy of Sciences
103:13116-13120.
Sitters, H., J. Di Stefano, F. J. Christie, M. Swan, and A. York. 2015. Bird functional
diversity decreases with time since disturbance: does patchy prescribed fire enhance
ecosystem function? Ecological Applications.
Smith, A. L., D. Blair, L. McBurney, S. C. Banks, P. S. Barton, W. Blanchard, D. A.
Driscoll, A. M. Gill, and D. B. Lindenmayer. 2014. Dominant Drivers of Seedling
Establishment in a Fire-Dependent Obligate Seeder: Climate or Fire Regimes?
Ecosystems 17:258-270.
Smucker, K. M., R. L. Hutto, and B. M. Steele. 2005. Changes in bird abundance after
wildfire: importance of fire severity and time since fire. Ecological Applications
15:1535-1549.
Stuart-Smith, K., I. T. Adams, and K. W. Larsen. 2002. Songbird communities in a pyrogenic
habitat mosaic. International Journal of Wildland Fire 11:75-84.
Taylor, R. S., S. J. Watson, D. G. Nimmo, L. T. Kelly, A. F. Bennett, and M. F. Clarke. 2012.
Landscape-scale effects of fire on bird assemblages: does pyrodiversity beget
biodiversity? Diversity and Distributions 18:519-529.
Thompson, J. R., and T. A. Spies. 2010. Factors associated with crown damage following
recurring mixed-severity wildfires and post-fire management in southwestern Oregon.
Landscape Ecology 25:775-789.
Turner, M. G., W. H. Romme, and D. B. Tinker. 2003. Surprises and lessons from the 1988
Yellowstone fires. Frontiers in Ecology and the Environment 1:351-358.
Watson, S. J., R. S. Taylor, D. G. Nimmo, L. T. Kelly, M. F. Clarke, and A. F. Bennett. 2012.
The influence of unburnt patches and distance from refuges on post-fire bird
communities. Animal Conservation:n/a-n/a.
Weir, J. M. H., E. A. Johnson, and K. Miyanishi. 2000. Fire frequency and the spatial age
mosaic of the mixed-wood boreal forest in western Canada. Ecological Applications
10:1162-1177.
Westerling, A. L., H. G. Hidalgo, D. R. Cayan, and T. W. Swetnam. 2006. Warming and
earlier spring increase western US forest wildfire activity. Science 313:940-943.
32
33
Chapter 2. Identifying the location of fire refuges in wet
forest ecosystems
Berry, L. E., Driscoll, D. A., Stein, J. A., Blanchard, W., Banks, S. C., Bradstock, R. A., &
Lindenmayer, D. B. (2015). Identifying the location of fire refuges in wet forest ecosystems.
Ecological Applications, 25:8, 2337-2348
34
35
2.1 Abstract
The increasing frequency of large, high-severity fires threatens the survival of old-growth
specialist fauna in fire-prone forests. Within topographically diverse montane forests, areas
which experience less severe or fewer fires compared with those prevailing in the landscape
may present unique resource opportunities enabling old-growth specialist fauna to survive.
Statistical landscape models which identify the extent and distribution of potential fire
refuges may assist land managers to incorporate these areas into relevant biodiversity
conservation strategies.
We used a case study in an Australian wet montane forest to establish how predictive fire
simulation models can be interpreted as management tools to identify potential fire refuges.
We examined the relationship between the probability of fire refuge occurrence as predicted
by an existing fire refuge model and fire severity experienced during a large wildfire. We also
examined the extent to which local fire severity was influenced by fire severity in the
surrounding landscape. We used a combination of statistical approaches including
generalised linear modelling, variogram analysis and receiver operating characteristics and
area under the curve analysis (ROC AUC).
We found that the amount of unburnt habitat and the factors influencing the retention and
location of fire refuges varied with fire conditions. Under extreme fire conditions, the
distribution of fire refuges was limited to only extremely sheltered, fire-resistant regions of
the landscape. During extreme fire conditions, fire severity patterns were largely determined
by stochastic factors that could not be predicted by the model. When fire conditions were
moderate, physical landscape properties appeared to mediate fire severity distribution.
Our study demonstrates that land managers can employ predictive landscape fire models to
identify the broader climatic and spatial domain within which fire refuges are likely to be
present. It is essential that within these envelopes, forest is protected from logging, roads and
36
other developments so that the ecological processes related to the establishment and
subsequent use of fire refuges are maintained.
37
2.2 Introduction
Landscape-scale high severity fire can alter ecosystem structure and extent across large areas
(Bradstock et al. 2005, Bowman et al. 2009). The increasing frequency of these events is
predicted to continue with climate change (McKenzie et al. 2004). This presents a challenge
to faunal conservation in fire-prone ecosystems, as species’ survival becomes dependent on
the limited distribution of suitable habitat in fire-modified landscapes (Driscoll et al. 2010).
However, within the extent of large fires, local variation in fire severity may preserve critical
resources for fauna that depend on unburnt habitat for foraging and denning (Mackey et al.
2002). These fire refuges may facilitate species survival in-situ following extensive wildfires
(Whelan 1995, Mackey et al. 2002, Robinson et al. 2013).
Intact habitat patches within the boundaries of large fires may provide essential resources to
facilitate species survival until the surrounding landscape can be successfully recolonized
(Stuart-Smith et al. 2002, Bradstock et al. 2005, Cook and Holt 2006, Castro et al. 2010). The
importance of refuges in facilitating survival will vary between species, and is dependent on
whether refuges provide critical resources which are absent from the surrounding landscape
(Robinson et al. 2013). Fire refuges may be especially important for fauna which are
dependent on mature vegetation features, such as tree-hollows for nesting or denning (Banks
et al. 2011b). The likelihood of a location acting as a refuge will depend on individual species
characteristics such as competitive behaviour and dispersal ability (Brown et al. 2013). Fire
refuges may ensure that ecosystem functions provided by species remain in the landscape
(Nugent et al. 2014). These functions may remain absent for successive generations if
recolonization occurs gradually from ex-situ areas (Banks et al. 2011a).
The occurrence of unburnt refuges may depend on two sets of processes. Refuge
establishment may occur as a result of stochastic fire behaviours unique to individual events
(Robinson et al. 2013). Alternatively, refuge formation may be attributable to deterministic
38
processes influenced by physical variation in the landscape (Lindenmayer et al. 1999,
Bradstock et al. 2010, Robinson et al. 2013). Both stochastic and deterministic refuges may
enable the short-term persistence of fauna by sheltering individuals from the immediate
effects of fire (Leonard et al. 2014). Deterministic refuges may enable the survival of species
sensitive to short fire return intervals (Robinson et al. 2014). Such areas enable important
biological legacies to remain in the landscape (Franklin et al. 2000).
Deterministic fire refuges can form in response to topographic characteristics such as
elevation and aspect, and fire-vegetation interactions such as vegetation type, stand age and
fire return interval (Mackey et al. 2002). In fire-prone ecosystems, interactions between fire
and topography can be a dominant driver of the distribution and extent of different
vegetation communities (Wood et al. 2011). For example, within topographically diverse
montane forest landscapes, fire-sensitive vegetation communities are generally restricted to
sheltered gullies and areas of lower elevation (Lindenmayer et al. 2009b). However, under
extreme fire conditions, the physical and topographic attributes of the landscape may exert
less of an influence on fire severity patterns as a wider range of fuels become available to
fires (Turner and Romme, 1994).
Under extreme fire conditions, the distribution of potential fire refuges may be limited to only
the most sheltered parts of the landscape (Mackey et al. 2002). Following a large fire in
Victoria (SE Australia), only ~1% of the total area within the fire boundary presented unburnt
refuge areas > 1 ha in size (Leonard et al. 2014). The conservation of rare, old-growth
dependent species in fire-prone montane forests may be dependent on the retention of larger
areas of intact, unburnt habitat (Lindenmayer et al. 2013). Therefore, it is essential that land
managers are able to predict the occurrence of potential fire refuges, in order to incorporate
them into relevant biodiversity management strategies.
39
Contemporary fire management planning rarely includes consideration of the mechanisms,
such as fire refuges, which may allow species to persist in landscapes following large-scale
wildfires (Clarke 2008). The primary objective of most fire management efforts in montane
forests is to preserve property and infrastructure (DSE 2012). Intense land-use practices, such
as industrial clear-fell logging, can compound the negative effects of fire on biodiversity
(Lindenmayer et al. 2011). However, management practices which encourage connectivity
between habitat patches within production forests may have positive biodiversity outcomes
(Lindenmayer 1994). The inclusion of fire refuges in land management planning may greatly
increase biodiversity retention following landscape-scale fires (Robinson et al. 2013).
Statistical landscape models which predict the occurrence of potential fire refuges, may help
land managers to identify and protect areas of the landscape of high conservation value
(Mackey et al. 2012).
A small number of studies have used models in an attempt to predict the potential distribution
of fire refuges (Camp et al. 1997, Mackey et al. 2002, Wood et al. 2011). These models are
developed from a number of landscape-level variables such as vegetation type, climatic
conditions, fuel loads, soil wetness, and topography (Gill et al. 1987). However, these
predictive models are often based upon a suite of theoretical assumptions. These include
setting fire weather conditions as constant (Bradstock et al. 2010) and overlooking the
influence of land-use practices on fire behaviour at the landscape scale (Taylor et al. 2014a).
Models predicting the outcome of large wildfires are rarely evaluated using data collected
following actual fire events.
In this study we compared the outcomes of a predictive fire model with fire severity data
collected following a large wildfire. Mackey et al. (2002) developed a predictive model of
fire refuges in the forests of the Victorian Central Highlands, Australia. In February 2009, the
40
Kilmore East - Murrindindi fire complex burnt ~ 250,000ha of this region (Leonard et al.
2014, Robinson et al. 2014), providing a unique opportunity to test the earlier predictions
about fire refuges. We asked (1) how do areas in the landscape predicted to act as fire refuges
mediate the severity of large fires? And (2) how does the predicted distribution of fire refuges
vary under different fire conditions? We expected to identify a positive relationship between
the modelled probability of refuge occurrence and the scale and the presence of low severity
fire. We also expected regions of high severity, crown fire to be correlated with a lower
probability of refuge occurrence.
Table. 1. Explanation of topographic and vegetative properties at each end of the
predicted refuge class probability scale derived from Mackey et al. (2002).
Predicted Refuge Class Topographic and vegetative characteristics
1
Low percentile mean fire interval (<100
years), low probability of multi-agedness
(<25%). lower mean TWI, higher elevation
percentile, higher mean annual temperature
9
90-100% percentile mean fire interval (>500
years), high probability of multi-agedness
(>65%), higher mean TWI, lower elevation
percentile, lower mean annual temperature
41
2.3 Methods
Study area, fire conditions and fire severity data
Mackey et al. (2002) modelled the probability of fire refuge occurrence in the Maroondah
and O’Shannassy water catchments located in the Victorian Central Highlands (VCH), north-
east of Melbourne, Australia (see Appendix. 1). This region was chosen as it contains strong
environmental gradients and topographically variable areas of high relief upon which to
calculate model projections. We limited our analyses of the Mackey et al, (2002) model to
areas within the boundaries of O’Shannassy and Maroondah water catchments. Within each
catchment, only areas within the extent of the 2009 fire boundary were analysed (Fig. 1). This
allowed the potentially confounding effects of logging and other land-uses to be minimized,
because the catchments are largely unlogged and uncleared.
The 2009 Black Saturday fires occurred following a period of protracted drought (Teague et
al. 2010). Wind speeds during these fires reached 57 kilometres per hour (Tolhurst et al.
2010). The interaction between a period of prolonged drought, consecutive days of
temperatures exceeding 43°C and large stands of predominantly single-aged 1939 regrowth
forest (the dominant forest age class in both catchments) created conditions conducive for
high intensity crown-fires (Teague et al. 2010, Taylor et al. 2014b). Each catchment was
subject to fires burning under different weather conditions as measured by the McArthur
Forest Fire Danger Index (FFDI; Noble et al. 1980). The O’Shannassy water catchment was
burnt during a ‘Catastrophic’ weather period (ie. FFDI > 100), categorized by rapidly
moving, uncontrollable fire (Teague et al. 2010). The Maroondah catchment was burnt by a
slower moving, ‘Moderate’ class fire ( i.e. FFDI < 10) during the period following a
Southerly weather change in the evening of the 7th February prior to midnight, which brought
strong winds, high humidity and low temperatures (Price and Bradstock 2012, Engel et al.
2013). This provided an opportunity to test the performance of the Mackey et al. (2002)
42
model under different fire conditions. We tested the Mackey et al (2002) model using data
from fire severity maps produced by the Department of Sustainability and Environment
(DSE), Victoria. Fire severity maps were produced at a scale of 1:25,000 from SPOT satellite
imagery using the Normalised Burn Ratio (NBR) index (DSE 2009).
Short summary of the Mackey et al. (2002) modelling approach
Mackey et al, (2002) combined survey data of vegetation community composition at sites
distributed widely across the Maroondah and O’Shannassy water catchments and pre-existing
GIS layers to generate spatial predictions of potential fire refuge occurrence. These spatial
predictions were expressed as a map describing the probability of a location remaining
unburnt (Fig. 1). Vegetation survey data from long-term research sites in the study region
were used to compare the spatial distribution of vegetation types to environmental gradients,
such as elevation, slope and aspect. Mapped GIS data for the O’Shannassy and Maroondah
catchments enabled these comparisons to be projected across the landscape. The presence or
absence of different forest types was correlated with a series of spatially explicit
environmental gradients using topographic environmental domain analysis (TEDA), a GIS-
based data analysis technique. The gradients were: Mean annual temperature, elevation
percentile, short-wave radiation, topographic-wetness index, elevation, aspect, catchment
area, elevation difference from mean and slope. Forest type was classified according to
species composition and stand age. The TEDA results provided a model of the probability
that a location supports old-growth forest. These estimates were converted to estimates of the
mean interval between stand-replacing fires (Johnson and Gutsell 1994). The gridded
probabilities generated by the multi-agedness model derived from analysis of the site-based
data were combined with the longest mean fire interval models produced from the TEDA
analyses to predict the probability of a location being a refuge for arboreal marsupials
(Mackey et al. 2002). The Mackey et al, (2002) fire refuge probability modelling procedure
43
produced a raster grid of refuge potential with cells attributed values of increasing probability
scaled from 1 to 9, where 1 corresponds to a low probability of the location remaining
unburnt and 9 corresponds to a high probability of the location remaining unburnt (Table. 1).
Ground-truthing the remotely sensed DSE fire severity map
We independently ground-truthed the accuracy of the DSE fire severity maps using field
observations of fire severity obtained from fire-affected long-term research sites
(Lindenmayer et al. 2014a). This step was necessary to quantify the accuracy of the DSE fire
severity mapping. The NBR accurately classifies areas of high severity fire, which are
characterised by substantial changes in canopy structure (Cocke et al. 2005). However, the
NBR approach to fire severity mapping may underestimate understorey burn severity when
the above canopy remains intact (Roy et al. 2006). To quantify the extent of misclassification,
ground-truthing sites were selected across the study region, to account for site-specific
variation in topography, vegetation and local fire conditions. We calculated the proportion of
sites where fire severity was correctly identified by the DSE fire severity maps, and the
proportions of sites where the measures were different by one and two categories. To reduce
the likelihood of fire severity misclassification influencing the outcomes of our analyses, we
pooled DSE severity categories 4 and 5 (understorey burn with canopy intact and both
understorey and canopy intact) for our analyses of low severity fire (Table 3). For further
details of the ground-truthing process see Lindenmayer at al. (2010).
Spatial Dependence
We conducted Moran’s I tests for spatial-autocorrelation in the fire severity maps using the
‘Spatial Autocorrelation’ tool in ArcGIS 10.1 (ESRI, 2012). To address any spatial
dependence in our logistic regression models, we included a spatially lagged response
variable (SLRV) as an auto-covariate (Haining 2003). To determine the appropriate scale of
SLRV to use within each catchment, we measured the influence of SLRVs at different scales
44
on the spatial dependence of 2009 fire severity using variogram analysis. The SLRV was
calculated as the mean fire severity of the points surrounding each grid cell using the ‘focal
mean’ function in ArcGIS 10.1 (ESRI, 2012). The ‘sill’ of a variogram is the semi-variance
value at which the fitted line plateaus. The variogram ‘range’ is defined as the distance at
which the sill is reached. The range is the greatest distance at which a point can be considered
related to its surroundings. Spatial dependence is evident when a clear sill is reached within
the range considered in each variogram. We calculated the SLRV at different scales to test
the extent to which spatial dependence should be considered. These were; the total areas of
the surrounding 4, 8, 120 and 2600 cells. This allowed us to consider the spatial influence of
the surrounding cells at 20m (surrounding 4 or 8 cells), 100m and 500m on the fire severity
of each focal cell (20 m2). A 500m measure is consistent with the SLRV approach described
in Price and Bradstock (2012), who calculated that the mean gully width was ~500 m across
all of the Victorian Central Highlands fire complexes. Moran’s I examines global spatial
auto-correlation across each data layer. Whereas, our variogram analysis (Table. 2) examined
auto-correlation using a SLRV at different local levels (20 m, 100 m and 500 m).
Generalised linear models
The variogram analyses indicated a high level of spatial dependence in fire severity within
each catchment, at the 20 m scale (Table. 2). Therefore, to achieve independence between our
sample points, we used a sub-set of our data points. Based on the results of our variogram
analysis, we selected each point at least 40 m apart. This is a common method for accounting
for spatial dependence in ecological data (Haining 2003). We used binomial generalised
linear models to determine the relationship between fire severity and refuge probability class.
We fitted crown fire and low severity fire as response variables and predicted refuge class as
the predictor variable (Table. 3). Analyses were conducted in the R statistical environment (R
Core Team 2012).
45
Receiver operating characteristics (ROC) and area under the curve (AUC) analysis
To visualise the performance of the predicted probability of fire refuge occurrence as a
successful classifier of fire severity (as ‘Crown fire” and “Low Severity fire”), we
constructed Receiver Operating Characteristics (ROC) graphs (Fawcett 2006). ROC was used
over simple classification accuracy measures as it enabled the comparison of different
classification systems (Hand and Till, 2001). ROC is preferred over cross-validation
techniques because, for cross-validation to occur, an arbitrary threshold needs to be selected
from the qualifying data to determine if a site is ‘occupied’ or not (Price and Ferrier, 2000).
We used area under the curve (AUC) analysis to test whether the model will rank a randomly
chosen positive instance higher than a randomly chosen negative instance (Fawcett 2006). An
AUC value of 1 can be interpreted as a 100% prediction rate, whereas, an AUC value of 0.5
indicates an equal number of successful and unsuccessful classifications (Worster et al.
2006). AUC has been described as a misleading measure in assessing the performance of
predictive distribution models (Lobo et al. 2008). It is therefore necessary to interpret these
results in unison with the auto-logistic regression models. Each ROC refuge probability class
figure must be interpreted independently, as the analysis ignores goodness-of-fit, p-values
and spatial dependence (Lobo et al. 2008).
We constructed ROC graphs and used the AUC to determine the ability of each refuge
probability class to categorize crown and low severity fire. To do this, we binomially
reclassified each refuge class. The class of interest was reclassified as ‘1’ (cases) and all
others as ‘0’ (controls). We then calculated the ROC using the package ‘RORC’ in R
development software (Sing et al. 2005). Refuge probability class was fitted as the predictor
variable with crown fire and low severity fire fitted separately as the response variable. This
determined the extent to which each refuge probability class accurately classified both crown
and low severity fire. This was repeated for both of the water catchments we targeted for
46
study. Used in unison with the auto logistic models, this approach enabled the identification
of individual refuge classes that reliably predicted areas of crown or low severity fire. The
ROC AUC analysis enabled individual refuge classes which were strong predictors of crown
and low severity fire to be identified. We predicted that refuge class 9 (high probability of a
location being unburnt) would be a strong predictor of low severity fire and refuge class 1
(low probability of a location being unburnt) would be a strong predictor of crown fire. We
predicted refuge classes 2 to 8 to be weaker predictors of crown and low severity fire.
Figure 1. Map displaying the probability of fire refuge occurrence in the Maroondah
(left) and O'Shannassy (right) catchments. Figure adapted from Mackey et al. 2002.
Green represents areas of low fire severity (potential fire refuges), where red represents
areas of high fire severity (dominant trees killed).
47
Figure 2. Fire severity distribution in the Maroondah (left) and O'Shannassy (right)
catchments. Data taken from DSE (2009) SPOT satellite imagery. Figure shows extent
of the 2009 fires within each catchment only, unburnt areas are not included.
48
2.4 Results
Our ground-truthing of the remotely sensed DSE fire severity map data indicated that 81% of
grid cells were accurately classified. Of the sites incorrectly classified, 14% were by a
misclassification distance of one category and 3.6% by two categories.
Moran’s I tests for Spatial Autocorrelation
Fire severity across the O’Shannassy water catchment was highly spatially dependent
(Moran’s Index: 0.79, Expected index: 0, Variance 0, z-score 1837.08, p-value < 0.001).
Probability of fire refuge occurrence, as derived from the Mackey et al. (2002) model, also
was highly spatially dependent (Moran’s Index: 0.7, Expected index -0, z-score 1640.84, p-
value <0.001).
Variograms
The variogram analysis indicated that both ‘focal mean 4’ and ‘focal mean 8’ (the mean
values of the surrounding 4 and 8 cells, each cell was 20 m x 20 m) SLRVs effectively
accounted for spatial dependence in the O’Shannassy catchment (Table. 2). The values for
the Maroondah catchment indicated a similarity in fire severity values throughout the
landscape, which was independent of localized spatial dependence (See appendix 2. for
variogram figures).
Generalised linear models
In the O’Shannassy catchment, probability of crown fire was highest (~48%) in refuge class 1
and lowest in refuge class 9 (~2%) (Figure. 3). There was a non-linear response to crown fire
between refuge classes 2 and 8 (Figure. 3). The highest probability for low severity fire was
found in refuge class 9 (~92%). The lowest probability of low severity fire was recorded in
refuge class 1 (~10%). There was a non-linear response to low severity fire between refuge
classes 2 and 8 (Figure. 3).
49
In the Maroondah catchment probability of crown fire was highest (~9%) in refuge class 1
and lowest in refuge class 9 (~0%) (Figure. 3). The low probabilities of crown fire in the
Maroondah catchment were related to the relatively low frequency of crown fire experienced.
The probability of each refuge class experiencing low severity fire was similar across refuge
classes 2- 9 in the Maroondah catchment (Figure. 3). Refuge class 1 experienced the lowest
probability of low severity fire (~80%). The high probabilities of low severity across all
refuge classes in the Maroondah catchment was related to the relatively high frequency of
low severity fire experienced (Figure. 4).
ROC AUC
The ROC AUC analyses for the O’Shannassy catchment indicated that refuge probability
class 1 accurately classified crown fire distribution (Figure. 5). Refuge probability class 9
accurately classified the distribution of low severity fire in the O’Shannassy catchment
(Figure. 6). No individual refuge probability class in the Maroondah catchment accurately
classified either crown fire or low severity fire (See Appendix 3).
50
Table 2. Summary of range and sill values for variograms using spatially lagged
response variables at different scales to account for spatial dependence. In the
O’Shannassy catchment, ‘Focal mean 4’ and ‘Focal mean 8’ (fire severity in the
neighbouring cells, within 20 m) effectively accounted for spatial dependence. In the
Maroondah catchment, fire severity was independent of local spatial dependence.
Catchment Fire severity SLRV Range (m) Sill
O’Shannassy Crown Fire None >6000 0.3
Fm4 1500 0.12
Fm8 1500 0.12
Fm120 2500 0.18
Fm2600 >6000 0.25
O’Shannassy Low Severity None >6000 0.25
Fm4 1000 0.06
Fm8 1000 0.06
Fm120 2,800 0.13
Fm2600 >6000 0.25
Maroondah Crown Fire None 6000 0.085
Fm4 6000 0.035
Fm8 6000 0.035
Fm120 6000 0.085
Fm2600 6000 0.085
Maroondah Low Severity None 6000 0.158
Fm4 6000 0.048
Fm8 6000 0.05
Fm120 6000 0.158
Fm2600 6000 0.158
Table 3. List of fire severity response variables and spatially lagged predictor variables
used in auto-logistic models. Each grid cell used to construct the SLRV measured 20 m2.
Variable Description
DSE fire severity categories 1. Crown burn
2. Crown scorch
3. Moderate crown scorch
4. Light or no crown scorch, understory burnt
5. No crown scorch, no understory burn
Crown fire 1 = DSE fire severity classes 1+2
0= classes 3-5
Low severity fire 1= DSE fire severity classes 4+5
0= classes 1-3
Spatially lagged response variable (SLRV) Focal mean (mean fire severity of
surrounding cells)
fm4= surrounding 4 cells
fm8= surrounding 8 cells
fm120=surrounding 120 cells
fm2600= surrounding 2600 cells
51
Figure 3. Probability of crown and low severity fire occurrence per predicted refuge
class in the O’Shannassy and Maroondah water catchments. The X axis indicates
Refuge Probability Class as taken from the Mackey et al. (2002) model.
0.0
0.2
0.4
1 2 3 4 5 6 7 8 9
Fire Refuge Probability Class
Pro
babili
ty o
f cro
wn f
ire
O’Shannassy crown fire
0.4
0.6
0.8
1.0
1 2 3 4 5 6 7 8 9
Fire Refuge Probability Class
Pro
ba
bili
ty o
f lo
w s
eve
rity
fire
O’Shannassy low severity fire
0.000
0.025
0.050
0.075
0.100
1 2 3 4 5 6 7 8 9
Fire Refuge Probability Class
Pro
babili
ty o
f cro
wn f
ire
Maroondah crown fire
0.80
0.85
0.90
0.95
1.00
1 2 3 4 5 6 7 8 9
Fire Refuge Probability Class
Pro
ba
bili
ty o
f lo
w s
eve
rity
fire
Maroondah low severity fire
52
Figure 4. The frequency of refuge probability grid squares for each class (top) and
frequency of each fire severity class (bottom) for each water catchment. Both water
catchments were predicted to return high frequencies of potential refuge areas
following fire. The observed frequency of fire severity in both catchments indicates that
the Maroondah catchment was exposed to predominantly low severity fire. The inverse
was observed in the O’Shannassy catchment. Note that the frequency of extreme high
severity fire in the O’Shannassy catchment was relatively low.
53
Figure 5. ROC curves displaying crown fire classification accuracy for each refuge
probability class in the O’Shannassy catchment Specificity represent the false positive
rate. Sensitivity represents the true positive rate. The grey line indicates a random
response and the black line the performance of each refuge class in accurately
predicting crown fire. AUC denotes Area Under the Curve.
AUC: 0.61 AUC: 0.38 AUC: 0.38
AUC: 0.44 AUC: 0.55 AUC: 0.42
AUC: 0.53 AUC: 0.46 AUC: 0.31
54
Figure 6. ROC curves displaying low severity fire classification accuracy for each refuge
probability class in the O’Shannassy catchment.
AUC: 0.43 AUC: 0.53 AUC: 0.62
AUC: 0.58 AUC: 0.45 AUC: 0.53
AUC: 0.45 AUC: 0.49 AUC: 0.79
55
2.5 Discussion
Fire refuges may mitigate the detrimental effects of large fires on fauna habitat, by providing
resources unavailable in the surrounding burnt landscape (Robinson et al. 2013).
Management actions that preserve potential fire refuges are relevant to biodiversity
conservation in montane forests globally, as the scale and frequency of natural and
anthropogenic disturbances increases (Lindenmayer et al. 2014b). We used a case study to
determine how models which predict the distribution of potential fire refuges can be
interpreted by land managers to identify fire refuge areas to target for management of
biodiversity values in fire prone forests. Our findings indicate that in extreme fire conditions,
the presence of fire refuges is limited to extremely sheltered parts of the landscape. The high
variability in fire severity in areas with moderate probabilities of being a fire refuge is
indicative of the central role played by fire weather in determining post-fire outcomes in
extreme conditions. It is essential that within potential fire refuge envelopes, detrimental
land management practices are minimised, and where possible, areas are protected to enable
the ecological processes relevant to the establishment and subsequent use of fire refuges to be
maintained (Lindenmayer and McCarthy 2002).
Do predicted fire refuges mediate the severity of large fires?
Our study found that modelled fire refuges were strong predictors of fire severity. The
occurrence of potential fire refuges was limited to areas with an extremely high probability of
refuge occurrence (refuge class 9). These fire refuges are characterised by deep, sheltered
topography in mesic gullies, and late-successional vegetation communities (Mackey et al.
2002). These deterministic properties sufficiently moderated fire severity, enabling the
persistence of ecologically significant habitat features, such as large hollow-bearing trees
(Taylor and Skinner 1998, Lindenmayer et al. 2012b). This is comparable to findings in the
56
Boreal forests of Canada and Alaska, where vegetation types of relatively low flammability
were associated with areas of low severity fire (Burton et al. 2008). Our findings suggest that
extremes of topography and wetness were the principal contributing factors to fire refuge
retention. These regions contribute to the establishment of landscape-wide variation in fire
severity, which may facilitate species’ survival in-situ (Robinson et al. 2013, Leonard et al.
2014).
How does the predicted distribution of fire refuges vary under different fire conditions?
Under extreme fire conditions, fire severity was highly variable in all but the most
confidently predicted refuge classes. In intermediate refuge classes (2-8), the effects of minor
topographic or vegetative variation on exposed slopes had a minimal influence on fire
severity. In the sub-alpine forests of North America, fire intensity and crown fire initiation
were strongly related to weather conditions immediately preceding or during the fire (Bessie
and Johnson 1995). Areas classified less confidently on the refuge probability scale ( refuge
classes 2-8) were more likely to be located on more exposed slopes (Mackey et al. 2002). Fire
severity in these areas was primarily influenced by weather conditions on the day of the fire
than by their physical and topographic properties. It is likely that the highly variable nature of
fire weather was responsible for the range of fire severity responses observed across these
moderate predicted classes (Bradstock et al. 2010, Price and Bradstock 2012, Sharples et al.
2012).
During moderate fire conditions, fire severity appeared to be topographically mediated, with
little evidence of any effects of fire weather. Forest stands which experienced moderate
severity or understorey burns only, may lose foliage but are unlikely to be killed by fire
(Chafer et al. 2004). These areas may still present critical resources necessary to the survival
of many specialist forest species (Smith and Lindenmayer 1988). Therefore, following a brief
57
period where canopy recovery may occur, stands burnt at moderate severity may continue to
provide vital resources that facilitate faunal persistence (Smucker et al. 2005).
Management implications
Our study demonstrates that landscape managers can use predictive fire models constructed
from digital elevation models, vegetation community distribution and fire history maps to
reliably identify fire refuges. The relatively limited distribution of these refuges increases the
need for management actions to ensure their protection (Leonard et al. 2014).
To ensure the ecological processes relevant to their establishment and subsequent use by
fauna are maintained, fire managers need to plan for the spatial outcomes of large fires. Our
variogram analyses indicate that under extreme fire conditions the occurrence of low severity
fire was spatially dependent on the fire severity in the surrounding landscape, up to 1 km
(Table. 2). Intense land uses such as logging can increase fire severity in different forest types
(Thompson et al. 2007, Krawchuk and Cumming 2009). Recently logged forests burned at
higher severity than older forest stands (Taylor et al. 2014a). Additionally, clear-fell and
salvage logging practices reduce the quality and extent of habitat across large areas and have
the potential to fragment populations (Hutto and Gallo 2006, Lindenmayer et al. 2009a).
Therefore, we recommend that logging activities should be relocated from areas within a
buffer distance from potential fire refuges. The size of these buffers should be based upon the
known ranges and dispersal habits of the old-growth dependent fauna which may use fire
refuges (Lindenmayer and Possingham 1996, Pope et al. 2004). Practices which encourage
habitat connectivity within these disturbed landscapes may have positive biodiversity
outcomes (Gibbons and Lindenmayer 1996, Lindenmayer et al.2000, Lindenmayer et al.
2006).
58
2.6 Conclusions
Following large-scale wildfires in montane forests, areas of the landscape persist and may act
as fire refuges. These areas are ecologically significant, as they can facilitate the presence of
old-growth dependent species within extensively burnt landscapes (Whelan 1995, Mackey et
al. 2002, Robinson et al. 2013). Following crown fires under extreme conditions, fire refuges
will only occur in the most sheltered parts of the landscape. To maintain the processes
leading to the establishment and subsequent use of fire refuges it is essential that land
management practices which may escalate fire risk and reduce species' use of refuges, such as
logging are excluded from potential refuge areas. Our findings demonstrate that land
management agencies can employ predictive landscape models as decision-making tools to
map the distribution of fire refuge envelopes enabling their prioritization as areas of
significant conservation value.
59
2.7 References
Banks, S. C., M. Dujardin, L. McBurney, D. Blair, M. Barker, and D. B. Lindenmayer.
2011a. Starting points for small mammal population recovery after wildfire:
recolonisation or residual populations? Oikos 120:26-37.
Banks, S. C., E. J. Knight, L. McBurney, D. Blair, and D. B. Lindenmayer. 2011b. The
Effects of Wildfire on Mortality and Resources for an Arboreal Marsupial: Resilience
to Fire Events but Susceptibility to Fire Regime Change. Plos One 6.
Bessie, W., and E. Johnson. 1995. The relative importance of fuels and weather on fire
behavior in subalpine forests. Ecology 76:747-762.
Bowman, D., J. K. Balch, P. Artaxo, W. J. Bond, J. M. Carlson, M. A. Cochrane, C. M.
D'Antonio, R. S. DeFries, J. C. Doyle, S. P. Harrison, F. H. Johnston, J. E. Keeley, M.
A. Krawchuk, C. A. Kull, J. B. Marston, M. A. Moritz, I. C. Prentice, C. I. Roos, A.
C. Scott, T. W. Swetnam, G. R. van der Werf, and S. J. Pyne. 2009. Fire in the Earth
System. Science 324:481-484.
Bradstock, R. A., M. Bedward, A. M. Gill, and J. S. Cohn. 2005. Which mosaic? A landscape
ecological approach for evaluating interactions between fire regimes, habitat and
animals. Wildlife Research 32:409-423.
Bradstock, R. A., K. A. Hammill, L. Collins, and O. Price. 2010. Effects of weather, fuel and
terrain on fire severity in topographically diverse landscapes of south-eastern
Australia. Landscape Ecology 25:607-619.
Brown, S. M., K. A. Harrisson, R. H. Clarke, A. F. Bennett, and P. Sunnucks. 2013. Limited
Population Structure, Genetic Drift and Bottlenecks Characterise an Endangered Bird
Species in a Dynamic, Fire-Prone Ecosystem. Plos One 8:e59732.
Burton, P. J., M. Parisien, J. A. Hicke, R. J. Hall, and J. T. Freeburn. 2008. Large fires as
agents of ecological diversity in the North American boreal forest. International
Journal of Wildland Fire 17:754-767.
Camp, A., C. Oliver, P. Hessburg, and R. Everett. 1997. Predicting late-successional fire
refugia pre-dating European settlement in the Wenatchee Mountains. Forest Ecology
and Management 95:63-77.
Castro, J., G. Moreno-Rueda, and J. A. Hodar. 2010. Experimental Test of Postfire
Management in Pine Forests: Impact of Salvage Logging versus Partial Cutting and
Nonintervention on Bird-Species Assemblages. Conservation Biology 24:810-819.
Chafer, C. J., M. Noonan, and E. Macnaught. 2004. The post-fire measurement of fire
severity and intensity in the Christmas 2001 Sydney wildfires. International Journal of
Wildland Fire 13:227-240.
Clarke, M. F. 2008. Catering for the needs of fauna in fire management: science or just
wishful thinking? Wildlife Research 35:385-394.
Cocke, A. E., P. Z. Fulé, and J. E. Crouse. 2005. Comparison of burn severity assessments
using Differenced Normalized Burn Ratio and ground data. International Journal of
Wildland Fire 14:189-198.
Cook, W. M., and R. D. Holt. 2006. Fire frequency and mosaic burning effects on a tallgrass
prairie ground beetle assemblage. Biodiversity and Conservation 15:2301-2323.
Driscoll, D. A., D. B. Lindenmayer, A. F. Bennett, M. Bode, R. A. Bradstock, G. J. Cary, M.
F. Clarke, N. Dexter, R. Fensham, G. Friend, M. Gill, S. James, G. Kay, D. A. Keith,
C. MacGregor, J. Russell-Smith, D. Salt, J. E. M. Watson, R. J. Williams, and A.
York. 2010. Fire management for biodiversity conservation: Key research questions
and our capacity to answer them. Biological Conservation 143:1928-1939.
60
DSE. 2009. Remote sensing guideline for assessing landscape-scale fire severity in Victoria’s
forest estate. Unpublished technical manual., Department of Sustainability and
Environment, Melbourne.
DSE. 2012. Code of Practice for Bushfire Management on Public Land. Victorian
Government Department of Sustainability and Environment, Melbourne.
Engel, C. B., T. P. Lane, M. J. Reeder, and M. Rezny. 2013. The meteorology of Black
Saturday. Quarterly Journal of the Royal Meteorological Society 139:585-599.
Fawcett, T. 2006. An introduction to ROC analysis. Pattern recognition letters 27:861-874.
Franklin, J. F., D. Lindenmayer, J. A. MacMahon, A. McKee, J. Magnuson, D. A. Perry, R.
Waide, and D. Foster. 2000. Threads of continuity. Conservation in Practice 1:8-17.
Gibbons, P., and D. Lindenmayer. 1996. Issues associated with the retention of hollow-
bearing trees within eucalypt forests managed for wood production. Forest Ecology
and Management 83:245-279.
Gill, A., K. Christian, P. Moore, and R. Forrester. 1987. Bushfire incidence, fire hazard and
fuel reduction burning. Australian Journal of Ecology 12:299-306.
Haining, R. P. 2003. Spatial data analysis: theory and practice. Cambridge University Press.
Hutto, R. L., and S. M. Gallo. 2006. The effects of postfire salvage logging on cavity-nesting
birds. Condor 108:817-831.
Johnson, E. A., and S. L. Gutsell. 1994. Fire frequency models, methods and interpretations.
Advances in ecological research 25.
Krawchuk, M. A., and S. G. Cumming. 2009. Disturbance history affects lightning fire
initiation in the mixedwood boreal forest: observations and simulations. Forest
Ecology and Management 257:1613-1622.
Leonard, S. W., A. F. Bennett, and M. F. Clarke. 2014. Determinants of the occurrence of
unburnt forest patches: Potential biotic refuges within a large, intense wildfire in
south-eastern Australia. Forest Ecology and Management 314:85-93.
Lindenmayer, D. 1994. Wildlife Corridors and the Mitigation of Logging Impacts on Fauna
in Wood-Production Forests in South-Eastern Australia: a Review. Wildlife Research
21:323-340.
Lindenmayer, D., D. Blair, L. McBurney, and S. Banks. 2010. Forest Phoenix: How a Great
Forest Recovers After Wildfire. CSIRO PUBLISHING.
Lindenmayer, D., J. Franklin, and J. Fischer. 2006. General management principles and a
checklist of strategies to guide forest biodiversity conservation. Biological
Conservation 131:433-445.
Lindenmayer, D., B. Mackey, I. Mullen, M. McCarthy, A. Gill, R. Cunningham, and C.
Donnelly. 1999. Factors affecting stand structure in forests–are there climatic and
topographic determinants? Forest Ecology and Management 123:55-63.
Lindenmayer, D., and M. A. McCarthy. 2002. Congruence between natural and human forest
disturbance: a case study from Australian montane ash forests. Forest Ecology and
Management 155:319-335.
Lindenmayer, D., and H. Possingham. 1996. Modelling the inter-relationships between
habitat patchiness, dispersal capability and metapopulation persistence of the
endangered species, Leadbeater's possum, in south-eastern Australia. Landscape
Ecology 11:79-105.
Lindenmayer, D. B., W. Blanchard, L. McBurney, D. Blair, S. Banks, G. E. Likens, J. F.
Franklin, W. F. Laurance, J. A. Stein, and P. Gibbons. 2012a. Interacting factors
driving a major loss of large trees with cavities in a forest ecosystem. Plos One
7:e41864.
61
Lindenmayer, D. B., W. Blanchard, L. McBurney, D. Blair, S. C. Banks, D. Driscoll, A. L.
Smith, and A. M. Gill. 2013. Fire severity and landscape context effects on arboreal
marsupials. Biological Conservation 167:137-148.
Lindenmayer, D. B., W. Blanchard, L. McBurney, D. Blair, S. C. Banks, D. A. Driscoll, A. L.
Smith, and A. M. Gill. 2014a. Complex responses of birds to landscape-level fire
extent, fire severity and environmental drivers. Diversity and Distributions 20:467-
477.
Lindenmayer, D. B., R. J. Hobbs, G. E. Likens, C. J. Krebs, and S. C. Banks. 2011. Newly
discovered landscape traps produce regime shifts in wet forests. Proceedings of the
National Academy of Sciences of the United States of America 108:15887-15891.
Lindenmayer, D. B., M. L. Hunter, P. J. Burton, and P. Gibbons. 2009a. Effects of logging on
fire regimes in moist forests. Conservation Letters 2:271-277.
Lindenmayer, D. B., W. F. Laurance, and J. F. Franklin. 2012b. Global decline in large old
trees. Science 338:1305-1306.
Lindenmayer, D. B., W. F. Laurance, J. F. Franklin, G. E. Likens, S. C. Banks, W. Blanchard,
P. Gibbons, K. Ikin, D. Blair, and L. McBurney. 2014b. New policies for old trees:
averting a global crisis in a keystone ecological structure. Conservation Letters 7:61-
69.
Lindenmayer, D. B., C. R. Margules, and D. B. Botkin. 2000. Indicators of biodiversity for
ecologically sustainable forest management. Conservation Biology 14:941-950.
Lindenmayer, D. B., J. T. Wood, D. Michael, M. Crane, C. MacGregor, R. Montague-Drake,
and L. McBurney. 2009b. Are gullies best for biodiversity? An empirical examination
of Australian wet forest types. Forest Ecology and Management 258:169-177.
Lobo, J. M., A. Jiménez‐Valverde, and R. Real. 2008. AUC: a misleading measure of the
performance of predictive distribution models. Global Ecology and Biogeography
17:145-151.
Mackey, B., S. Berry, S. Hugh, S. Ferrier, T. D. Harwood, and K. J. Williams. 2012.
Ecosystem greenspots: identifying potential drought, fire, and climate-change micro-
refuges. Ecological Applications 22:1852-1864.
Mackey, B., D. Lindenmayer, M. Gill, M. McCarthy, and J. Lindesay. 2002. Wildlife, Fire
and Future Climate: A Forest Ecosystem Analysis. CSIRO publishing, Collingwood.
Manly, B. F. 2006. Randomization, bootstrap and Monte Carlo methods in biology. CRC
Press.
McCarthy, M. A., and D. B. Lindenmayer. 1998. Multi-aged mountain ash forest, wildlife
conservation and timber harvesting. Forest Ecology and Management 104:43-56.
McKenzie, D., Z. e. Gedalof, D. L. Peterson, and P. Mote. 2004. Climatic change, wildfire,
and conservation. Conservation Biology 18:890-902.
Noble, I., A. Gill, and G. Bary. 1980. McArthur's fire‐danger meters expressed as equations.
Australian Journal of Ecology 5:201-203.
Nugent, D. T., S. W. Leonard, and M. F. Clarke. 2014. Interactions between the superb
lyrebird (Menura novaehollandiae) and fire in south-eastern Australia. Wildlife
Research 41:203-211.
Pope, M. L., D. B. Lindenmayer, and R. B. Cunningham. 2004. Patch use by the greater
glider (Petauroides volans) in a fragmented forest ecosystem. I. Home range size and
movements. Wildlife Research 31:559-568.
Price, O. F., and R. A. Bradstock. 2012. The efficacy of fuel treatment in mitigating property
loss during wildfires: Insights from analysis of the severity of the catastrophic fires in
2009 in Victoria, Australia. Journal of Environmental Management 113:146-157.
R Core Team, R. 2012. R: A language and environment for statistical computing.
62
Robinson, N. M., S. W. Leonard, E. G. Ritchie, M. Bassett, E. K. Chia, S. Buckingham, H.
Gibb, A. F. Bennett, and M. F. Clarke. 2013. REVIEW: Refuges for fauna in fire‐
prone landscapes: their ecological function and importance. Journal of Applied
Ecology 50:1321-1329.
Robinson, N. M., S. W. J. Leonard, A. F. Bennett, and M. F. Clarke. 2014. Refuges for birds
in fire-prone landscapes: The influence of fire severity and fire history on the
distribution of forest birds. Forest Ecology and Management 318:110-121.
Roy, D. P., L. Boschetti, and S. N. Trigg. 2006. Remote sensing of fire severity: assessing the
performance of the normalized burn ratio. Geoscience and Remote Sensing Letters,
IEEE 3:112-116.
Sharples, J. J., R. H. McRae, and S. R. Wilkes. 2012. Wind–terrain effects on the propagation
of wildfires in rugged terrain: fire channelling. International Journal of Wildland Fire
21:282-296.
Sing, T., O. Sander, N. Beerenwinkel, and T. Lengauer. 2005. ROCR: visualizing classifier
performance in R. Bioinformatics 21:3940-3941.
Smith, A., and D. Lindenmayer. 1988. Tree hollow requirements of Leadbeater's possum and
other possums and gliders in timber production ash forests of the Victorian Central
Highlands. Wildlife Research 15:347-362.
Smucker, K. M., R. L. Hutto, and B. M. Steele. 2005. Changes in bird abundance after
wildfire: importance of fire severity and time since fire. Ecological Applications
15:1535-1549.
Stuart-Smith, K., I. T. Adams, and K. W. Larsen. 2002. Songbird communities in a pyrogenic
habitat mosaic. International Journal of Wildland Fire 11:75-84.
Taylor, A. H., and C. N. Skinner. 1998. Fire history and landscape dynamics in a late-
successional reserve, Klamath Mountains, California, USA. Forest Ecology and
Management 111:285-301.
Taylor, C., M. A. McCarthy, and D. B. Lindenmayer. 2014a. Non‐linear effects of stand age
on fire severity. Conservation Letters.
Taylor, D., M. Tharapos, and S. Sidaway. 2014b. Downward accountability for a natural
disaster recovery effort: Evidence and issues from Australia's Black Saturday. Critical
Perspectives on Accounting.
Taylor, R. S., S. J. Watson, D. G. Nimmo, L. T. Kelly, A. F. Bennett, and M. F. Clarke. 2012.
Landscape-scale effects of fire on bird assemblages: does pyrodiversity beget
biodiversity? Diversity and Distributions 18:519-529.
Teague, B., R. McLeod, and S. Pascoe. 2010. Final report, 2009 Victorian bushfires royal
commission. Parliament of Victoria, Melbourne Victoria, Australia.
Thompson, J. R., T. A. Spies, and L. M. Ganio. 2007. Reburn severity in managed and
unmanaged vegetation in a large wildfire. Proceedings of the National Academy of
Sciences 104:10743-10748.
Tolhurst, K., W. Street, and V. Creswick. 2010. Report on fire danger ratings and public
warning. Melbourne, Australia: Victorian Bushfires Royal Commission.
Whelan, R. J. 1995. The ecology of fire. Cambridge University Press.
Wood, S. W., B. P. Murphy, and D. M. J. S. Bowman. 2011. Firescape ecology: how
topography determines the contrasting distribution of fire and rain forest in the south-
west of the Tasmanian Wilderness World Heritage Area. Journal of Biogeography
38:1807-1820.
Worster, A., J. Fan, and S. Upadhye. 2006. Understanding receiver operating characteristic
(ROC) curves. Cjem 8:19-20.
63
2.8 Appendix 1
Map of study region in South Eastern Australia, indicating fire severity within the 2009
Kinglake West and Kilmore-Murrindindi fire complexes. Blue shapes indicate
Maroondah (left) and O’Shannassy (right) water catchments.
64
Map displaying the incidence of fire severity class ‘5’ (canopy and understorey intact) in
the O’Shannassy (right) and Maroondah (left) catchments.
Map displaying the Incidence of fire severity class '4' (Canopy intact/ partially intact,
understorey burnt) in the Maroondah (left) and O'Shannassy (right) catchments.
65
Map displaying the spatially lagged response variable (SLRV) ‘Fm4’ for Maroondah
(left) and O'Shannassy (right) catchments. The SLRV ‘Fm4’ was constructed from the
mean fire severity of the surrounding 4 points of each grid cell.
66
2.9 Appendix 2
O’Shannassy crown fire variograms
67
O’Shannassy low severity fire variograms
68
Maroondah crown fire variograms
69
Maroondah low severity fire variograms
70
2.10 Appendix 3
ROC AUC curves for the ability of the model to predict crown fire and low severity fire in the
Maroondah water catchment
Figure 3. ROC curves displaying crown fire classification accuracy for each refuge
probability class in the Maroondah catchment. P.Ref.1, AUC: 0.5181, P.Ref.2, AUC:
0.5042, P.Ref.3, AUC: 0.5042, P.Ref.4, AUC: 0.4949, P.Ref.5, AUC: 0.474, P.Ref.6,
AUC: 0.4648, P.Ref.7, AUC: 0.4727, P.Ref.8, AUC: 0.4613, P.Ref.9, AUC: 0.4691
71
Figure 4. . ROC curves displaying low severity fire classification accuracy for each
refuge probability class in the Maroondah catchment. P.Ref.1, AUC: 0.4699, P.Ref.2,
AUC: 0.4937, P.Ref.3, AUC: 0.5007, P.Ref.4, AUC: 0.5162, P.Ref.5, AUC: 0.5544,
P.Ref.6, AUC: 0.5604, P.Ref.7, AUC: 0.5616, P.Ref.8, AUC: 0.5745, P.Ref.9, AUC:
0.5587
72
73
Chapter 3. The use of topographic fire refuges by the
greater glider (Petauroides volans) and the mountain
brushtail possum (Trichosurus cunninghami)
following a landscape-scale fire.
Berry, L. E., Driscoll, D. A., Banks, S. C., & Lindenmayer, D. B. (2015). The use of
topographic fire refuges by the greater glider (Petauroides volans) and the mountain brushtail
possum (Trichosurus cunninghami) following a landscape-scale fire. Australian Mammalogy,
37(1), 39-45.
74
75
3.1 Abstract
We examined the abundance of arboreal marsupials in topographic fire refuges after a major
fire in a stand-replacing crown-fire forest ecosystem. We surveyed arboreal marsupial
abundance across 48 sites in rainforest gullies burnt to differing extents by the 2009 fires in
the Mountain Ash (Eucalyptus regnans) forests of the Victorian Central Highlands, Australia.
The greater glider (Petauroides volans) was less abundant within the extent of the 2009 fire.
The mountain brushtail possum (Trichosurus cunninghami) was more abundant within the
extent of the 2009 fire, particularly within unburnt peninsulas protruding into burnt areas
from unburnt edges. Our results indicate that fire refuges may facilitate the persistence of
some species within extensively burnt landscapes. Additional work should seek to clarify this
finding and identify the demographic mechanisms underlying this response.
76
3.2 Introduction
The frequency and intensity of large fires is increasing globally (Bowman et al. 2009). The
effects of these fires on biodiversity are compounded by climate change-induced shifts in
local and regional fire regimes (Krawchuk et al. 2009). Changing fire regimes can alter the
distribution and availability of essential resources for animals across large spatial extents
(Haslem et al. 2011). These processes have the potential to threaten fauna whose life-cycles
are intrinsically connected to the availability of old-growth habitat features (Lindenmayer et
al. 2010a). As a consequence, determining how the spatial outcomes of large fires influence
the persistence of natural populations has become a priority in contemporary fire
management and research (Driscoll et al. 2010b).
In topographically diverse landscapes, deep sheltered gullies may support fire regimes
different to those prevailing elsewhere in more exposed parts of the landscape (Leonard et al.
2014). Areas which experience a lower probability of fire occurrence or lower intensity fire
are more likely to preserve mature vegetation structures that are often associated with faunal
survival, such as tree-hollows (Mackey et al. 2012). These fire refuges may enable species
survival in otherwise burnt landscapes, and in turn facilitate subsequent recolonization of
adjacent, regenerating areas (Robinson et al. 2013). The retention of unburnt patches
following extensive fires can be instrumental in the survival of faunal groups, such as small
mammals (Pereoglou et al. 2011). However, determining the appropriate spatial attributes of
potential fire refuges from which to develop effective fire management plans has received
insufficient attention to date (Bradstock et al. 2005; Clarke 2008).
The spatial characteristics of unburnt remnants may determine their performance as fire
refuges (Driscoll et al. 2010b). Species with a limited ability to disperse through, and forage
77
in, recently burnt habitat will require sufficient resource and mate availability to persist as
viable populations in situ (Mackey et al. 2002). For species with a greater dispersal capacity,
a network of intact patches may be necessary to maintain ecological processes at the
bioregional scale (Mackey et al. 2012). For other species, fine-scale landscape heterogeneity
may provide greater resource availability and potentially increase population viability
(Bradstock et al. 2005). However, current fire management plans rarely consider how the
spatial outcomes of large fires will influence biodiversity conservation (Clarke 2008). In their
recent study of the factors determining fire refuge distribution in a topographically varied
forest, Leonard et al. (2014) found that potential fire refuges were retained in only 1% of the
burnt landscape and occurred without management interventions, such as fuel-reduction
burning. Do these naturally occurring fire refuges have the potential to enable species to
survive following extensive fires?
We conducted a pilot study to determine the potential conservation value of fire refuges for
arboreal marsupials in stand-replacing crown-fire forest ecosystems of the Victorian Central
Highlands, Australia. We examined how the landscape context, severity and size of potential
fire refuges influenced the abundance of arboreal marsupials. Our study posed two questions.
First, how did the overall effects of fire in the landscape influence the distribution of arboreal
marsupials? Secondly, how did this vary with fire severity, potential fire refuge size, and
landscape situation?
78
3.3 Methods
Study area
Our study was conducted in the Victorian Central Highlands, south-eastern Australia. The
region is topographically diverse and dominated by stands of Mountain Ash (Eucalyptus
regnans) and Alpine Ash (Eucalyptus delegatensis) forest. Mountain Ash forests are adapted
to infrequent large stand-replacing crown fires, with mean intervals of between 75 to 150
years (McCarthy et al. 1999). These forests are located 120 km north-east of Melbourne and
cover an area of approximately 60 km × 80 km between 37°20'–37 ° 55'S and 145° 30'–146°
20'E (Lindenmayer et al. 2014a). See Lindenmayer et al. (2011a) for a full description of the
region including land use and climate. The region is home to a diverse fauna of arboreal
marsupials including the endangered Leadbeater’s possum (Gymnobelideus leadbeateri), the
vulnerable yellow-bellied glider (Petaurus australis), the mountain brushtail possum
(Trichosurus cunninghami), greater glider (Petauroides volans), sugar glider (Petaurus
breviceps), feathertail Glider (Acrobates pygmaeus), common ringtail possum
(Pseudocheirus peregrinus), and eastern pygmy possum (Cercartetus nanus). See
Lindenmayer et al. (2013b) for a description of the arboreal marsupial fauna and their key
ecological attributes. The region was part of the ‘Kilmore-Murrindindi Complex’ burnt
during the 2009 ‘Black Saturday’ bushfires. For a full description of the fire conditions, and
effects see Teague et al. (2010).
Study design and survey methods
We established 48 sites across 6 fire severity classes (Fig. 1). Potential fire refuges were un-
common following the 2009 fires (Leonard et al. 2014). The total number of sites was
therefore limited by the number of replicates available. All sites were located in wet gullies.
These sites were predominantly 1939 regrowth from the Black Friday fires. All sites had not
79
been previously commercially logged, although limited historical selective logging may have
occurred (Lindenmayer et al. 2011a). Sites were selected to maximise similarity in elevation,
topographic position and vegetation type. The Black Saturday fires produced a landscape
mosaic (sensu (Bennett et al. 2006)) with gullies burnt at different fire severities (Bradstock
et al. 2012). Rainforest gullies are less likely to burn at high severity during bushfires
(Lindenmayer et al. 2009d). Therefore, gullies have a greater potential to act as fire refuges
than other parts of the landscape (Mackey et al. 2012).
Within the extent of the fire, we established the following classes; large >3 ha unburnt
patches (maximum size 30 ha), small ~1 ha unburnt patches (both canopy and understorey
unburnt, and sites completely isolated by surrounding forest burnt at high severity), unburnt
peninsulas (connected to the unburnt edge of the fire, but protruding into burnt forest, 3-30
ha), moderate severity sites (canopy partially intact, understorey burnt, >3ha) and high
severity sites (both canopy and understorey burnt, >3ha), (Table 1). Beyond the extent of the
fires, we established 17 comparison sites in gullies unaffected by the 2009 fires. Sites were
spaced at least 1 km apart over a total distance of 52 km (see Figure 1).
We conducted spotlight surveys for arboreal marsupials between January and March 2013.
We surveyed each site for 30 minutes. Each site consisted of a 100 m transect located in the
centre of each patch along the creek line. We searched all habitat and trees up to 50 m either
side of each transect. We surveyed multiple treatments on each night and varied the time at
which different treatments were surveyed to reduce any potential temporal bias. To determine
whether suitable denning habitat was present, we counted the total number of hollow bearing
trees within 50m of each transect. Due to the known affinity between silver wattle and
Leadbeater’s possum we counted the number total number of wattles present within 10 m of
80
each transect (Smith 1984). As the greater glider forages within the canopy, we estimated
canopy cover directly overhead at 0 m, 50 m and 100 m along the transect (Lindenmayer et
al. 1990a). Canopy cover scores were averaged at the site level for analysis. We present the
means and standard errors for each structural habitat variable in Figure 2.
Table 1. The proportional occurrence of each species observed at burnt, unburnt patch
and continuous unburnt (beyond the extent of the 2009 fire) sites. The number of sites
occupied by each species within each class is given in brackets.
Original
Site type
3 sample
site type
2
sample
site
type
number
of sites
Mountain
brushtail
possum
Greater
glider
Sugar
glider
Leadbeater’s
possum
High
severity Burnt
Within
fire 7 0.14 (1) 0.00 (0) 0.00 (0) 0.00 (0)
Moderate
severity Burnt
Within
fire 6 0.17 (1) 0.00 (0) 0.00 (0) 0.00 (0)
Small
patches Patch
Within
fire 5 0.00 (0) 0.20 (1) 0.20 (1) 0.20 (1)
Large
patches Patch
Within
fire 6 0.33 (2) 0.17 (1) 0.00 (0) 0.00 (0)
Peninsulas Patch Within
fire 7 0.57 (4) 0.14 (1) 0.00 (0) 0.00 (0)
Unburnt
continuous
Unburnt
continuous
Outside
fire 17 0.18 (3) 0.29 (5) 0.18 (3) 0.00 (0)
Analysis
Due to the low numbers of observations across our original class groups, we re-classified the
classes for the analysis (Table 2). To determine if species present within the extent of the fire
were more abundant in unburnt patches within the extent of the fire we re-classified sites into
three categories; unburnt patches within the extent of the fire, burnt sites and unburnt sites
beyond the extent of the fire. To determine the overall impacts of fire in the landscape on
species presence, we re-classified sites into two categories; sites within the extent of the fire
and sites located externally to the fire boundary. We used an analysis of variance (ANOVA)
to determine whether the number of hollow-bearing trees, wattles and the proportion of
81
canopy cover differed significantly between our re-classified groups (Iversen and Norpoth
1987).
We analysed the data using independent two-sample and k-sample permutation tests,
(sampling permutation distribution 5000 times). During a permutation test, the distribution of
the original data between classes s is tested against multiple permutations of the data with
treatment labels re-arranged. Permutation tests provide an efficient approach to determining
significance when data are not normally distributed or samples are small (Manly, 1997). The
permutation test assumes that observations are exchangeable under the null hypothesis. The
two-sampled permutation test calculates the P-value from the proportion of sampled
permutations where the difference in means was greater or equal to the means from the
original treatments. We used the two-sample permutation test to calculate the significance
level for the overall impact of fire in the landscape on species presence (original treatments
re-classified into two groups). The k-sample permutation test compares the randomly selected
means from the re-classified permutations with the original treatment means, and can be used
to obtain p-values where more than two treatment categories are used. We used a k-sample
test to determine whether species abundance varied significantly within our original treatment
design and the three levelled re-classification of the data. The null distribution of the test
statistic was determined using Monte-Carlo resampling. The analysis was conducted using
the ‘coin’ package (Zeileis et al. 2008) in the R statistical environment (R Core Team 2012).
82
3.4 Results
We detected 32 individuals from four species across our 48 sites. We recorded only two
species, the greater glider and the mountain brushtail possum in sufficient numbers for
statistical analyses. We detected a significant response for the mountain brushtail possum to
our full design classes (Table. 2, P = 0.042). The mountain brushtail possum appeared more
abundant within the extent of the 2009 fire and proportionally more abundant in unburnt
peninsula gullies at the edge of the fire than within unburnt forest (Table 1). The greater
glider was significantly less abundant at sites within the extent of the fire than at unburnt sites
(Table 2, P = 0.038). The sugar glider and Leadbeater’s possum were detected only very
rarely (Table 1). We found no significant response to patch size for the mountain brushtail
possum or the greater glider. The greater glider, sugar glider and Leadbeater’s possum were
absent from sites burnt at moderate and high severity (Table 1).There were significantly more
hollow bearing trees at sites beyond the extent of the fire (Table 3). Canopy cover differed
significantly between classes (Table 3) and was higher at unburnt sites (Figure 2). There
were as significantly greater number of wattles at burnt sites (Figure 2).
83
Figure 2. Box-plots displaying differences in habitat structures between site types within
each re-classified treatment group. Fs1 = high severity fire, fs3 = moderate severity fire,
lp = large patch, pg = unburnt penninsula, sp = small patch, uc = unburnt control
84
Table. 2. Results of permutation tests. T1 represents the full treatment design, T2 the
three levelled re-classification of the treatments (“burnt”, “unburnt patch”, ‘unburnt
continuous”). T3 the two-levelled re-classification of the treatments (“within fire”,
“beyond fire”). Mountain brushtail and Greater glider response to T1 and T2 was
tested using approximative K-Sample Permutation Tests (Max T). Response to T3 was
tested using an approximative 2-Sample Permutation Test (Z).
Treatment Mountain brushtail possum Greater glider
Max T / Z P Max T / Z P
T1 2.99 0.042 2.09 0.240
T2 1.53 0.288 2.09 0.090
T3 -1.06 0.386 2.09 0.038
Table 3. Relationship between habitat structure and treatment groups. Table displays
results of ANOVA.
Treatment Hollow bearing trees Canopy cover % Number of wattles
F P F P F P
T1 1.20 0.324 8.34 <0.001 5.53 <0.001
T2 2.37 0.105 18.19 <0.001 12.42 <0.001
T3 4.13 0.048 31.57 <0.001 0.92 0.342
85
3.5 Discussion
This study presents a preliminary attempt to identify whether fire refuges are an ecologically
important outcome of large fires for arboreal marsupials. The apparent refuge value of mesic
gullies in burnt landscape varied between species. Of the two species analysed, the mountain
brushtail possum had a positive response to a particular kind of topographic refuge (unburnt
peninsulas connected to larger areas of unburnt forest), whereas the greater glider had a
negative response to the overall effects of fire in the landscape. Our results support findings
that current declines in the abundance of the greater glider, are partially attributable to the
effects of large-scale fires (Lindenmayer et al. 2013b). Given the estimated home range of
this species and the non-overlapping range of males (mean = 2.6 ± 0.8 ha), it is unlikely that
small (~ 1 ha), isolated islands of vegetation embedded within the extent of large stand-
replacing fires will provide sufficient resources to maintain viable populations of this species
in-situ (Pope et al. 2004). This is potentially the case for most species of arboreal marsupial
in these forests. The decline of the greater glider within the extent of the 2009 fires mirrors
similar trends identified by Lindenmayer et al. (2013b) for Leadbeater’s possum
(Gymnobelideus leadbeateri) and the sugar glider (Petaurus breviceps).
The higher abundance of the mountain brushtail possum within unburnt gullies protruding
into the fire from the edge than in those beyond its extent, could be indicative of the process
of landscape re-colonization by the species from unburnt edges (Banks et al. 2011a).
Alternatively, these findings might be associated with a positive edge response, with
individuals concentrated in areas of high foliage density presented by regrowth forest, and the
availability of mature habitat features such as tree hollows (Harding and Gomez 2006).
Compared to other arboreal marsupial species, the mountain brushtail possum was the least
adversely affected by the 2009 fires (Lindenmayer et al. 2013b). The mountain brushtail
86
possum showed no evidence of mortality in radio-collared individuals in the short term
following fire (Banks et al. 2011c). Some post-fire specialist species travel large distances to
access resources available under post-fire conditions (Muona and Rutanen 1994). These
patterns may be driven by the spatial variation in resource availability between post-fire burnt
and unburnt areas. These patterns of movement may be observed in the short term when
species which survive a fire move from resource-poor burnt habitat to resource-rich refuges
or as longer-term recolonisation of burnt areas as habitat suitability improves. Some studies
have suggested that the former (i.e. long-distance movements) is uncommon, and that
individuals are likely to remain within established territories after disturbance (Vernes and
Pope 2001; Dalerum et al. 2007; Lyet et al. 2009; Sanz-Aguilar et al. 2011).
Recent studies have highlighted the importance of fire refuges for taxa such as birds (Taylor
et al. 2012; Robinson et al. 2014) and small mammals (Pereoglou et al. 2011). However, in
Mountain Ash forests, conservation of the arboreal marsupial fauna is intrinsically linked to
the availability of tree hollows for denning (Banks et al. 2011d; Lindenmayer et al. 2014a).
In our study, the sugar glider and Leadbeater’s possum were recorded in small unburnt
patches but were absent from burnt sites. However, the low occurrence of these species
prevents any meaningful conclusions. Given the complex movement patterns and social
interactions of species such as Leadbeater’s possum (Lindenmayer and Possingham 1996)
and the large area requirements of the yellow-bellied glider (Craig 1985), it is unlikely that
small, isolated unburnt forest fragments would provide sufficient resources to facilitate the
long-term persistence of these species in-situ. Following the 2009 Black Saturday bushfires,
potential fire refuges were present in less than 1% of the total area burned (Leonard et al.
2014). The limited availability of potential refuge areas may reduce their ability to act as the
primary mechanism of conserving arboreal marsupials vulnerable to the effects of fire within
87
burnt landscapes. The greater glider and the mountain brushtail are the only species known to
persist in burnt mountain ash forests, the former within or near areas of intact canopy and the
later in burnt areas presenting intact tree hollows (Lindenmayer et al. 2013b). Therefore, it is
likely that these species may be able to recolonise the burnt landscape from in-situ refuges.
For other arboreal marsupial species, eventual movement into burnt landscapes from adjacent
large areas of intact forest appears the most likely source of recolonisation. We suggest that
further research is needed to determine the conservation role of fire refuges in forested
systems, particularly those subject to different fire regimes, where refuges may be more
prevalent.
In their study of arboreal marsupial responses to fire severity in crown-fire montane forests,
Lindenmayer et al. (2013b) concluded that future management actions should focus on the
conservation of large areas of unburnt forest with hollow-bearing trees. Our observations of
greater glider distribution support these recommendations. However, we caution that due to
the limited number of individuals observed, our results should be interpreted with care and
future work is required before specific conclusions and recommendations can be made.
However, our study revealed a number of interesting findings. In particular, we highlight the
need for a more developed understanding of how post-fire habitat patterns facilitate species
survival within the burnt landscape and subsequent recolonisation. Although we observed
limited numbers of arboreal marsupials in fire refuges, whether these patches will enable the
in-situ survival of these species and become eventual sources of recolonisation remains
unknown. We suggest further research is required to determine if fire refuges may support
viable populations of these species in-situ in the long-term until subsequent recolonisation of
the surrounding regenerated forest can occur. Subsequent research should document the
88
patterns of dispersal into burnt areas and subsequent landscape recolonisation of these species
from potential topographic refuges.
89
3.6 Acknowledgements
This manuscript benefited from the comments of two anonymous reviewers. LB would like to
thank Wade Blanchard and Jeff Wood for their statistical advice and Lachlan McBurney and
David Blair for their help in the field. LB was supported through the Australian Research
Council Discovery grant program. Funding for the project was provided by the Fenner School
of Environment and Society. This project was conducted in compliance with Australian
National University Animal Ethics Protocol number: A2012/42.
90
3.7 References
Banks, SC, Dujardin, M, McBurney, L, Blair, D, Barker, M, Lindenmayer, DB (2011a)
Starting points for small mammal population recovery after wildfire: recolonisation or
residual populations? Oikos 120, 26-37.
Banks, SC, Knight, EJ, McBurney, L, Blair, D, Lindenmayer, DB (2011b) The effects of
wildfire on mortality and resources for an arboreal marsupial: resilience to fire events
but susceptibility to fire regime change. Plos One 6, e22952.
Banks, SC, Lindenmayer, DB, McBurney, L, Blair, D, Knight, EJ, Blyton, MDJ (2011c) Kin
selection in den sharing develops under limited availability of tree hollows for a forest
marsupial. Proceedings of the Royal Society B-Biological Sciences 278, 2768-2776.
Bennett, AF, Radford, JQ, Haslem, A (2006) Properties of land mosaics: implications for
nature conservation in agricultural environments. Biological Conservation 133, 250-
264.
Bowman, D, Balch, JK, Artaxo, P, Bond, WJ, Carlson, JM, Cochrane, MA, D'Antonio, CM,
DeFries, RS, Doyle, JC, Harrison, SP, Johnston, FH, Keeley, JE, Krawchuk, MA,
Kull, CA, Marston, JB, Moritz, MA, Prentice, IC, Roos, CI, Scott, AC, Swetnam,
TW, van der Werf, GR, Pyne, SJ (2009) Fire in the Earth System. Science 324, 481-
484.
Bradstock, R, Cary, G, Davies, I, Lindenmayer, D, Price, O, Williams, R (2012) Wildfires,
fuel treatment and risk mitigation in Australian eucalypt forests: insights from
landscape-scale simulation. Journal of Environmental Management 105, 66-75.
Bradstock, RA, Bedward, M, Gill, AM, Cohn, JS (2005) Which mosaic? A landscape
ecological approach for evaluating interactions between fire regimes, habitat and
animals. Wildlife Research 32, 409-423.
Clarke, MF (2008) Catering for the needs of fauna in fire management: science or just
wishful thinking? Wildlife Research 35, 385-394.
Craig, S (1985) Social organization, reproduction and feeding behaviour of a population of
yellow-bellied gliders, Petaurus australis (Marsupialia: Petauridae). Wildlife Research
12, 1-18.
Dalerum, F, Boutin, S, Dunford, JS (2007) Wildfire effects on home range size and fidelity of
boreal caribou in Alberta, Canada. Canadian Journal of Zoology-Revue Canadienne
De Zoologie 85, 26-32.
Driscoll, DA, Lindenmayer, DB, Bennett, AF, Bode, M, Bradstock, RA, Cary, GJ, Clarke,
MF, Dexter, N, Fensham, R, Friend, G, Gill, M, James, S, Kay, G, Keith, DA,
MacGregor, C, Russell-Smith, J, Salt, D, Watson, JEM, Williams, RJ, York, A (2010)
Fire management for biodiversity conservation: Key research questions and our
capacity to answer them. Biological Conservation 143, 1928-1939.
Harding, EK, Gomez, S (2006) Positive edge effects for arboreal marsupials: an assessment
of potential mechanisms. Wildlife Research 33, 121-129.
Haslem, A, Kelly, LT, Nimmo, DG, Watson, SJ, Kenny, SA, Taylor, RS, Avitabile, SC,
Callister, KE, Spence-Bailey, LM, Clarke, MF, Bennett, AF (2011) Habitat or fuel?
Implications of long-term, post-fire dynamics for the development of key resources
for fauna and fire. Journal of Applied Ecology 48, 247-256.
Iversen, GR, Norpoth, H (1987) 'Analysis of variance.' (Sage:
Krawchuk, MA, Moritz, MA, Parisien, M-A, Van Dorn, J, Hayhoe, K (2009) Global
Pyrogeography: the Current and Future Distribution of Wildfire. Plos One 4, e5102.
91
Leonard, SW, Bennett, AF, Clarke, MF (2014) Determinants of the occurrence of unburnt
forest patches: Potential biotic refuges within a large, intense wildfire in south-eastern
Australia. Forest Ecology and Management 314, 85-93.
Lindenmayer, D, Cunningham, R, Tanton, M, Smith, A, Nix, H (1990) Habitat requirements
of the mountain brushtail possum and the greater glider in the montane ash-type
eucalypt forests of the central highlands of Victoria. Wildlife Research 17, 467-478.
Lindenmayer, D, Possingham, H (1996) Modelling the inter-relationships between habitat
patchiness, dispersal capability and metapopulation persistence of the endangered
species, Leadbeater's possum, in south-eastern Australia. Landscape Ecology 11, 79-
105.
Lindenmayer, DB, Barton, PS, Lane, PW, Westgate, MJ, McBurney, L, Blair, D, Gibbons, P,
Likens, GE (2014) An empirical assessment and comparison of species-based and
habitat-based surrogates: A case study of forest vertebrates and large old trees. Plos
One 9, e89807.
Lindenmayer, DB, Blanchard, W, McBurney, L, Blair, D, Banks, SC, Driscoll, D, Smith, AL,
Gill, AM (2013) Fire severity and landscape context effects on arboreal marsupials.
Biological Conservation 167, 137-148.
Lindenmayer, DB, Hobbs, RJ, Likens, GE, Krebs, CJ, Banks, SC (2011) Newly discovered
landscape traps produce regime shifts in wet forests. Proceedings of the National
Academy of Sciences 108, 15887-15891.
Lindenmayer, DB, Steffen, W, Burbidge, AA, Hughes, L, Kitching, RL, Musgrave, W,
Smith, MS, Werner, PA (2010) Conservation strategies in response to rapid climate
change: Australia as a case study. Biological Conservation 143, 1587-1593.
Lindenmayer, DB, Wood, JT, Michael, D, Crane, M, MacGregor, C, Montague-Drake, R,
McBurney, L (2009) Are gullies best for biodiversity? An empirical examination of
Australian wet forest types. Forest Ecology and Management 258, 169-177.
Lyet, A, Cheylan, M, Prodon, R, Besnard, A (2009) Prescribed fire and conservation of a
threatened mountain grassland specialist: a capture-recapture study on the Orsini's
viper in the French alps. Animal Conservation 12, 238-248.
Mackey, B, Berry, S, Hugh, S, Ferrier, S, Harwood, TD, Williams, KJ (2012) Ecosystem
greenspots: identifying potential drought, fire, and climate-change micro-refuges.
Ecological Applications 22, 1852-1864.
Mackey, B, Lindenmayer, D, Gill, M, McCarthy, M, Lindesay, J (2002) 'Wildlife, Fire and
Future Climate: A Forest Ecosystem Analysis.' (CSIRO publishing: Collingwood)
McCarthy, MA, Gill, MA, Lindenmayer, DB (1999) Fire regimes in mountain ash forest:
evidence from forest age structure, extinction models and wildlife habitat. Forest
Ecology and Management 124, 193-203.
Muona, J, Rutanen, I (1994) The short term impact of fire on the beetle fauna in boreal
forests. Annales Zoologici Fennici 31, 109-11.
Pereoglou, F, Macgregor, C, Banks, SC, Ford, F, Wood, J, Lindenmayer, DB (2011) Refuge
site selection by the eastern chestnut mouse in recently burnt heath. Wildlife Research
38, 290-298.
Pope, ML, Lindenmayer, DB, Cunningham, RB (2004) Patch use by the greater glider
(Petauroides volans) in a fragmented forest ecosystem. I. Home range size and
movements. Wildlife Research 31, 559-568.
R Core Team, R (2012) R: A language and environment for statistical computing.
Robinson, NM, Leonard, SW, Ritchie, EG, Bassett, M, Chia, EK, Buckingham, S, Gibb, H,
Bennett, AF, Clarke, MF (2013) REVIEW: Refuges for fauna in fire‐prone
landscapes: their ecological function and importance. Journal of Applied Ecology 50,
1321-1329.
92
Robinson, NM, Leonard, SWJ, Bennett, AF, Clarke, MF (2014) Refuges for birds in fire-
prone landscapes: The influence of fire severity and fire history on the distribution of
forest birds. Forest Ecology and Management 318, 110-121.
Sanz-Aguilar, A, Anadon, JD, Gimenez, A, Ballestar, R, Gracia, E, Oro, D (2011) Coexisting
with fire: The case of the terrestrial tortoise Testudo graeca in mediterranean
shrublands. Biological Conservation 144, 1040-1049.
Smith, A (1984) Diet of Leadbeaters Possum, Gymnobelideus Leadbeateri (Marsupialia).
Wildlife Research 11, 265-273.
Taylor, RS, Watson, SJ, Nimmo, DG, Kelly, LT, Bennett, AF, Clarke, MF (2012) Landscape-
scale effects of fire on bird assemblages: does pyrodiversity beget biodiversity?
Diversity and Distributions 18, 519-529.
Teague, B, McLeod, R, Pascoe, S (2010) Final report, 2009 Victorian bushfires royal
commission. Parliament of Victoria, Melbourne Victoria, Australia
Vernes, K, Pope, LC (2001) Stability of nest range, home range and movement of the
northern bettong (Bettongia tropica) following moderate-intensity fire in a tropical
woodland, north-eastern Queensland. Wildlife Research 28, 141-150.
Zeileis, A, Wiel, MA, Hornik, K, Hothorn, T (2008) Implementing a class of permutation
tests: The coin package. Journal of Statistical Software 28, 1-23.
93
Figure 1. Map of study site location in the Victorian Central Highlands highlighting the region burnt in the 2009 Kilmore-Murrindindi
fire complex
94
95
Chapter 4. Bird use of fire refuges is contingent on
landscape context and the spatial extent of mixed
severity fire
Berry, L. E., Driscoll, D. A., Banks, S. C., & Lindenmayer, D. B. (2015) Bird use of fire
refuges is contingent on landscape context and the spatial extent of mixed severity fire.
Diversity and Distributions. UNDER REVIEW
96
97
4.1 Abstract
Context. Areas which experience fire regimes differing from those prevailing in the
landscape may act as refuges for fauna within extensively burnt montane forest systems.
There is currently a limited understanding of how the spatial attributes of these areas and
their interactions with abiotic and biotic gradients influence their ability to act as refuges.
Objectives We conducted a replicated landscape-scale observation study to determine
whether mesic gullies acted as refuges for forest birds following a large fire in the Mountain
Ash forests of south-eastern Australia.
Methods We used model selection to establish the influence on bird species and trait
occurrence of (1) patch size, connectivity and fire severity, (2) the proportion of forest burnt
at low, moderate and high severity within 300 m or 3 km of each site and (3) their
interactions with vegetation structure, elevation, precipitation and topography.
Results We found 4 bird species and 22 traits occurred more frequently in unburnt forest
patches. We found bird occurrence was both positively and negatively related to fire severity
proportions in the surrounding landscape at different scales. We found responses to the
spatial outcomes of fire were contingent on the availability of particular biotic and abiotic
gradients.
Conclusions Unburnt mesic gullies can function as important fire refuges for forest birds.
Bird use of these refuges is dependent on fire severity proportions in the surrounding
landscape. Developing contingent theory by examining ecological interactions between fire-
induced habitat patterns and environmental gradients is key to unravelling complex faunal
responses to fire.
98
4.2 Introduction
Large-scale crown-fires are a major form of disturbance in forested ecosystems globally
(Kasischke et al. 1995; Thonicke et al. 2001; Bond and Keeley 2005; Lindenmayer et al.
2014b). Wildfires can radically alter the distribution, structure, extent and availability of key
habitat features (Cocking et al. 2013; Clarke et al. 2014; Smith et al. 2014). These impacts
can lead to severe range contractions and local extirpations or extinctions of species whose
life cycles are intrinsically connected to the availability of mature habitat components or
provide essential resources for early successional specialists (Nappi and Drapeau 2009;
Lindenmayer et al. 2013a). However, large fires rarely burn homogeneously (Taylor et al.
2012; Leonard et al. 2014). In montane forest systems, the interaction of complex
topography, fuels and weather during fires, often leads to the establishment of heterogeneous
landscape mosaics, with areas burnt at different fire severities (Lindenmayer et al. 1999;
Bradstock et al. 2005; Alexander et al. 2006; Lentile et al. 2006; Wood et al. 2011; Mackey
et al. 2012; Leonard et al. 2014; Taylor et al. 2014a). Areas with certain topographic, edaphic
or vegetation characteristics which influence the distribution of fire severities, such as deep
sheltered mesic gullies, may burn at lower severity than the surrounding landscape or retain
unburnt habitat (Berry et al. 2015b). These areas may act as fire refuges for species that need
unburnt forest, providing resources which govern species distribution in post-fire landscapes
(Robinson et al. 2013; Cullinane-Anthony et al. 2014).
The conservation value of fire refuges for fauna in fire prone landscapes is increasingly
recognized (Robinson et al. 2014; Berry et al. 2015b). In non-forest ecosystems, the use of
unburnt habitat patches by fauna has been linked with; predation pressure (Yarnell et al.
2008), the availability of contrasting food resources (Watson et al. 2012), and the availability
of complex habitat structures required for denning (Pereoglou et al. 2011). However, the
99
ability of fire refuges to provide these functions for fauna is likely to be related to the ability
of each species to disperse to, and persist within, unburnt patches (Driscoll et al. 2014).
These processes are likely to be governed by the spatial attributes of unburnt remnants, such
as size and broader fire mosaic patterns within landscapes (Bradstock et al. 2005; Clarke
2008). However, understanding of how the spatial attributes of fire mosaic components
influence the presence and distribution of fauna in fire-affected forest systems remains
limited (Driscoll et al. 2010b).
A small number of studies have examined bird responses to unburnt residuals or fire refuges
following large-scale fires, but few have quantified whether the spatial attributes of these
residuals influence their ability to act as refuges for fauna (Barlow et al. 2006; Kelly et al.
2012; Taylor et al. 2012; Watson et al. 2012; Lindenmayer et al. 2013b; Lindenmayer et al.
2014b; Robinson et al. 2014; Sitters et al. 2014). In fragmented systems, the probability of
species using patches increases with size and decreases with isolation (Lindenmayer and
Fischer 2006). Larger patches commonly support more resources and interior habitat than
smaller patches (Bender et al. 1998; Major et al. 2001). Fragmented habitat patches
connected to larger contiguous areas of unfragmented habitat are more easily colonized
following disturbance and often present higher species occurrence than more isolated patches
(Lindenmayer 1994; Haas 1995). However, recently burnt landscapes differ from traditional
patch-matrix systems (sensu Forman 1995), due to the complex mosaics of habitat burnt at
different severities, compounded by environmental gradients, and the distribution of
ecologically important biological legacies (Robinson et al. 2013). Species responses to the
spatial attributes of fire refuges are likely to be related to their ability to tolerate the inter-
patch matrix and life history attributes such as dispersal ability, feeding guild and nesting
habit.
100
Landscape-scale patterns of fire severity are increasingly being recognised as important
determinants of species distributions (Bradstock et al. 2005; Driscoll et al. 2010b; Nimmo et
al. 2013; Lindenmayer et al. 2014b). Landscapes with increased fine-scale heterogeneity in
fire age classes may support higher total bird species richness due to increased niche
availability (Sitters et al. 2015). However, the distribution of old-growth specialist birds in
burnt landscapes is more likely to be governed by the availability of large areas of intact
unburnt late-successional forest (Taylor et al. 2012; Berry et al. 2015c). In some cases, large
fires may benefit early successional specialists by creating new habitat (Swanson et al. 2010),
particularly in systems where fire is actively supressed (Nappi and Drapeau 2009). In recently
burnt landscapes, increasing isolation (distance from the unburnt edge) decreases bird
abundance and species richness (Watson et al. 2012; Berry et al. 2015c). Therefore, the
relative isolation of unburnt landscape components may influence their ability to act as fire
refuges for forest birds.
Bird responses to the spatial outcomes of large fires have been described as complex
(Lindenmayer et al. 2014). To unravel this complexity, it is necessary to develop contingent
theory which examines how disturbance patterns, landscape patterns and environmental
gradients interact to influence species distributions (Driscoll and Lindenmayer 2012). The
ability of mesic gullies to act as refuges for birds may be contingent on the availability of key
vegetation structures, suitable environmental gradients, and the spatial extent of different fire
severities in the surrounding landscape. Fire severity describes the effects of fire on the loss
or change of organic matter above or below ground, such as changes in vegetation structure
(Keeley 2009). However, the effects of fire severity on bird distributions may differ when
specific vegetation structures, such as tree hollows, fruiting bodies, or dense regrowth are
locally available (Lindenmayer et al. 2009c; Perry et al. 2011). Similarly, bird response to the
101
area of different levels of fire severity at the site level may be influenced by regional patterns
of fire severity. For example, local occupancy of unburnt forest patches may be higher when
the regional availability of unburnt forest is low, as some species capitalise on the limited
availability of high value resources (Lindenmayer et al. 2013; Berry et al. 2015a). Some
species with high fidelity to sites within particular environmental envelopes, such as areas of
high elevation, may occur only in areas of low or high severity which also fit within these
envelopes (Lindenmayer et al. 2014).
We identified mesic gullies as landscape components likely to function as fire refuges based
on previous development and validation of landscape models (Mackey et al. 2002; Berry et
al. 2015b). We aimed to establish whether the spatial attributes of mesic gullies enable these
areas to maintain bird species and functional traits within landscapes extensively impacted by
large-scale fires. To address these aims, we completed a detailed empirical study of the 2009
Black Saturday bushfires, which occurred in Victoria, Australia and produced diverse
mosaics of mixed severity fire. Large areas of mountain ash, Eucalyptus regnans, forest were
burnt mostly homogeneously at high severity, whilst other parts of the landscape produced
diverse mosaics of interspersed low and high severity fire (Leonard et al. 2014). The diverse
range of fire mosaics surrounding gully forests enabled our study to quantify whether mesic
gullies act as fire refuges for forest birds and identify key refuge characteristics.
We asked three key questions; 1) does the size, connectivity and fire severity of fire-affected
mesic gullies influence their ability to act as refuges for forest birds? 2) How does the amount
of particular fire severities in the surrounding landscape at different scales influence the use
of potential fire refuges by birds? 3) How are bird responses to the spatial outcomes of fire
influenced by vegetation structure and environmental gradients in the surrounding landscape?
102
By examining these three questions, our study was able to establish how bird distributions
and persistence within burnt landscapes is influenced by a range of interacting factors,
providing a thorough test of the concept that fire refuges constitute desirable habitat within a
hostile landscape of non-habitat.
103
4.3 Methods
Study Region
We conducted our study in Victorian Central Highlands, approximately 120 km north-east of
Melbourne, Australia. The region covers an area of approximately 60 km x 80 km between
37°20'–37 ° 55'S and 145° 30'–146° 20'E. The region is topographically mountainous with
local variation in elevation among our study sites up to 1000 m. Mean annual temperature
varies from 7.8 °C to 13.4 °C (Smith et al. 2014). Vegetation in our study area was
dominated by mountain ash (Eucalyptus regnans) and alpine ash (Eucalyptus delegatensis)
forests. Mountain ash trees are obligate seeders, which grow to 100 m in height and are
subject to landscape-mediated mean fire intervals of between 30 – 300 years (Smith et al.
2014). Within these forests, all of our sites were located in mesic gullies with cool-temperate
rainforest vegetation communities characterised by myrtle beech (Nothofagus cunninghamii),
southern sassafras (Atherosperma moschatum), silver wattle (Acacia dealbata) and tree ferns
(Dicksonia sp.). Our study sites were also located within a 72,000 ha region of ash forest
burnt by the 2009 Black-Saturday bushfires (Gibbons et al. 2012b).
Site selection and study design
We selected 33 field sites using remotely-sensed fire severity maps and aerial photography
obtained in the weeks following the 2009 fires (DSE, 2009). See appendix 1 for a map of site
locations. We ground-truthed each potential site and included it in the study if site-level
vegetation accurately reflected the classifications indicated by the fire severity maps. We
selected sites in 5 categories; large discrete unburnt forest patches (>10 ha), small discrete
unburnt forest patches (<10ha), unburnt forest connected to the unburnt edge, forest burnt at
moderate severity (understorey burnt with canopy intact) and forest burnt at high severity
(burnt sites with fire-killed trees). All of our sites were located in mesic gullies within the
104
extent of the 2009 fires, as these were the only parts of the landscape where sufficient
replicates of size, connectivity and severity were consistently available.
We classified fire severity into 3 categories; 1) unburnt forest located within the extent of the
2009 fire, 2) moderate severity, where understorey structure and vegetation had been
consumed, but where the canopy remained intact and 3) high severity sites, where most
overstorey trees were killed by the fire. We calculated the area and proportion of forest burnt
at each of these severities within 300 m and 3 km surrounding each of our 33 study sites
using ArcMap 9.3 (ESRI 2008), based on fire severity data produced by the Victorian
Government from SPOT satellite imagery following the 2009 fires (DSE 2009, see appendix
2 for map of site locations). The 300 m and 3 km spatial scales were chosen as they
encompassed both potential localised bird foraging movements and wider dispersal
movements across the landscape.
Bird surveys
We established one 200 m transect at each of our 33 study sites. All transects were placed at
the bottom of each gully running parallel with the creek line, within no more than 10 m of the
creek. We conducted three twenty- minute point counts along each transect at 0 m, 100 m and
200 m (Pyke and Recher 1983). Bird species recorded at the previous point were not recorded
at subsequent points unless calls were from a distinctly different individual. All bird surveys
were completed by the same observer. Surveys were conducted to coincide with the peak
period of bird activity (Lindenmayer et al. 2014b) between October and December in 2014,
five years after fire. Surveys at each site were repeated on two different days and were
conducted within two hours of dawn to coincide with peak activity periods. We pooled our
data across these two visits to give a frequency of recording for each species, the number of
detections out of 6 observations at each site.
105
Bird life history attributes are linked with responses to vegetation structure and landscape
patterns (Hansen and Urban 1992). We classified bird species into different trait groups for
analysis, using the bird trait database compiled by Lindenmayer et al. (2009d) in the same
study region to assign the appropriate traits to each species. Trait groups were based on
feeding guild (granivores, omnivores, insectivores, carnivores, nectarivores, frugivores,
folivores), foraging habit (arboreal, multiple, foliage, ground, bark, wood, pounce, hawk,
bush carnivore), movement habit (resident, sedentary, migratory, part migratory), nest type
(hollow, bowl, dome, purse, multiple, cup) and absolute wing length (< 100 mm, 100 -200
mm, > 200mm). For a list of common and Latin bird names please see Appendix 2.
Due to the large area covered by our study, we included covariates in our analyses to account
for variation in topography and environmental conditions between sites. We used variables
identified by Lindenmayer et al. (2009d), in an earlier pre-fire study of birds, as important
determinants of bird species distribution in the Victorian Central Highlands. These included
elevation (m), Topographic Wetness Index (Beven and Kirkby 1979), annual mean
precipitation (mm) and vegetation structure. As the Victorian Central Highlands is subject to
extensive clear-fell logging, we interspersed replicates of each of our different site classes
across landscapes with various extents of recent and historical logging to account for logging-
induced regional variation in bird community distribution.
Vegetation surveys
We surveyed vegetation structure at the 0 m, 100 m and 200 m points of each transect. At
each point, we established a 20 m x 20 m quadrat. Within each quadrat we visually estimated
canopy cover (%) and midstorey cover %. We counted the number of wattles and the number
106
of regrowth mountain ash and alpine ash saplings within each quadrat. Using a clinometer,
we measured maximum canopy height (m) and maximum mid-storey height (m) within each
quadrat. All vegetation surveys were completed by the same observer. At the site level, we
counted the number of hollow bearing trees and the number of tree hollows per tree for all
trees within 20 m each side of the 200 m transect. We also measured the diameter at breast
height (cm) of all mountain ash and alpine ash trees present within the same 40 m x 200 m
area, to calculate mean tree diameter.
Statistical Analysis
To address our three key research questions, we used binomial generalised linear models of
species and trait occurrence. We fitted bird species and trait occurrence (number of times
presented out of six observations) per site as response variables. We fitted treatment class, the
proportion of fire severities in the surrounding landscape, vegetation structure (as the first
two axis of a Principal Components analysis), and environmental co-variates as predictor
variables. To avoid over-fitting models, we fitted a maximum of two explanatory variables
and their interaction in each model. Each model represented a test of one of our questions.
We fitted a total of 223 models for each species and trait group.
We identified the best models for each species and trait using a model selection approach
based on Akaike Information Criterion corrected for low sample size (AICc) (Burnham and
Anderson 2004). AIC-based model selection uses an information theory approach to
determine the relative ability of a given model to explain the observed patterns, from a suite
of potential models (Symonds and Moussalli 2011). Using this approach, we were able to
compare how treatment class, surrounding fire severity, vegetation structure and
environmental covariates influenced bird use of potential fire refuges. To determine the
107
influence of predictors within each set of best models on bird species and trait occurrence, we
calculated the coefficients and standard errors of all predictors for models with an AICc score
less than 2 units greater than the best model. We considered each predictor to have an
influential effect on bird and trait occurrence where the absolute value of the coefficient was
greater than double the standard error. If a response variable occurred more than once in the
list of ‘best’ models, we reported the coefficient and standard error for the highest ranked
model in which it first occurred. Where treatment class appeared in the list of best models, we
used a Wald test to determine if bird and trait occurrence differed between each class.
To address our research questions, we fitted treatment class separately in our models at 5
levels, 4 levels, 3 levels and 2 levels respectively. In addition to our full study design, we
included condensed versions of treatment class in our candidate models, where A) the
original five levels were reclassified into 4 levels, where small and large patches were
amalgamated to compare bird occurrence between isolated and connected unburnt gullies, B)
3 levels, where small patches, large patches and unburnt gullies connected to unburnt forest
beyond the extent of the fire were grouped to form a comparison of fire severities and C) 2
levels, where both moderate and high severity sites formed one group and small, large and
unburned gullies connected to edge formed another, to compare bird response to burnt and
unburnt mesic gullies.
We fitted vegetation structure into our models as the first two components of a principal
components analysis (PCA) of vegetation structure (Dunteman 1989). To interpret the
ecological relevance of each component, we calculated the Pearson’s correlation coefficient
between vegetation principal components 1 and 2 and each of the vegetation variables.
108
We did not include an analysis of bird detection in this study as recent work has shown that
adjusting occupancy models for non-detection can be as misleading as ignoring non-detection
completely (Welsh et al. 2015). The potential for birds to disperse between repeat surveys
violates the key assumption of single season occupancy models, that populations are closed
populations between surveys (MacKenzie et al. 2002). We chose to instead control for
differences in bird detection between sites using a replicated landscape-scale study design
(Banks‐Leite et al. 2014), accounting for temporal heterogeneity by conducting surveys on
multiple days, accounting for local spatial heterogeneity by sampling each site at multiple
points over a 200 m area, and reducing the likelihood of error between observers by using a
single observer for all surveys (Lindenmayer et al. 2009c).
Figure 1. Examples of species responding positively (a) and negatively (b) to the
availability of unburnt patches. Plots show the probability of occurrence of each species
in burnt and patch treatment class, with 95% confidence intervals (bars). Both
responses were significant at the 0.05 level.
109
4.4 Results
We recorded 52 bird species, representing 29 individual traits from five groups during our
survey period across our 33 study sites. We first present results of bird species and trait
responses to treatment class. We then identify responses to the proportion of fire severities at
local (300 m) and regional (3 km) scales. We then present results where species and trait
responses to treatment class and fire severity proportions were contingent on the proportion
of other fire severities at different scales, vegetation structure, elevation, topographic wetness
index and precipitation.
Table 1. Pearson’ s correlation coefficients between vegetation variables and first two
axes of a Principal Components (PC) Analysis of differences in vegetation structure
between sites.
Variable PC1 PC2
Canopy cover 0.56 0.00
Canopy Height 0.74 -0.02
Regrowth stem density -0.32 0.47
Regrowth stem height -0.34 0.32
Wattles -0.14 0.45
Mean tree diameter 0.89 0.22
Mid cover 0.66 -0.70
Mid height 0.78 -0.34
Total hollow bearing trees 0.37 0.19
Total tree hollows 0.53 0.29
Hollows per tree 0.26 0.26
Site vegetation characteristics
The first two components of the principal components analysis of vegetation structure used in
our model selection process were associated with contrasting vegetation structures and
accounted for 82.88% of observed variation in vegetation structure between sites (Table 1).
Higher values for PC1 were observed in moderately burnt sites (estimate = 114.14, P =
0.013), large patches (estimate = 104.68, P =0.022), small patches (estimate = 236.37, P =
110
<0.001) and unburnt gullies connected to the edge (estimate = 148.83, P = <0.001), than high
severity burnt sites (estimate = -114.72, P = <0.001). PC 2 was not related to our site types.
Does the size, connectivity and fire severity of fire affected mesic gullies influence their
ability to act as refuges for forest birds?
We found that bird occurrence was explained by the presence or absence of unburnt forest at
the site level. However, we found no difference in bird species or trait occurrence between
unburnt patches of different size or connectivity as represented by our treatment classes.
Treatment class categorized as two levels (unburnt patches, burnt forest) was the only
variable in the set of best models for 22 trait groups and 4 species (Fig. 1). Of these, we found
that ground foragers, sedentary birds, wood foragers and the brown thornbill occurred
significantly more frequently in unburnt patches than in burnt forest (Table. 2).
Table 2. Wald test output indicating significant differences in species and trait
occurrences between burnt forest and unburnt forest.
Wald test Burnt Unburnt Response X2 P estimate SE estimate SE
brown thornbill 7.2 0.007 -3.64 0.72 2.03 0.76 pilotbird 2.9 0.091 -3.22 0.59 1.11 0.66
fruit feeding 3.7 0.055 -2.68 0.46 18.88 3505.51 ground foraging 5.7 0.017 -2.17 0.37 1.03 0.43
sedentary 4.4 0.036 -1.61 0.30 0.76 0.36 wood foragers 5.7 0.017 -2.32 0.40 1.08 0.45 purse nesters 3.0 0.085 -4.343 1.01 1.83 1.06
Our model selection process indicated that the best models for 32 bird species and 20 traits,
included treatment class and covariate related to either vegetation structure or topography.
From these models, 11 bird species and 8 traits showed significant responses to treatment
class categorized as two levels (Table 3). Of these responses, we found an additional 4
species and 1 trait which also had significant responses to either vegetation structure or the
amount of low or moderate severity fire within 3 km (Table. 3). The occurrence of the
111
crescent honeyeater was positively influenced by vegetation principal component 1. The
occurrence of the red wattle bird was positively related to increasing topographic wetness.
The occurrence of the eastern spinebill was positively influenced by higher proportions of
low severity fire within 3 km. The occurrence of the willy wagtail was positively related to
vegetation principal component 2. Plant feeders were more likely to occur at sites with a
lower proportion of the landscape within the surrounding 3 km, subject to moderate severity
fire.
Treatment class with three levels (high severity, moderate severity and low severity) was the
sole variable in the set of best models for six trait groups and one species, although there
were no significant effects of treatment class and the Wald test results for each species
indicated no significant difference in occurrence between treatments. Treatment class
(categorised into 4 levels) was present as the sole variable in a model within 2 AICc of the
‘best’ model for the flame robin (Wald test on 3 terms excluding understory burn = X2 = 7.1,
df = 1, P= 0.0079, dev expl = 59.94%). Treatment class with 5 levels was not present in the
set of best models for any trait group or species. We found no significant interactions
between treatment class and the covariates.
112
Table 3. Bird responses to treatment class categorised into two levels and the effects of co-variates. Table shows wald test output and
coefficients and standard errors for models within 2 AICc of the ‘best’ model. Covariates include vegetation principal component 1
(PC1) and component 2 (PC2), the proportion of low severity (low), moderate severity (mod), high severity (high) within 3 km,
topographic wetness index (twi), precipitation (prec) and elevation (elev).
burnt unburnt Wald test covariate Interaction response est SE est SE X2 P name est S.E est S.E
crescent honeyeater -4.06 1.01 1.99 1.09 3.4 0.067 PC1 -0.001 0.01 - - olive whistler -7.01 2.96 5.50 2.97 3.4 0.064 PC1 0.005 0.01 -2.46 0.01 lyrebird -1.36 0.47 1.36 0.51 7.1 0.007 PC1 0.001 0.01 0.01 0.01 golden whistler -1.53 0.30 1.47 0.37 15.5 <0.001 PC2 0.001 0.01 - - striated thornbill -2.08 0.45 1.98 0.49 16.5 <0.001 PC2 -0.003 0.01 0.01 0.18 brown thornbill -7.14 2.89 5.38 2.91 3.4 0.065 low 0.001 0.01 0.00 0.01 eastern spinebill -3.76 0.72 0.75 0.87 4.90 0.027 low 0.026 0.013 - - grey currawong -1.62 0.52 -0.53 0.62 3.80 0.050 twi -0.045 0.126 0.05 0.16 willy wagtail -62.6 7191 16.6 5847 0.00 1.000 PC2 0.033 0.015 - - pilotbird -3.07 0.60 1.47 0.71 0.02 0.890 mod 0.032 0.017 - - gang gang cockatoo -123 2598 26.1 4391 13.3 0.001 PC1 0.001 0.001 - - bark foragers -1.62 0.31 0.48 0.40 6.00 0.015 PC2 0.002 0.003 - - plant feeders -1.13 0.27 -0.76 0.44 3.0 0.081 mod -0.212 0.02 - - wood foragers -2.54 0.42 1.39 0.48 8.3 0.004 PC2 0.005 0.01 - - ground foragers -2.08 0.39 0.88 0.48 3.90 0.049 PC1 -0.001 0.002 - - nectar feeders -21.4 3485 18.7 3485 3.20 0.074 PC1 -0.006 0.007 0.00 0.01 nectar feeders -21.7 3502 18.6 3502 4.30 0.037 low -0.02 0.020 - - nectar feeders -20.2 3502 18.7 3502 4.10 0.043 high 0.03 0.038 - - nectar feeders -21.7 3505 18.8 3505 3.60 0.058 mod -0.01 0.021 - - plant feeders 0.48 0.77 -0.64 0.41 4.40 0.037 high 0.02 0.035 - - plant feeders -1.00 0.26 -0.51 0.39 3.90 0.049 low -0.01 0.017 - - pounce foragers -2.66 1.17 0.21 0.39 5.70 0.017 prec 0.001 0.002 - - pounce forages -5.06 3.42 0.20 0.39 5.70 0.017 elev -0.001 0.001 - - various nesters -3.31 0.73 -0.59 1.08 3.00 0.083 PC1 -0.001 0.003 - -
113
Table 4. The influence of the proportion of fire severities at local and region scales
and vegetative and environmental co-variates on the occurrence of bird species
and traits. Covariates include vegetation principal component 1 (PC1) and
component 2 (PC2), the proportion of low severity (3kL), moderate severity (3kM),
high severity (3kH) within 3 km and the proportion of moderate severity (300M)
within 300 m.
fire severity variable co-variate response Cov 1 est S.E Cov 2 est S.E
Laughing kookaburra 300M 0.033 0.011 - - - Yellow faced honeyeater 300M 0.036 0.015 - - - Sacred kingfisher 300M 0.042 0.018 - - - Eastern spinebill 3kL 0.048 0.018 - - - Grey shrike thrush 3kM 0.041 0.014 PC1 <0.001 0.001 shining bronze cuckoo 3kH 0.27 0.14 PC1 0.01 0.005 Horsfield’s bronze cuckoo
3kL 1.635 0.826 3kH 3.226 1.622
Table 5. Table displaying responses to the proportion of fire severities at different
scales that were contingent on vegetation structure (PC1, PC2), the proportion of
low (300L) moderate (300M) and high (300H) severity fire within 300 m, the
proportion of low severity (3kL), moderate severity (3kM), high severity (3kH)
within 3 km, elevation (elev), topographic wetness (twi) and precipitation (prec).
fire severity variable covariate interaction response Cov 1 est S.E Cov 2 est S.E est S.E
flame robin 300H 0.023 0.005 PC1 0.012 0.005 <0.001 <0.001 Brown headed honeyeater
300H 0.057 0.023 300M -0.024 0.025 -0.002 0.001
striated pardalote
300M 0.402 0.133 elev 0.009 0.003 <0.001 <0.001
Grey currawong 3kM -0.01 0.025 twi -0.098 0.128 0.018 0.008 Pied currawong 3kM 0.053 0.021 300L 0.101 0.041 0.002 0.001 bassian thrush 3kM -0.08 0.072 PC2 0.010 0.011 -0.002 <0.001 Lewin’s honeyeater
3kM -1.63 0.587 prec 0.001 0.004 0.001 <0.001
White-browed scrubwren
300L 0.022 0.009 PC1 -0.005 0.002 <0.001 <0.001
white-naped honeyeater
300L -0.02 0.022 PC2 0.008
0.006 <0.001 <0.001
Pink robin 3kL -0.10 0.046 300H -0.042 0.013 0.002 0.001 fan-tailed cuckoo
3kL 0.01 0.016 twi -0.060 0.086 -0.022 0.008
114
How does the amount of particular fire severities in the surrounding landscape at
different scales influence the use of potential fire refuges by birds?
The best models for 33 species and 14 traits contained a measure of the proportion of
the landscape subject to different levels of fire severity, a covariate and their
interactions. Of these models, 7 species displayed significant responses to the
proportion of fire severity within the landscape (Table. 4). Horsfield’s bronze cuckoo
and the shining bronze cuckoo were more likely to occur when the proportion of high
severity fire within the surrounding 3 km of a site was higher (Table. 4). The occurrence
of the grey shrike thrush was positively related to higher proportions of the landscape
subject to moderate severity fire (Fig 3). The sacred kingfisher, laughing kookaburra
and yellow-faced honeyeater were more likely to occur at sites with high proportions of
the forest within 300 m of site subject to moderate severity fire (Fig 4). The eastern
spinebill was more likely to occur at sites with a higher proportion of low severity fire
within the surrounding 3 km (Fig 5).
We found the best models, with the lowest AIC scores for 4 species included a
significant response to fire severity proportion and an insignificant interaction (Table.
5). The occurrence of Lewin’s Honeyeater was lower in landscapes with a greater
amount of forest within 3 km burnt at moderate severity. The striated pardalote,
occurred more frequently at sites surrounded by a greater amount of forest within 300
m, which was burnt at moderate severity. The occurrence of the flame robin was higher
when sites had a higher amount of forest burnt within 300 m burnt at high severity. The
occurrence of the white browed scrub-wren was higher at sites which were surrounded
by larger amounts of unburnt forest within 300 m (Fig 6).
115
Figure 2. Examples of species responding positively (a) and negatively (b) to the
increasing proportion of high severity fire within a buffer area with a radius of 300
m from each site. Plots a + b show the probability of occurrence (bold line) of each
species treatment class at a site. The shaded grey and broken dotted lines represent
95% confidence intervals. Plot (c) show the interacting effects of vegetation
principal component 1 and the proportion of high severity fire within 300 m on the
occurrence of the flame robin. Plot (d) shows the interacting effects of the
proportion of moderate severity fire and high severity fire within 300 m on the
occurrence of the brown headed honeyeater.
116
Figure 3. Examples of species with positive (a) and negative (b) relationships with
the proportion of moderate severity fire within a buffer zone with a 3 km radius
from each site. The bold line shows predicted trends in occurrence. The grey
shading and broken dotted lines represent 95% confidence intervals. Plot (c) shows
the how the response of the pied currawong to the proportion of moderate severity
fire within 3 km varies depending on the proportion of low severity fire within 300
m. Plots (d) and (e) show how the occurrence of the Bassian thrush at sites with
differing proportions of moderate severity fire within 3 km is influenced by
vegetation principal component 1 + 2.
Figure 4. Examples of species with positive (a) and negative (b) relationships with
the proportion of moderate severity fire within a buffer zone with a 300 m radius
from each site. The bold line shows predicted trends in occurrence. The grey
shading and broken dotted lines represent 95% confidence intervals. Plot (c)
shows how the response of the striated pardalote to the proportion of moderate
severity fire within 300 m was influenced by site elevation.
117
Figure 5. Example of species with a higher probability of occurrence at sites with
high proportions of low severity fire within a 3 km radius buffer (a). The bold line
shows predicted trends in occurrence. The grey shading and broken dotted lines
represent 95% confidence intervals. Plot (b) shows how the response of the pink
robin the proportion of low severity fire within 3 km was influenced by the
proportion of high severity fire within 300 m. Plot (c) shows how the response of
the fan tailed cuckoo to the proportion of low severity fire within 3 km was
influenced by the topographic position (TWI) of a site.
Figure 6. Example of a species with a lower probability of occurrence at sites with
a high proportion of low severity fire within a 300 m (a). The bold line shows
predicted trends in occurrence. The grey shading and broken dotted lines
represent 95% confidence intervals. The response of the white-browed scrub wren
(b) and the white-naped honeyeater (c) to the proportion of low severity fire within
300 m was influenced by vegetation principal component 1 (b) and 2 (c).
118
How are bird responses to fire refuges influenced by the interaction of the spatial
outcomes of fire, vegetation and environmental gradients?
Our results revealed that for 11 species, responses to treatment class and fire severity
proportion were contingent on the amount of the landscape subject to contrasting fire
severities at different scales, vegetation structure and environmental gradients (Table.
5). The flame robin increased in occurrence with increasing amounts of high severity
fire within 300 m, but only when vegetation mature old-growth structures represented
by vegetation principal component 1 were scarce (Fig 2.C). The occurrence of the
brown headed honeyeater decreased when the proportion of high severity fire within
300 m was high, but this response reversed when the proportion of moderate severity
fire within 300 m was high (Fig 2.D). The pied currawong occurred more frequently at
sites with larger amounts of moderate severity fire within 3 km, this response was
amplified when sites had lower amounts of low severity fire within 300 m (Fig 3.C).
Lewin’s honeyeater declined in occurrence at sites with large amounts of moderate
severity fire within 3 km when precipitation was low, but increased at sites which also
had annual precipitation greater than 1700 mm (Fig 3. D). The bassian thrush only
displayed increased occurrence at sites with high proportions of moderate severity fire
within 3 km when vegetation structures associated with principal component 2 were
absent (Fig 3. E). The occurrence of the striated pardalote was higher at sites with high
proportions of moderate severity fire within 300 m when at sites of low elevation, and
the opposite response at sites of high elevation (Fig 4.C). The pink robin was more
likely to occur at sites that had high proportions of high severity fire within 300 m, but
only when the proportion of unburnt forest within 3 km was high. However, when the
proportion of unburnt forest within 3 km was low, the pink robin occurred only at sites
with low proportions of high severity fire within 300 m (Fig 3. B). The fan-tailed
119
cuckoo exhibited a positive response to high proportions of moderate severity fire
within 3 km at sites with low scores of TWI and a negative response at sites with high
scores of TWI (Fig 5. C). The white-browed scrub wren increased in occurrence at sites
with large amounts of low severity fire within 300 m when vegetation structures
associated with principal component 1 were present, and declined when these structures
were absent (Fig 6. B). The white naped honeyeater declined in occurrence at sites with
large amounts of low severity fire within 300 m, but increased when vegetation
structures associated with principal component 2 were absent (Fig 6. C).
120
4.5 Discussion
Fire refuges have been identified as an important mechanism influencing the
distribution and persistence of fauna within extensively burnt landscapes (Mackey et al.
2002). However, very few studies have examined this relationship or identified
desirable refuge characteristics (Robinson et al. 2013). Identifying landscape
components which constitute faunal refuges is a conservation priority as forested
ecosystems globally experience increasing disturbance frequency (Overpeck et al. 1990;
Dale et al. 2001b; Bowman et al. 2009).
We found that unburnt mesic gullies facilitated the retention of forest birds within
extensively burnt montane forest landscapes. We found that many species responded
positively to the occurrence of intact forest patches regardless of their size or
connectivity to the unburnt edge. However, the ability of unburnt mesic gullies to
support many species within the landscape was contingent on appropriate proportions of
fire severities at different scales in the surrounding landscape, elevation, precipitation,
topographic position and the availability of particular vegetation structures.
Does the size, connectivity and fire severity of fire affected mesic gullies influence their
ability to act as refuges for forest birds?
We found that unburnt mesic gullies facilitated the persistence of birds within
extensively burnt landscapes irrespective of their size and connectivity. The absence of
a patch size effect suggests that bird use of unburnt forest within the extent of fire was
unrelated to attributes commonly found in larger patches, such as increased niche
availability, decreased competition pressure and the absence of edge effects (Bender et
al. 1998). Areas of low severity fire retain important habitat features such as large old
121
trees with hollows, mature flowering canopies and complex multi-story forest structures
(Lindenmayer et al. 1990b; Fulé and Laughlin 2007). After fires, wet forest gullies
support more large logs and dead trees, (which provide food resources for birds), than
slopes (Bassett et al. 2015). In turn, this is likely to explain the increased occurrence of
insectivores, ground foragers, wood foragers and bark foragers in unburnt gullies. The
activities of birds belonging to these foraging guilds such as Lyrebirds, which turn over
soil and bury leaf litter, may in turn, reduce fuel loads in these areas, leading to a
positive feedback loop where these areas may have an increased chance of persisting
through future fires (Nugent et al. 2014).
The impacts of fire-induced habitat fragmentation on fauna are often less severe than in
other fragmentation scenarios (such as within agricultural landscapes) due the relative
hospitability of the recently burnt habitat (Berry et al. 2015c). Although fire refuges
provide essential habitat for many species, we found three species occurred more often
in recently burnt forest. Many of the patch-favouring species observed in this study also
were observed at very low levels of occurrence in the recently burnt habitat. This effect
may be attributed to the rapid change in forest structures in regenerating Mountain Ash
stands in the five years following fire. For example, small-bodied birds such as the
brown thornbill and large-billed scrub wren are able to recolonize burnt forests two to
three years following fire due to the availability of dense foliage provided by
regenerating Mountain Ash (Lindenmayer et al. 2014b).
How does the amount of particular fire severities in the surrounding landscape at
different scales influence the use of potential fire refuges by birds?
122
Bird occurrence in mesic gullies was dependent on the spatial extent of fire severities at
different scales in the surrounding landscape. We found that bird dispersal ability was
likely to explain bird responses to different scales of fire severities in the landscape We
found three birds commonly associated with canopy foraging (Table. 4) occurred more
frequently when higher proportions of intact canopy were locally available (within 300
m). Canopy foraging birds in Mountain Ash forests are known to disperse between
areas of forest when the canopy is in flower (Lindenmayer et al. 2010b). Due to the
relative scarcity of intact canopy within the extent of the 2009 fire, it is likely that larger
areas of unburnt canopy may be more attractive to birds capable of moving throughout
the burnt landscape than smaller areas of canopy. Similarly, the preference of the
eastern spinebill for sites with a greater proportion of unburnt forest within 3 km, can be
attributed to the ability of the species to migrate large distances to areas where mature
flowering plants are abundant for feeding (Chan et al. 1990). These movement patterns
may be common for nectar feeding birds as flowering events may remain absent from
burnt forest for several years following fire. Conversely, we found two species of
cuckoo favoured extensively burnt landscapes. These cuckoos parasitise the nests of
small passerines, which are abundant in the dense regrowth vegetation present fire years
after fire (Langmore and Kilner 2007; Lindenmayer et al. 2014b). Our findings suggest
that within post-fire landscapes, mesic gullies are less likely to be important as refuges
for birds with the dispersal ability to reach desirable habitat features. However, the
presence of these birds within the landscape is likely to depend on the sufficient spatial
availability of these structures within their dispersal range
.
How are bird responses to the spatial outcomes of fire influenced by vegetation
structure and environmental gradients in the surrounding landscape?
123
Our study demonstrated that bird response to the spatial outcomes of fire was contingent
on the availability of particular vegetation structures, rainfall gradients, elevation
gradients and topographic positions. We found that areas which may otherwise not
support the occurrence of a species, may act as refuges under certain conditions. For
example, two birds associated with mature unburnt forest (the white-eared honeyeater
and brown-headed honeyeater) occurred more frequently in mesic gullies burnt at high
severity when the local and regional proportion of moderate severity fire (intact canopy)
was high. This suggests that mobile birds commonly associated with mature forest (such
as honeyeaters) are able to use extensively burnt habitat when mature canopy is
available within their movement range in the surrounding landscape (Green 1993;
Stanley and Lill 2002).
We found that for some species, response to the proportion of unburnt forest was
dependent upon the availability of key vegetation structures. This demonstrates that the
use of unburnt fire refuges may be influenced by the historical patterns of disturbance or
land use responsible for differences in vegetation structure (McCarthy and Lindenmayer
1998; Lindenmayer et al. 2000a). Our results demonstrate that although the spatial
outcomes of fire are an important determinant of bird occurrence, their relevance to
conservation is further dependent on the location of these patterns in the landscape in
relation to desirable environmental gradients and vegetation structures.
124
4.6 Management Recommendations
Fire refuges are increasingly recognised as an important mechanism of faunal survival
following the occurrence of large fires (Watson et al. 2012; Robinson et al. 2013;
Cullinane-Anthony et al. 2014; Berry et al. 2015a). These refuges are likely to become
of greater conservation value as the frequency, severity and intensity of large fires
increases with global climate change (Cary et al. 2012). Current fire management
policies focused primarily on asset protection are unlikely to yield an improvement in
the creation, retention and subsequent use of refuges in fire-prone forests (Clarke 2008).
In particular, our study demonstrates that policies such as ‘blackout burning’ (where
patches left unburnt within the extent of large fires are routinely burnt by fire managers
to reduce the risk of additional conflagrations) may severely impact the likelihood of
birds surviving and persisting in burnt landscapes.
In Mountain Ash forests, fire management practices which aim to manipulate the spread
of fire are ineffective due to the moist nature of the vegetation or the danger presented
by high fuel loads (Lindenmayer 2009d). Consequently, we suggest that fire
management planning in Mountain Ash forests should limit future manipulations to
forest structure to ensure that the natural processes which lead to the establishment and
subsequent use of refuges by fauna are maintained (Berry et al. 2015b). This includes
minimising the extent of land practices, such as logging, which can influence the spatial
outcomes of large fires.
Our study demonstrates that fire refuges perform an important ecological role in
facilitating the persistence of species and functional traits within extensively disturbed
ecosystems. Consequently, unburnt forest patches embedded within burnt landscapes
125
should be managed for their conservation values and protected from future
anthropogenic disturbance, such as timber harvesting, blackout burning and road
building.
Our study presents a key advance in that the effects of fire-induced habitat patterns on
the distribution of fauna is contingent on their spatial relationships with key biotic and
abiotic landscape patterns. Bird response to the amount of forest burnt at different fire
severities was not uniform across the landscape. We found that bird responses to the
proportion of unburnt canopy, intact forest or high severity fire can differ depending on
the availability or absence of key habitat components or at extremes of environmental
gradients such as elevation and rainfall. This finding has significant implications for
future biodiversity management planning in mountainous forest systems. To produce
ecologically beneficial fire patterns, land managers must develop strategies to produce
mosaics of mixed severity fire which overlap with a range of biotic and abiotic
gradients. It is particularly important that areas which support extremes of these
gradients within landscapes (ie, old-growth vegetation, high elevations and low
elevation), and often support rare species, are given priority in land management
planning as potentially valuable refuge areas.
126
4.7 Acknowledgements
LB thanks Karen Ikin and Jeff Wood for their statistical advice, Lachlan Mc Burney and
David Blair for their support with fieldwork and Clive Hilliker for his help in preparing
our figures for publication. This project was conducted in accordance with Australian
National University animal ethics permit number A2012/42.
127
4.8 References
Alexander, JD, Seavy, NE, Ralph, CJ, Hogoboom, B (2006) Vegetation and
topographical correlates of fire severity from two fires in the Klamath-Siskiyou
region of Oregon and California. International Journal of Wildland Fire 15,
237-245.
Banks‐Leite, C, Pardini, R, Boscolo, D, Cassano, CR, Püttker, T, Barros, CS, Barlow, J
(2014) Assessing the utility of statistical adjustments for imperfect detection in
tropical conservation science. Journal of Applied Ecology, 51, 849-859.
Barlow, J, Peres, CA, Henriques, L, Stouffer, P, Wunderle, J (2006) The responses of
understorey birds to forest fragmentation, logging and wildfires: an Amazonian
synthesis. Biological Conservation 128, 182-192.
Bassett, M, Chia, EK, Leonard, SWJ, Nimmo, DG, Holland, GJ, Ritchie, EG, Clarke,
MF, Bennett, AF (2015) The effects of topographic variation and the fire regime
on coarse woody debris: Insights from a large wildfire. Forest Ecology and
Management 340, 126-134.
Bender, DJ, Contreras, TA, Fahrig, L (1998) Habitat loss and population decline: a
meta-analysis of the patch size effect. Ecology 79, 517-533.
Berry, LE, Driscoll, DA, Stein, JA, Blanchard, W, Banks, SC, Bradstock, RA,
Lindenmayer, DB (2015a) Identifying the location of fire refuges in wet forest
ecosystems. Ecological Applications
Berry, LE, Lindenmayer, DB, Driscoll, DA (2015b) Large unburnt areas, not small
unburnt patches, are needed to conserve avian diversity in fire‐prone landscapes.
Journal of Applied Ecology 52, 486-495.
Berry, LE, Driscoll, DA, Banks, SC, Lindenmayer, DB (2015c) The use of topographic
fire refuges by the greater glider (Petauroides volans) and the mountain brushtail
possum (Trichosurus cunninghami) following a landscape-scale fire. Australian
Mammalogy 37, 39-45.
Beven, K, Kirkby, M (1979) A physically based, variable contributing area model of
basin hydrology. Hydrological Sciences Journal 24, 43-69.
Bond, WJ, Keeley, JE (2005) Fire as a global ‘herbivore’: the ecology and evolution of
flammable ecosystems. Trends in Ecology & Evolution 20, 387-394.
Bowman, D, Balch, JK, Artaxo, P, Bond, WJ, Carlson, JM, Cochrane, MA, D'Antonio,
CM, DeFries, RS, Doyle, JC, Harrison, SP, Johnston, FH, Keeley, JE,
Krawchuk, MA, Kull, CA, Marston, JB, Moritz, MA, Prentice, IC, Roos, CI,
Scott, AC, Swetnam, TW, van der Werf, GR, Pyne, SJ (2009) Fire in the Earth
System. Science 324, 481-484.
Bradstock, RA, Bedward, M, Gill, AM, Cohn, JS (2005) Which mosaic? A landscape
ecological approach for evaluating interactions between fire regimes, habitat and
animals. Wildlife Research 32, 409-423.
Burnham, KP, Anderson, DR (2004) Multimodel inference understanding AIC and BIC
in model selection. Sociological methods & research 33, 261-304.
Cary, G, Bradstock, RA, Gill, AM, Williams, RJ (2012) Global change and fire regimes
in Australia. Flammable Australia: fire regimes, biodiversity and ecosystems in
a changing world 149-169.
Chan, K, Ford, H, Ambrose, S (1990) Ecophysiological Adaptations of the Eastern
Spinebill <I>Acanthorhynchus tenuirostris</I> to a High Altitudinal Winter
Environment. Emu 90, 119-122.
Clarke, MF (2008) Catering for the needs of fauna in fire management: science or just
wishful thinking? Wildlife Research 35, 385-394.
128
Clarke, PJ, Knox, KJE, Bradstock, RA, Munoz-Robles, C, Kumar, L (2014) Vegetation,
terrain and fire history shape the impact of extreme weather on fire severity and
ecosystem response. Journal of Vegetation Science 25, 1033-1044.
Cocking, MI, Varner, JM, Knapp, EE (2013) Long-term effects of fire severity on oak–
conifer dynamics in the southern Cascades. Ecological Applications 24, 94-107.
Cullinane-Anthony, BL, Seefelt, NE, Gregory Corace Iii, R, Kashian, DM, Gehring,
TM (2014) Influence of residual forest patches on post-fire bird diversity
patterns in jack pine-dominated ecosystems of northern Lower Michigan. Forest
Ecology and Management 331, 93-103.
Dale, VH, Joyce, LA, McNulty, S, Neilson, RP, Ayres, MP, Flannigan, MD, Hanson,
PJ, Irland, LC, Lugo, AE, Peterson, CJ, Simberloff, D, Swanson, FJ, Stocks, BJ,
Michael Wotton, B (2001) Climate Change and Forest Disturbances. BioScience
51, 723-734.
Driscoll, DA, Banks, SC, Barton, PS, Ikin, K, Lentini, P, Lindenmayer, DB, Smith, AL,
Berry, LE, Burns, EL, Edworthy, A (2014) The trajectory of dispersal research
in conservation biology. Systematic review. Plos One 9, e95053.
Driscoll, DA, Lindenmayer, DB (2012) Framework to improve the application of theory
in ecology and conservation. Ecological Monographs 82, 129-147.
Driscoll, DA, Lindenmayer, DB, Bennett, AF, Bode, M, Bradstock, RA, Cary, GJ,
Clarke, MF, Dexter, N, Fensham, R, Friend, G, Gill, M, James, S, Kay, G,
Keith, DA, MacGregor, C, Russell-Smith, J, Salt, D, Watson, JEM, Williams,
RJ, York, A (2010) Fire management for biodiversity conservation: Key
research questions and our capacity to answer them. Biological Conservation
143, 1928-1939.
DSE (2009) Remote sensing guideline for assessing landscape-scale fire severity in
Victoria’s forest estate. Unpublished technical manual., Department of
Sustainability and Environment, Melbourne.
Dunteman, GH (1989) 'Principal components analysis.' (Sage:
ESRI, E (2008) ArcMap 9.3. ESRI® ArcGIS 9, 1999-2008.
Forman, RTT (1995) 'Land mosaics. The ecology of landscapes and regions.'
(Cambridge University Press: Cambridge)
Fulé, PZ, Laughlin, DC (2007) Wildland fire effects on forest structure over an
altitudinal gradient, Grand Canyon National Park, USA. Journal of Applied
Ecology 44, 136-146.
Gibbons, P, van Bommel, L, Gill, AM, Cary, GJ, Driscoll, DA, Bradstock, RA, Knight,
E, Moritz, MA, Stephens, SL, Lindenmayer, DB (2012) Land Management
Practices Associated with House Loss in Wildfires. Plos One 7,
Green, R (1993) Avian Seed Dispersal in and Near Subtropical Rainforests. Wildlife
Research 20, 535-557.
Haas, CA (1995) Dispersal and use of corridors by birds in wooded patches on an
agricultural landscape. Conservation Biology 9, 845-854.
Hansen, AJ, Urban, DL (1992) Avian response to landscape pattern: the role of species'
life histories. Landscape Ecology 7, 163-180.
Kasischke, ES, Christensen Jr, N, Stocks, BJ (1995) Fire, global warming, and the
carbon balance of boreal forests. Ecological Applications 437-451.
Keeley, JE (2009) Fire intensity, fire severity and burn severity: a brief review and
suggested usage. International Journal of Wildland Fire 18, 116-126.
Kelly, LT, Nimmo, DG, Spence-Bailey, LM, Taylor, RS, Watson, SJ, Clarke, MF,
Bennett, AF (2012) Managing fire mosaics for small mammal conservation: a
landscape perspective. Journal of Applied Ecology 49, 412-421.
129
Langmore, N, Kilner, R (2007) Breeding site and host selection by Horsfield's bronze-
cuckoos, Chalcites basalis. Animal Behaviour 74, 995-1004.
Lentile, LB, Smith, FW, Shepperd, WD (2006) Influence of topography and forest
structure on patterns of mixed severity fire in ponderosa pine forests of the
South Dakota Black Hills, USA. International Journal of Wildland Fire 15, 557-
566.
Leonard, SW, Bennett, AF, Clarke, MF (2014) Determinants of the occurrence of
unburnt forest patches: Potential biotic refuges within a large, intense wildfire in
south-eastern Australia. Forest Ecology and Management 314, 85-93.
Lindenmayer, D (1994) Wildlife Corridors and the Mitigation of Logging Impacts on
Fauna in Wood-Production Forests in South-Eastern Australia: a Review.
Wildlife Research 21, 323-340.
Lindenmayer, D, Blair, D, McBurney, L, Banks, S, Stein, J, Hobbs, R, Likens, G,
Franklin, J (2013a) Principles and practices for biodiversity conservation and
restoration forestry: a 30 year case study on the Victorian montane ash forests
and the critically endangered Leadbeater's Possum. Australian Zoologist 36,
441-460.
Lindenmayer, D, Cunningham, R, Donnelly, C, Franklin, J (2000) Structural features of
old-growth Australian montane ash forests. Forest Ecology and Management
134, 189-204.
Lindenmayer, D, Mackey, B, Mullen, I, McCarthy, M, Gill, A, Cunningham, R,
Donnelly, C (1999) Factors affecting stand structure in forests–are there climatic
and topographic determinants? Forest Ecology and Management 123, 55-63.
Lindenmayer, DB (2009d) Old forest, new perspectives-Insights from the Mountain
Ash forests of the Central Highlands of Victoria, south-eastern Australia. Forest
Ecology and Management 258, 357-365.
Lindenmayer, DB, Blanchard, W, McBurney, L, Blair, D, Banks, SC, Driscoll, D,
Smith, AL, Gill, AM (2013b) Fire severity and landscape context effects on
arboreal marsupials. Biological Conservation 167, 137-148.
Lindenmayer, DB, Blanchard, W, McBurney, L, Blair, D, Banks, SC, Driscoll, DA,
Smith, AL, Gill, AM (2014) Complex responses of birds to landscape-level fire
extent, fire severity and environmental drivers. Diversity and Distributions 20,
467-477.
Lindenmayer, DB, Fischer, J (2006) 'Habitat fragmentation and landscape change: an
ecological and conservation synthesis.' (Island Press:
Lindenmayer, DB, MacGregor, C, Wood, JT, Cunningham, RB, Crane, M, Michael, D,
Montague-Drake, R, Brown, D, Fortescue, M, Dexter, N, Hudson, M, Gill, AM
(2009c) What factors influence rapid post-fire site re-occupancy? A case study
of the endangered Eastern Bristlebird in eastern Australia. International Journal
of Wildland Fire 18, 84-95.
Lindenmayer, DB, Norton, TW, Tanton, MT (1990) Differences between wildfire and
clearfelling on the structure of montane ash forests of Victoria and their
implications for fauna dependent on tree hollows. Australian forestry 53, 61-68.
Lindenmayer, DB, Wood, JT, McBurney, L, Michael, D, Crane, M, MacGregor, C,
Montague-Drake, R (2010) Comparing bird species richness and assemblage
composition between montane ash eucalypt forest and cool temperate rainforests
- an empirical study from Victoria, south-eastern Australia. Emu 110, 109-117.
MacKenzie, DI, Nichols, JD, Lachman, GB, Droege, S, Andrew Royle, J, Langtimm,
CA (2002) Estimating site occupancy rates when detection probabilities are less
than one. Ecology 83, 2248-2255.
130
Mackey, B, Berry, S, Hugh, S, Ferrier, S, Harwood, TD, Williams, KJ (2012)
Ecosystem greenspots: identifying potential drought, fire, and climate-change
micro-refuges. Ecological Applications 22, 1852-1864.
Mackey, B, Lindenmayer, D, Gill, M, McCarthy, M, Lindesay, J (2002) 'Wildlife, Fire
and Future Climate: A Forest Ecosystem Analysis.' (CSIRO publishing:
Collingwood)
Major, RE, Christie, FJ, Gowing, G (2001) Influence of remnant and landscape
attributes on Australian woodland bird communities. Biological Conservation
102, 47-66.
McCarthy, MA, Lindenmayer, DB (1998) Multi-aged mountain ash forest, wildlife
conservation and timber harvesting. Forest Ecology and Management 104, 43-
56.
Nappi, A, Drapeau, P (2009) Reproductive success of the black-backed woodpecker
(Picoides arcticus) in burned boreal forests: Are burns source habitats?
Biological Conservation 142, 1381-1391.
Nimmo, D, Kelly, L, Spence-Bailey, L, Watson, S, Taylor, R, Clarke, M, Bennett, A
(2013) Fire Mosaics and Reptile Conservation in a Fire‐Prone Region.
Conservation Biology 27, 345-353.
Nugent, DT, Leonard, SW, Clarke, MF (2014) Interactions between the superb lyrebird
(Menura novaehollandiae) and fire in south-eastern Australia. Wildlife Research
41, 203-211.
Overpeck, JT, Rind, D, Goldberg, R (1990) Climate-induced changes in forest
disturbance and vegetation. Nature 343, 51-53.
Pereoglou, F, Macgregor, C, Banks, SC, Ford, F, Wood, J, Lindenmayer, DB (2011)
Refuge site selection by the eastern chestnut mouse in recently burnt heath.
Wildlife Research 38, 290-298.
Perry, DA, Hessburg, PF, Skinner, CN, Spies, TA, Stephens, SL, Taylor, AH, Franklin,
JF, McComb, B, Riegel, G (2011) The ecology of mixed severity fire regimes in
Washington, Oregon, and Northern California. Forest Ecology and Management
262, 703-717.
Pyke, G, Recher, HF (1983) Censusing Australian birds: a summary of procedures and a
scheme for standardisation of data presentation and storage. Methods of
censusing birds in Australia. Bulletin 153, 55-63.
Robinson, NM, Leonard, SW, Ritchie, EG, Bassett, M, Chia, EK, Buckingham, S, Gibb,
H, Bennett, AF, Clarke, MF (2013) REVIEW: Refuges for fauna in fire‐prone
landscapes: their ecological function and importance. Journal of Applied
Ecology 50, 1321-1329.
Robinson, NM, Leonard, SWJ, Bennett, AF, Clarke, MF (2014) Refuges for birds in
fire-prone landscapes: The influence of fire severity and fire history on the
distribution of forest birds. Forest Ecology and Management 318, 110-121.
Sitters, H, Christie, FJ, Di Stefano, J, Swan, M, Penman, T, Collins, PC, York, A (2014)
Avian responses to the diversity and configuration of fire age classes and
vegetation types across a rainfall gradient. Forest Ecology and Management
318, 13-20.
Sitters, H, Di Stefano, J, Christie, FJ, Swan, M, York, A (2015) Bird functional
diversity decreases with time since disturbance: does patchy prescribed fire
enhance ecosystem function? Ecological Applications
Smith, AL, Blair, D, McBurney, L, Banks, SC, Barton, PS, Blanchard, W, Driscoll, DA,
Gill, AM, Lindenmayer, DB (2014) Dominant Drivers of Seedling
131
Establishment in a Fire-Dependent Obligate Seeder: Climate or Fire Regimes?
Ecosystems 17, 258-270.
Stanley, MC, Lill, A (2002) Avian fruit consumption and seed dispersal in a temperate
Australian woodland. Austral Ecology 27, 137-148.
Swanson, ME, Franklin, JF, Beschta, RL, Crisafulli, CM, DellaSala, DA, Hutto, RL,
Lindenmayer, DB, Swanson, FJ (2010) The forgotten stage of forest succession:
early-successional ecosystems on forest sites. Frontiers in Ecology and the
Environment 9, 117-125.
Symonds, MR, Moussalli, A (2011) A brief guide to model selection, multimodel
inference and model averaging in behavioural ecology using Akaike’s
information criterion. Behavioral Ecology and Sociobiology 65, 13-21.
Taylor, C, McCarthy, MA, Lindenmayer, DB (2014) Non‐linear effects of stand age on
fire severity. Conservation Letters
Taylor, RS, Watson, SJ, Nimmo, DG, Kelly, LT, Bennett, AF, Clarke, MF (2012)
Landscape-scale effects of fire on bird assemblages: does pyrodiversity beget
biodiversity? Diversity and Distributions 18, 519-529.
Thonicke, K, Venevsky, S, Sitch, S, Cramer, W (2001) The role of fire disturbance for
global vegetation dynamics: coupling fire into a Dynamic Global Vegetation
Model. Global Ecology and Biogeography 10, 661-677.
Watson, SJ, Taylor, RS, Nimmo, DG, Kelly, LT, Clarke, MF, Bennett, AF (2012) The
influence of unburnt patches and distance from refuges on post-fire bird
communities. Animal Conservation 15, 499-507.
Welsh, AH, Lindenmayer, DB, Donnelly, CF (2015) Fitting and Interpreting Occupancy
Models. Plos One 8, e52015.
Wood, SW, Murphy, BP, Bowman, DMJS (2011) Firescape ecology: how topography
determines the contrasting distribution of fire and rain forest in the south-west of
the Tasmanian Wilderness World Heritage Area. Journal of Biogeography 38,
1807-1820.
Yarnell, RW, Metcalfe, DJ, Dunstone, N, Burnside, N, Scott, DM (2008) The impact of
fire on habitat use by the short-snouted elephant shrew (Elephantulus
brachyrhynchus) in North West Province, South Africa. African Zoology 43, 45-
52.
132
4.9 Appendix 1
133
4.10 Appendix 2
Species list and latin names
Species Scientific Name
Australian King-Parrot Alisterus scapularis
Bassian Thrush Zoothera lunulata
Black-faced Cuckoo-shrike Coracina novaehollandiae
Brown Thornbill Acanthiza pusilla
Brown-headed Honeyeater Melithreptus brevirostris
Brush Cuckoo cacomantis variolosus
Cicadabird Coracina tenuirostris
Crescent Honeyeater Phylidonyris pyrrhopterus
Crested Shrike tit Falcunculus frontatus
Crimson Rosella Platycercus elegans
Eastern Spinebill Acanthorhynchus tenuirostris
Eastern Whipbird Psophodes olivaceus
Eastern Yellow Robin Eopsaltria australis
Fan tailed Cuckoo Cacomantis flabelliformis
Flame Robin Petroica phoenicea
Forest Raven Corvus tasmanicus
Gang gang Cockatoo Callocephalon fimbriatum
Golden Whistler Pachycephala pectoralis
Brown Goshawk Accipiter fasciatus
Grey Currawong Strepera versicolor
Grey Fantail Rhipidura albiscapa
Grey Shrike thrush Colluricincla harmonica
Horsfield's Bronze-Cuckoo Chalcites baslis
Sacred Kingfisher Todiramphus sanctus
Large-billed Scrubwren Sericornis magnirostra
Laughing Kookaburra Dacelo novaeguineae
Lewin's Honeyeater Meliphaga lewinii
Olive Whistler Pachycephala olivacea
Pallid Cuckoo Cuculus pallidus
Pied Butcherbird Cracticus nigrogularis
Pied Currawong Strepera graculina
Pilotbird Pycnoptilus floccosus
Pink Robin Petroica rodinogaster
Red Wattlebird Anthochaera carunculata
Rose Robin Petroica rosea
Rufous Fantail Rhipidura rufifrons
Satin Bowerbird Ptilonorhynchus violaceus
Shining BronzeCuckoo Chalcites lucidus
Silvereye Zosterops lateralis
Painted Buttonquail Turnix varius
Spotted Pardalote Pardalotus punctatus
Striated Pardalote Pardalotus striatus
Striated Thornbill Acanthiza lineata
Superb Fairy wren Malurus cyaneus
Superb Lyrebird Menura novaehollandiae
134
White browed Scrubwren Sericornis frontalis
White eared Honeyeater Lichenostomus leucotis
White naped Honeyeater Melithreptus lunatus
White throated Treecreeper Cormobates leucophaea
Willie Wagtail Rhipidura leucophrys
Yellow faced Honeyeater Lichenostomus chrysops
Yellow-tailed Black-Cockatoo Calyptorhynchus funereus
135
Appendix 3
List of binomial generalised linear model equations used in model selection. Type = 5
level treatment design, Type2 = 4 level treatment design, Severity = 3 level treatment
design, Type3 = 2 level treatment design, B300 = within the surrounding 300 m, B3K =
within the surrounding 3 km. PC1 = vegetation principal component 1, PC2 =
vegetation principal component 2, twi = topographic wetness index, elev = elevation,
precip = precipitation, highsev = high severity fire, lowsev = low severity fire. “:” =
indicates an interaction. “bird and trait occurrence” = indicates response variable.
Models
t0m0 bird and trait occurrence ~1
t0m1 bird and trait occurrence ~type+type2
t0m2 bird and trait occurrence ~type+type3
t0m3 bird and trait occurrence ~type+severity
t0m4 bird and trait occurrence ~type2+type3
t0m5 bird and trait occurrence ~type2+severity
t0m6 bird and trait occurrence ~type3+severity
t1m1 bird and trait occurrence ~type
t1m2 bird and trait occurrence ~type+PC1
t1m3 bird and trait occurrence ~type+PC1+type:PC1
t1m4 bird and trait occurrence ~type+precip
t1m5 bird and trait occurrence ~type+precip+type:precip
t1m6 bird and trait occurrence ~type+twi
t1m7 bird and trait occurrence ~type+twi+type:twi
t1m8 bird and trait occurrence ~type+PC1
t1m9 bird and trait occurrence ~type+PC1+type:PC1
t1m11 bird and trait occurrence ~type+PC2
t1m12 bird and trait occurrence ~type+PC2+type:PC2
t1m13 bird and trait occurrence ~type+b3khighsev
t1m14 bird and trait occurrence ~type+b3khighsev+b3khighsev:type
t1m15 bird and trait occurrence ~type+b3kmodsev
t1m16 bird and trait occurrence ~type+b3kmodsev+b3kmodsev:type
t1m17 bird and trait occurrence ~type+b3klowsev
t1m18 bird and trait occurrence ~type+b3klowsev+b3klowsev:type
t1m19 bird and trait occurrence ~type+b300highsev
t1m20 bird and trait occurrence ~type+b300highsev+b300highsev:type
t1m21 bird and trait occurrence ~type+b300modsev
t1m22 bird and trait occurrence ~type+b300modsev+b300modsev:type
t1m23 bird and trait occurrence ~type+b300lowsev
t1m24 bird and trait occurrence ~type+b300lowsev+b300lowsev:type
t2m1 bird and trait occurrence ~type2
t2m2 bird and trait occurrence ~type2+PC1
t2m3 bird and trait occurrence ~type2+PC1+type2:PC1
t2m4 bird and trait occurrence ~type2+precip
t2m5 bird and trait occurrence ~type2+precip+type2:precip
t2m6 bird and trait occurrence ~type2+twi
136
t2m7 bird and trait occurrence ~type2+twi+type2:twi
t2m8 bird and trait occurrence ~type2+PC1
t2m9 bird and trait occurrence ~type2+PC1+type2:PC1
t2m11 bird and trait occurrence ~type2+PC2
t2m12 bird and trait occurrence ~type2+PC2+type2:PC2
t2m13 bird and trait occurrence ~type2+b3khighsev
t2m14 bird and trait occurrence ~type2+b3khighsev+b3khighsev:type2
t2m15 bird and trait occurrence ~type2+b3kmodsev
t2m16 bird and trait occurrence ~type2+b3kmodsev+b3kmodsev:type2
t2m17 bird and trait occurrence ~type2+b3klowsev
t2m18 bird and trait occurrence ~type2+b3klowsev+b3klowsev:type2
t2m19 bird and trait occurrence ~type2+b300highsev
t2m20 bird and trait occurrence ~type2+b300highsev+b300highsev:type2
t2m21 bird and trait occurrence ~type2+b300modsev
t2m22 bird and trait occurrence ~type2+b300modsev+b300modsev:type2
t2m23 bird and trait occurrence ~type2+b300lowsev
t2m24 bird and trait occurrence ~type2+b300lowsev+b300lowsev:type2
t3m1 bird and trait occurrence ~type3
t3m2 bird and trait occurrence ~type3+PC1
t3m3 bird and trait occurrence ~type3+PC1+type3:PC1
t3m4 bird and trait occurrence ~type3+precip
t3m5 bird and trait occurrence ~type3+precip+type3:precip
t3m6 bird and trait occurrence ~type3+twi
t3m7 bird and trait occurrence ~type3+twi+type3:twi
t3m8 bird and trait occurrence ~type3+PC1
t3m9 bird and trait occurrence ~type3+PC1+type3:PC1
t3m11 bird and trait occurrence ~type3+PC2
t3m12 bird and trait occurrence ~type3+PC2+type3:PC2
t3m13 bird and trait occurrence ~type3+b3khighsev
t3m14 bird and trait occurrence ~type3+b3khighsev+b3khighsev:type3
t3m15 bird and trait occurrence ~type3+b3kmodsev
t3m16 bird and trait occurrence ~type3+b3kmodsev+b3kmodsev:type3
t3m17 bird and trait occurrence ~type3+b3klowsev
t3m18 bird and trait occurrence ~type3+b3klowsev+b3klowsev:type3
t3m19 bird and trait occurrence ~type3+b300highsev
t3m20 bird and trait occurrence ~type3+b300highsev+b300highsev:type3
t3m21 bird and trait occurrence ~type3+b300modsev
t3m22 bird and trait occurrence ~type3+b300modsev+b300modsev:type3
t3m23 bird and trait occurrence ~type3+b300lowsev
t3m24 bird and trait occurrence ~type3+b300lowsev+b300lowsev:type3
t4m1 bird and trait occurrence ~severity
t4m2 bird and trait occurrence ~severity+PC1
t4m3 bird and trait occurrence ~severity+PC1+severity:PC1
t4m4 bird and trait occurrence ~severity+precip
t4m5 bird and trait occurrence ~severity+precip+severity:precip
t4m6 bird and trait occurrence ~severity+twi
t4m7 bird and trait occurrence ~severity+twi+severity:twi
137
t4m8 bird and trait occurrence ~severity+PC1
t4m9 bird and trait occurrence ~severity+PC1+severity:PC1
t4m11 bird and trait occurrence ~severity+PC2
t4m12 bird and trait occurrence ~severity+PC2+severity:PC2
t4m13 bird and trait occurrence ~severity+b3khighsev
t4m14 bird and trait occurrence ~severity+b3khighsev+b3khighsev:severity
t4m15 bird and trait occurrence ~severity+b3kmodsev
t4m16 bird and trait occurrence ~severity+b3kmodsev+b3kmodsev:severity
t4m17 bird and trait occurrence ~severity+b3klowsev
t4m18 bird and trait occurrence ~severity+b3klowsev+b3klowsev:severity
t4m19 bird and trait occurrence ~severity+b300highsev
t4m20 bird and trait occurrence ~severity+b300highsev+b300highsev:severity
t4m21 bird and trait occurrence ~severity+b300modsev
t4m22 bird and trait occurrence ~severity+b300modsev+b300modsev:severity
t4m23 bird and trait occurrence ~severity+b300lowsev
t4m24 bird and trait occurrence ~severity+b300lowsev+b300lowsev:severity
t5m1 bird and trait occurrence ~b3khighsev
t5m2 bird and trait occurrence ~b3khighsev+PC1
t5m3 bird and trait occurrence ~b3khighsev+PC1+b3khighsev:PC1
t5m4 bird and trait occurrence ~b3khighsev+precip
t5m5 bird and trait occurrence ~b3khighsev+precip+b3khighsev:precip
t5m6 bird and trait occurrence ~b3khighsev+twi
t5m7 bird and trait occurrence ~b3khighsev+twi+b3khighsev:twi
t5m8 bird and trait occurrence ~b3khighsev+PC1
t5m9 bird and trait occurrence ~b3khighsev+PC1+b3khighsev:PC1
t5m11 bird and trait occurrence ~b3khighsev+PC2
t5m12 bird and trait occurrence ~b3khighsev+PC2+b3khighsev:PC2
t5m13 bird and trait occurrence ~b3khighsev+b3kmodsev
t5m14 bird and trait occurrence ~b3khighsev+b3kmodsev+b3kmodsev:b3khighsev
t5m15 bird and trait occurrence ~b3khighsev+b3klowsev
t5m16 bird and trait occurrence ~b3khighsev+b3klowsev+b3klowsev:b3khighsev
t5m17 bird and trait occurrence ~b3khighsev+b300highsev
t5m18 bird and trait occurrence ~b3khighsev+b300highsev+b300highsev:b3khighsev
t5m19 bird and trait occurrence ~b3khighsev+b300modsev
t5m20 bird and trait occurrence ~b3khighsev+b300modsev+b300modsev:b3khighsev
t5m21 bird and trait occurrence ~b3khighsev+b300lowsev
t5m22 bird and trait occurrence ~b3khighsev+b300lowsev+b300lowsev:b3khighsev
t6m1 bird and trait occurrence ~b3kmodsev
t6m2 bird and trait occurrence ~b3kmodsev+PC1
t6m3 bird and trait occurrence ~b3kmodsev+PC1+b3kmodsev:PC1
t6m4 bird and trait occurrence ~b3kmodsev+precip
t6m5 bird and trait occurrence ~b3kmodsev+precip+b3kmodsev:precip
t6m6 bird and trait occurrence ~b3kmodsev+twi
t6m7 bird and trait occurrence ~b3kmodsev+twi+b3kmodsev:twi
t6m8 bird and trait occurrence ~b3kmodsev+PC1
t6m9 bird and trait occurrence ~b3kmodsev+PC1+b3kmodsev:PC1
t6m11 bird and trait occurrence ~b3kmodsev+PC2
t6m12 bird and trait occurrence ~b3kmodsev+PC2+b3kmodsev:PC2
138
t6m13 bird and trait occurrence ~b3kmodsev+b3khighsev
t6m14 bird and trait occurrence ~b3kmodsev+b3khighsev+b3kmodsev:b3khighsev
t6m15 bird and trait occurrence ~b3kmodsev+b3klowsev
t6m16 bird and trait occurrence ~b3kmodsev+b3klowsev+b3klowsev:b3kmodsev
t6m17 bird and trait occurrence ~b3kmodsev+b300highsev
t6m18 bird and trait occurrence ~b3kmodsev+b300highsev+b300highsev:b3kmodsev
t6m19 bird and trait occurrence ~b3kmodsev+b300modsev
t6m20 bird and trait occurrence ~b3kmodsev+b300modsev+b300modsev:b3kmodsev
t6m21 bird and trait occurrence ~b3kmodsev+b300lowsev
t6m22 bird and trait occurrence ~b3kmodsev+b300lowsev+b300lowsev:b3kmodsev
t7m1 bird and trait occurrence ~b3klowsev
t7m2 bird and trait occurrence ~b3klowsev+PC1
t7m3 bird and trait occurrence ~b3klowsev+PC1+b3klowsev:PC1
t7m4 bird and trait occurrence ~b3klowsev+precip
t7m5 bird and trait occurrence ~b3klowsev+precip+b3klowsev:precip
t7m6 bird and trait occurrence ~b3klowsev+twi
t7m7 bird and trait occurrence ~b3klowsev+twi+b3klowsev:twi
t7m8 bird and trait occurrence ~b3klowsev+PC1
t7m9 bird and trait occurrence ~b3klowsev+PC1+b3klowsev:PC1
t7m11 bird and trait occurrence ~b3klowsev+PC2
t7m12 bird and trait occurrence ~b3klowsev+PC2+b3klowsev:PC2
t7m13 bird and trait occurrence ~b3klowsev+b3khighsev
t7m14 bird and trait occurrence ~b3klowsev+b3khighsev+b3khighsev:b3khighsev
t7m15 bird and trait occurrence ~b3klowsev+b3kmodsev
t7m16 bird and trait occurrence ~b3klowsev+b3kmodsev+b3klowsev:b3kmodsev
t7m17 bird and trait occurrence ~b3klowsev+b300highsev
t7m18 bird and trait occurrence ~b3klowsev+b300highsev+b300highsev:b3klowsev
t7m19 bird and trait occurrence ~b3klowsev+b300modsev
t7m20 bird and trait occurrence ~b3klowsev+b300modsev+b300modsev:b3klowsev
t7m21 bird and trait occurrence ~b3klowsev+b300lowsev
t7m22 bird and trait occurrence ~b3klowsev+b300lowsev+b300lowsev:b3klowsev
t8m1 bird and trait occurrence ~b300highsev
t8m2 bird and trait occurrence ~b300highsev+PC1
t8m3 bird and trait occurrence ~b300highsev+PC1+b300highsev:PC1
t8m4 bird and trait occurrence ~b300highsev+precip
t8m5 bird and trait occurrence ~b300highsev+precip+b300highsev:precip
t8m6 bird and trait occurrence ~b300highsev+twi
t8m7 bird and trait occurrence ~b300highsev+twi+b300highsev:twi
t8m8 bird and trait occurrence ~b300highsev+PC1
t8m9 bird and trait occurrence ~b300highsev+PC1+b300highsev:PC1
t8m11 bird and trait occurrence ~b300highsev+PC2
t8m12 bird and trait occurrence ~b300highsev+PC2+b300highsev:PC2
t8m13 bird and trait occurrence ~b300highsev+b3khighsev
t8m14 bird and trait occurrence ~b300highsev+b3khighsev+b300highsev:b3khighsev
t8m15 bird and trait occurrence ~b300highsev+b3kmodsev
t8m16 bird and trait occurrence ~b300highsev+b3kmodsev+b300highsev:b3kmodsev
t8m17 bird and trait occurrence ~b300highsev+b3klowsev
t8m18 bird and trait occurrence ~b300highsev+b3klowsev+b300highsev:b3klowsev
139
t8m19 bird and trait occurrence ~b300highsev+b300modsev
t8m20 bird and trait occurrence
~b300highsev+b300modsev+b300modsev:b300highsev
t8m21 bird and trait occurrence ~b300highsev+b300lowsev
t8m22 bird and trait occurrence
~b300highsev+b300lowsev+b300lowsev:b300highsev
t9m1 bird and trait occurrence ~b300modsev
t9m2 bird and trait occurrence ~b300modsev+PC1
t9m3 bird and trait occurrence ~b300modsev+PC1+b300modsev:PC1
t9m4 bird and trait occurrence ~b300modsev+precip
t9m5 bird and trait occurrence ~b300modsev+precip+b300modsev:precip
t9m6 bird and trait occurrence ~b300modsev+twi
t9m7 bird and trait occurrence ~b300modsev+twi+b300modsev:twi
t9m8 bird and trait occurrence ~b300modsev+PC1
t9m9 bird and trait occurrence ~b300modsev+PC1+b300modsev:PC1
t9m11 bird and trait occurrence ~b300modsev+PC2
t9m12 bird and trait occurrence ~b300modsev+PC2+b300modsev:PC2
t9m13 bird and trait occurrence ~b300modsev+b3khighsev
t9m14 bird and trait occurrence ~b300modsev+b3khighsev+b300modsev:b3khighsev
t9m15 bird and trait occurrence ~b300modsev+b3kmodsev
t9m16 bird and trait occurrence ~b300modsev+b3kmodsev+b300modsev:b3kmodsev
t9m17 bird and trait occurrence ~b300modsev+b3klowsev
t9m18 bird and trait occurrence ~b300modsev+b3klowsev+b300modsev:b3klowsev
t9m19 bird and trait occurrence ~b300modsev+b300highsev
t9m20 bird and trait occurrence
~b300modsev+b300highsev+b300modsev:b300highsev
t9m21 bird and trait occurrence ~b300modsev+b300lowsev
t9m22 bird and trait occurrence
~b300modsev+b300lowsev+b300lowsev:b300modsev
t10m1 bird and trait occurrence ~b300lowsev
t10m2 bird and trait occurrence ~b300lowsev+PC1
t10m3 bird and trait occurrence ~b300lowsev+PC1+b300lowsev:PC1
t10m4 bird and trait occurrence ~b300lowsev+precip
t10m5 bird and trait occurrence ~b300lowsev+precip+b300lowsev:precip
t10m6 bird and trait occurrence ~b300lowsev+twi
t10m7 bird and trait occurrence ~b300lowsev+twi+b300lowsev:twi
t10m8 bird and trait occurrence ~b300lowsev+PC1
t10m9 bird and trait occurrence ~b300lowsev+PC1+b300lowsev:PC1
t10m11 bird and trait occurrence ~b300lowsev+PC2
t10m12 bird and trait occurrence ~b300lowsev+PC2+b300lowsev:PC2
t10m13 bird and trait occurrence ~b300lowsev+b3khighsev
t10m14 bird and trait occurrence ~b300lowsev+b3khighsev+b300lowsev:b3khighsev
t10m15 bird and trait occurrence ~b300lowsev+b3kmodsev
t10m16 bird and trait occurrence ~b300lowsev+b3kmodsev+b300lowsev:b3kmodsev
t10m17 bird and trait occurrence ~b300lowsev+b3klowsev
t10m18 bird and trait occurrence ~b300lowsev+b3klowsev+b300lowsev:b3klowsev
t10m19 bird and trait occurrence ~b300lowsev+b300highsev
140
t10m21 bird and trait occurrence ~b300lowsev+b300modsev
t10m22 bird and trait occurrence
~b300lowsev+b300modsev+b300lowsev:b300modsev
141
Chapter 5. Fire severity patterns alter spatio-
temporal movements and habitat utilization by an
arboreal marsupial, the Mountain Brushtail
Possum Trichosurus Cunninghamii
Berry, L. E., Lindenmayer, D. B., Driscoll, D. A., Dennis, T. & Banks, S. C. (2015) Fire
severity alters spatio-temporal movements and habitat utilization by an arboreal
marsupial, the Mountain Brushtail Possum (Trichosurus Cunninghami).
International Journal of Wildland Fire. In Press.
142
143
5.1 Abstract
Understanding how severe wildfires influence faunal movement is essential for
predicting how changes in fire frequency will affect ecosystems. Our study examined
the effects of fire severity distribution on spatial and temporal variation in the
movement patterns of the Australian arboreal mammal, the mountain brushtail possum,
Trichosurus cunninghami. We used GPS telemetry to characterise the movements of 18
possums within landscapes burnt to differing extents by a large wildfire. We analysed
relationships between movement, landscape structure and resource availability.
We identified a temporal change in movement patterns in response to fire. In unburnt
landscapes, individuals moved greater distances early and late in the night and had less
overlap in the areas used for foraging and denning, than in high severity burnt
landscapes. We also found habitat selection was dependent on the spatial context of fire
in the surrounding landscape. Our results suggested non-significant trend for smaller
home ranges in high-severity burnt landscapes.
Forest systems recently burnt at high severity may provide suitable habitat for species
such as the Mountain brushtail possum, if protected from subsequent disturbance, such
as salvage logging. However, spatial and temporal patterns of habitat use and selection
differ considerably between burnt and undisturbed landscapes. The spatial outcomes of
ecological disturbances such as wildfires have the potential to alter the behaviour and
functional roles of fauna within recently burnt forests.
144
5.2 Introduction
Landscape-scale wildfires are a major cause of ecological disturbance and are a globally
important determinant of biodiversity patterns in forested systems (Bond and Keeley
2005; Bowman et al. 2009). Fires alter forest structure and create complex mosaics of
habitat burnt at differing severities across landscapes (Bradstock et al. 2005). Within
montane forests, interactions between topography, fuels and weather during fires often
produce mixed-severity fire patterns that present heterogeneous resource opportunities
to fauna (Lentile et al. 2006; Perry et al. 2011). However, increasingly extreme weather
conditions and widespread land-use changes generate favourable conditions for large-
scale, high-severity crown-fires (Bradstock et al. 2002; Bond et al. 2005; Chapin et al.
2011). Such crown-fire events may result in large homogeneous areas of forest burnt at
high-severity, and reduce the likelihood of potential refuges persisting within burnt
landscapes (Berry et al. 2015b). The loss and spatial redistribution of essential habitat
components poses a major challenge to faunal conservation efforts in fire-affected
forests (Lindenmayer et al. 2013b; Nugent et al. 2014). Identifying faunal responses to
the spatial patterns of habitat produced by fire will facilitate improved management of
fire-affected ecosystems for biodiversity conservation.
Quantifying faunal movement patterns within heterogeneous landscapes is central to
identifying the primary drivers of biotic responses to key ecological processes such as
habitat disturbance (Nathan et al. 2008). Faunal movement patterns can vary in relation
to temporal changes in abiotic and biotic environmental conditions such as drought
(Yospin et al. 2015) or seasonal differences in predator abundance (Börger et al. 2006).
Similar changes to patterns of animal movement are expected following fire due to
substantial alterations in forest structure, floristic composition and the spatiotemporal
distribution of key resources (Clarke 2008). Changes in home-range areas in response to
145
disturbance-mediated habitat redistribution may have considerable impacts on important
ecosystem processes such as predation, pollination, seed dispersal and herbivory.
Fire can have positive impacts for some wildlife species in forests by creating diverse
complex, multi-aged habitat mosaics (Bradstock et al. 2005; Yospin et al. 2015). In
montane ecosystems, areas where fire regimes differ from those prevailing in the
surrounding landscape often present unique resources unavailable in the surrounding
burnt forest, such as mature canopies (Mackey et al. 2012). Some animal species are
able to exploit the contrasting resource opportunities presented by both burnt and
unburnt areas (Kelly et al. 2012; Taylor et al. 2012; Berry et al. 2015c). For these
species, persistence within burnt landscapes is likely to depend on the ability to adapt
spatio-temporal foraging patterns to the challenges and opportunities presented by novel
habitat patterns, such as changes in the spatial distribution of competition and predation
pressures as a consequence of resource scarcity and open habitat structure (Torre and
Díaz 2004).
Few examples exist of studies that explore how species’ movement behaviour and home
range use are influenced by fire. Following large fires, some mammals may disperse to
unburnt areas which present desirable features for denning and feeding (Hutto and Gallo
2006). For example, during an experimental low-severity burn, Northern Bettongs,
Bettongia tropica, relocated home ranges from areas of grassy tussocks and logs to
rocky areas and remnant patches of vegetation (Vernes and Pope 2001). Similarly,
following a mixed-severity burn in a desert ecosystem, small mammals utilized a
mosaic of burnt and unburnt habitat and had sufficient dispersal ability to locate high-
quality habitat patches within the burnt zone (Parr and Andersen 2006). However, some
animal species respond positively to the novel resource opportunities presented by fire.
For example, the Florida panther, Puma concolor, was found to actively disperse to
146
patches of Pinus forest burnt the previous year, as these areas had higher abundance of
the key prey species, white-tailed deer, Odocoileus virginianus (Dees et al. 2001). To
our knowledge, however, no studies have considered how spatio-temporal faunal
movement dynamics vary within landscapes that have been burnt to differing extents by
wildfires.
We exploited the opportunity presented by a large-scale wildfire to examine the
relationship between fire-induced patterns of vegetation cover and the movement
patterns of the Mountain brushtail possum, Trichosurus cunninghami in Victoria, SE
Australia. During 2009, the Black Saturday bushfires burnt over 74,000 ha of mountain
ash forest in the Victorian Central Highlands, 120 km northeast of Melbourne (Teague
et al. 2010; Robinson et al. 2014). The 2009 fires dramatically altered the distribution
and abundance of many arboreal marsupials in these forests (Lindenmayer et al. 2013b).
The Mountain brushtail possum was the only species present in significant numbers
throughout landscapes that were burnt to differing extents by the 2009 fires. This
species is nocturnal and dependent on large hollow-bearing trees for denning
(Lindenmayer et al. 1998). Previous studies have found that Mountain brushtail
possums favour den trees with a large number of cavities that are not surrounded by
dense vegetation (Lindenmayer et al. 1996).
In this study, we address three key questions: 1) How do fire-severity patterns influence
the home-ranges of T. cunninghami?; 2) Within home-ranges, do individuals select
particular areas burnt at different fire severities relative to availability?; and 3) Are there
differences in movement patterns within home-ranges among animals occupying
landscapes characterized by different fire-severity patterns? We predict that individuals
will alter their spatial and temporal movement patterns in fire-affected landscapes in
response to the availability and redistribution of foraging and denning resources. We
147
expect movement response patterns to depend on whether high-severity fire restricts or
promotes habitat values. If high-severity fires limit resource availability, we expect to
observe larger home ranges and preferential selection of low-severity burned areas
within extensively burnt landscapes as individuals maximise the availability of spatially
limited resources within their ranges. If high severity fire promotes resource availability
we expect to observe smaller home ranges, non-preferential habitat selection and
feeding and foraging activity to occur in mutual space due to the ubiquitous availability
of resources across fire severity categories.
148
5.3 Methods
Study Region
During 2009, a large wildfire burnt ~74,000 ha of mountain ash forest (Eucalyptus
regnans) in Victorian Central Highlands, approximately 120 km north-east of
Melbourne, Australia (Gibbons et al. 2012b). The region is topographically diverse with
local variation in elevation up to 1000 m. Mean annual temperature varies from 7.8 °C
to 13.4 °C. Vegetation in our study area was dominated by mountain ash and alpine ash
(Eucalyptus delegatensis). Mountain ash trees are obligate-seeders (Smith et al. 2014),
which grow to 100 m in height and are subject to mean fire intervals of 30 – 300 years
(Lindenmayer 2009).
Figure 1. Map of study sites displaying replicate blocks of three landscape classes
within the extent of the 2009 Black Saturday wildfire in the Victorian Central
Highlands, Australia. White areas indicate unburnt forest.
149
Site selection and study design
To investigate whether movement behaviour of T. cunninghami varied in response to
altered patterns of resource distribution, we established a replicated landscape-scale
telemetry study (Figure 1). We identified three distinct landscape classes five years
following a large wildfire, using remotely-sensed fire-severity maps and aerial
photography (DSE, 2009). These were: 1) landscapes mostly burnt at high fire-
severities; 2) landscapes representative of a range of fire-severities (mixed severity);
and 3) landscapes burnt mostly at low fire-severities (see Table 1 for fire-severity
classifications). We defined individual landscape sampling units as circles with a radius
of 1 km. We established three replicates of each landscape class across areas of
mountain ash forest affected by the 2009 fires to give a total of nine sites. All of the
sites were located within National Parks or closed water-catchment areas that are
protected from clearfell logging. To account for sex-specific differences in movement
patterns, we deployed GPS collars on one adult female and one adult male at each site,
giving a total of 18 animals.
Trapping protocol and collar deployment
We trapped animals within our nine field sites between 3rd February and 7th April 2014.
At each site we established two 120 m transects placed 50 m apart. We set traps baited
with apples at 20 m intervals along each transect (12 traps per site). We deployed large
(~120 x 40 x 40 cm) wire cage traps that use a treadle mechanism to close the trap. Each
trap was covered with a hessian sack to serve as shelter for the captured animals. When
possible, traps were placed on or near habitat features commonly used by T.
cunninghami, such as hollow-bearing trees and large fallen logs (Lindenmayer et al.
1990a). We checked each trap before first light, and reset traps where animals trapped
150
were recaptures or unsuitable for collar deployment. We continued trapping at each site
until one adult female and one adult male (as determined by both weight and tooth
wear-based estimation of age class (Winter 1980)) had been captured. Following release
of collared animals, traps were removed from each site. After the two-week deployment
period, collared animals were radio-racked to den trees using a VHF receiver and traps
were placed around the base of the den tree to capture the individual and remove the
collar.
Table 1. The mean percentages of each fire-severity category within each
landscape class.
Fire-severity category
Landscape Class
High mixed Low
1. Crown burn 7.23 1.19 0.02
2. Crown scorch 49.97 33.37 1.70
3. Moderate crown scorch 26.07 39.12 7.67
4. Light or no crown scorch, understorey burnt 8.80 19.98 30.06
5. No crown scorch, no understorey burn 7.92 6.34 60.55
We sedated trapped animals using an intra-muscular injection of Zoletil [Tiletamine as
hydrochloride (250 mg) and Zolazepam as hydrochloride (250 mg)] (Viggers and
Lindenmayer 1995). The injection was directed into the gluteal muscles or the muscles
of the cranial thigh. We identified each individual trapped using Trovan ID100
microchips implanted with a Trovan Deluxe implanter. To ensure collared individuals
were not overly disparate in condition, we determined the sex of each animal, measured
the body length, tail length, head length, ear length, foot length, testicular length, tooth
151
age wear class (Winter 1980), checked for pouch young and weighed all individuals
trapped.
We fitted the 18 possums with Sirtrack GPS collars and Australian Telemetry Solutions
VHF units that recorded at 15-min fixed intervals for a period of two weeks. This two
week period was chosen to maximise the number of collar deployments within the
logistical constraints of the study. Collars began recording after sunset at 1900 and
recorded in cycles of 11 hours on and 13 hours off.
Vegetation surveys
To establish the differences in habitat structure between landscape classes burnt at
different fire severities, we conducted vegetation surveys surrounding den trees being
used by individual possums. We identified the location of a den tree from each
individual one week after GPS collar deployment using a VHF radio receiver. We
conducted a 1-ha gridded vegetation survey comprised of 5 x 5 400-m2 cells, placing the
den tree in the centre of the grid. Within each grid cell, we visually estimated the
percentage cover of litter, canopy and grasses, hollow-bearing trees, the total number of
tree hollows, mountain ash trees in three different size classes (<50 cm, 51 cm – 1 m, >1
m diameter breast height), Nothofagus, silver wattles, montane wattles, tree ferns and
Sassafras. We chose to measure these plant species based on the analysis of T.
cunninghami diet conducted by Seebeck et al. (1984).
Analysis
Prior to analyses, we converted the coordinates collected from the GPS collars from the
datum WGS84 to projected coordinates system GDA 1994 MGA zone 55 in ArcMap.
Home-range isopleths were calculated as the minimum area covered by 99%, 75% and
152
50% of the total GPS fixes for each individual. These scales are commonly used to
differentiate between maximum range (99%), foraging range (75%) and core denning
range (50%) (Börger et al. 2006). We calculated the isopleths using the open source
software package Geospatial Modelling Environment (GME; (Beyer 2012). We used
linear models and Wald tests to identify differences in the area of home-range isopleths
(99%, 75%, and 50%) between each landscape class.
We used resource-selection analyses to determine whether individuals preferred areas of
particular fire severity given the availability of each fire-severity class within their home
ranges, and whether the selection of fire-severity categories differed among landscape
classes (Millspaugh et al. 2006). We tested these relationships by comparing how the
proportions of landscape classes varied within the utilization distributions of
individuals, using the approach outlined by (Marzluff et al. 2004). Briefly, this method
involved calculating the kernel-density estimates (KDE) of the area within the 99%
isopleth of each individual’s home-range area, using the software package GME. We
tested this relationship using quasi-Poisson generalised linear mixed models of the
density estimates of possum locations fitted as the response variable, and fire severity,
landscape class and their interaction fitted as predictor variables; individuals were fitted
as a random effect. We used a quasi-Poisson distribution to account for over-dispersion.
This analysis was conducted using the ‘lme4’ package in R statistical software (R-
Development-Core-Team 2005).
We used a generalised additive model to determine whether there was temporal
variation in step lengths, the distance covered between each GPS fix, throughout each
night for individuals in landscapes burnt to differing proportions. We fitted step length
as the response variable, and time of night, landscape class and their interaction as
predictor variables. We accounted for spatial-autocorrelation by fitting time of the night
153
at a response variable, which is a surrogate measure of GPS fixes that are close in space.
We calculated step length from our GPS data using the ‘movement path metrics’
function in GME. The generalised additive model was assessed using the ‘mgvc’
package (Wood 2001) in R.
We tested whether vegetation structure differed between landscape classes (high,
moderate, low) using an analysis of variance (ANOVA). We used Wald Tests to
evaluate differences in vegetation structure and composition between landscape classes.
154
5.4 Results
Vegetation characteristics
We found significant differences in vegetation characteristics among landscape classes
(Table. 2). Burnt landscapes were characterized by the absence of canopy cover, live
Mountain Ash >1 m in diameter, and rainforest plants such as Sassafras and
Nothofagus. Burnt landscapes had high grass cover and high numbers of Silver Wattle
and juvenile Mountain Ash saplings. Unburnt landscapes were characterized by
rainforest vegetation structures, high canopy cover, and greater numbers of large
Mountain Ash trees.
Table 2. Differences in vegetation structure and composition between landscape
classes. All responses except canopy and grass cover represent counts of each
vegetation component within 1 ha of each den tree.
Response
Model Wald B:M Wald B:U Wald M:U
F P X2 P X2 P X2 P
Canopy cover (%) 21.9 <0.001 40.1 <0.001 123.7 <0.001 43.8 <0.001
Stags 0.613 0.55 28.7 <0.001 34.7 <0.001 1.2 0.54
Hollows 4.154 0.037 66.4 <0.001 75.3 <0.001 8.3 0.016
ash <50 cm 7.894 0.005 38.2 <0.001 38.2 <0.001 15.8 <0.001
ash > 1m 11.47 0.001 32 <0.001 6836 <0.001 22.9 <0.001
Nothofagus 9.064 0.003 6.4 0.041 42.4 <0.001 18.1 <0.001
Silver wattle 79.15 <0.001 324.2 <0.001 326 <0.001 158.3 <0.001
Montane wattle 6.635 0.009 2.9 0.23 28.5 <0.001 13.3 0.001
Tree ferns 5.755 0.014 44.5 <0.001 87.5 <0.001 11.5 0.003
Sassafrass 2.418 0.123 9.4 0.009 10.6 0.014 4.8 0.089
Grass cover (%) 10.86 0.001 62.2 <0.001 62.3 <0.001 21.7 <0.001
Differences in home-range area among landscape classes
We found no significant difference in core, foraging and maximum home-range area
between landscape classes (Table. 3). There was a non-significant trend for larger home
ranges in unburnt landscapes than in mixed landscapes, and smaller home ranges in
burnt landscapes (Figure, 2). When fitted as a continuous variable, there was a non-
significant trend for smaller 99 % home-range areas with higher percentages of high-
155
severity fire within the surrounding 1-km buffer zone (R2 = 0.1216, F = 3.354, D = 16,
P = 0.086). Similar non-significant trends were observed for 75% home range sizes (R2
= 0.1064, F = 3.025, DF = 16, P = 0.1012) and 50% home range sizes (R2 = 0.07243, F
= 2.327, DF = 16, P = 0.1466).
Table 3 Results from ANOVA predicting differences in home-range area between
landscape classes, indicating no siginificant difference in home ranges between
landscapes burnt to differing fire-severity proportions.
Model R2 F P DF
Maximum range (99%) 0.002 1.010 0.418 14
Foraging range (75%) 0.015 1.088 0.387 14
Core range (50%) -0.048 0.738 0.547 14
Figure 2. Differences in home-range areas of mountain brushtail possums
inhabiting landscapes burnt to differing fire-severity proportions. Plots show 99%
(left), 75% (center) and 50% home-ranges (right). Error bars represent 95%
confidence intervals.
156
Figure 3. Variation in the usage densities of each fire severity category by possums
within each landscape class. The probability density function represents the
average use of each fire severity class relative to availability, within the 99% home
range isopleths for all possums within each landscape class. Fire severity is shown
on the x-axis, with 1 = high severity fire (mountain ash killed by fire), 2= high
severity fire (some mountain ash present epicormic growth following fire), 3 =
moderate severity (mountain ash canopy intact, understorey burnt), 4= moderate
severity fire (some midstorey intact) 5 = low-severity burnt forest (forest structure
unaffected). Error bars represent 95% confidence intervals.
Differences in fire-severity choice among landscape classes
We used generalised linear mixed modelling to quantify differences in the use of grid
cells of each fire-severity category within each landscape class, relative to availability.
We found a significant interaction between the density of use of each fire-severity
category and landscape class. In low-severity landscapes, the lowest severity category
was associated with the highest usage, indicating that individuals avoided burnt zones
within patchy, low-severity burnt forest (Figure 3, left). In mixed landscapes, highest
GPS fix densities were observed in severity categories 3 and 4, representing moderately
burnt forest with burnt understorey and intact canopy (Figure 3, center). In mixed
157
landscapes, severity category 2 was under-used relative to availability. In high-severity,
burnt landscapes, the highest probability density function scores were recorded in
moderately burnt grid cells (Figure 3, right). In high-severity burnt landscapes, high-
severity areas were under-utilized relative to availability. Moreover, in high-severity
burnt landscapes, there was no evidence that the possums selected areas of unburnt
forest, despite the presence of these areas within burnt landscapes.
Differences in temporal movement patterns between landscape classes
We found that movement patterns differed throughout the night between individuals
inhabiting different landscape classes (Figure, 4). Possums in low-severity landscapes
exhibited three distinct movement phases, with long step-lengths tending to occur early
and late in the night, and a period characterized by shorter-distance movements
occurring during the middle of activity periods (Figure, 4). Individuals in high-severity
landscapes exhibited a similar three-phase pattern of movement, with less distinction
between periods than possums in low-severity landscapes. Possums displayed a delay in
activity in high-severity landscapes, having shorter step lengths at the start and end of
the night than possums in low-severity landscapes. Individual variation in mixed-
severity landscapes was too great to detect any clear trend.
We found a significant relationship between step length and time of night in landscape
classes of low-severity burns (estimate = 14.10, S. E = 6.51, T= 2.17, P = 0.03) and
high-severity burns (estimate = 31.33, S.E = 4.37, T= 7.17, P = <0.001). We found a
significant interaction between time and low-severity burnt landscapes (estimate = 7.75,
S.E = 8.37, T = 3.01, P = 0.007), but no significant relationship between time and step
length in mixed landscapes.
158
Using the results of the generalised additive models, we identified three phases of
movement activity in unburnt landscapes: (1) early in the night when animals became
active and moved larger distances between sequential GPS fixes (phase 1, proportion of
activity periods 0-0.5); (2) intermediate time periods, when distances between fixes
were shorter (phase 2, 0.5-0.7), and: (3) late time periods when step distances between
fixes were greater (phase 3, 0.7-1). We divided the fixes of each individual into these
three phase groupings and calculated the maximum home-range area of each movement
phase using a 99% isopleth. As distinct movement phases were not clearly defined in
mixed and high severity landscapes, we applied the three phase approach visible in low
severity landscapes to examine the occurrence of temporal variation in space use in
landscapes with greater proportions of disturbance. We examined the relationship
between total home-range area across all three phases and the proportions of range
overlap between phases using a linear model to establish whether the likelihood of an
individual foraging and denning in the same area was related to landscape class.
Between landscape classes, we found a significant relationship in the proportion of total
home-range areas that were shared between different movement phases (R2 = 0.27, F =
4.16, DF = 15, P = 0.037). We found total home-range area shared between phases
differed among landscape classes (Wald = 98.5, P = <0.001). There was greater overlap
in home-range area between each movement phase in burnt landscapes than in unburnt
landscapes (Figure 5). The comparatively low overlap in unburnt landscapes indicated
that individuals were utilizing different areas during each movement phase (Figure 6).
159
Figure 4. Variation in mean step length throughout the night (y-axis) among
different landscape classes. ‘Step length’ is the beeline distance travelled by an
individual between sequential GPS fixes. The y-axis represents the smoothed
parameter for step length derived from the GAM; higher scores on the y-axis
indicate greater distances travelled. The x-axis represents the proportion of time
throughout the night. Solid lines indicate values of mean step length relative to
time for all individuals in each landscape class over the duration of the study.
Dotted lines represent 95% confidence intervals.
160
Figure 5. The spatial overlap between temporal movement phases (with 95%
confidence intervals), indicating different movement phases overlap in high-
severity burnt forest, but overlap considerably less in low-severity burnt forest.
Figure 6. Examples of the utilisation distributions of mountain brushtail possums
between landscape classes. The figure depicts kernel-density estimates of the
utilization distribution of individuals. The figure shows examples of individuals in
high-severity (left), mixed-severity (centre) and low-severity (right) landscape
classes.
161
5.5 Discussion
Our results support the postulate that areas of high severity fire can provide key habitat
components for the mountain brushtail possum five years after a major wildfire. We
found that the amount of forest burnt at different severities within the landscape
influenced home-range area, habitat selection, and temporal patterns of space use. Our
findings demonstrate that landscapes recovering from recent high-severity crown-fires
can contain valuable habitat for a generalist mammal. We identified temporal
differences in the movement patterns of nocturnal animals inhabiting landscapes
affected to differing extents by high-severity fire. Furthermore, we identified differences
in the area of spatial overlap used during three distinct movement phases by animals in
landscapes subject to differing burn severity. We found that the use of particular fire
severity classes differed depending on the context of severity in the surrounding
landscape.
The influence of fire-severity patterns on movement area
Mountain brushtail possums had a tendency to have greater movement distances in
landscapes with higher proportions of unburnt forest, although our sample size was
likely to be too small to detect a significant effect (Figure 1). We suggest this result was
driven by two factors: (1) high foliage density and resource abundance within forest
burnt at high severity; and (2) the low density of key food plants, such as Silver Wattle
Acacia dealbata, in unburnt forest (Figure 6). Our findings are in contrast to those of the
effects of fire on Northern Bettongs, which do not exhibit differences in home-range
area or show temporal variation in movement or foraging patterns in response to fire
(Vernes and Pope 2001). Smaller foraging areas in high-severity landscapes may be
associated with higher predation pressure. Four years following a major fire,
162
populations of common ringtail possums (Pseudocheirus peregrinus) had not recovered
to pre-fire levels due to high rates of predation by lace monitors (Varanus varius) and
red foxes (Vulpes vulpes) (Russell et al. 2003). Alternatively, smaller foraging areas
may be a response to competitive pressures from other mountain brushtail possums,
which may be increase in abundance in severely burnt landscapes due to high
availability of foliage and hollow-bearing trees (Lindenmayer et al. 2013b).
Fires can impair the ability of fauna to perform key ecosystem functions in forested
landscapes. For example, in Sumatran rainforests, fire lowered the reproductive success
of Simangs, Symphalangus syndactylus, for multiple successive generations, with long-
term consequences for seed dispersal in recently burnt rainforests (O'Brien et al. 2003).
The intensive use of small areas of severely burnt forest may have implications for the
successional dynamics of forest vegetation (Smith et al. 2014). It is likely that Mountain
brushtail possums, which forage intensively on plant foliage (Seebeck et al. 1984), have
an important role in nutrient cycling and the process of vegetation thinning in these
rapidly developing early-successional forests.
The fire-severity patterns on habitat selection
We found that possum use of forested areas burnt at differing severity varied
substantially depending on the surrounding landscape context. In landscapes of high fire
severity, we identified a preference by the possums for areas of moderate-severity forest
where intact canopy was present. Similar uses of areas of contrasting vegetation
structure by Mmuntain brushtail possums have been observed in recently logged
landscapes, where strips of remnant undisturbed forest provide desirable microhabitat
features such as tree ferns, Dicksonia antartica, and silver wattle (Lindenmayer et al.
1994). In severely burnt mountain ash forests, areas of unburnt canopy present essential
163
food resource opportunities for foliage-foraging marsupials such as the greater glider,
Petauroides volans (Lindenmayer et al. 1990a; Berry et al. 2015a).
We found no evidence that possums consistently selected areas of high or low fire-
severity among landscape classes. This result suggests that, five years after fire, the
potential fitness benefits provided by contrasting severities within disturbed landscapes
are limited (Berry et al. 2015a). Similar patterns of habitat selection following fire have
been observed in savannah ungulates which preferred to graze on high-severity burnt
grasslands with low tree cover, due to lower predation risk and increased access to high
nutrient grasses (Klop et al. 2007). Our results suggest that fire management practices
that lead to retention of intact mid-storey and canopy features may facilitate the use of
extensively burnt landscapes by some fauna (Leonard et al. 2014)
.
The influence of fire severity on temporal patterns of space use
We found that in landscapes burnt at low severity, individuals exhibited three distinct
movement phases characterised by substantial differences in step lengths. Periods of
consistently long step lengths (i.e., higher movement rates) are indicative of
transitioning/commuting movements between social, feeding or nesting events (Votier
et al. 2011), whereas, shorter step lengths are often interpreted as behaviour associated
with fine-scale habitat use or social interactions (Guilford et al. 2008; Shamoun-
Baranes et al. 2012). The occurrence of three distinct nightly phases of movement is
likely an indication of individuals in low-severity landscapes foraging and denning in
separate areas (Welsh et al. 1998). Such an interpretation is further supported by our
analysis of range overlap during each foraging period between landscape classes (Figure
5). The lack of a distinct three-phase movement pattern in high and mixed-severity
landscapes may be due to the influence of dense regrowth of vegetation on the physical
164
ability of mountain brushtail possums to effectively move large distances between
foraging periods (With et al. 1999). Lower overlap between denning and foraging areas
in unburnt landscapes is likely due to the dense mid-storey and understorey foliage in
late-successional forest, and the relatively open nature of forest canopies surrounding
stag trees (Gibbons and Lindenmayer 1996; Lindenmayer et al. 2011c; Collins et al.
2012). Additionally, the observed delay in the initiation of activity periods in burnt
landscapes may represent an effort to avoid detection by predators that is facilitated
through later onset of darkness because of a reduction in canopy cover (Hebblewhite
and Haydon 2010).
Refuges and the value of severely burnt forest
We found no evidence that refugia from fire were needed for a foliage-foraging
mammal to persist within an extensively burnt landscape five years later. Banks et al.
(2015) reported that the mountain brushtail possum remained in both burnt and unburnt
areas following the 2009 fire, but survival rates were higher in fine-scale refuges than in
burnt areas during the two-years immediately following the fires. We found landscapes
burnt mostly at high fire-severity had higher numbers of tree hollows within 1 ha of
occupied den trees, than the other landscape classes. Previous studies have found that
mountain brushtail possums are likely to occupy several den trees and den with multiple
individuals at sites with high hollow abundance (Banks et al. 2013; Blyton et al. 2014),
although differences in use of den trees have not previously been attributed to the
effects of fire (Banks et al. 2011c). Our findings were consistent with the understanding
of resource use and den-site flexibility displayed by this species (Banks et al. 2011c).
The response of the mountain brushtail possum appear similar to the short-term impacts
of fire on tree-use by koalas, Phascolarctos cinereus, which were found to utilize burnt
165
trees for feeding within several months after fire (Matthews et al. 2007). Our study
suggests that fire refuges are not necessary for the long-term survival of mammals that
can persist in recently burnt forest in the years following fire. However, we
acknowledge that refuges may be more important as sources of hollow trees under
frequent fire regimes (Mackey et al. 2002).
166
Conclusions/ Management Implications
Our study demonstrates that early successional forested landscapes previously burnt by
high severity crown fires provides important habitat for mammals. These findings are
also applicable in high productivity forests globally, where fires are followed by rapid
tree growth (Ne'eman and Izhaki 1998b; Williams 2000; Johnstone and Chapin III
2006). Our results indicate that early successional forest provides habitat for some forest
mammals. However, current land-management policies in burnt forests globally
overlook their ecological value and instead focus on extracting as much economic value
as possible through post-fire salvage logging (Hutto and Gallo 2006; Lindenmayer and
Noss 2006; Schmiegelow et al. 2006). The removal of key habitat structures and
biological legacies such hollow bearing-trees from landscapes recently disturbed by fire
will result in altered animal and plant successional dynamics within these systems
(Lindenmayer and Ough 2006). These recommendations are consistent with those for
the conservation of the black-backed woodpecker, Picoides arcticus in North American
Boreal forests subject to large-scale salvage logging operations where previous studies
have suggested that future habitat availability is dependent on the retention of large
areas of forest representative of all burn severities, which contain high quality, old-
growth den trees (Nappi et al. 2010; Nappi and Drapeau 2011).
Immediately following large scale wildfires, prior to the re-establishment of a dense
understorey, the persistence of many forest mammals within extensively burnt
landscapes may be dependent on the occurrence of unburnt forest refuges (Mackey et al.
2002). It is essential that these refuge areas be protected from clearing and degradation
to ensure that they continue to support animal populations within the landscape in the
event of large, crown-fires (Robinson et al. 2014; Berry et al. 2015a).
167
5.6 Acknowledgements
We thank Jeff Wood for providing statistical advice. We also thank Lachie Mc Burney
and David Blair for their helpful suggestions and support in the field as well as Brian
Tew and many field volunteers for their assistance in collecting the data. This project
was funded by the Australian Academy of Science through the Margaret Middleton
Fund Award for Endangered Australian Vertebrates and by an ARC Discovery Grant
held by DBL. The project was conducted in accordance with ANU animal ethics permit
number: A2014/01.
168
5.7 References
Banks, SC, Knight, EJ, McBurney, L, Blair, D, Lindenmayer, DB (2011) The Effects of
Wildfire on Mortality and Resources for an Arboreal Marsupial: Resilience to
Fire Events but Susceptibility to Fire Regime Change. Plos One 6, e22952.
Banks, SC, Lindenmayer, DB, Wood, JT, McBurney, L, Blair, D, Blyton, MD (2013)
Can individual and social patterns of resource use buffer animal populations
against resource decline? Plos One 8, e53672.
Banks, SC, Lorin, T, Shaw, RE, McBurney, L, Blair, D, Blyton, MD, Smith, AL,
Pierson, JC, Lindenmayer, DB (2015) Fine‐scale refuges can buffer
demographic and genetic processes against short‐term climatic variation and
disturbance: a 22‐year case study of an arboreal marsupial. Molecular Ecology
24, 3831-3845.
Berry, LE, Driscoll, DA, Banks, SC, Lindenmayer, DB (2015a) The use of topographic
fire refuges by the greater glider (Petauroides volans) and the mountain
brushtail possum (Trichosurus cunninghami) following a landscape-scale fire.
Australian Mammalogy 37, 39-45.
Berry, LE, Driscoll, DA, Stein, JA, Blanchard, W, Banks, SC, Bradstock, RA,
Lindenmayer, DB (2015b) Identifying the location of fire refuges in wet forest
ecosystems. Ecological Applications 25, 2337-2348.
Berry, LE, Lindenmayer, DB, Driscoll, DA (2015c) Large unburnt areas, not small
unburnt patches, are needed to conserve avian diversity in fire‐prone landscapes.
Journal of Applied Ecology 52, 486-495.
Beyer, HL (2012) 'Geospatial Modelling Environment.' http://www. spatialecology.
com/gme. Blyton, MD, Lindenmayer, DB, Banks, SC (2014) Maternal lineages best explain the
associations of a semisocial marsupial. Behavioral Ecology 25, 1212-1222.
Bond, WJ, Keeley, JE (2005) Fire as a global ‘herbivore’: the ecology and evolution of
flammable ecosystems. Trends in Ecology & Evolution 20, 387-394.
Bond, WJ, Woodward, FI, Midgley, GF (2005) The global distribution of ecosystems in
a world without fire. New Phytologist 165, 525-538.
Börger, L, Franconi, N, De Michele, G, Gantz, A, Meschi, F, Manica, A, Lovari, S,
Coulson, T (2006) Effects of sampling regime on the mean and variance of
home range size estimates. Journal of Animal Ecology 75, 1393-1405.
Bowman, D, Balch, JK, Artaxo, P, Bond, WJ, Carlson, JM, Cochrane, MA, D'Antonio,
CM, DeFries, RS, Doyle, JC, Harrison, SP, Johnston, FH, Keeley, JE,
Krawchuk, MA, Kull, CA, Marston, JB, Moritz, MA, Prentice, IC, Roos, CI,
Scott, AC, Swetnam, TW, van der Werf, GR, Pyne, SJ (2009) Fire in the Earth
System. Science 324, 481-484.
Bradstock, RA, Bedward, M, Gill, AM, Cohn, JS (2005) Which mosaic? A landscape
ecological approach for evaluating interactions between fire regimes, habitat and
animals. Wildlife Research 32, 409-423.
Bradstock, RA, Williams, JE, Gill, AM (2002) 'Flammable Australia: the fire regimes
and biodiversity of a continent.' (Cambridge University Press:
Chapin, FS, Matson, PA, Vitousek, P (2011) 'Principles of terrestrial ecosystem
ecology.' (Springer Science & Business Media)
Clarke, MF (2008) Catering for the needs of fauna in fire management: science or just
wishful thinking? Wildlife Research 35, 385-394.
169
Collins, L, Bradstock, RA, Tasker, EM, Whelan, RJ (2012) Can gullies preserve
complex forest structure in frequently burnt landscapes? Biological
Conservation 153, 177-186.
Dees, CS, Clark, JD, Van Manen, FT (2001) Florida panther habitat use in response to
prescribed fire. The Journal of Wildlife Management 65, 141-147.
Flannigan, MD, Stocks, BJ, Wotton, B (2000) Climate change and forest fires. Science
of the Total Environment 262, 221-229.
Gibbons, P, Lindenmayer, D (1996) Issues associated with the retention of hollow-
bearing trees within eucalypt forests managed for wood production. Forest
Ecology and Management 83, 245-279.
Gibbons, P, van Bommel, L, Gill, AM, Cary, GJ, Driscoll, DA, Bradstock, RA, Knight,
E, Moritz, MA, Stephens, SL, Lindenmayer, DB (2012) Land Management
Practices Associated with House Loss in Wildfires. Plos One 7, e29212.
Hutto, RL, Gallo, SM (2006) The effects of postfire salvage logging on cavity-nesting
birds. Condor 108, 817-831.
Johnstone, JF, Chapin III, FS (2006) Effects of soil burn severity on post-fire tree
recruitment in boreal forest. Ecosystems 9, 14-31.
Kelly, LT, Nimmo, DG, Spence-Bailey, LM, Taylor, RS, Watson, SJ, Clarke, MF,
Bennett, AF (2012) Managing fire mosaics for small mammal conservation: a
landscape perspective. Journal of Applied Ecology 49, 412-421.
Lentile, LB, Smith, FW, Shepperd, WD (2006) Influence of topography and forest
structure on patterns of mixed severity fire in ponderosa pine forests of the
South Dakota Black Hills, USA. International Journal of Wildland Fire 15, 557-
566.
Leonard, SW, Bennett, AF, Clarke, MF (2014) Determinants of the occurrence of
unburnt forest patches: Potential biotic refuges within a large, intense wildfire in
south-eastern Australia. Forest Ecology and Management 314, 85-93.
Lindenmayer, D (2009) 'Forest pattern and ecological process: a synthesis of 25 years of
research.' CSIRO publishing, Canberra.
Lindenmayer, D, Cunningham, R, Donnelly, C, Triggs, B, Belvedere, M (1994) The
conservation of arboreal marsupials in the montane ash forests of the central
highlands of Victoria, south-eastern Australia V. Patterns of use and the
microhabitat requirements of the mountain brushtail possum Trichosurus
caninus ogilby in retained linear habitats (wildlife corridors). Biological
Conservation 68, 43-51.
Lindenmayer, D, Cunningham, R, Tanton, M, Smith, A, Nix, H (1990) Habitat
requirements of the mountain brushtail possum and the greater glider in the
montane ash-type eucalypt forests of the central highlands of Victoria. Wildlife
Research 17, 467-478.
Lindenmayer, D, Noss, R (2006) Salvage logging, ecosystem processes, and
biodiversity conservation. Conservation Biology 20, 949-958.
Lindenmayer, D, Ough, K (2006) Salvage logging in the montane ash eucalypt forests
of the Central Highlands of Victoria and its potential impacts on biodiversity.
Conservation Biology 20, 1005-1015.
Lindenmayer, D, Welsh, A, Donnelly, C (1998) The use of nest trees by the mountain
brushtail possum (Trichosurus caninus)(Phalangeridae: Marsupialia). V.
Synthesis of studies. Wildlife Research 25, 627-634.
Lindenmayer, D, Welsh, A, Donnelly, C, Cunningham, R (1996) Use of Nest Trees by
the mountain brushtail possum (Trichosurus Caninus)(Phalangeridae:
170
Marsupialia). Ii. Characteristics of Occupied Trees. Wildlife Research 23, 531-
545.
Lindenmayer, DB, Blanchard, W, McBurney, L, Blair, D, Banks, SC, Driscoll, D,
Smith, AL, Gill, AM (2013) Fire severity and landscape context effects on
arboreal marsupials. Biological Conservation 167, 137-148.
Lindenmayer, DB, Wood, J, McBurney, L, Michael, D, Crane, M, Macgregor, C,
Montague-Drake, R, Gibbons, P, Banks, SC (2011) Cross-sectional vs.
longitudinal research: a case study of trees with hollows and marsupials in
Australian forests. Ecological Monographs 81, 557-580.
Mackey, B, Berry, S, Hugh, S, Ferrier, S, Harwood, TD, Williams, KJ (2012)
Ecosystem greenspots: identifying potential drought, fire, and climate-change
micro-refuges. Ecological Applications 22, 1852-1864.
Mackey, B, Lindenmayer, D, Gill, M, McCarthy, M, Lindesay, J (2002) 'Wildlife, Fire
and Future Climate: A Forest Ecosystem Analysis.' (CSIRO publishing:
Collingwood)
Marzluff, JM, Millspaugh, JJ, Hurvitz, P, Handcock, MS (2004) Relating resources to a
probabilistic measure of space use: forest fragments and Steller's jays. Ecology
85, 1411-1427.
Matthews, A, Lunney, D, Gresser, S, Maitz, W (2007) Tree use by koalas
(Phascolarctos cinereus) after fire in remnant coastal forest. Wildlife Research
34, 84-93.
Millspaugh, JJ, Nielson, RM, McDonald, L, Marzluff, JM, Gitzen, RA, Rittenhouse,
CD, Hubbard, MW, Sheriff, SL (2006) Analysis of resource selection using
utilization distributions. Journal of Wildlife Management 70, 384-395.
Nappi, A, Drapeau, P (2011) Pre-fire forest conditions and fire severity as determinants
of the quality of burned forests for deadwood-dependent species: the case of the
black-backed woodpecker. Canadian journal of forest research 41, 994-1003.
Nappi, A, Drapeau, P, Saint-Germain, M, Angers, VA (2010) Effect of fire severity on
long-term occupancy of burned boreal conifer forests by saproxylic insects and
wood-foraging birds. International Journal of Wildland Fire 19, 500-511.
Nathan, R, Getz, WM, Revilla, E, Holyoak, M, Kadmon, R, Saltz, D, Smouse, PE
(2008) A movement ecology paradigm for unifying organismal movement
research. Proceedings of the National Academy of Sciences 105, 19052-19059.
Ne'eman, G, Izhaki, I (1998) Stability of pre‐and post‐fire spatial structure of pine trees
in Aleppo pine forest. Ecography 21, 535-542.
Nugent, DT, Leonard, SW, Clarke, MF (2014) Interactions between the superb lyrebird
(Menura novaehollandiae) and fire in south-eastern Australia. Wildlife Research
41, 203-211.
O'Brien, TG, Kinnaird, MF, Nurcahyo, A, Prasetyaningrum, M, Iqbal, M (2003) Fire,
demography and the persistence of siamang (Symphalangus syndactylus:
Hylobatidae) in a Sumatran rainforest. Animal Conservation 6, 115-121.
Parr, CL, Andersen, AN (2006) Patch mosaic burning for biodiversity conservation: a
critique of the pyrodiversity paradigm. Conservation Biology 20, 1610-1619.
Perry, DA, Hessburg, PF, Skinner, CN, Spies, TA, Stephens, SL, Taylor, AH, Franklin,
JF, McComb, B, Riegel, G (2011) The ecology of mixed severity fire regimes in
Washington, Oregon, and Northern California. Forest Ecology and Management
262, 703-717.
R-Development-Core-Team (2005) R: A language and environment for statistical
computing. R foundation for Statistical Computing
171
Robinson, NM, Leonard, SWJ, Bennett, AF, Clarke, MF (2014) Refuges for birds in
fire-prone landscapes: The influence of fire severity and fire history on the
distribution of forest birds. Forest Ecology and Management 318, 110-121.
Russell, B, Smith, B, Augee, M (2003) Changes to a population of common ringtail
possums (Pseudocheirus peregrinus) after bushfire. Wildlife Research 30, 389-
396.
Schmiegelow, FK, Stepnisky, DP, Stambaugh, CA, Koivula, M (2006) Reconciling
salvage logging of boreal forests with a natural‐disturbance management model.
Conservation Biology 20, 971-983.
Seebeck, J, Warneke, R, Baxter, B (1984) Diet of the bobuck, Trichosurus caninus
(Ogilby)(Marsupialia: Phalangeridae) in a mountain forest in Victoria. Possums
and gliders 145-154.
Smith, AL, Blair, D, McBurney, L, Banks, SC, Barton, PS, Blanchard, W, Driscoll, DA,
Gill, AM, Lindenmayer, DB (2014) Dominant drivers of seedling establishment
in a fire-dependent obligate seeder: Climate or Fire Regimes? Ecosystems 17,
258-270.
Taylor, RS, Watson, SJ, Nimmo, DG, Kelly, LT, Bennett, AF, Clarke, MF (2012)
Landscape-scale effects of fire on bird assemblages: does pyrodiversity beget
biodiversity? Diversity and Distributions 18, 519-529.
Teague, B, McLeod, R, Pascoe, S (2010) Final report, 2009 Victorian bushfires royal
commission. Parliament of Victoria, Melbourne Victoria, Australia
Torre, I, Díaz, M (2004) Small mammal abundance in Mediterranean post-fire habitats:
a role for predators? Acta Oecologica 25, 137-142.
Vernes, K, Pope, LC (2001) Stability of nest range, home range and movement of the
northern bettong (Bettongia tropica) following moderate-intensity fire in a
tropical woodland, north-eastern Queensland. Wildlife Research 28, 141-150.
Viggers, K, Lindenmayer, D (1995) The use of tiletamine hydrochloride and zolazepam
hydrochloride for sedation of the mountain brushtail possun Trichosurus caninus
Ogilby (Phalangeridae: Marsupialia). Australian veterinary journal 72, 215-216.
Votier, SC, Grecian, WJ, Patrick, S, Newton, J (2011) Inter-colony movements, at-sea
behaviour and foraging in an immature seabird: results from GPS-PPT tracking,
radio-tracking and stable isotope analysis. Marine biology 158, 355-362.
Welsh, A, Lindenmayer, D, Donnelly, C, Ruckstuhl, A (1998) Use of nest trees by the
mountain brushtail possum (Trichosurus caninus)(Phalangeridae: Marsupialia).
IV. Transitions between den trees. Wildlife Research 25, 611-625.
Williams, PR (2000) Fire-stimulated rainforest seedling recruitment and vegetative
regeneration in a densely grassed wet sclerophyll forest of north-eastern
Australia. Australian Journal of Botany 48, 651-658.
Winter, J (1980) Tooth wear as an age index in a population of the brush-tailed possum,
Trichosurus vulpecula (Kerr). Wildlife Research 7, 359-363.
Yospin, GI, Wood, SW, Holz, A, Bowman, DM, Keane, RE, Whitlock, C (2015)
Modeling vegetation mosaics in sub-alpine Tasmania under various fire regimes.
Modeling Earth Systems and Environment 1, 1-10.
172
5.8 Appendix 1.
Demographic characteristics for each study animal.
Landscape
Class
Sex
(F/M)
Body
Length
(cm)
Tail
length
(cm)
Head
Length
(mm)
Age Index
(1-5)
Weight
(g)
High Severity F 81 30.5 81.5 5 3375
High Severity M 82 35 90 4 2725
Mixed Severity F 83.5 35 91 3 3500
Mixed Severity M 86 36 93.5 3 3250
Low Severity F 83 32.5 92 3 3225
Low Severity M 81 32 100 3 3125
High Severity F 90 37 104 5 3725
High Severity M 80 35 86 2 2775
Mixed Severity F 82 35.5 79 5 3250
Mixed Severity M 78 34 93 4 3100
Low Severity F 87.5 38.5 88.5 3 3300
Low Severity M 81.5 37 88 3 2900
High Severity F 86 38 90 3 3525
High Severity M 81 35 95 5 3450
Mixed Severity F 84 34.5 90 4 3815
Mixed Severity M 82 38 90 4 3525
Low Severity F 86.5 37.5 88.5 4 3500
Low Severity M 84 35 91 4 3175
173
Chapter 6. Spatially managing fire in forests for
biodiversity: concepts, current practices and
future challenges
Berry, L. E., Driscoll, D. A., Banks, S. C and Lindenmayer, D.B. Spatially managing
fire in forests for biodiversity: concepts, current practices and future challenges, In
preparation for publication in Frontiers in Ecology and the Environment.
174
175
6.1 Abstract
Within the fire ecology literature, it is becoming increasingly recognized that the spatial
patterns generated by wildfires have a significant influence on the conservation of
biodiversity. This is particularly relevant to the fire-prone tall forest systems of south-
eastern Australia and the Pacific Northwest of the United States of America.
Many spatially-focused ecological studies conclude with suggested fire management
recommendations to maintain or improve the ecological value of fire-affected
landscapes. However, these research findings are rarely integrated into decision-making
processes within fire management organizations or translated into applied outcomes.
We employed a qualitative research approach to identify the barriers to and enablers of
spatially managing fire for biodiversity. We then developed a conceptual framework to
achieve the integration of spatial approaches to fire into management. We conducted
structured interviews with experts in fire and biodiversity management and research
working in fire-prone forest ecosystems in the Pacific Northwest United States and
south-eastern Australia. The trans-pacific nature of our study enabled us to access a
broad range of views on the spatial management of fire for biodiversity. We aimed to 1)
establish whether a spatial approach to fire was currently being employed in
biodiversity management, to establish the barriers to spatially managing fire for
biodiversity and 2) develop a conceptual approach to incorporating spatial fire concepts
into current management and research frameworks.
We identified that spatial approaches to fire management must co-exist within a
complex system of social and ecological feedbacks between landscapes, academic
176
research, socio-political land management systems and environmental pressures. Our
findings suggest that spatially managing fire can be achieved through a number of
refinements to existing processes. These steps relate to developing community
understanding of fire science, improving the relevance of fire research outputs to land
management, amending existing government policies and approaches, and refining
management tools, structures, scales and monitoring to meet biodiversity and fire risk
objectives
177
6.2 Introduction
Large-scale forest fires occur globally and commonly create heterogeneous patterns of
fire. The repeat occurrence of fire over time produces complex landscape fire mosaics
(Bradstock et al. 2005). These mosaics consist of a range of differing fire regimes with
areas varying in fire frequency, intensity, seasonality and type (Gill 1975). Within fire-
prone systems, many components of biodiversity are intrinsically adapted to particular
fire regimes. The maintenance of tolerable disturbance regimes is a fundamental goal
for biodiversity conservation in systems subject to extensive large-scale disturbance
events. However, environmental pressures such as land clearing, invasive species and
climate change will likely alter the global distribution and frequency of fire and promote
novel fire conditions (Brook et al. 2008).
In crown-fire forest systems, fires occurring in increasingly extreme weather conditions
are predicted to produce more homogeneous patterns of high severity fire (Berry et al.
2015). Forest fires which occur under extreme conditions often burn large areas, for
example the 2009 Black Saturday fires in Victoria, Australia burnt over 450,000 ha
(Cruz et al. 2012), whilst the 2013 Rim fire in California, United States burnt 104,200
ha (Harris and Taylor 2015). Additionally, centuries of successful fire suppression
activities in North America have altered fire regime distribution in many systems (Noss
et al. 2006). This is particularly evident in drier forest types such as mixed conifer
forests, which are adapted to small-scale, short interval (11- 16 years ), mixed and low-
severity fire, but are now experiencing large high-severity fire with return intervals
greater than 200 years (Taylor and Skinner 2003). Given these challenges to global
forest fire management, alternative perspectives which incorporate key spatial concepts
178
are required to reduce the detrimental impacts of altered fire regimes on biodiversity,
ecosystems and communities.
The importance of the spatial patterns of fire in landscapes
Fire is an inherently spatial phenomenon. In montane forest systems, the interaction
between topography, weather and fuels leads to the occurrence of mixed severity fire
regimes across landscapes (Perry et al. 2011). Following fire, topographically mediated
heterogeneous patterns of fire severity will have a major influence on the ecology of
recently burnt forests (Leonard et al. 2014; Berry et al. 2015). Diverse patterns of fire
may lead to increased niche opportunities for species, thereby increasing biodiversity in
disturbed landscapes (Bradstock et al. 2005). However, the spatial arrangement of post-
fire landscape components, such as biological legacies and areas of contrasting fire
severity will have a major influence on the ability of biota to survive and persist
following large-scale fires (Robinson et al. 2013).
Over the last decade, knowledge of the influence of spatial fire patterns has been
building (Bradstock et al. 2005). This body of knowledge relates to multiple areas
including fuels, biodiversity, ecosystem resilience, fire management, landscape
dynamics and earth system processes. This increasing body of work suggests that by
implementing a spatial approach to fire in addition to current temporal management
practices, multiple desirable outcomes for biodiversity and risk reduction can be
achieved. Through allowing fire regimes to occur along natural gradients and
controlling the size, shape and landscape context of applied burning, land managers can
maximize the ecological and social benefits of wildland fire. Tailoring the spatial
configurations of burns to the resident biota of forested landscapes will reduce the
detrimental impacts of fire, such as habitat loss and fragmentation effects on species and
179
will maximize the beneficial aspects of fire, such as resource heterogeneity and habitat
connectivity (Parr and Andersen 2006; Clarke 2008). Furthermore, by strategically
managing the spatial distribution of temporal fire treatments, managers may create
heterogeneous landscape fire mosaics which may satisfy risk-based objectives by
reducing the severity and extent of unplanned wildfires (Bradstock et al. 2005).
Managing landscapes for beneficial spatial patterns of fire
The management of fire-prone forests is one of the most contentious natural resource
issues world-wide, including in the United States and Australia (Lindenmayer 1995;
Noss et al. 2006). Forested ecosystems are a fundamental component of key earth
system processes such as carbon, water, nutrient, and air cycling and represent a
substantial proportion of global biodiversity (Bonan 2008). Forests are also a key
natural resource, with high social and economic values. Fire is a key component of
forest ecology globally, and is a major contributing factor to the distribution of forest
types (Bond et al. 2005). Globally, land management agencies are increasingly
recognizing the importance of managing fire to restore or conserve ecological values
(Ryan et al. 2013). However, the management of fire-prone forest systems must occur
within socio-economic and political constraints (Yoder et al. 2004). Additionally,
managers must deal with the challenges of deploying fire prescriptions within highly
modified and dynamic contemporary landscapes (Fernandes et al. 2013). Whilst there is
a firm base understanding of fire ecology, there are still key knowledge gaps,
particularly relating to applied fire management, faunal responses and the spatial aspects
of fire (Driscoll et al. 2010b). Ecological fire management approaches are more
commonly developed around temporal concepts such as minimum fire return intervals
and in some cases, on creating temporally diverse landscape mosaics (Clarke 2008).
180
Our study aimed to identify the barriers to spatially managing fire within contemporary
land management and academic research settings. We used a qualitative interview-
based approach (Patton 2002) to ask whether it is possible to successfully manage the
spatial patterns of fire for biodiversity outcomes in fire-prone forest landscapes of
south-eastern Australia and the Pacific Northwest USA within current research and land
management frameworks and socio-political contexts. We used our findings to develop
a theoretical framework which highlights the key areas which influence our ability to
spatially manage fire. Using this framework, we identified the steps needed to
implement spatially-based fire management approaches. The trans-pacific nature of our
study enabled comparisons to be made surrounding key issues in the spatial
management of wildfires for biodiversity. Both countries have a strong history of active
fire suppression, widespread and large-scale logging industries, increasing inhabitation
of wildland areas and declining land management budgets (Agee 1993; Russell-Smith et
al. 2003). Additionally, both regions have recently experienced destructive large-scale
fires which have resulted in reactionary policy changes and approaches to wildfire
management (Cruz et al. 2012; Harris and Taylor 2015).
181
6.3 Methods
Social science methods can provide insights into the social, policy and management
settings within which research outcomes are interpreted and applied (Bosomworth et al.
2015). We conducted open-ended structured interviews with sixteen fire management
practitioners and sixteen fire ecology researchers working in fire and biodiversity
management fields in the Pacific Northwest United States and south-eastern Australia
according to the methods outlined in Patton (2002). To gain the broadest possible range
of views, we interviewed managers at different career stages, including those working at
either operational or planning levels with a local or regional focus from multiple land
management agencies. We selected researchers for their expertise in forest and fire
management from a broad range of research institutions and that were also
representative of various career stages. To enable comparisons between groups and
specifically address our research questions, we limited our participant pool to those
working within fire-prone tall forests. The tall wet-forest ecosystems of the Pacific
Northwest share similar fire regimes with wet-forest types in south-eastern Australia.
Both regions also support drier foothills types suited to shorter intervals of lower
severity fire (Swanson et al. 2010).
The interviews aimed to gather information on the challenges to managing landscapes
for spatially beneficial patterns of fire for biodiversity. The interviews were structured
around several key areas in fire ecology. These included eliciting attitudes to; priorities
in fire management and research, the current state of fire knowledge, current fire
management practices, management institutional structures and frameworks, current fire
policies, knowledge transfer and integration, community engagement in fire
management issues and professional relationships. We asked land managers and
182
researchers questions. Interviews were approximately one hour in length. All interviews
were digitally recorded and subsequently transcribed in full. In accordance with human
ethics guidelines, all personal identifiers were removed from the transcripts. To identify
key themes, barriers and enablers within the data, we conducted manual content analysis
in accordance with the approach described in Patton (2002). To validate each identified
theme, we used the triangulation method outlined by Creswell and Miller (2000), where
only themes which occurred between multiple participants were recorded and then
cross-referenced in the literature for corroborating evidence. We grouped responses to a
key set of overarching core questions into agreements, disagreement or no clear opinion
and calculated the proportion of managers and researchers with each response. Finally,
we interpreted the themes, barriers and enablers elicited from the content analysis to
identify steps to achieving spatially based ecological fire management and to develop a
conceptual systems diagram.
183
6.4 Results
Summary of views to key issues in fire ecology and management
We found several key issues in fire ecology were similarly viewed by land managers
and researchers (Figure. 1). Both groups considered current land management agencies
to be employing ecological principles as part of their burning programs. However, both
groups strongly agreed that fire management was currently under resourced to achieve
its aims. Similarly, both groups strongly agreed that current fire management tools and
practices could be used to create or manage for desirable patterns of fire in the
landscape. Both groups agreed that forested systems in the Pacific Northwest and south-
eastern Australia currently experience unsuitable fire regimes. However, a minority of
managers (37.5%) and researchers (35%) considered current levels of prescribed fire to
be sufficient.
We also identified three key areas where managers and researchers had divergent views.
We observed opposing views between managers (37.5%) and researchers (85%)
concerning whether ecological and asset protection burning practices were
complementary. Only a small number of researchers (20%) and managers (37.5%)
agreed that the spatial arrangement of fire was currently considered in fire management
plans. We found that researchers (100%) and managers (70%) identified collaborations
with land managers an important component of achieving spatially based fire
management. However, researchers (100%) and managers (50%) differed in their level
of regular external collaborations.
184
Figure 1. Contrasting views on key issues relating to the spatial management of
fire between the 32 participants
Barriers and enablers to employing a spatial approach to fire management in forested
systems
We found wide-ranging views relating to barriers and enablers to employing a spatial
approach to fire management in forested ecosystems (Table 1). We grouped barriers and
enablers under five themes which emerged from the data; knowledge, integration,
community, policy and institutional. Within each theme, we identified categories which
describe specific aspects influencing spatial fire management.
Barriers relating to the development and use of ecological knowledge could be divided
into two categories. ‘Fundamental science’ barriers identified by researchers included a
lack of scientific consensus on core concepts and poor research tools for investigating
complex spatio-temporal landscape patterns. Land managers identified as barriers, the
infancy of the field, the theoretical rather than applied nature of current understanding,
and research projects not framed within a management context. We categorized barriers
relating to the use of existing research knowledge by management agencies as
0 50 100
I have personal relationships with managers / scientists
It is important to involve land managers in research
Institution structures aid ecological fire managent
Ecological and asset protection practices are complitmentary
The current extent of ecological burning is sufficient
There is currently too much fire in forests
Current applied tools can be used to manage fire spatially
We currently manage for spatial patterns of fire
There is sufficient literature to manage fire for biodiversity
Ecological fire management is appropriately funded
Ecological fire management is currenly practiced
Percentage of particpants in agreementresearchers managers
185
‘institutional barriers’. Both groups identified the lack of area-specific knowledge, the
absence of monitoring beyond first level fire effects and inadequate integrated modeling
tools as reasons for current spatial planning decisions being based largely on intuition.
Barriers relating to knowledge integration could be categorized as ‘institutional’ or
‘policy’ related. Both groups identified the difference in cultures between management
and research, the delayed uptake of research and the lack of policy makers with
scientific backgrounds as barriers to the integration of spatial fire research into land
management.
We identified a complex system of prioritization, policy development and
implementation was responsible for the interpretation and application of spatial fire
knowledge. This complex system presented some key barriers to achieving spatially
based fire management. Within this system we divided these barriers into three core
components, ‘community’, ‘policy’, and ‘institutional’. Community barriers were those
identified by participants to be dependent on public values, education and restrictions
imposed on management actions. These included comments such as the lack of value
placed on burnt forests, the contentious nature of fire management, and societal
constraints on windows for applied burning due to air pollution restrictions. Institutional
barriers were divided into three categories; prioritization, operational and
organizational. Prioritization barriers were those relating to the decision-making context
within which knowledge is applied, including a focus on fire suppression and close
associations between land management and resource extraction agencies. Operational
barriers concerned the ability of agencies to manage fires spatially due to lack of
resourcing, an absence of a whole-of-range scale approach to fire and limited windows
for applied burning. Organizational barriers were those relating to structures and
186
processes within agencies. These included the limited scope for adaptive learning within
top-down hierarchically-based institutions and the over-riding of long-term land
management structures by emergency management structures during fire events. Policy-
related barriers were classified as ‘legislative’ or ‘systematic’. Legislative barriers were
those relating to current policies, such as mandated area burn targets, endangered
species legislation and black-out burning. Systematic barriers were those related to
current policy development frameworks and included the lack of manager involvement
in policy creation and inappropriately lengthy intervals between reviews of major over-
arching fire policies.
In addition to these system-based barriers, both land managers and researchers
identified influential environmental factors within each of the six major themes
identified. These included landscape-level factors such as varying spatial-temporal fire
regimes between systems, contrasting habitat requirements of key species and
threatening processes such as habitat loss, climate change and invasive species. We
visualized the relationships between these barriers and landscapes within a conceptual
diagram (Figure 2).
187
Table 1. List of barriers and enablers to the spatial management of fire within each
of six core themes identified from structured interviews with land managers and
researchers.
Theme Barriers Enablers
Knowledge Fundamental science
- Emerging field of research
- Mostly theoretical not practical
- Poor ecological models/ research
approach
- Knowledge still contentious within
research community
-Literature in formats not relevant to
management
Institutional knowledge
- Not location specific
- Only first order effects monitored
- Managing spatial patterns based on
intuition
- Don’t know what a desirable mosaic
looks like
- Lack accurate integrated modeling
system to predict effects of fire
- rapidly developing field
- Understanding currently
ahead of practices
Integration
Institutional
- Cultural differences
between land managers and
researchers
-delayed uptake of knowledge by
agencies
Policy
-Disconnect because policy makers
are non-scientists
- Personal relationships
between practitioners and
researchers deliver science
directly to the ground.
- Internal knowledge
brokers
- Land grant colleges
funded for extension
activities
- Knowledge brokerage
schemes
- Involving land managers
in research creates relevant
knowledge products
community Values
- Highly contentious activity
- Values based decisions are a societal
responsibility
Education
- Public sees no value in burnt forests
- Lack of understanding of fire
ecology and suppression
188
Restrictions
- Social constraints on prescribed
burning
- Air quality issue prohibit restrict
windows for applied burning
Policy
Legislative
- Mandated area targets
- Asset protection and life
preservation number one priority
- Endangered species legislation can
prohibit beneficial restoration
activities
- Blackout burning reduces internal
heterogeneity
Systematic
- Managers work within constraints of
government policy
- Landscape strategies reviewed at
inappropriately long intervals
Institutional Prioritization
- Focus on suppression
- Poor monitoring and evaluation
frameworks
- Decisions based on intuition
- Management institutions aligned
with resource institutions
Operational
- High workloads on managers
- Do not manage at sufficiently large
spatial scales
- Insufficient resources
- Land ownership
- Tools not employed at sufficient
scales
- Windows (Seasons and conditions)
for planned burning limited
Organizational
- During fires emergency management
structures override other long term
planning structures
- Coordination between highly
specialized individuals challenging
- Hierarchical top down structures
inhibit adaptive learning
- Burning and biodiversity
officers
-Current tools effective at
creating desirable patterns
- Prescribed fire could be
used effectively
- Use mechanical treatments
when burns
unsuitable/prohibited
189
Figure 2. A conceptual diagram which illustrates the relationships between
research, complex socio-political priority, decisions and implementation systems,
environmental factors and fire landscapes.
Res
earc
h policy
management community
En
vir
on
men
tal
Fac
tors
Text Box 1. An illustration of the conceptual framework using examples from the
2009 Black Saturday fires
The outcomes of the 2009 Black Saturday bushfires in south-eastern Australia were
influenced by environmental factors such as; a period of extreme weather following a
prolonged drought (Cruz et al. 2012), and the impacts of anthropogenic land uses (Taylor
et al. 2014). These factors interacted with pre-existing landscape patterns (such as the
vegetation patterns and historical patterns of disturbance) to produce novel landscape
patterns and fire effects. The 2009 fires resulted in the death of 173 people, substantial
property damage and the consumption of ~450,000 hectares across the state of Victoria
(Teague et al. 2010). The fire outcomes prompted a Royal Commission into the causes
and consequences of the events and built community pressure on policy makers. This
resulted in the implementation of a spatially undefined broad-area burning policy, which
mandated that 5% of public lands be burnt for hazard reduction. As a consequence,
management institutions began implementing the policies which influenced patterns of
fire on the land. Following the 2009 fires, researchers began understanding the impacts of
the consequences of the fire on biodiversity and fire risk and also into the effectiveness of
broad acre burning targets (Gibbons et al. 2012; Lindenmayer et al. 2014). As a
consequence of the interaction between research outputs, management experience and
community understanding over time, the broad-area targets were abandoned in 2015 in
favour of a risk-based approach to fire management and asset protection. These new
strategies will involve a spatially distinct application of fire from the previous policy.
190
6.5 Discussion
The importance of the spatial patterns produced by fire for beneficial ecological and
social outcomes is increasingly recognized (King et al. 2008; Stephens et al. 2008;
Driscoll et al. 2010b). However, spatial approaches to fire management in forested
systems are rarely implemented, with the emphasis placed on managing landscapes for
temporal aspects of fire (Bradstock et al. 2005; Parr and Andersen 2006; Clarke 2008).
Our study employed a trans-pacific qualitative interview-based research approach to
identify key barriers to the implementation of spatial approaches to fire management.
We found that spatial approaches to fire management must co-exist within a complex
system of social and ecological feedbacks between landscapes, academic research,
socio-political land management systems and environmental pressures. Through
identifying knowledge, management, environmental and socio-political barriers, we
developed a list of strategic actions required in each component of this system to
achieve the integration of spatial approaches to fire management into current paradigms
(see Table. 2).
The role of ecological knowledge in the spatial management of fire
Ecological research has a core role in forming the knowledge base upon which policy
goals and management practices can be developed (Russell-Smith et al. 2015). For
many conservation scientists, the practical implementation of their findings is a
fundamental part of their activity (Arlettaz et al. 2010). However, ecological knowledge
does not flow linearly from research through management into practices and must co-
exist within complex socio-political value systems (Christensen et al. 1996). Scientific
knowledge has been identified as one component of a complex feed-back system
191
involving both policy and public perception which drives decision-making in
conservation (Martín-López et al. 2009). In this context, developing management and
policy guidelines from scientific knowledge relating to desirable spatial fire patterns to
both beneficial ecological diversity and achieve social risk objectives is inherently
complex due to the variability in fire behavior and effects between different systems.
Achieving beneficial patterns of fire requires a shift in perspective from seeking blanket
approaches which can be applied to all systems, to locally specific knowledge and
management structures which can maximize outcomes. For example, the restoration of
spatial heterogeneity between forests adapted to low or mixed severity fire regimes and
those which experience stand-replacing high severity crown fires requires
fundamentally different approaches (Agee 2002; Thompson and Spies 2010). In systems
adapted to short-term intervals of low severity fire, such as mixed conifer forests, the
application of fire in a fine-scale patchwork of treatments at regular intervals, coupled
with the absence of fire suppression can restore desirable fire regimes (Stephens 1998).
In contrast, the restoration of desirable fire regimes in crown-fire systems requires a
high level of fire suppression and the exclusion of fire promoting land-uses to avoid
increasing fire return intervals which can drive forests into alternative successional
stages (Lindenmayer et al. 2011b). It has been suggested that improved integrated
modeling products which produce guidelines for landscape patterns can aid managers in
the design and implementation of desirable landscape patterns (Opdam et al. 2001).
Our findings support the concept that the successful integration of ecological knowledge
into management requires a shift in focus from producing knowledge transfer products,
to developing systems for the integration of scientific and local knowledge between
192
researchers and land managers (Raymond et al. 2010). We identified attitudes to the
contemporary body of ecological fire knowledge, which limits its ability to inform fire
management decision-making. In our study, the conceptual nature of current
publications and rapid developments in spatial understanding of fire in a generalized
context were identified as major barriers preventing the application of current
knowledge. Whilst researchers stressed the importance of generalizable research and
modeling products to aid decision-making, land managers stressed their need for highly
specific local information relating to desirable fire configurations, burn sizes and
locations. Similar studies into the science and policy interface have found that
developing discourse between managers and practitioners, and their increased
awareness of the sociological context surrounding their work may help lead to a more
fruitful integration of contemporary science in forest management (Rist et al. 2015).
The role of communities in the spatial management of fire
Fire management practice and policy is driven by community perceptions of fire as a
detrimental process and social values related to asset protection and fire suppression
(Kauffman 2004). Our findings demonstrate that public engagement with fire science
knowledge is an important aspect of achieving spatially-based fire management. Social
aversion to risk rather than ecological rationales drive decisions relating to prescribed
fire application and management (Ryan et al. 2013). For example, the occurrence of
destructive large wildfires can result in broad-scale reactionary policies, such as the
mandated area- based policies implemented in Victoria Australia following the ‘Black
Saturday’ fires (Clode and Elgar 2014). Such values-based policies often ignore more
nuanced risk-based management approaches, such as strategically placing fuel reduction
burns within the urban wildland interface to maximize their efficacy in property
193
protection (McCaw 2013). Additionally, negative public attitudes towards the
consequences of applied fire management, such as reduction in air quality can lead to
legislation which restricts the ability of such actions to be applied effectively
(Schweizer and Cisneros 2014).
The role of policy in the spatial management of fire
Current fire management policies in the Northwest United States and south-eastern
Australia are focused on asset protection, fuel reduction and fire suppression (Spies et
al. 2012). This approach presents major challenges to restoring ecological patterns of
fire within forested landscapes. During large fires, emergency management structures
often override long-term management plans in order to successfully extinguish fire and
reduce risks to communities (Houtman et al. 2013). These emergency management
approaches often employ tactics such as blackout burning, which homogenize post-fire
landscape structures and reduce the ability of managers to maintain spatio-temporal
heterogeneity in ecological systems (Backer et al. 2004). Consequently, policy
approaches which treat large fires with a disaster response mentality re-enforce barriers
to implementing ecological fire management and perpetuate societal fear of fire
(Kauffman 2004).
The role of management agencies in the spatial management of fire
Our study identified the organizational structures, priorities and resourcing of
management agencies as institutional barriers to the employing a spatial approach to fire
management. We found that agencies established primarily to manage landscapes for
biodiversity outcomes were mandated to prioritize asset protection and social risk
reduction within their fire management plans. Balancing these objective within reserve
194
systems is a considerable challenge and requires new data quantifying trade-offs
between asset protection and conservation goals to improve fire policy and practices
(Driscoll et al. 2010a). However, participants in our study identified funding and
resources provided solely for asset protection aims of fire management, restricted the
feasibility of implementing ecological burning programs. Despite the lack of available
resourcing, we found that both managers and researchers agreed unanimously that
current fire management tools could be used to create ecologically beneficial fire
patterns. Improving ecological resilience to fire by restoring beneficial fire regimes will
likely produce desirable outcomes in risk reduction for less than is currently invested in
fire response and suppression budgets (Loehle 2004). Investing in ecosystem services is
often viewed as a win-win situation in decision-making relating to environmental and
developmental objectives, due to the creation of positive feedbacks (De Groot et al.
2010). In this regard, our results suggest linking ecological and asset protection
outcomes could overcome funding and resourcing issues in fire management.
The role of environmental factors in the spatial management of fire
Addressing the socio-political factors involved in land management will not be
sustainable in the long term without concurrent management of key threating
environmental processes (Brook et al. 2008). Climate change, invasive species and
habitat destruction are altering the distribution of fire regimes within forested systems
globally (Dale et al. 2001). If the impacts of these threats continue on current
trajectories, future fire management strategies will likely need considerably different
approaches than those currently advocated.
195
Key steps to incorporating a spatial approach to fire management within current
frameworks
Managing desirable spatial patterns of fire within fire-prone systems is gaining
increasing traction as a viable approach to biodiversity conservation as altered fire
regimes emerge as a key threatening process (Driscoll et al. 2010b). Previous studies
have identified barriers to knowledge integration between science and land management
as challenges to achieving the uptake and communication of science (Tress et al. 2007).
Our findings suggest that spatially managing fire can be achieved within current socio-
political land management systems through a number of refinements to existing
processes (Table 2). These steps relate to developing community understanding of fire
science, improving the relevance of fire research outputs to land management,
amending existing government policy approaches and refining management tools,
structures, scales and monitoring to meet biodiversity and fire risk objectives. Steps
aimed at addressing knowledge and community barriers to spatial fire management can
be met with responses primarily from the scientific community. Our suggested solutions
to barriers in management and policy areas do not require substantial investment and
can be addressed during regular cycles of review.
196
Table 2. Key steps within four major areas to achieving spatially based ecological
fire management.
Steps to achieving spatially based ecological fire management
Research which addresses the spatial assumptions currently
made by managers
Knowledge
Improved spatial modeling planning tools to inform managers
where to burn
Research partnerships between researchers and managers to
identify key spatial questions
Policy refinement to recognize the value of ecological and fire
risk values of spatial managing fire patterns
Policy
Government policies relating to spatial burning strategies
reviewed by managers and researchers to ensure relevance to
local systems
Cessations of broad-area burning and a move towards more
restricted and targeted fuel reduction and suppression
measures
Mandated monitoring beyond first level fire effects to enable
the consequences of spatial outcomes to inform future
management plans
Management
Separation of long-term management structures from
emergency management to prevent spatial planning from being
overridden
Consideration of larger scale perspectives of fire to include
patterns and strategies at the landscape and range level
Application of techniques other than prescribed burning to
manage fire, to introduce spatial complexity outside of
available suitable burning windows
Educate community on importance of spatially locating fire
treatments in particular the importance of targeted and
localized fire suppression actions
Community
Communicate and highlight the value of burnt forests and
importance of mosaics
197
Figure 3. Photographs depicting fire prone forest landscapes in the Victorian
Central Highlands, Australia. A) industrial harvesting within forests burnt by the
2009 ‘Black Saturday’ wildfires. B) An example of an unburnt fire refuge within a
sheltered mesic gully in the Armstrong water catchment. C) Aerial image of the
complex spatial patterns resulting from the temporal occurrence of clear-fell
harvesting within a mixed severity fire mosaic. D) Aerial image depicting
homogeneous high severity patterns occurring under extreme fire weather
conditions (top half of image) and mixed severity fire patterns occurring during
moderate fire weather conditions (bottom half of image). Credits: A and B, David
Blair, C and D, Department of Sustainability and Environment, Victoria,
Australia.
A
.
B
.
C
.
D
.
198
Acknowledgements
The authors acknowledge the de-identified participants of the study who gave their time
to provide the data used in this study. LB would like to thank Claudia Benham for her
guidance with the data analysis. LB would like to thank Fiona Tew for her support in
the field. This study was funded through a travelling fellowship award from the ARC
Center of Excellence for Environmental Decisions. This project was conducted in
accordance with ANU human ethics permit number: 2014/672.
199
References
Agee, J. K. 1993. Fire ecology of Pacific Northwest forests. Island Press.
Agee, J. K. 2002. The fallacy of passive management managing for firesafe forest
reserves. Conservation in Practice 3:18-26.
Arlettaz, R., M. Schaub, J. Fournier, T. S. Reichlin, A. Sierro, J. E. M. Watson, and V.
Braunisch. 2010. From Publications to Public Actions: When Conservation
Biologists Bridge the Gap between Research and Implementation. BioScience
60:835-842.
Backer, D. M., S. E. Jensen, and G. McPherson. 2004. Impacts of fire‐suppression
activities on natural communities. Conservation Biology 18:937-946.
Berry, L. E., D. A. Driscoll, J. A. Stein, W. Blanchard, S. C. Banks, R. A. Bradstock,
and D. B. Lindenmayer. 2015. Identifying the location of fire refuges in wet
forest ecosystems. Ecological Applications.
Bonan, G. B. 2008. Forests and climate change: forcings, feedbacks, and the climate
benefits of forests. Science 320:1444-1449.
Bond, W. J., F. I. Woodward, and G. F. Midgley. 2005. The global distribution of
ecosystems in a world without fire. New Phytologist 165:525-538.
Bosomworth, K., J. Handmer, and R. Thornton. 2015. The role of social science in the
governance and management of wildland fire. International Journal of Wildland
Fire 24:151-152.
Bradstock, R. A., M. Bedward, A. M. Gill, and J. S. Cohn. 2005. Which mosaic? A
landscape ecological approach for evaluating interactions between fire regimes,
habitat and animals. Wildlife Research 32:409-423.
Brook, B. W., N. S. Sodhi, and C. J. Bradshaw. 2008a. Synergies among extinction
drivers under global change. Trends in Ecology & Evolution 23:453-460.
Christensen, N. L., A. M. Bartuska, J. H. Brown, S. Carpenter, C. D'Antonio, R.
Francis, J. F. Franklin, J. A. MacMahon, R. F. Noss, D. J. Parsons, C. H.
Peterson, M. G. Turner, and R. G. Woodmansee. 1996. The Report of the
Ecological Society of America Committee on the Scientific Basis for Ecosystem
Management. Ecological Applications 6:665-691.
Clarke, M. F. 2008. Catering for the needs of fauna in fire management: science or just
wishful thinking? Wildlife Research 35:385-394.
Clode, D. and M. A. Elgar. 2014. Fighting Fire with Fire: Does a Policy of Broad-Scale
Prescribed Burning Improve Community Safety? Society & Natural Resources
27:1192-1199.
Creswell, J. W. and D. L. Miller. 2000. Determining validity in qualitative inquiry.
Theory into practice 39:124-130.
Cruz, M. G., A. L. Sullivan, J. S. Gould, N. C. Sims, A. J. Bannister, J. J. Hollis, and R.
J. Hurley. 2012. Anatomy of a catastrophic wildfire: The Black Saturday
Kilmore East fire in Victoria, Australia. Forest Ecology and Management
284:269-285.
Dale, V. H., L. A. Joyce, S. McNulty, R. P. Neilson, M. P. Ayres, M. D. Flannigan, P. J.
Hanson, L. C. Irland, A. E. Lugo, and C. J. Peterson. 2001. Climate change and
forest disturbances: climate change can affect forests by altering the frequency,
intensity, duration, and timing of fire, drought, introduced species, insect and
pathogen outbreaks, hurricanes, windstorms, ice storms, or landslides.
BioScience 51:723-734.
De Groot, R. S., R. Alkemade, L. Braat, L. Hein, and L. Willemen. 2010. Challenges in
integrating the concept of ecosystem services and values in landscape planning,
management and decision making. Ecological Complexity 7:260-272.
200
Driscoll, D. A., D. B. Lindenmayer, A. F. Bennett, M. Bode, R. A. Bradstock, G. J.
Cary, M. F. Clarke, N. Dexter, R. Fensham, G. Friend, M. Gill, S. James, G.
Kay, D. A. Keith, C. MacGregor, H. P. Possingham, J. Russel-Smith, D. Salt, J.
E. M. Watson, D. Williams, and A. York. 2010a. Resolving conflicts in fire
management using decision theory: asset-protection versus biodiversity
conservation. Conservation Letters 3:215-223.
Driscoll, D. A., D. B. Lindenmayer, A. F. Bennett, M. Bode, R. A. Bradstock, G. J.
Cary, M. F. Clarke, N. Dexter, R. Fensham, G. Friend, M. Gill, S. James, G.
Kay, D. A. Keith, C. MacGregor, J. Russell-Smith, D. Salt, J. E. M. Watson, R.
J. Williams, and A. York. 2010b. Fire management for biodiversity
conservation: Key research questions and our capacity to answer them.
Biological Conservation 143:1928-1939.
Fernandes, P. M., G. M. Davies, D. Ascoli, C. Fernández, F. Moreira, E. Rigolot, C. R.
Stoof, J. A. Vega, and D. Molina. 2013. Prescribed burning in southern Europe:
developing fire management in a dynamic landscape. Frontiers in Ecology and
the Environment 11:e4-e14.
Gibbons, P., L. van Bommel, A. M. Gill, G. J. Cary, D. A. Driscoll, R. A. Bradstock, E.
Knight, M. A. Moritz, S. L. Stephens, and D. B. Lindenmayer. 2012. Land
management practices associated with house loss in wildfires. Plos One 7.
Gill, A. M. 1975. Fire and the Australian flora: a review. Australian forestry 38:4-25.
Harris, L. and A. Taylor. 2015. Topography, fuels, and fire exclusion drive fire severity
of the rim fire in an old-growth mixed-conifer forest, Yosemite National Park,
USA. Ecosystems 18:1192-1208.
Houtman, R. M., C. A. Montgomery, A. R. Gagnon, D. E. Calkin, T. G. Dietterich, S.
McGregor, and M. Crowley. 2013. Allowing a wildfire to burn: estimating the
effect on future fire suppression costs. International Journal of Wildland Fire
22:871-882.
Kauffman, J. B. 2004. Death Rides the Forest: Perceptions of fire, land use, and
ecological eestoration of western forests, Conservation Biology 18: 878-882.
King, K. J., R. A. Bradstock, G. J. Cary, J. Chapman, and J. B. Marsden-Smedley. 2008.
The relative importance of fine-scale fuel mosaics on reducing fire risk in south-
west Tasmania, Australia. International Journal of Wildland Fire 17:421-430.
Leonard, S. W., A. F. Bennett, and M. F. Clarke. 2014. Determinants of the occurrence
of unburnt forest patches: Potential biotic refuges within a large, intense wildfire
in south-eastern Australia. Forest Ecology and Management 314:85-93.
Lindenmayer, D. 1995. Forest disturbance, forest wildlife conservation and the
conservative basis for forest management in the mountain ash forests of
Victoria—comment. Forest Ecology and Management 74:223-231.
Lindenmayer, D. B., W. Blanchard, L. McBurney, D. Blair, S. C. Banks, D. A. Driscoll,
A. L. Smith, and A. M. Gill. 2014. Complex responses of birds to landscape-
level fire extent, fire severity and environmental drivers. Diversity and
Distributions 20:467-477.
Lindenmayer, D. B., R. J. Hobbs, G. E. Likens, C. J. Krebs, and S. C. Banks. 2011.
Newly discovered landscape traps produce regime shifts in wet forests.
Proceedings of the National Academy of Sciences of the United States of
America 108:15887-15891.
Loehle, C. 2004. Applying landscape principles to fire hazard reduction. Forest Ecology
and Management 198:261-267.
201
Martín-López, B., C. Montes, L. Ramírez, and J. Benayas. 2009. What drives policy
decision-making related to species conservation? Biological Conservation
142:1370-1380.
McCaw, W. L. 2013. Managing forest fuels using prescribed fire – A perspective from
southern Australia. Forest Ecology and Management 294:217-224.
Noss, R. F., J. F. Franklin, W. L. Baker, T. Schoennagel, and P. B. Moyle. 2006.
Managing fire-prone forests in the western United States. Frontiers in Ecology
and the Environment 4:481-487.
Opdam, P., R. Foppen, and C. Vos. 2001. Bridging the gap between ecology and spatial
planning in landscape ecology. Landscape Ecology 16:767-779.
Parr, C. L. and A. N. Andersen. 2006. Patch mosaic burning for biodiversity
conservation: a critique of the pyrodiversity paradigm. Conservation Biology
20:1610-1619.
Patton, M. Q. 2002. Qualitative interviewing. Qualitative research and evaluation
methods 3:344-347.
Perry, D. A., P. F. Hessburg, C. N. Skinner, T. A. Spies, S. L. Stephens, A. H. Taylor, J.
F. Franklin, B. McComb, and G. Riegel. 2011. The ecology of mixed severity
fire regimes in Washington, Oregon, and Northern California. Forest Ecology
and Management 262:703-717.
Raymond, C. M., I. Fazey, M. S. Reed, L. C. Stringer, G. M. Robinson, and A. C.
Evely. 2010. Integrating local and scientific knowledge for environmental
management. Journal of Environmental Management 91:1766-1777.
Rist, L., C. Shackleton, L. Gadamus, F. S. Chapin, III, C. M. Gowda, S. Setty, R.
Kannan, and R. U. Shaanker. 2015. Ecological Knowledge Among
Communities, Managers and Scientists: Bridging Divergent Perspectives to
Improve Forest Management Outcomes. Environmental Management:1-16.
Robinson, N. M., S. W. Leonard, E. G. Ritchie, M. Bassett, E. K. Chia, S. Buckingham,
H. Gibb, A. F. Bennett, and M. F. Clarke. 2013. REVIEW: Refuges for fauna in
fire‐prone landscapes: their ecological function and importance. Journal of
Applied Ecology 50:1321-1329.
Russell-Smith, J., D. Lindenmayer, I. Kubiszewski, P. Green, R. Costanza, and A.
Campbell. 2015. Moving beyond evidence-free environmental policy. Frontiers
in Ecology and the Environment 13:441-448.
Russell-Smith, J., C. Yates, A. Edwards, G. E. Allan, G. D. Cook, P. Cooke, R. Craig,
B. Heath, and R. Smith. 2003. Contemporary fire regimes of northern Australia,
1997–2001: change since Aboriginal occupancy, challenges for sustainable
management. International Journal of Wildland Fire 12:283-297.
Ryan, K. C., E. E. Knapp, and J. M. Varner. 2013. Prescribed fire in North American
forests and woodlands: history, current practice, and challenges. Frontiers in
Ecology and the Environment 11:e15-e24.
Schweizer, D. and R. Cisneros. 2014. Wildland fire management and air quality in the
southern Sierra Nevada: Using the Lion Fire as a case study with a multi-year
perspective on PM2.5 impacts and fire policy. Journal of Environmental
Management 144:265-278.
Spies, T. A., D. B. Lindenmayer, A. M. Gill, S. L. Stephens, and J. K. Agee. 2012.
Challenges and a checklist for biodiversity conservation in fire-prone forests:
Perspectives from the Pacific Northwest of USA and Southeastern Australia.
Biological Conservation 145:5-14.
202
Stephens, S. L. 1998. Evaluation of the effects of silvicultural and fuels treatments on
potential fire behaviour in Sierra Nevada mixed-conifer forests. Forest Ecology
and Management 105:21-35.
Stephens, S. L., D. L. Fry, and E. Franco-Vizcaino. 2008. Wildfire and spatial patterns
in forests in Northwestern Mexico: The United States wishes it had similar fire
poblems. Ecology and Society 13: 1538-1547.
Swanson, M. E., J. F. Franklin, R. L. Beschta, C. M. Crisafulli, D. A. DellaSala, R. L.
Hutto, D. B. Lindenmayer, and F. J. Swanson. 2010. The forgotten stage of
forest succession: early-successional ecosystems on forest sites. Frontiers in
Ecology and the Environment 9:117-125.
Taylor, A. H. and C. N. Skinner. 2003. Spatial pattterns and controls on historical fire
regimes and forest structure in the Klamath Mountains. Ecological Applications
13:704-719.
Taylor, C., M. A. McCarthy, and D. B. Lindenmayer. 2014. Non‐linear effects of stand
age on fire severity. Conservation Letters. 7: 355-370.
Teague, B., R. McLeod, and S. Pascoe. 2010. Final report, 2009 Victorian bushfires
royal commission. Parliament of Victoria, Melbourne Victoria, Australia.
Thompson, J. R. and T. A. Spies. 2010. Factors associated with crown damage
following recurring mixed-severity wildfires and post-fire management in
southwestern Oregon. Landscape Ecology 25:775-789.
Tress, G., B. Tress, and G. Fry. 2007. Analysis of the barriers to integration in
landscape research projects. Land Use Policy 24:374-385.
Yoder, J., D. Engle, and S. Fuhlendorf. 2004. Liability, incentives, and prescribed fire
for ecosystem management. Frontiers in Ecology and the Environment 2:361-
366.
203
Chapter 7. Conclusions
Landscape-scale wildfires are a major form of ecosystem disturbance and are a driver of
biodiversity in forested systems worldwide (Stephens et al. 2014). Fires in
topographically diverse forest systems often produce heterogeneous burn patterns
(Burton et al. 2009; Barros et al. 2013). Within the extent of large fires, parts of the
landscape which experience fire regimes differing from those prevailing in the
surrounding forest may act as refuges, influencing the survival, persistence and
distribution of biota (Robinson et al. 2013). However, desirable refuge attributes and the
importance of the interaction between refuge characteristics, surrounding patterns and
environmental gradients are poorly understood (Driscoll et al. 2010). Identifying how
the spatial outcomes of disturbance determine the distribution of species within fire-
prone forests addresses key knowledge gaps in fire ecology and is a core component of
developing effective conservation strategies to mitigate the effects of altered fire
regimes on biodiversity (Bradstock et al. 2005; Clarke 2008).
My thesis aimed to quantify the ecological role of fire refuges in fire affected forests
using a range of techniques to develop scientific theory and ultimately provide evidence
to support conservation and land management planning. I explored the spatial aspects
of fire refuge ecology with a series of papers addressing three key areas; 1) the factors
governing the occurrence and distribution of refuges, 2) the influence of refuge type and
spatial context on the distribution of fauna, 3) the mechanisms underpinning faunal
response to post-fire landscape patterns. The final chapter of the thesis addressed how
understanding of the spatial consequences of fire can be integrated into contemporary
fire management for improved biodiversity outcomes. Collectively, the works included
204
in this thesis provide key advances in spatial fire ecology and develop our theoretical
understanding of fire refuges.
Fire refuge occurrence
The occurrence of potential fire refuges is an inherent outcome large-scale high
intensity crown fires in topographically diverse montane forests. Under moderate fire
conditions, fire refuge occurrence is mediated by topography. As fire conditions become
more extreme, refuge occurrence will become restricted to topographically sheltered
areas and determined by stochastic factors such as fire weather in more exposed areas.
To maintain the processes leading the establishment and subsequent use of fire refuges,
it is essential that land management practices which may alter the behaviour of fire in
forested landscapes and affect species use of refuges are excluded from the landscape.
My findings demonstrate that land management agencies can employ predictive
landscape models as decision-making tools to map the distribution of fire refuge
envelopes.
The influence of fire refuges and post-fire spatial patterns on faunal distributions
Fire refuges have been defined as areas within burnt landscapes which support fire
regimes differing from those prevailing in the surrounding landscape (Mackey et al.
2012). Core theory on species response to habitat fragmentation in other contexts
suggest that species use of fire refuges may be dependent on three types of attributes;
patch quality, patch size and shape and landscape context (Lindenmayer and Fischer
2006). However, post-fire landscapes vary from other fragmented systems (such as
agricultural landscapes) due to potentially higher matrix hospitability, complex spatial
205
and temporal patterns of fire severity, and a diverse array of successional gradients. I
predicted that these factors would influence the ecological role of fire refuges in
recently burnt forests.
My study of bird response to the spatial attributes and landscape context of fire refuges
on bird distribution suggests that for many species, refuges are unlikely to appear as
isolated islands in an inhospitable sea. The concepts of ecosystems Greenspots, and fire
skips and isolates suggest that intact habitat remnant in disturbed forests may support
species unable to persist in the surrounding landscape (Burton et al. 2008; Mackey et al.
2012). Our findings indicate that a small number of dispersal-limited birds are able to
occupy these areas given sufficient patch size. However, we found that the presence of
areas of intact canopy which had experienced understorey fire were an important
determinant of occurrence for many birds. The findings detailed in chapter 4 are likely
to be amplified within crown-fire systems burnt during extreme conditions due to the
scarce availability of areas of mature intact canopy within the extent of fires (Leonard et
al. 2014). My findings suggest that for some species, these types of ephemeral refuges
determined by stochastic processes during the previous fire, contribute to patterns of
resource heterogeneity which are fundamental to the occurrence of many species in
systems which experience extensively high severity burns. In this regard, the
consideration of potential fire refuge areas should be developed from areas with long-
interval fire regimes which support old-growth habitat features, to also include
stochastically determined fire patterns which we have found are a core factor in
supporting higher species occurrences in burnt landscapes.
206
Due to the ability of most species to tolerate the dense regrowth habitat in the recently
burnt matrix, fire refuges were more likely to be represented by the interaction of local
fire severity and surrounding landscape context for forest birds. I found that birds which
favoured unburnt habitat were able to persist in burnt forest if sufficient unburnt habitat
was available in the surrounding landscape. For many species, the increased availability
of unburnt forest in the surrounding landscape increased occurrence. This suggests that
at a larger scale, regional features, such as mountains which may influence the spatial
outcomes of fire at larger scales should also be considered as contributing a similar role
to faunal persistence as patch-scale fire refuges (Gill and Bradstock 1995). In this
regard, fire refuge theory should be expanded to include the influence of fire shadows
driving desirable fire patterns for fauna at a landscape scale.
Faunal responses to intact habitat islands within disturbed landscapes often differ in fire
affected systems due to the relative hospitability of the matrix (Driscoll et al. 2013;
Berry et al. 2015). In many cases, fires promote resource availability for species (Nappi
and Drapeau 2009). In my study, many species benefited from the resource
heterogeneity provided by the fine scale availability of both high-severity burnt and
unburnt habitat features. Localised heterogeneity in resources is a major driver of
diversity in many systems. In burnt landscapes, species are often able to exploit novel
resource opportunities presented by areas of contrasting fire severity (Berry et al. 2015).
In some cases, localised resource heterogeneity, such as the availability of tree hollows
in burnt forest and intact canopy in adjacent unburnt forest can enable species
commonly associated with mature forest to persist within disturbed systems. This is an
example of how fire refuges for some species should be considered as the availability of
207
habitat structures and features necessary for survival, within an adequate spatial range,
with less emphasis on discrete patches or fire severity.
For many birds, increased occurrence in fire-affected landscapes was contingent on the
interaction of appropriate local and regional fire severity context with desirable
environmental gradients. An important contribution of this thesis is the ability of intact
forest patches, intact canopy or areas of fine-scale resource heterogeneity, to function as
fire refuges is dependent on their relationship to key environmental gradients within the
landscape. Bird response to the amount of forest burnt at different fire severities was not
uniform across the landscape. I found that bird responses to the proportion of unburnt
canopy, intact forest or high severity fire can differ depending on the availability or
absence of key habitat components or at extremes of environmental gradients such as
high elevation and rainfall.
Collectively, the findings in chapters 3 and 4 have significant implications for future
biodiversity management planning in montane forest systems. These studies
demonstrate that fire refuges perform an important ecological role in facilitating the
persistence of species and functional traits within extensively disturbed ecosystems.
Consequently, unburnt forest patches embedded within burnt landscapes should be
managed for their conservation values and protected from anthropogenic disturbance
such as timber harvesting, blackout burning and road building. To produce ecologically
beneficial fire patterns, land managers must develop strategies to produce mosaics of
mixed severity fire which overlap with a range of biotic and abiotic gradients. It is
particularly important that areas which support extremes of these gradients within
landscapes (ie, old-growth vegetation, high elevations and low elevation), often
208
associated with rarer species, are given priority in land management planning as
potentially valuable refuge areas.
Faunal response mechanisms
The spatial extent of fire severities altered the spatio-temporal movement patterns of a
forest marsupial. Our study of the impacts of fire severity heterogeneity on the
movement patterns and dynamics of the mountain brushtail possum, Trichosurus
cunninghami demonstrates that early successional forested landscapes burnt by high-
severity crown-fires may provide important habitat for mammals. These findings also
may be applicable in high productivity forests globally, where fire events are followed
by rapid tree growth (Ne'eman and Izhaki 1998a; Williams 2000; Johnstone and Chapin
2006). My findings indicate that early successional forest does not represent non-habitat
for some forest mammals. However, current land management policies in burnt forests
globally overlook their ecological value and instead focus on recovering as much
economic value as possible through post-fire salvage logging (Hutto and Gallo 2006;
Lindenmayer and Noss 2006; Schmiegelow et al. 2006). The removal of key habitat
structures and biological legacies such hollow bearing trees from landscapes recently
disturbed by fire may result in altered animal and plant successional dynamics within
these systems (Lindenmayer and Ough 2006)
209
Incorporating spatial fire science into fire management practices
The integration spatial fire ecology principles into land management must occur within
a complex system of feedbacks between research, socio-political management systems,
dynamic landscapes and external environmental factors. Spatially managing fire can be
achieved within this framework by developing community understanding of fire
science, improving the relevance of fire research outputs to land management,
amending existing government policy approaches and refining management tools,
structures, scales and monitoring to meet biodiversity and fire risk objectives.
Emerging questions, study limitations and opportunities for further research
An important component of fire refuge theory not examined in this thesis was the role
of temporal variation. Our studies, conducted five years after a major wildfire could not
determine the influence of fire refuges on the immediate survival of species. However,
these aspects have been noted elsewhere (Weir et al. 2000; Bradstock et al. 2005;
Clarke 2008; Kelly et al. 2012). As the burnt forest matrix surrounding fire refuges
recovers over time, it is likely that the role of refuges on species survival and
distributions will diminish (Robinson et al. 2013). In the long-term absence of fire,
refuge areas may still contribute valuable sources of habitat heterogeneity, particularly
for species which require habitat features such as tree hollows which form in the long-
term absence of fire and must persist through multiple fire events. To quantify the
temporal dynamics of refuges on fauna, it is necessary to take a long-term approach.
Examining how the spatial outcomes of large fires influence faunal distributions over
time and their interactions with other spatio-temporal disturbances in forested systems,
such as logging are priorities for future research.
210
References
Barros, A. M., J. Pereira, M. A. Moritz, and S. L. Stephens. 2013. Spatial
characterization of wildfire orientation patterns in California. Forests 4:197-217.
Berry, L. E., D. B. Lindenmayer, and D. A. Driscoll. 2015. Large unburnt areas, not
small unburnt patches, are needed to conserve avian diversity in fire‐prone
landscapes. Journal of Applied Ecology 52:486-495.
Bradstock, R. A., M. Bedward, A. M. Gill, and J. S. Cohn. 2005. Which mosaic? A
landscape ecological approach for evaluating interactions between fire regimes,
habitat and animals. Wildlife Research 32:409-423.
Burton, P. J., M.-A. Parisien, J. A. Hicke, R. J. Hall, and J. T. Freeburn. 2009. Large
fires as agents of ecological diversity in the North American boreal forest.
International Journal of Wildland Fire 17:754-767.
Burton, P. J., M. Parisien, J. A. Hicke, R. J. Hall, and J. T. Freeburn. 2008. Large fires
as agents of ecological diversity in the North American boreal forest.
International Journal of Wildland Fire 17:754-767.
Clarke, M. F. 2008. Catering for the needs of fauna in fire management: science or just
wishful thinking? Wildlife Research 35:385-394.
Driscoll, D. A., S. C. Banks, P. S. Barton, D. B. Lindenmayer, and A. L. Smith. 2013.
Conceptual domain of the matrix in fragmented landscapes. Trends in Ecology
& Evolution 28:605-613.
Driscoll, D. A., D. B. Lindenmayer, A. F. Bennett, M. Bode, R. A. Bradstock, G. J.
Cary, M. F. Clarke, N. Dexter, R. Fensham, G. Friend, M. Gill, S. James, G.
Kay, D. A. Keith, C. MacGregor, J. Russell-Smith, D. Salt, J. E. M. Watson, R.
J. Williams, and A. York. 2010. Fire management for biodiversity conservation:
Key research questions and our capacity to answer them. Biological
Conservation 143:1928-1939.
Gill, A., and R. Bradstock. 1995. 'Extinction of biota by fires', in Conserving
biodiversity: threats and solutions, Chipping Norton, Australia: 309-322.
Hutto, R. L., and S. M. Gallo. 2006. The effects of postfire salvage logging on cavity-
nesting birds. Condor 108:817-831.
Johnstone, J., and F. S. Chapin, III. 2006. Effects of soil burn severity on post-fire tree
recruitment in boreal forests. Ecosystems 9:14-31.
Kelly, L. T., D. G. Nimmo, L. M. Spence-Bailey, R. S. Taylor, S. J. Watson, M. F.
Clarke, and A. F. Bennett. 2012. Managing fire mosaics for small mammal
conservation: a landscape perspective. Journal of Applied Ecology 49:412-421.
Leonard, S. W., A. F. Bennett, and M. F. Clarke. 2014. Determinants of the occurrence
of unburnt forest patches: Potential biotic refuges within a large, intense wildfire
in south-eastern Australia. Forest Ecology and Management 314:85-93.
Lindenmayer, D., and R. Noss. 2006. Salvage logging, ecosystem processes, and
biodiversity conservation. Conservation Biology 20:949-958.
Lindenmayer, D., and K. Ough. 2006. Salvage logging in the montane ash eucalypt
forests of the Central Highlands of Victoria and its potential impacts on
biodiversity. Conservation Biology 20:1005-1015.
Lindenmayer, D. B., and J. Fischer. 2006. Habitat fragmentation and landscape change:
an ecological and conservation synthesis. Island Press.
Mackey, B., S. Berry, S. Hugh, S. Ferrier, T. D. Harwood, and K. J. Williams. 2012.
Ecosystem greenspots: identifying potential drought, fire, and climate-change
micro-refuges. Ecological Applications 22:1852-1864.
211
Nappi, A., and P. Drapeau. 2009. Reproductive success of the black-backed
woodpecker (Picoides arcticus) in burned boreal forests: Are burns source
habitats? Biological Conservation 142:1381-1391.
Ne'eman, G., and I. Izhaki. 1998. Stability of pre- and post-fire spatial structure of pine
trees in Aleppo pine forest. Ecography 21:535-542.
Robinson, N. M., S. W. Leonard, E. G. Ritchie, M. Bassett, E. K. Chia, S. Buckingham,
H. Gibb, A. F. Bennett, and M. F. Clarke. 2013. REVIEW: Refuges for fauna in
fire‐prone landscapes: their ecological function and importance. Journal of
Applied Ecology 50:1321-1329.
Schmiegelow, F. K., D. P. Stepnisky, C. A. Stambaugh, and M. Koivula. 2006.
Reconciling salvage logging of boreal forests with a natural‐disturbance
management model. Conservation Biology 20:971-983.
Stephens, S. L., N. Burrows, A. Buyantuyev, R. W. Gray, R. E. Keane, R. Kubian, S.
Liu, F. Seijo, L. Shu, and K. G. Tolhurst. 2014. Temperate and boreal forest
mega-fires: characteristics and challenges. Frontiers in Ecology and the
Environment 12:115-122.
Weir, J. M. H., E. A. Johnson, and K. Miyanishi. 2000. Fire frequency and the spatial
age mosaic of the mixed-wood boreal forest in western Canada. Ecological
Applications 10:1162-1177.
Williams, P. R. 2000. Fire-stimulated rainforest seedling recruitment and vegetative
regeneration in a densely grassed wet sclerophyll forest of north-eastern
Australia. Australian Journal of Botany 48:651-658.
212