Fire frequencies for Western Sydney’s woodlands: indications from vegetation dynamics
Penelope J. Watson
June 2005
Submitted for the degree of Doctor of Philosophy at the University of Western Sydney, Sydney, Australia
SUMMARY
Although the importance of fire management for biodiversity conservation is increasingly being recognised, little is known about the relationship between fire regimes and plant diversity in Australia’s temperate grassy woodland ecosystems. This project sought to address this gap in the woodlands of Western Sydney’s Cumberland Plain. Aspects of vegetation dynamics were investigated through six studies, mostly in shale-based Cumberland Plain Woodland (CPW) remnants. Results indicate that fire frequency profoundly affects both vegetation composition and structure.
The influence of fire cycles was most readily apparent in the shrub layer. A survey in CPW remnants with differing fire histories found a markedly higher abundance of Bursaria spinosa in sites where fire frequency was low, to the point where this species dominated the landscape. Other native shrubs, particularly obligate seeders, were most abundant in sites burnt once or twice a decade. Findings were consistent with predictions based on fire-related attributes of individual shrub species.
Themeda australis dominated the ground layer in high and moderate fire frequency sites, but not where fire frequency was low. A study of woodland microhabitats found that fire frequency did not affect ground layer species richness or composition directly, however open patches, patches around trees and patches under Bursaria varied in species composition. Thus fire frequency is likely to affect ground layer composition indirectly, through its influence on the shrub layer.
Findings from the six studies were synthesized into a state and transition model which allows exploration of management actions. Interfire intervals between 4 and 12 years are predicted to maintain Themeda woodland with both Bursaria thickets and open areas, and obligate seeder shrubs. Variable intervals across time and space within these thresholds should maintain much of the landscape at fuel levels compatible with property protection; fuel loads in CPW peak well below those in woodlands on sandstone. Low fire frequency remnants dominated by Bursaria retain many conservation values, but are likely to support lower abundances of obligate seeder shrubs and open patch herbs, and to be more weed-prone, than remnants burnt once or twice a decade. Experimentation with one or two short interfire intervals may be appropriate in long unburnt CPW.
i TABLE OF CONTENTS
Summary ...... i
Table of contents ...... ii
Acknowledgements ...... ix
Abbreviations ...... xii
Chapter 1 Introduction to the project...... 1 1.1 Project goal and focus ...... 1 1.2 Temperate woodlands...... 3 1.2.1 Description, location, conservation...... 3 1.2.2 Abiotic influences ...... 4 1.2.3 Introducing Western Sydney’s woodlands ...... 5 1.3 Plant species responses to burning ...... 6 1.3.1 Post-fire regeneration modes ...... 6 1.3.2 Reproduction and fire...... 7 1.3.3 Life history stages ...... 8 1.3.4 Plant responses to fire on the Cumberland Plain ...... 9 1.3.5 Study question concerning plant responses to fire...... 10 1.4 Fire frequency and woodland shrubs...... 10 1.4.1 Effects of frequent fire ...... 10 1.4.2 Effects of infrequent fire...... 12 1.4.3 Shrubs and fire frequency on the Cumberland Plain ...... 12 1.4.4 Study question concerning shrubs and fire frequency ...... 13 1.5 ‘Encroachment’ in grassy vegetation ...... 13 1.5.1 Encroachment across the planet...... 13 1.5.2 Encroachment on the Cumberland Plain...... 14 1.5.3 Study questions concerning encroachment...... 14 1.6 Fire and woodland trees...... 14 1.6.1 Eucalypt recruitment...... 14 1.6.2 Tree survival ...... 15 1.6.3 Fire and trees on the Cumberland Plain...... 16 1.6.4 Study question concerning fire and trees ...... 16 1.7 Fire, grasses and herbs...... 16 1.7.1 Grassland research ...... 16 1.7.2 Grassy woodland research ...... 17 1.7.3 Fire, grasses and herbs on the Cumberland Plain ...... 18 1.7.4 Study questions concerning fire, grasses and herbs...... 19 1.8 Guidelines for the use of fire in conservation management ...... 20 1.8.1 Recommended fire frequency thresholds...... 21 1.8.2 Guidelines for management of grassy woodlands ...... 22 1.8.3 Guidelines for managing fire on the Cumberland Plain ...... 22 1.9 Fire management at the urban fringe...... 23 1.9.1 Fuel accumulation on the Cumberland Plain ...... 24 1.10 Methodological approach ...... 24 1.11 Thesis outline...... 25
ii Chapter 2 The study area: Western Sydney’s Cumberland Plain...... 27 2.1 Location and landform...... 27 2.2 Geology and soils...... 28 2.3 Climate ...... 30 2.3.1 Climate averages...... 30 2.3.2 Rainfall during the project...... 31 2.4 Vegetation ...... 32 2.4.1 Cumberland Plain Woodland...... 34 2.4.2 Castlereagh Woodland...... 36 2.5 Management history...... 38 2.5.1 Aboriginal management ...... 38 2.5.2 European management ...... 40 2.6 Current status of vegetation ...... 41 2.7 Study sites ...... 43 2.7.1 Locating remnants ...... 43 2.7.2 Identifying fire history...... 44
Chapter 3 Shrub vital attributes ...... 45 3.1 Introduction...... 45 3.1.1 Vital attributes and plant functional types...... 45 3.1.2 Interval sensitivities...... 46 3.1.3 Study aims, boundaries and questions...... 46 3.2 Methods...... 48 3.2.1 Documenting post-fire regeneration modes ...... 48 3.2.2 Documenting juvenile periods...... 48 3.2.3 Consolidating observations...... 51 3.2.4 Comparison with NSW Database...... 52 3.2.5 Assigning vital attributes...... 52 3.2.6 Characterising shrub species in broad vegetation types ...... 53 3.2.7 Identifying key fire response species and thresholds of concern ...... 54 3.2.8 Threatened species...... 55 3.3 Results...... 55 3.3.1 Regeneration modes ...... 55 3.3.2 Seedling lines...... 55 3.3.3 Juvenile periods ...... 59 3.3.4 Comparison with NSW Database...... 59 3.3.5 Assigned vital attributes ...... 60 3.3.6 Characterisation of broad vegetation types ...... 60 3.3.7 Key fire response species in Cumberland Plain woodlands...... 65 3.3.8 Key fire response species in Castlereagh woodlands ...... 67 3.3.9 Threatened species...... 69 3.4 Discussion ...... 70 3.4.1 Reliability and availability of data...... 70 3.4.2 Post-fire seedling establishment ...... 72 3.4.3 Primary juvenile periods of obligate seeders...... 73 3.4.4 Characteristics of the Cumberland Plain shrub flora...... 73 3.4.5 Vital characteristics and fire interval domains ...... 75 3.4.6 Fire interval domain in Cumberland Plain woodlands ...... 77 3.4.7 Fire interval domain in Castlereagh woodlands ...... 79 3.5 Conclusions...... 81
iii Chapter 4 Fire frequency and woodland landscapes: shrubs, trees and grasses ...... 82 4.1 Introduction ...... 82 4.1.1 Fire frequency and woodland structure...... 82 4.1.2 Fire frequency and shrub species types...... 84 4.1.3 Fire frequency and grasses...... 85 4.1.4 Fire frequency and exotics ...... 85 4.1.5 Study aims, approach, and questions ...... 86 4.2 Methods ...... 88 4.2.1 Fire frequency categories and study sites ...... 88 4.2.2 Sampling and data collection ...... 92 4.2.3 Vegetation parameters...... 94 4.2.4 Data analysis ...... 99 4.3 Results ...... 99 4.3.1 Bursaria ...... 99 4.3.2 Other native shrubs ...... 102 4.3.3 Exotic shrubs...... 107 4.3.4 Trees...... 108 4.3.5 Grasses ...... 114 4.4 Discussion...... 118 4.4.1 Bursaria ...... 118 4.4.2 Other native shrubs ...... 120 4.4.3 Exotic shrubs...... 123 4.4.4 Trees...... 123 4.4.5 Grasses ...... 126 4.5 Conclusion...... 128
Chapter 5 Fire frequency and woodland landscapes: spatial pattern, shrub species composition, associations and grazing...... 129 5.1 Introduction ...... 129 5.1.1 Spatial pattern ...... 129 5.1.2 Species composition...... 129 5.1.3 Associations ...... 130 5.1.4 Grazing pressure ...... 131 5.1.5 Study questions ...... 131 5.2 Methods ...... 132 5.2.1 Spatial pattern ...... 132 5.2.2 Shrub species composition...... 133 5.2.3 Associations ...... 134 5.2.4 Grazing pressure ...... 136 5.3 Results ...... 137 5.3.1 Spatial pattern ...... 137 5.3.2 Shrub species composition...... 141 5.3.3 Associations ...... 145 5.3.4 Grazing pressure ...... 147 5.4 Discussion...... 148 5.4.1 Spatial pattern ...... 148 5.4.2 Shrub species composition...... 148 5.4.3 Bursaria in relation to other shrubs...... 149
iv 5.4.4 Tree density and recruitment ...... 150 5.4.5 Grazing pressure...... 151 5.5 Conclusion ...... 153
Chapter 6 Post-fire regeneration of Bursaria spinosa ...... 155 6.1 Introduction...... 155 6.1.1 Bursaria spinosa and fire ...... 155 6.1.2 Study aims and limitations ...... 156 6.2 Methods...... 157 6.2.1 Study site and burn ...... 157 6.2.2 Tagging plants in plots ...... 157 6.2.3 Assessing post-fire recovery...... 159 6.2.4 Bursaria excavation ...... 160 6.2.5 Data analysis...... 160 6.3 Results...... 161 6.3.1 Resprouts ...... 161 6.3.2 Seedlings...... 163 6.3.3 Flowering...... 164 6.3.4 Overall plant numbers ...... 164 6.3.5 Growth...... 165 6.3.6 Rainfall ...... 166 6.3.7 Excavation ...... 167 6.4 Discussion ...... 167 6.4.1 Status of findings...... 167 6.4.2 Does fire kill Bursaria?...... 168 6.4.3 How does Bursaria reproduce?...... 169 6.4.4 When does Bursaria reproduce?...... 169 6.4.5 When does Bursaria grow?...... 171 6.4.6 Why is Bursaria more abundant in sites where fire is relatively infrequent: towards a model ...... 171 6.5 Conclusion ...... 173
Chapter 7 Fire and herb reproduction ...... 174 7.1 Introduction...... 174 7.2 Methods...... 175 7.2.1 Study sites and fires...... 175 7.2.2 Sampling and data collection...... 176 7.2.3 Univariate data analysis...... 177 7.2.4 Multivariate data analysis...... 178 7.3 Results...... 179 7.3.1 Nurragingy forbs...... 179 7.3.2 Nurragingy grasses and graminoids ...... 182 7.3.3 Bare ground and seedlings at Nurragingy ...... 184 7.3.4 Composition of species reproducing at Nurragingy...... 186 7.3.5 Glenmore Park forbs...... 187 7.3.6 Glenmore Park grasses and graminoids...... 190 7.3.7 Bare ground and seedlings at Glenmore Park ...... 192 7.3.8 Composition of species reproducing at Glenmore Park ...... 194
v 7.4 Discussion...... 195 7.4.1 Status of findings ...... 195 7.4.2 Flowering and fruiting ...... 196 7.4.3 Seedling emergence ...... 199 7.4.4 Towards an understanding of ground layer recruitment ...... 201 7.5 Conclusion...... 204
Chapter 8 Fire frequency and woodland microhabitats...... 205 8.1 Introduction ...... 205 8.1.1 Microhabitats and herbs...... 205 8.1.2 Fire frequency and herbs...... 207 8.1.3 Study aims and questions...... 208 8.2 Methods ...... 209 8.2.1 Study sites ...... 209 8.2.2 Sampling and data collection ...... 211 8.2.3 Data analysis ...... 216 8.3 Results ...... 218 8.3.1 Native and exotic species richness...... 218 8.3.2 Native species groupings ...... 221 8.3.3 Species composition...... 225 8.3.4 Taxa characterising microhabitats ...... 231 8.3.5 Correlations with Themeda and grazing ...... 235 8.4 Discussion...... 236 8.4.1 Herbs and microhabitat ...... 236 8.4.2 Herbs and fire frequency...... 240 8.4.3 Diversity across sites...... 243 8.5 Conclusion...... 245
Chapter 9 Fuel accumulation...... 246 9.1 Introduction ...... 246 9.1.1 Vegetation as fuel...... 246 9.1.2 Modelling fuel accumulation ...... 247 9.1.3 Variation between vegetation types ...... 248 9.1.4 Fuel components and patchiness...... 249 9.1.5 Study aims...... 250 9.2 Methods ...... 251 9.2.1 Study sites and times-since-fire ...... 251 9.2.2 Sampling and data collection ...... 252 9.2.3 Fuel load determination ...... 253 9.2.4 Curve fitting ...... 253 9.2.5 Contribution of components...... 254 9.2.6 Time-since-fire comparisons...... 254 9.2.7 Fire frequency comparisons...... 255 9.2.8 Effects of overstorey cover ...... 256 9.2.9 Statistical tests for comparisons...... 256 9.3 Results ...... 257 9.3.1 Fuel accumulation curves...... 257 9.3.2 Contribution of components...... 260
vi 9.3.3 Time-since-fire comparisons ...... 261 9.3.4 Fire frequency comparisons...... 263 9.3.5 Effects of overstorey cover...... 264 9.4 Discussion ...... 265 9.4.1 Fuel accumulation in CPW...... 265 9.4.2 CPW and sandstone woodlands...... 266 9.4.3 Fuel components in CPW...... 267 9.4.4 Effects of overstorey cover...... 270 9.4.5 Litter depth and bare ground ...... 270 9.5 Conclusion ...... 271
Chapter 10 Synthesis and implications...... 272 10.1 Introduction...... 272 10.2 Key findings in Cumberland Plain Woodland ...... 274 10.2.1 CPW shrubs and fire...... 274 10.2.2 CPW trees and fire...... 276 10.2.3 CPW herbs and fire...... 277 10.2.4 CPW exotics and fire...... 278 10.2.5 CPW vegetation as fuel ...... 279 10.2.6 Ecology informing management ...... 279 10.3 A state and transition model for CPW ...... 280 10.3.1 Description of states ...... 281 10.3.2 Maintaining states...... 281 10.3.3 Transitions ...... 286 10.4 Conservation values and fire management in CPW...... 288 10.4.1 Conservation value of states...... 288 10.4.2 Value of Bursaria spinosa...... 289 10.4.3 Value of open grassy woodland...... 289 10.4.4 Management of grassy woodland with shrubs ...... 289 10.4.5 Management of Bursaria-dominated woodland ...... 290 10.4.6 Management of grassy Themeda woodland ...... 290 10.4.7 Importance of active management...... 291 10.5 Managing weeds ...... 291 10.6 Fire interval domains for Castlereagh woodlands...... 292 10.6.1 Lower thresholds ...... 293 10.6.2 Upper thresholds...... 293 10.7 Managing fire on the urban fringe ...... 295 10.7.1 Intervals for bushland conservation...... 295 10.7.2 Intervals for property protection...... 296 10.7.3 Managing for multiple aims ...... 297 10.8 Implications for setting fire frequency thresholds ...... 299 10.8.1 Fire management goals...... 299 10.8.2 Setting lower thresholds ...... 300 10.8.3 Setting upper thresholds ...... 300 10.9 Relevance of CPW recommendations for grassy woodlands elsewhere .....303 10.10 Conclusion ...... 305
References...... 307
vii Appendices ...... 333
App. 1 Climate data for Cumberland Plain and comparison weather stations.... 334 App. 2 Rainfall during the project...... 336 App. 3 Vital attributes, plant functional types and sensitivity to disturbance regimes...... 337 App. 4 Regeneration modes of Cumberland Plain shrub species...... 342 App. 5 Juvenile periods of Cumberland Plain shrub species ...... 354 App. 6 Comparison of Cumberland Plain data and data in NSW Flora Fire Response Database...... 370 App. 7 Fire-related attributes of two Cumberland Plain vegetation types ...... 375 App. 8 Species in landscape study sites ...... 383 App. 9 Summary of landscape study findings...... 386 App. 10 Grazing-intolerant species ...... 388 App. 11 Dry weight, litter depth, bare ground and understorey cover, by data point ...... 389 App. 12 Fuel components by data point...... 390
viii ACKNOWLEDGEMENTS
A successful industry-linked Australian Post-graduate Award (APAI) application, submitted in 2000 by Charles Morris from the University of Western Sydney (UWS) and Grahame Douglas from the New South Wales Rural Fire Service (RFS), outlined the broad direction of the project. Dr Morris subsequently became my principal supervisor, with co-supervision provided by Mr Douglas and Ross Bradstock from the NSW Department of Environment and Conservation (DEC). I would like to heartily thank all three. Charles’ knowledge of experimental design, ecological principles and scientific writing informed my work throughout. He must also hold a record for availability; his open door contributed substantially to the ease and enjoyment of the project. Ross’ expertise in fire ecology translated into invaluable discussions as the project was planned, as results began to come in and as conclusions were drawn. Charles and Ross both read and commented on all thesis chapters. Grahame introduced me to the Cumberland Plain, discussed findings with enthusiasm and put me in touch with local land and fire managers. As industry partners the RFS made this project possible, and their support is gratefully acknowledged.
As the project spanned three years and many sites, inevitably it also encompassed many people, particularly land managers and Rural Fire Service personnel. Apart from their practical contributions, many of those mentioned below were the source of fruitful discussions. I thank them all for their part in the learning and enjoyment that has characterised the project for me.
Various folks helped me locate remnants and identify fire history. Some are acknowledged in the text (see in particular Table 4.1), and in the paragraphs below. Others included John Bennett, John Foster, Joe Pearce, Linda Pearson, John Rota and Geoff Young. Particular thanks to those who introduced me to Western Sydney’s bushland in person, including Mark Anderson and others from Blacktown Council, Wayne Butcher, Stephanie Clark, Samantha Clarke, Rob Corby, John Diamond, Clinton Jessop-Smith, Peter Lister, Beth Michie, Peter Mobbs, John Pearson, Craig Sampson, Kathy Ware, David Warren and Malcolm Wells.
Study site managers were uniformly supportive and enthusiastic. Particular thanks to National Parks and Wildlife’s Jonathan Sanders, Tanya Leary, Col Davidson, Paul Hardey, Sarah Hill and Des Ayallew, to Marina Peterson and her colleagues at the Department of Defence, to the Blacktown Council bushcare team including Glenn White and Claire Nuttgens, to Ian Lovelock and his staff at Air Services Australia, to Jane MacCormick and her colleagues from the Sydney Catchment Authority, to Diana Picone at Bankstown Council and Marlene Spinks at Penrith. These people welcomed me to their remnants, helped organise access and talked with me about their bush. Staff at the Royal Botanic Gardens deserve particular mention. Lotte von Richter and Debra Little from the Mt Annan office, and Doug Benson and Jocelyn Howell from the Herbarium, provided ecological insights, botanical advice, and practical assistance in relation to all three studies which utilised the Mt Annan woodlands. The concept and methods for the Bursaria recovery study reported in Chapter 6 were developed through discussions with these staff members. Lotte monitored plots on an informal basis throughout that study, but particularly at the beginning. Lotte and Deb assisted with three-monthly assessments. Doug, and particularly Lotte and Deb, worked with me in
ix the field. Field work at Mt Annan was a joy and I thank them all, as well as Natural Area Manager Peter Cuneo.
The contribution of Jilske de Bruin to the fuel load study reported in Chapter 8 is also gratefully acknowledged. Jilske completed an exchange to UWS from Wageningen University in the Netherlands, during first semester 2004. By this time the fuel load study had been underway for some time; some data had been collected (see Table 9.1) however more field sampling was needed. Initially with my guidance and assistance but subsequently without, Jilske collected additional samples and carried out the associated laboratory work. She then conducted her own analysis of the data, and produced a report (de Bruin 2004). I later reanalysed the data, taking a somewhat different approach. The results, words and ideas in Chapter 8 are therefore my own. Jilske’s work, however, inevitably informed my thinking. Without her participation, the fuel load aspect of the project could not have been adequately addressed, and I am very thankful for the intelligence and hard work she brought to this study.
The contribution of Ray Richie to the fuel load study is also acknowledged, with thanks. Ray customised an Excel spreadsheet to fit exponential curves to the fuel load data.
As well as Jilske and the Mt Annan staff mentioned above, many others assisted in the field over the course of the project, often on a voluntary basis. Without them the project would have been impossible, and I cannot thank them enough for their time, effort and excellent company. Assistants on the UWS payroll included Venesa Brusic and Janet Long. These two long-suffering women spent many hours crawling through Bursaria thickets. They were stalwarts, became friends, and have my gratitude for life. Others who kindly helped in the field included Burhan Amiji, Angela Baker, Malcolm Boncales, Ross Bradstock, Jenny Bui, David Crossley, Alex Debono, Kelly Lachman, Tanya Leary, John Llewelyn, David Loose, Beth Michie, Claire Nuttgens and others at Bankstown Council, Waminda Parker, Diana Picone, Marlene Spinks, Annie Storey, Cate Storey, Cuong Tran, Ilkka Vanha-Majamaa, Emma Walter, Bill Watson, Edna Watson, Adam White and Peter Wood. Judy Christie from Greening Australia helped locate volunteers.
While most of the photos illustrating this document are my own, some owe their existence to Venesa Brusic, Annie Storey or Lotte von Richter, and are used with their permission. Mark Tozer kindly drafted the maps, using DEC data on GIS.
The University of Western Sydney has supported my candidature through its students and staff. Many thanks to my fellow PhD students, particularly those in the Ecology Research Group including Paul Thomas, Peter Nichols, Monique de Barse and Jennifer Fitzgerald. Your ideas, company and practical support have been a boon. School, College and Research Office staff helped me through the administrative maze: Grace Brown, Julie Elven and Vince Wyatt are three amongst many to whom I am indebted. The consummately organised Burhan Amiji helped equip me for field and lab. Academic staff who welcomed me to the intellectual life of the University included Shelley Burgin, Peter Cornish and Pauline Ross.
People outside the University also helped me develop ideas and skills used in the project. Ian Lunt from Charles Sturt University kindly commented on the project proposal. David Keith and Lisa Holman from DEC helped get me started on multi-
x variate analysis using PRIMER. Liz Tasker, also from DEC, discussed her work on fire in forests. David Robertson lent me his thesis on grassy woodland in Victoria. Harry Recher exchanged e-mails about his research on birds on the Cumberland Plain. Kevin Wale from the Threatened Species Unit in DEC helped organise the appropriate permissions to enable the research to proceed. Andrew Stanton discussed Aboriginal history and management. Others too numerous to name provided encouragement and a sounding board.
Finally, the support of my family and friends is acknowledged and warmly appreciated. My partner David has been through it all himself, helped steady me throughout the process and provided many useful insights. Annie and Cate have encouraged their mother’s unusual career path, including this latest escapade, with unstinting generosity and love. Thanks too to Nicky, Imogen and my parents; to my Brisbane friends who I deserted to take up the scholarship; and to my Sydney friends old and new, particular Karen Thumm whose successfully completed PhD has been an inspiration.
Cumberland Plain Woodland at Holsworthy
xi ABBREVIATIONS
CP Cumberland Plain (see Chapter 2)
CPW Cumberland Plain Woodland (see Section 2.4.1)
DBH diameter at breast height
DEC New South Wales Department of Environment and Conservation
FRG Forest Red Gum Reserve, an unburnt site sampled for the herb reproduction study reported in Chapter 7
NPWS National Parks and Wildlife Service of New South Wales
NS not significant
NSW New South Wales
PFT plant functional type (see Appendix 3)
RBG Royal Botanic Gardens, Sydney
RFS New South Wales Rural Fire Service
S.E. standard error
UWS University of Western Sydney
xii CHAPTER 1 INTRODUCTION TO THE PROJECT
1.1 Project goal and focus
Fire is a widespread and recurring disturbance which has shaped the vegetation of Australia, Africa and to some extent that of other vegetated continents (Bond 1997, Bowman 2003, Bond et al. 2005). Plants in communities subject to periodic burning exhibit traits that enable them to regenerate after fire (Gill 1981); in fact, the life histories of many Australian plant species are cued to fire (Bradstock et al. 2002). This does not mean, however, that species in fire-prone communities will persist under any fire regime. Frequency of burning can strongly influence community composition (Cary and Morrison 1995, Watson and Wardell-Johnson 2004), while fire intensity (Williams et al. 1999, Morrison and Renwick 2000) and season (Clark 1988, Tolhurst 1996b) also affect plant species. Both frequent and infrequent burning can have negative consequences for plant diversity.
The importance of fire management for biodiversity conservation is increasingly being recognised, both in Australia and around the world. Ecological knowledge which can potentially inform fire management is also growing. Although the impacts of fire on plants and animals are complex, broad principles can be extrapolated from the existing research, and practical guidelines for fire and land managers are starting to appear (Reid et al. 1996, Watson 2001a, NPWS 2004c).
Appropriate fire frequencies for plant species conservation vary widely between ecosystems (Bond 1997, Watson 2001a, NPWS 2004c). However ecological research which provides insight into the role of fire is not equally available for all vegetation types, and research has not necessarily focussed on areas where the need is greatest, even in Australia’s most populated areas.
Amongst Sydney’s sprawling western suburbs are remnants of a mosaic of vegetation types, most of which are now listed as endangered under the New South Wales Threatened Species Conservation (TSC) Act 1995 (Tozer 2003). Relative to the
1 surrounding sandstone vegetation, all have received scant attention from researchers with an interest in fire. This research project was designed to redress that imbalance.
The overall project goal was to develop a greater understanding of the relationship between fire regimes and plant diversity in the woodlands of Western Sydney’s Cumberland Plain, and to draw out implications for conservation management. The project focused primarily on the role of fire frequency in the grassy woodlands found on Western Sydney’s shale-based plains and low-relief hillsides. The project comprised six studies. Knowledge of ecological processes gained through these studies was used to develop a model of vegetation dynamics which can be used to inform decision- making within an adaptive management framework.
This chapter reviews the literature on a number of topics germane to the project, summarizing what is known in general, what is known about Cumberland Plain vegetation, and where knowledge gaps exist. Broad study questions are introduced in the course of the review. These questions are elaborated in the introductions to subsequent chapters, where specific hypotheses will be articulated.
Topics addressed in the literature review, and in the project itself, are briefly outlined below, together with a preview of the sorts of issues with which this thesis will be concerned:
• Temperate woodlands (Section 1.2, all chapters but particularly Chapters 2, 3 and 10). What are they? How do they differ? What might these differences mean for fire regimes? • Species responses to burning (Section 1.3, Chapter 3). How do individual plant species respond to fire? What can an understanding of the fire-related characteristics of plant species tell us about the impacts of fire interval domains? • Fire frequency and woodland shrubs (Section 1.4, Chapters 4, 5 and 6). What do we find when we look at shrub distributions in relation to fire frequency in the field? Are these findings congruent with what might be expected from an understanding of species characteristics? • ‘Encroachment’ in grassy vegetation (Section 1.5, Chapters 4, 5 and 6). Increases in woody plant density in grassy vegetation have been reported from all over the world. On the Cumberland Plain, some woodland areas are open and grassy, while elsewhere single-species shrub thickets cover the landscape. Might frequency of burning explain these differences? • Fire and woodland trees (Section 1.6, Chapters 4 and 5). Woodland eucalypts form a pool of suppressed lignotuberous seedlings, some of which will eventually grow through the sapling stage to join the population of adult trees.
2 Does frequent fire influence recruitment into the canopy by disrupting this process? • Fire, grasses and herbs (Section 1.7, Chapters 7 and 8). Fire mediates the dynamics of grasses and forbs in temperate grasslands; does it do likewise in temperate grassy woodlands? A woodland ground layer may be influenced by other structural elements; trees and shrubs may create a range of microhabitats. Does this have implications for the effects of fire on woodland grasses and herbs? • Management guidelines (Section 1.8, Chapter 10). Vegetation management guidelines may, or may not, address the issue of fire frequency in woodlands. Is there a consensus as to what constitutes best practice? • Fire management at the urban fringe (Section 1.9, Chapters 9 and 10). Those responsible for urban bushland remnants need to manage to minimise the risk posed by fire to life and property, as well as for biodiversity conservation. To what extent are these goals compatible?
The penultimate segment of this chapter sets the scene for the data chapters by outlining the methodological approach adopted. Finally, a brief description of remaining thesis chapters is provided.
1.2 Temperate woodlands 1.2.1 Description, location, conservation
The term ‘woodland’ is used in Australia to describe vegetation which contains trees whose projective foliage cover is under 30% (Specht 1970, Yates and Hobbs 2000). Woodlands have three structural layers: an overstorey of trees, a mid storey of shrubs, and a ground layer of grasses interspersed with forbs. Some temperate woodlands have a predominantly shrubby understorey, where others are predominantly grassy (Clarke 2000, Yates and Hobbs 2000). Grassy woodlands are structurally similar to savannas (Gillison 1994), which have been defined as “tropical or near-tropical seasonal ecosystems with a continuous herbaceous layer, usually dominated by grasses or sedges, and a discontinuous layer of trees and/or shrubs” (Skarpe 1992).
Eucalypt woodlands were once widespread in Eastern Australia (Sivertsen and Clarke 2000, Yates and Hobbs 2000). In New South Wales (NSW) they occurred over much of the State west of the Great Dividing Range, and also in some sub-coastal regions (Sivertsen and Clarke 2000). Coastal Valley Grassy Woodlands (Keith 2004) were
3 found in rain-shadow areas with fine-textured soils and low-relief topography, including the Cumberland Plain in the Sydney Basin, and the valleys of the Hunter, Clarence and Araluen Rivers. Woodlands across New South Wales formed a mosaic with areas of treeless grassland, and, in wetter areas, with forests (Yates and Hobbs 2000).
Relative to other vegetation types, Australia’s temperate woodlands are not well conserved. From the early days of European settlement, they have been subject to intensive management. Many areas were cleared for crops and towns. Woodlands grazed by domestic animals were often ‘improved’ through addition of exotic species. Few intact remnants remain, and those that do have inevitably been modified (Silvertsen 1993, Prober and Thiele 1995, Sivertsen and Clarke 2000, Yates and Hobbs 2000).
1.2.2 Abiotic influences
Woodlands are found across a wide range of climatic, soil and hydrological conditions. In Australia, temperate woodlands are most prevalent in areas with a rainfall between 200 and 800mm (Yates and Hobbs 2000).
Grassy woodlands tend to occur on relatively nutrient-rich, deep, clay or loam soils, while a shrub-dominated understorey is more common in nutrient-poor, shallow, sandy soils (Beadle 1962, Specht 1970, Ashton 1976, Prober 1996, P.J. Clarke 2003). Grass dominance in clay soils where water infiltration is relatively difficult may be a function of grass species’ competitive superiority in accessing the soil moisture in the upper soil layer (van Langevelde et al. 2003).
As rainfall decreases, tree height and woodland structure change. In low rainfall areas, trees are widely separated, low in stature and have a low bole to crown ratio. However “as soil moisture increases, the structure of the formation approaches that of a forest; the trees are up to 20-25 metres tall and tend to have the more compact crowns and the extended boles characteristic of forest form” (Specht 1970:55). Continuity of the grass sward also varies with rainfall. Grassy woodlands in high rainfall areas have a continuous ground layer, while in regions where rainfall is more erratic, the grass layer is discontinuous in space and time. This difference has a profound effect on the ability of the vegetation to support fire: in high rainfall areas, grassy woodlands may have sufficient fuel to burn every year, while semi-arid woodlands only have the wherewithal occasionally (Bond 1997, Gill et al. 2002).
4 In grassy woodlands, grass and herb species composition shifts from south to north across NSW and into Queensland, as temperatures increase and season of rainfall changes from winter to summer dominance (Bowman 2003a). Native species richness also increases along this gradient (Prober 1996, McIntyre and Martin 2001, P.J. Clarke 2003). The additional productivity generated by summer rainfall is reflected in landholder attitudes and practices: fire is commonly used in subcoastal northern woodlands to encourage grass production, but is less frequently employed in the south (McIntyre and Martin 2001, Noble and Grice 2002), and on the cooler New England Tablelands (P.J. Clarke 2003, but see Kitchin 2001).
Fire is thought to have played a role in the spread of temperate woodlands in Australia as climatic conditions became drier in the latter part of the Pleistocene (Hobbs 2002). Aboriginal burning also influenced woodlands (Kohen and Downing 1992, Bowman 1998, Gott 1999). However “few studies have been conducted on the fire-related dynamics of temperate woodland communities in Australia” (Hobbs 2002: 311). While this project focussed on the woodlands of the Cumberland Plain, the relevance of its findings to temperate grassy woodlands elsewhere will also be considered.
1.2.3 Introducing Western Sydney’s woodlands
Western Sydney hosts both grassy and shrubby woodlands. Unlike the sandstone of the surrounding plateaux, which typically weathers to low nutrient, sandy soils, Western Sydney’s shale produces moderately fertile clays. Sydney’s shale-based woodlands, which tend to be grassy, are collectively known as Cumberland Plain Woodland, or CPW. The second major vegetation grouping of the Western Sydney basin, Castlereagh Woodland, has a more shrubby understorey. Castlereagh Woodland is found on unconsolidated tertiary sediments (Tozer 2003). While this project focussed primarily on CPW, some work on the fire-related characteristics of Castlereagh Woodland shrubs was also carried out.
Western Sydney’s woodlands sit at the high end of the woodland spectrum in terms of rainfall, which averages approximately 800 – 900 mm per annum, and falls predominately in summer. Trees in CPW tend to be tall and straight, with high bole to crown ratios. The grass layer is often continuous. While most of the plant diversity in
5 CPW is found in the ground layer, this vegetation type also contains a number of shrub species. One shrub in particular, Bursaria spinosa1, is very abundant.
A description of the study area follows in Chapter 2.
1.3 Plant species responses to burning
Fire ecology research in Australian temperate grassy woodlands is in its infancy; in fact ecological research of any kind in this vegetation type is sparse (Yates and Hobbs 2000, Hobbs 2002, Prober and Thiele 2004). However the role of fire in two other Australian grassy ecosystems, Victoria’s low elevation grasslands and Northern Australia’s tropical savannas, has been the subject of considerable research. Some Australian temperate shrubby woodlands are also well studied. The discussion in this section and those that follow draws on these sources.
There has been considerable research, in a number of countries including Australia, into the responses of plant species to fire. We thus have quite a good picture of fire-related life history characteristics, and of the role of those characteristics in assisting – or failing to assist – species to persist within fire cycles. There has been little documentation, however, of the life history responses of plants on the Cumberland Plain.
Plants vary in the basic mechanism by which they regenerate after a fire, in their reproductive strategies, and in the timing of life-history stages.
1.3.1 Post-fire regeneration modes
Plant species in fire-prone vegetation types have two main ways of retaining their position in the community. Gill (1981) classified plants as ‘non-sprouters’ or ‘sprouters,’ on the basis of whether mature plants just subject to 100% leaf scorch die or survive fire. Most adults of sprouting species, also called ‘resprouters’ regrow from
1 Local authorities James et al. (1999) list this taxon as Bursaria spinosa var. spinosa. Tozer (2003), another local expert, omits the variety. Harden (1992:69-70) lists five varieties of B. spinosa occurring in NSW, three of them in the Central Coast region which includes Western Sydney. She notes that varieties are “linked in some cases by morphological intermediates, and developmental factors also complicate recognition.” I will use either the specific name Bursaria spinosa when referring to this taxon, or the common name ‘Bursaria’, as that is what this shrub is generally called by CPW managers.
6 shoots after a fire. These shoots may come from root suckers, rhizomes, or lignotubers at the base of the plant, or from epicormic buds on stems (Gill 1981). On the other hand, adults of non-sprouting species, or ‘obligate seeders,’ die when their leaves are all scorched in a fire, and rely on regeneration from seed. Obligate seeder species generally produce more seed (Cowling et al. 1990, Lamont et al. 1998), and greater numbers of seedlings (Wark et al. 1987, Benwell 1998, Enright and Goldblum 1999) than resprouters, and seedling growth rates tend to be more rapid (Bell and Pate 1996, Benwell 1998, Enright and Goldblum 1999, Bell 2001).
These categories are not invariant. Survival rates in the field for both resprouters and obligate seeders change with fire intensity (Morrison and Renwick 2000). Some species exhibit different regeneration strategies in different environments (Williams et al. 1994, Benwell 1998).
1.3.2 Reproduction and fire
Fire provides conditions conducive to seedling growth. Shrubs, grass clumps, litter and sometimes canopy cover are removed, allowing increased light penetration to ground level and reducing competition for water and nutrients (Williams and Gill 1995, Morgan 1998a). Propagules need to be available for species to take advantage of this opportunity.
Some species hold their seeds in on-plant storage organs such as cones, and release them after a fire. These ‘serotinous’ taxa include species in the Proteaceae and Cupressaceae families, for example Banksia, Hakea and Callitris. Some eucalypts also release seed in response to fire (Noble 1982, Gill 1997). The degree to which seed release also occurs in the absence of fire varies between species (Enright et al. 1998).
A second group of species stores dormant seeds in the soil; dormancy requirements ensure germination occurs after fire. Heat promotes germination in legumes (Shea et al. 1979, Auld and O’Connell 1991, Clarke et al. 2000), while smoke plays a role for many species (Dixon et al. 1995, Roche et al. 1998, Morris 2000, Flematti et al. 2004). Many Sydney shrub taxa respond best to a combination of these two fire-related cues (Morris 2000, Thomas et al. 2003).
7 A third strategy is to create seeds rapidly after a fire, though fire-cued flowering. Xanthorrhoea species are a well-known example of this phenomenon (Harrold 1979, McFarland 1990), however shrubs such as Lomatia silaifolia (Denham and Whelan 2000), Telopea speciossima (Bradstock 1995) and Stirlingia latifolia (Bowen and Pate 2004) also flower almost exclusively in the years after a fire. Many grassland forbs exhibit this characteristic (Lunt 1994).
Seed dispersal is a vital factor in maintaining metapopulations of species which become locally extinct after a fire. However dispersal distances in Australian fire-prone vegetation may be limited to tens of metres or less in most species (Auld 1986, Keith 1996, Hammill et al. 1998). Some wind and animal-dispersed species do, however, occur in these environments, and may have a different relationship to fire cycles than other taxa (French and Westoby 1996).
1.3.3 Life history stages
The time taken by a species in various life stages affects its ability to persist in a fire- prone environment. Time from germination to death of adult plants, time to reproductive maturity and, for resprouters, time to fire tolerance are important variables, as are duration of seed viability and ability to recruit at different times post-fire.
The time from seed germination to reproductively-mature adult is known as a species’ ‘primary juvenile period’. Resprouting species also have a ‘secondary juvenile period’: the time taken for vegetative regrowth to produce viable seed (Morrison et al. 1996). The length of these periods differs between species, and may even differ within a species, depending on location (Gill and Bradstock 1992, Knox and Clarke 2004). Once flowering has occurred, it may take additional years before viable seed is produced, and even longer to accumulate an adequate seedbank (Wark et al. 1987, Bradstock and O’Connell 1988).
In resprouters, the primary juvenile period is often much longer than the secondary juvenile period, as well as being longer than the primary juvenile period in equivalent obligate seeders (Keith 1996, Benwell 1998). Resprouter seedlings are not immediately fire tolerant: it may take many years before lignotuber development or starch reserves are sufficient to allow the young plant to survive a fire (Bradstock and Myerscough 1988, Bell and Pate 1996).
8 The length of time seed remains viable is another important variable, but one about which not a great deal is known. It is clear, however, that species vary greatly (Keith 1996). The seedbanks of serotinous species are likely to be depleted more quickly than those of species with soil-stored seed, although much variation exists even here (Gill and Bradstock 1995, Morrison et al. 1996). Some species, generally those with soil- stored seeds, retain viable ungerminated seed through a fire and its post-fire period, allowing the species to regenerate after two fires within the juvenile period: Bossiaea laidlawiana, from south-west Western Australia, is an example (Christensen and Kimber 1975).
Finally, timing is important, and varies between species, in relation to how long after fire seedling recruitment is possible. For many species in fire-prone environments, recruitment is confined to the immediate post-fire period (Auld 1987, Zammit and Westoby 1987, Cowling et al. 1990, Vaughton 1998, Keith et al. 2002a), although this may vary between populations (Whelan et al. 1998) and with post-fire age (Enright and Goldblum 1999).
1.3.4 Plant responses to fire on the Cumberland Plain
Very little research on post-fire regeneration modes has been undertaken specifically on the Cumberland Plain, although modes of a number of species which occur on the Plain have been documented elsewhere (DEC 2002). Benson and Howell (2002) reported resprouting in 63 of 66 species observed post-fire in one CPW remnant (Mt Annan), however neither the names of these species, nor the methods by which their resprouting status was determined, were mentioned.
Fire-related germination cues, on the other hand, have been comparatively well studied in CPW, though not in Castlereagh Woodland (there does not appear to have been any fire ecology research in the Castlereagh vegetation types). Responses appear to be mixed. S. Clarke (2003) studied the effects of heat and smoke on seeds of 22 CPW grass species sourced from plants at two sites (Holsworthy and Mt Annan). Sixteen species (73%) responded to either heat or smoke, or to the two cues in combination, although in some cases the response was negative. Germination of the endangered shrub Pimelea spicata is enhanced by smoke (Willis et al. 2003). However at community level, two greenhouse studies of the effects of germination cues on
9 seedling emergence from soil seedbanks failed to find clear differences between heated and control treatments in abundance of germinants (Wood 2001, Hill and French 2003), although emergence was significantly higher after smoking (Wood 2001). Both studies, however, identified a number of individual species which responded positively to heat, including several legumes and the forb Dichondra repens. Many species in the standing vegetation did not emerge from the seedbank, and seedbank species richness was well below that above ground (Hill and French 2003).
I am not aware of any studies of juvenile period or longevity on the Cumberland Plain, although again these characteristics have been recorded elsewhere for a number of species (DEC 2002). However the importance of post-fire recruitment for many Cumberland Plain shrubs has been demonstrated at one CPW site (Holsworthy; Hill 2000, Hill and French 2004), where a study designed to assess the effects of fire and grazing by macropods has been underway for several years. Unplanned fires upset this study’s design, but allowed comparison between the effects of planned and unplanned fire. Species richness and abundance of shrubs was significantly greater in plots burnt 18 months earlier, whether by design or accident, than in unburnt plots. Seedling numbers were highest in plots burnt by summer wildfire. Later work confirmed an overall increase in species richness after fire.
1.3.5 Study question concerning plant responses to fire
Plant species characteristics can be used to define disturbance frequency domains compatible with the maintenance of particular suites of species (Noble and Slatyer 1980, Noble and Gitay 1996, Bradstock and Kenny 2003). This approach may prove particularly fruitful when applied to minimum interfire intervals, which need to exceed the primary juvenile period of slow-growing obligate seeders (Bradstock and Kenny 2003). The first question addressed by the project, therefore, was:
What fire-related attributes do shrub species in Cumberland Plain and Castlereagh woodlands possess, and what does this imply for fire interval domains?
This question is addressed in Chapter 3. This study represents one of a small number of attempts that have so far been made to define fire frequencies through analysis of the vital attributes of species in a local vegetation type; others include Bradstock and Kenny
10 (2003), and Kitchin (2001). As such it provided an opportunity to reflect on the process involved, and to make recommendations for its development.
1.4 Fire frequency and woodland shrubs
While there has been considerable research into the effects of fire frequency on shrubs in shrubby woodlands, grassy woodland shrubs are not well studied.
1.4.1 Effects of frequent fire
Field studies in Sydney sandstone heaths and woodlands have identified several shrub species, including the Proteaceous dominants Banksia ericifolia and Hakea teretifolia, which are eliminated or reduced in abundance on frequently burnt sites (Siddiqi et al. 1976, Nieuwenhuis 1987, Cary and Morrison 1995, Morrison et al. 1995a, Bradstock et al. 1997). These species tend to be heavy-seeded, serotinous obligate seeders. Their vulnerability relates to their relatively long juvenile periods – they take up to eight years to flower (Benson 1985) – lack of soil-stored seeds which could potentially survive through two fires, and short dispersal distances (Hammill et al. 1998). Serotinous Proteaceae species are, however, uncommon in temperate grassy woodlands (James et al. 1999, Clarke and Knox 2002). Short interfire intervals in shrubby woodlands on granite in the New England region also disadvantage some species, including a number of resprouters (Watson and Wardell-Johnson 2004). Demographic studies show some Sydney sandstone resprouters are also likely to decline under repeated short interfire intervals, as fire tolerance can take many years to develop (Bradstock and Myerscough 1988, Bradstock 1990). Repeated fires can weaken resprouting ability (Noble 1982).
1.4.2 Effects of infrequent fire
At the other end of the spectrum, field research has identified shrub species which are disadvantaged if fire is too infrequent (Bond 1980, Fox and Fox 1986, Morrison et al. 1996, Watson and Wardell-Johnson 2004). Some small shrubs and other low-growing plants get shaded out as bigger shrubs expand (Specht and Specht 1989, Vlok and Yeaton 2000), and the resulting disadvantage may be reflected in reduced recruitment when a fire does occur (Bond 1980, Cowling and Gxaba 1990, Keith and Bradstock
11 1994, McMahon et al. 1996, Tozer and Bradstock 2002). Bond and Ladd (2001), who studied this issue in a range of West Australian environments, found that resprouting dominants had a particularly profound influence on understorey species. Fox and Fox (1986) speculate that fire may be necessary to prevent senescence in a number of resprouters which they found reduced in abundance after a 12 year interfire interval. Fire-cued obligate seeder species may be at risk under low fire recurrence, as these plants often form even-aged stands after a fire (Auld 1987), and may die some years later as a group. These species are then dependent on fire occurring before soil-stored seed loses viability.
Shrub seedling establishment in the absence of fire is very infrequent in grassy woodlands on the New England tableland. Germination rates are low, as are survival rates of germinants. In this environment, shrub recruitment appears to be episodic and disturbance-driven (Clarke 2002). Recruitment is also heavily dependent on disturbance in shrubby woodland communities, with a large majority of species recruiting only, or mostly, after fire (Keith et al. 2002a).
1.4.3 Shrubs and fire frequency on the Cumberland Plain
Only one study on the Cumberland Plain has assessed the effects of fire frequency directly, and then only in a single location (Prospect Reservoir). Thomas (1994) compared an area that had been burnt on an annual or biannual basis for over twenty years, with an area which she believed had been unburnt during that time – although aerial photos from 1982 and 1986 suggest much of the infrequently burnt area may have been exposed to some fire. This study found few significant differences between the frequently and infrequently burnt areas, at community, group or individual species level. The trend for two shrub species, Bursaria spinosa and Pultenaea microphylla, was towards greater density where burning had been less frequent, however these trends were not significant. Thomas (1994:7-1) concluded that "probably the most important finding ... was that, despite a long-term history of hazard-reduction burning, very few fire effects could be discerned."
12 1.4.4 Study question concerning shrubs and fire frequency
This project extends the work of Thomas (1994) through a landscape scale study comparing vegetation characteristics in woodland sites subjected to low, medium and high fire frequencies over the last 20 years. The core question with respect to shrubs was:
Does shrub species richness or composition differ with fire frequency?
The influence of regeneration mode was examined as part of this study, which is reported in Chapters 4 and 5. Although not specifically designed for the purpose, the data provided a test of predictions flowing from the application of the vital attributes model – the first time, to my knowledge, that such a validation has been attempted.
1.5 ‘Encroachment’ in grassy vegetation 1.5.1 Encroachment across the planet
Over the last century, large and rapid increases in shrub and/or tree density in grassy vegetation have been reported from various parts of the world, including Africa (van Vegten 1983, Bond 1997), North America (Madany and West 1983, Hobbs and Mooney 1986, Archer et al. 1988), South America (San Jose and Farinas 1991), and Australia (eg Fensham and Fairfax 1996, Noble 1997, Lunt 1998a,b, Griffiths 2002). This thickening is generally considered problematic: reduction in forage for grazing animals has been a primary concern, however loss of open character and of biodiversity are also issues, as woody encroachment often favours a small number of plant species at the expense of others (eg Withers and Ashton 1977, Hobbs and Mooney 1986, Costello et al. 2000, Roques et al. 2001). Habitat for birds and other fauna can progressively disappear (Crowley and Garnett 1998, Russell-Smith 2002).
While the cause of this phenomenon is often not readily apparent, a reduction in fire frequency has been implicated in at least some cases (eg Fensham and Fairfax 1996, Lunt 1998a,b, Rocques et al. 2001). Experimental studies of fire frequency have confirmed that frequent burning favours grasses over shrubs, and vice versa, in a number of temperate woodland and forest communities (Bradfield 1981, Westfall et al. 1983, Birk and Bridges 1989, Lamb et al. 1992, House 1995).
13 1.5.2 Encroachment on the Cumberland Plain
The issue of shrub encroachment has not been formally studied on the Cumberland Plain. However observers have noted increases in cover of the dominant shrub, Bursaria spinosa (Benson and Howell 2002, Robinson 2003, H. Recher pers. comm. 2004). As mentioned above, Thomas (1994) detected a trend toward greater Bursaria density with reduced fire frequency at her study site, however the link with fire was not definitively established.
1.5.3 Study questions concerning encroachment
This study seeks to ascertain whether Bursaria abundance is linked to fire frequency. Its vital attribute status is discussed in Chapter 3, while in Chapter 4 the question
Does Bursaria spinosa frequency, density and/or dominance increase with decreasing fire frequency? is explored through survey work in sites with differing fire histories. The effects of fire on Bursaria are further investigated in Chapter 6, through a longitudinal study of resprouting and recruitment after a single fire, the question being:
How does Bursaria spinosa regenerate after a fire?
1.6 Fire and woodland trees
Resprouting eucalypt species dominate Eastern Australia’s woodlands. Adults generally regenerate after a fire from epicormic buds, although basal resprouting may also occur.
1.6.1 Eucalypt recruitment
Surprisingly little is known about the role of fire in woodland eucalypt recruitment. Eucalypt seeds are stored in the canopy; the extent to which live seed can accumulate in the soil appears to be negligible (Ashton 1979, Vlahos and Bell 1986, Read et al. 2000, Hill and French 2003). Recruitment in mallee eucalypts in north-western Victoria is limited to the period immediately after a fire (Wellington and Noble 1985). Fire may enhance recruitment opportunities by reducing the competition that seedlings would
14 otherwise experience from grasses or herbs (Noble 1980, Curtis 1990, Semple and Koen 2003), by killing some adult trees and thus creating gaps (Wellington and Noble 1985), by enhancing seedbed conditions (Clarke and Davison 2001), or by triggering sufficient seed release to cause ‘predator satiation’ of ants (Ashton 1979, Andersen 1988). As trees are long-lived organisms, low levels of recruitment may be sufficient to maintain populations. Eucalypt recruitment may be episodic, depending on the coincidence of seed availability, gap-creating disturbance, and rainfall (Wellington and Noble 1985, Curtis 1990, Clarke 2000).
While many eucalypt seedlings die within a year or two of establishment (Henry and Florence 1966, Wellington and Noble 1985, Clarke 2002), those that survive rapidly develop lignotubers which help them survive not only fire, but other disturbances such as drought and grazing (Curtis 1990, Clarke 2002). Suppressed lignotuberous seedlings can persist in the understorey for many years (Noble 1984), even in the face of regular burning (Henry and Florence 1966). When conditions are right, individuals grow through the sapling stage and join the adult population (Florence 1996).
1.6.2 Tree survival
The effect of fire on tree populations has been studied in savanna woodlands, particularly in Northern Australia and Africa. A number of studies link fluctuations in tree density to fire frequency or intensity, although there is also evidence of considerable stability in eucalypt populations. For example, in the Kapalga fire experiment in the Northern Territory, intense annual fires caused a reduction in tree stems, as did a wildfire after six years of fire exclusion. Mild annual burns, however, did not affect stem survival, and at whole tree level there was little difference between treatments (Williams et al. 1999). Frequent fire may limit tree recruitment by killing small diameter stems (Williams et al. 1999) and returning saplings to the basal- sprouting lignotuber pool.
Very long-term fire exclusion may cause a decline in dominant woodland eucalypts (Withers and Ashton 1977, Lunt 1998b).
15 1.6.3 Fire and trees on the Cumberland Plain
Very little is known about trees and fire on the Cumberland Plain. Observation, and data from elsewhere (DEC 2002), confirm that the dominant CPW and Castlereagh Woodland eucalypts conform to the general pattern for woodland trees in being epicormic resprouters. Hill and French (2004) report enhanced eucalypt seedling establishment after a summer wildfire at Holsworthy.
1.6.4 Study question concerning fire and trees
The effects of fire frequency on tree density were investigated as part of the study reported in Chapter 4. Thus the broad question addressed was:
Does tree density or recruitment into the canopy vary with fire frequency?
1.7 Fire, grasses and herbs 1.7.1 Grassland research
While little research has addressed the effects of fire on the ground layer in temperate grassy woodlands, fire-related dynamics in Victoria’s lowland grasslands have been extensively researched. Many species in this ecosystem are the same as, or similar to, those found in East Coast temperate grassy woodlands. Victoria’s grasslands are dominated by Themeda australis (Kangaroo Grass)2, which is also an important component of CPW. Between the tussocks formed by this species grow forbs and subdominant grasses (Tremont and McIntyre 1994, Kirkpatrick et al. 1995). The potential for competitive exclusion of interstitial species in the absence of periodic removal of Themeda biomass is a key theme in grassland fire ecology literature (Stuwe 1994). Studies have found that:
• Very frequently burnt grasslands support more native forb species than grasslands which have been grazed or left without management, and not regularly burnt (Stuwe and Parsons 1977).
2 I have used the designation Themeda australis throughout this document, as this name is used both locally (James et al. 1999), and throughout NSW (Harden 1993). Others, including the Victorian researchers whose work is cited here, prefer Themeda triandra, a name which recognises the close affiliation between the African and Australian forms.
16 • All perennial species resprout after fire: there are no obligate seeders, although there are some annual species (Lunt 1990, Morgan 1996). • Grassland species almost all flower within the first year after a fire (Lunt 1990, Morgan 1996, 1999). Flowering effort for many forb species is concentrated in the first year after a fire, dropping considerably in year two (Lunt 1994). • Themeda australis grows rapidly after fire (Morgan 1996, Lunt 1997b). By three years post-fire, gaps between Themeda tussocks have mostly disappeared (Morgan 1998a). • Species richness is significantly reduced in patches where Themeda is dense (Lunt and Morgan 1999). • Forb species in productive grasslands need gaps for seedlings to survive and grow (Hitchmough et al. 1996, Morgan 1998a). • Grassland perennial forbs tend not to have a large permanent store of seed in the soil (Morgan 1998b). Many species germinate easily and rapidly, and are not inhibited by darkness (Willis and Groves 1991, Lunt 1995b, Morgan 1998c), characteristics which imply that seedbanks tend to be rapidly depleted by germination. • Productivity in Themeda starts to decline at around five years post-fire, with serious collapse occurring by 12 years. Once this has happened, recovery of Themeda when a fire does occur is limited (Morgan and Lunt 1999).
These findings have led to the conclusion that grassland species are not only able to cope with frequent fire, but that fire or other disturbance is essential if forbs and less competitive grasses are to persist (Lunt and Morgan 2002). Frequent fire ensures that two of the three conditions for seedling establishment – gaps in the grass canopy and seed availability – are fulfilled. The third requirement, adequate moisture, may not be met after every fire, but it is argued that with relatively frequent fire, seeds, gaps and rainfall will coincide often enough to maintain forb populations (Morgan 1998a).
1.7.2 Grassy woodland research
The relevance of these findings for grassy woodlands has been questioned, on the grounds that competitive exclusion by dominant grasses is less likely where their abundance is in turn limited by competition from trees and shrubs (Stuwe 1994, D. Keith pers. comm. 2002). However the small amount of research into the effects of fire frequency on woodland grasses and forbs suggests there may be common ground with grasslands. Stewart (1999), working in northern NSW, found that the species richness of above-ground vegetation in grassy forest plots burnt every three years (eight years after treatment ceased), was greater than that in plots unburnt for 30 years. Only six of
17 the 36 species ’missing’ from the above-ground vegetation in unburnt plots were found in these plots’ seedbanks, suggesting that many species would not reappear if frequent burning were reinstated. Lunt (1995a, 1997a) compared frequently burnt anthropogenic grasslands (trees had been cut down many years prior to the study), with infrequently burnt and grazed grassy forests in Gippsland. The two environments supported quite different floras, although presumably derived originally from the same species pool. While the infrequently burnt forests were more species rich than the frequently burnt grasslands, the grassland sites had double the richness of native geophytes (forbs with underground storage organs). Again, seeds of species found in frequently-burnt grasslands were not stored in appreciable quantities in seedbanks of long-unburnt forest, or vice versa (Lunt 1997c). In savanna vegetation near Darwin, annual burning favours grasses and herbs: both species richness and cover was significantly greater in annually burnt quadrats relative to quadrats where fire had been excluded (Fensham 1990, Woinarski et al. 2004).
Seeds of grasses and forbs in grassy woodlands, like their grassland counterparts, tend to have limited dormancy (Clarke et al. 2000), restricted response to fire-related cues (Grant and Macgregor 2001), and short-term soil storage (Odgers 1999).
1.7.3 Fire, grasses and herbs on the Cumberland Plain
Thomas’ study included assessment of the ground layer (Thomas 1994). While the leguminous herb Glycine tabacina was significantly more abundant on frequently burnt plots, no other differences between fire treatments were detected. Benson and Howell (2002) also reported little change in vegetation composition over 15 years of monitoring at Mt Annan despite fires in the study area, although their work did not specifically address fire effects.
On the other hand, the studies of fire-related germination cues reported in Section 1.3.4 provide useful information about CPW herbs. This work suggests that these species, like their counterparts in grassy vegetation elsewhere, often do not form persistent soil seedbanks (Hill and French 2003). Similarly, while some species respond to fire-related germination cues, many do not (Wood 2001, S. Clarke 2003, Hill and French 2003). For example Themeda australis responded negatively to heat in both the Wood (2001)
18 and Hill and French (2003) studies, although Wood (2001) found a positive response to smoke in this species.
1.7.4 Study questions concerning fire, grasses and herbs
It is important that effects of fire on ground layer species in CPW is considered, as it is here that most of the plant diversity lies.
The comparative study reported in Chapter 4 provided an opportunity to investigate the effects of fire frequency on grasses at a moderately broad scale. The research question was:
Does grass cover or dominance differ with burning frequency?
Findings on senescence of Themeda in infrequently burnt grasslands meant effects on this taxon were of particular interest. This issue has not previously been addressed in Australian temperate grassy woodlands.
The finding that many herbaceous species in CPW do not respond to fire-related germination cues raises questions about how, or even whether, these species respond to fire. Casual post-fire observation, and a grassland study (Lunt 1994), suggested that reproduction in some CPW herbs might be linked to enhanced post-fire flowering, rather than to enhanced post-fire germination. The second question with respect to herbs addressed by the project, therefore, was:
Is flowering and fruiting in CPW herbs greater in the post-fire period than some years after a fire?
This question is taken up in Chapter 7 through comparison between recently burnt and unburnt areas. Seedling numbers were also compared, in order to provide an indication of whether herb recruitment is concentrated in the initial post-fire period. So far as I am aware, this question has not previously been investigated in an Australian grassy woodland ecosystem.
However the primary approach to the effects of fire on herbs in this project relates to the possibility of shrub encroachment noted above. Trees and shrubs often provide a somewhat different microhabitat for ground layer species than do open areas (eg Robertson 1985, Pieper 1990, Prober et al. 2002a). If fire frequency affects the
19 abundance of Bursaria, and ground layer species richness or composition differs between shaded and open microhabitats, herbaceous species will be indirectly, but surely, influenced by fire frequency. The effect of microhabitat on CPW herbs was therefore investigated in a separate study, which is reported in Chapter 8. This study was conducted in sites with varying fire histories, thus also allowing assessment of direct fire frequency effects. Broad questions were:
Does ground layer species richness, abundance or composition differ with fire frequency?
Do different microhabitats support different herbaceous species?
This approach to assessing the effects of fire on grassy woodland herbs has not been taken before.
1.8 Guidelines for the use of fire in conservation management
Over recent years, fire ecologists in a number of Australian jurisdictions have synthesised research findings into practical guidelines for land and fire managers. In New South Wales, the National Parks and Wildlife Service (NPWS, now the Department of Environment and Conservation, or DEC) has developed guidelines for mimimum and maximum interfire intervals for broad vegetation types, on the basis of plant life history characteristics (NPWS 2004c). Public land management agencies in Victoria have taken a similar approach (Friend et al. 2003). In southeast Queensland the Fire and Biodiversity Consortium has produced guidelines for the use of fire for biodiversity conservation in broad vegetation types characteristic of that region (Watson 2001a); these guidelines were based on published and unpublished ecological research, and were developed in consultation with local ecologists and land managers (Watson 2001b). In all jurisdictions, limitations in the knowledge base and the probable need for adjustment over time are acknowledged. In particular, lack of quantitative data on plant and seed longevity mean upper thresholds developed using life history characteristics are “largely based on assumptions and generalisations” (NPWS 2004c:1). The need to particularise guidelines for specific localities is also widely recognized.
20
1.8.1 Recommended fire frequency thresholds
Although the south-east Queensland (SEQ) and NSW guidelines suggest similar thresholds for shrubby woodlands and forests, they vary considerably in their recommendations for grassy woodlands. For shrubby dry sclerophyll forests, the NSW thresholds are 7 and 30 years (NPWS 2004c), while the SEQ guidelines suggest 7 to 25 year interfire intervals (Watson 2001a). However the NSW thresholds for grassy sclerophyll woodlands are 5 and 40 years (NPWS 2004c), while the SEQ recommendation is for fires at 3 to 6 year intervals (Watson 2001a).
Interfire intervals for grassy woodlands recommended by individual researchers have also varied widely. Robertson (1985) recommended burning every three years to retain Themeda-dominated understorey in open areas of a grassy woodland in Victoria, with other areas to be burnt every 5 or 6 years to encourage regeneration of woody shrubs. Kitchin (2001), using localised life history data for shrubs found in grassy woodlands and open forests on the eastern edge of the New England Tablelands in northern NSW, recommended intervals between 4 and 30 years.
These large difference may have several explanations. First, they may reflect the extent to which matters other than species responses were considered when guidelines were set. The NSW guidelines do not consider the effects of fire on structure, whereas in southeast Queensland, the observation that unburnt grassy woodlands around Brisbane were ‘thickening’ to the detriment of once-common Pretty-face Wallabies (Kington 1997) was considered, as were other reports of shrub encroachment into grassy vegetation. Potential problems of competitive exclusion within the grass layer were also taken into account (Watson 2001a). Second, they may reflect genuine differences between vegetation types: the NSW guidelines cover a very wide range of grassy woodland ecosystems (see Section 1.2). Third, they may reflect the rudimentary nature of our understanding of the role of fire in mediating diversity in temperate grassy woodlands. As one of the first studies to directly investigate the effects of fire frequency in one of these ecosystems, this project has the potential to contribute considerably to our knowledge base.
21 1.8.2 Guidelines for management of grassy woodlands
Guidelines addressing the management of temperate grassy vegetation for conservation purposes have been produced for north-western NSW (Nadolny et al. 2003) and for Victoria’s lowland plains and foothills (Barlow 1998). In 2002, McIntyre et al. published a book titled “Managing and Conserving Grassy Woodlands”, with linked chapters of varied authorship. The use of fire is one concern amongst many in all these documents.
Barlow (1998) notes the importance of fire in grasslands, but is more equivocal in relation to woodlands, due to concerns that fire will limit tree and shrub regeneration. He suggests a frequency between 5 and 10 years.
Nadolny et al. (2003:16) sidestep the fire frequency issue by recommending that “burning should not occur more frequently than the recommended fire regime for the community being conserved,” but without indicating what those frequencies might be. These authors list possible positive and negative consequences of burning, and note the need to consider local conditions and management aims.
Fire management concerns are briefly addressed in McIntyre et al. (2002) by McIntyre in a chapter on trees, and by Martin and Green, whose interest is wildlife conservation. Both authors note that current fire regimes often differ from those prior to European settlement, and that wildlife adapted to pre-settlement regimes and resulting vegetation structure may be disadvantaged if fire is excluded. However fire frequency is not directly addressed.
Both McIntyre (2002) and Nadolny et al. (2003) note that fire may play a positive role in limiting the density of woody vegetation.
1.8.3 Guidelines for managing fire on the Cumberland Plain
‘Changed fire regimes’ are an often-cited issue in Cumberland Plain vegetation, however management recommendations tend to be absent, vague, or little more than homilies on the dangers of too frequent fire (Rawling 1994, NPWS 1997). Little (2003:27-8) discusses fire as a “trigger” for regeneration of some native plant species, reports on pile burning and cautiously suggests “broader area patch burns” may be
22 important in maintaining CPW. A fire-free interval of 5-10 years is recommended, however no upper threshold is mentioned.
Ecologists who have studied the effects of fire on the Cumberland Plain also tend to be less than definitive. Thomas (1994) proposed variable interfire intervals between 5 and 10 years to maintain plant diversity at Prospect, but was unwilling to generalise beyond her study site. Hill and French (2004:28) believed “fire would be a useful management tool to promote species regeneration in the Cumberland Plain Woodland community,” while S. Clarke (2003:52) concluded that “The maintenance of the Cumberland Plain Woodlands at Holsworthy will depend on variable fire and grazing regimes to ensure species diversity by providing conditions favourable to all species periodically.”
All in all, current guidelines for management of grassy vegetation, both in Western Sydney and elsewhere in temperate Australia, give little practical guidance in relation to fire frequency. This highlights the need for ecological research. Tozer (2003) designated research to identify fire regimes appropriate for conservation of native biota one of three top management problems requiring urgent attention on the Cumberland Plain. The question:
What guidance for managers can be drawn from ecological research? is addressed in Chapter 10.
1.9 Fire management at the urban fringe
As well as attempting to retain the conservation values of the remnants for which they are responsible, land and fire managers need to consider the risk posed to human life and property by fire. From this standpoint, vegetation is fuel, and as time-since-fire progresses, fuel accumulates. The acceptable hazard level in the sclerophyll vegetation of southern Australia is generally considered to be between eight and fifteen tonnes per hectare (Gill et al. 1987, Raison et al. 1983, Simmons and Adams 1986, Fensham 1992, Tolhurst 1996a). However the rate at which fuel accumulates, and maximum hazard levels achieved, vary with vegetation type. There are some indications that levels achieved in shrubby vegetation tend to outweigh those in grassy ecosystems (Simmons and Adams 1999, Williams et al. 2002).
23 In Sydney’s sandstone vegetation, a conflict exists between the fire frequencies required if fuel reduction is to be an effective strategy for reducing risk to life and property at the urban/bushland interface, and frequencies appropriate for biodiversity conservation (Morrison et al. 1996, Bradstock and Gill 2001). Fuel loads in sandstone woodlands rise to over 30 tonnes per hectare, far above the level considered ‘safe’. ‘Unsafe’ loads are achieved within two years of a fuel reduction burn (Morrison et al. 1996).
1.9.1 Fuel accumulation on the Cumberland Plain
I am not aware of any documentation of fuel loadings on the Cumberland Plain. This project is therefore the first to address the questions:
How does fuel accumulate over time in CPW?
Do fire regimes compatible with bushland conservation in CPW overlap those needed to achieve protection from wildfire?
Fuel load accumulation is addressed in Chapter 9, while the question of balancing regimes for protection of conservation and property is dealt with in the final chapter.
1.10 Methodological approach
Knowledge of ecological processes can be advanced by a variety of research methods, each with its strengths and weaknesses (Diamond 1986). Landscape-scale processes such as fire need to be addressed, at some point, through “appropriately scaled field studies” (Carpenter 1996) which provide context, relevance and an ability to address issues of temporal and spatial scale (Diamond 1986, Carpenter 1996).
The six studies which made up this project were all field based. They were also descriptive rather than manipulative, relying on observations in fortuitously available areas which happened to differ in past exposure to fire. In particular, the effects of fire frequency were studied in remnants which had experienced low, medium and high fire frequencies over the last 20 years. ‘Natural experiments’ such as this are not the ideal form of experimental design for attributing cause and effect. However a manipulative study was not feasible for a variety of reasons, including the limited duration of project funding (three years), difficulties of organising multiple planned burns in a rare vegetation type, and the impossibility of ensuring exclusion of unplanned arson fires in
24 heavily populated Western Sydney. The place-for-time approach adopted here allowed consideration of the effects of sequences of fires over a relatively long time period, in remnants from across the Plain. ‘Natural experiments’ such as this are often the only practical way to study perturbations whose effects depend on patterns of disturbance over decades and/or across broad spatial scales (Diamond 1986), or whose potential for destruction limits their use in experimental design (eg high intensity fire).
1.11 Thesis outline
Chapter 2 provides an overview of the study region, Western Sydney’s Cumberland Plain. This chapter includes information on how sites for the current project, and their fire history, were sourced, and on rainfall during the years of the project.
Methods and findings of the six studies which made up the project are described in Chapters 3 to 9 (the largest study is reported in two chapters). Detailed study questions and their context are outlined in the introduction to each chapter. Discussion at each chapter’s end relates findings to previous work and draws out implications for an understanding of the dynamics of the woodlands of the Cumberland Plain.
Chapter 3 focuses on fire-related characteristics of shrub species in Cumberland Plain and Castlereagh woodlands. Parameters such as post-fire regeneration mode, time to first flowering, and time to senescence provide insight into the domain of interfire intervals compatible with the retention of shrub species in each community.
Chapters 4 and 5 report the findings of a landscape scale study comparing vegetation characteristics in woodland sites subjected to low, medium and high fire frequencies over the last 20 years. Where other project studies can be used to extrapolate the effects of fire frequency indirectly from plant species characteristics and documentation of the effects of a single fire, this study assesses the correlates of different fire frequencies directly. Chapter 4 covers relationships between fire frequency and shrub species richness and abundance, tree density and recruitment, and grass cover and dominance. Chapter 5 explores spatial pattern in trees and shrubs, shrub species composition, associations amongst vegetation components and the influence of grazing.
Chapter 6 documents post-fire regeneration of the dominant CPW shrub, Bursaria spinosa, in a remnant at Mt Annan.
25 Chapter 7 addresses the role of fire in ground layer species reproduction. Fire appears to play a part in regeneration of forbs in productive Victorian grasslands – does it also do so in CPW?
Ground layer species may vary between woodland microhabitats. If fire frequency in CPW influences the abundance of trees and/or shrubs, ground layer species composition and abundance may be affected indirectly. This possibility is explored in Chapter 8, along with direct effects of fire frequency on native and exotic forbs.
While the aforementioned studies focus on ecological issues, Chapter 9 investigates characteristics of CPW relevant to the protection of life and property. An equation for fuel load accumulation with time-since-fire is presented, along with an analysis of fuel load composition.
Finally, Chapter 10 summarises study findings, places them in the context of non- equilibrium paradigm concepts, and presents a state and transition model for Cumberland Plain Woodland. Fire frequency guidelines for Western Sydney’s woodlands are suggested, and implications for processes and ecosystems beyond the Cumberland Plain, considered.
26 CHAPTER 2 THE STUDY AREA: WESTERN SYDNEY’S CUMBERLAND PLAIN
2.1 Location and landform
Sydney’s Cumberland Plain lies roughly between the suburban centres of Parramatta and Liverpool, and the Blue Mountains (Figure 2.1). It consists of a shallow ellipse- shaped basin surrounded by higher country: the MacDonald Ranges and the Hornsby Plateau to the north and north-east, the Blue Mountains to the west and the Woronora Plateau to the south and south-east. The north-south axis of the Plain stretches a distance of some 75 kilometres. The centre of the Plain is traversed from east to west by the Western Motorway which runs for 30 kilometres between Parramatta and Penrith. The boundaries of the Plain lie between latitudes 33° and 35° S, and longitudes 150° and 152° E (Bannerman and Hazelton 1990, Tozer 2003, Table A1.1 in Appendix 1).
The topography of the Plain contrasts sharply with that of the dissected plateaux which surround it. Gently undulating ground rises from just above sea level in the east and north to an altitude of around 350 m in the south; most of the Plain is below 100 m asl (Benson 1992, Tozer 2003, Table A1.1 in Appendix 1). The Hawkesbury-Nepean River system drains most of the area. The river itself flows north and then north-east near the western boundary of the Plain, meeting the primary drainage channels from the body of the Plain, South Creek and Eastern Creek, near Windsor. The Georges River system drains the south east quarter of the area (Figure 2.1).
27
Figure 2.1. Map of the Cumberland Plain and surrounds showing areas of retained native vegetation (canopy cover > 10%). Orange, Cumberland Plain woodlands (Section 2.4.1; includes Shale-Gravel Transition Forest); pink, Castlereagh woodlands (Section 2.4.2); green, other native vegetation (source: M. Tozer, Department of Environment and Conservation, pers. comm. 2005). Towns and suburbs shown include those with weather stations listed in Appendices 1 and 2. Inset: approximate location of Sydney in the state of New South Wales.
2.2 Geology and soils
Most soils on the Cumberland Plain were formed from Bringelly Shale, part of the Wianamatta group laid down in the mid Triassic, some 200 million years BP. The surrounding Hawkesbury Sandstone predates the Wianamatta group; the Cumberland Plain is a sag block formed by uplifting of the coast and Blue Mountains. In these
28 higher areas, the sandstone was exposed in the uplift, and through erosion of the overlying shale (Bannerman and Hazelton 1990, James et al. 1999). Bringelly Shale consists of claystone and siltstone, carbonaceous claystone, laminite and fine to medium-grained lithic sandstone (Bannerman and Hazelton 1990, Tozer 2003).
Recently in geological time, in the late Tertiary about five million years BP, the Hawkesbury-Nepean and Georges Rivers flowed across the Plain in the vicinity of Castlereagh and Liverpool respectively. These ancestral rivers laid down alluvial sediments – gravel, sand, silt and clay. These unconsolidated sediments are now known as Tertiary alluvium (Bannerman and Hazelton 1990, James et al. 1999, Tozer 2003).
The soil landscapes of primary interest for the current project are those underlying existing Cumberland Plain and Castlereagh Woodland remnants (Section 2.4). Bannerman and Hazelton (1990) refer to these respectively as the Blacktown and Berkshire Park soils landscapes.
The Blacktown soil landscape covers 670 km2, and is the dominant soil landscape of the Cumberland Plain basin. It occurs in gently undulating country on Wianamatta shales. Local relief is up to 30 m, slopes usually do not exceed 5°. Soils are acidic, shallow to moderately deep hardsetting mottled texture contrast soils, with red and brown podzolic soils on crests grading to yellow podzolics on lower slopes and in drainage lines. These soils are of low to moderate fertility, and may or may not have a layer of friable brownish black loam overlaying the brown clay loam which otherwise occurs as an A2 horizon. These topsoils have a high to moderate organic matter content, and may become waterlogged when rainfall is high (Bannerman and Hazelton 1990).
Berkshire Park soils are found on the flat or gently undulating Tertiary alluvium terraces which mostly occur in the north-west segment of the Plain. The Hawkesbury-Nepean terraces were formed during three depositional phases. The lowest deposit, known as the St Mary’s formation, is overlain by Rickaby’s Creek gravel, which in turn is topped by Londonderry clay. All three formations are exposed in different locations, due to erosion. Soils are weakly pedal heavy orange clays and clayey sands, often containing ironstone nodules. Clay content generally increases with depth although erosion and local depositional cycles may sometimes reverse this trend. Lower horizons are often quite stony. Soils are deep, ranging up to 4.5 m. Berkshire Park soils are strongly acid and of low fertility (Bannerman and Hazelton 1990).
29 2.3 Climate 2.3.1 Climate averages
The Australian Bureau of Meteorology (1991:1) classifies the climate of the Sydney region as “temperate, with warm to hot summers, cool to cold winters and mainly reliable rainfall all year round.” However the Cumberland Plain is drier and subject to more extreme temperatures than Sydney’s coastal areas to the east, or the Blue Mountains to the west. Climate averages are available for a number of locations on the Plain (Bureau of Meteorology 2004, Table A1.1 in Appendix 1).
Mean annual rainfall on the Plain ranges from just over 900 mm at Parramatta and Bankstown on the eastern edge, to 801 mm at Richmond and 830 mm at Campbelltown further west (Appendix 1, Table A1.2). In contrast, Sydney City and Katoomba average 1217 mm and 1399 mm respectively. This is because moist airstreams, which generally have an easterly component, tend to expend their moisture either as they cross the coast, or on the windward slopes of the ranges (Bureau of Meteorology 1991).
Rainfall peaks between January and March, while late winter and spring (July to September) tend to be relatively dry (Table A1.2 in Appendix 1). Average within-year rainfall variability in Sydney, including on the Cumberland Plain, is classified as moderate relative to other parts of Australia, as the median rainfall of the driest month is generally 30-50% that of the wettest month. By contrast, in Adelaide and Brisbane this ratio is about 20%, falling to below 10% in Townsville and Perth. However within any particular year, rainfall totals from month to month are quite variable in Sydney, and rain tends to fall in concentrated bursts (Bureau of Meteorology 1991).
Rainfall variability between years in Sydney is also classified as moderate; it is much greater than in southern Victoria, but much less than in the interior of the continent. Even in the driest part of the Plain, rainfall is between 500 and 1000 mm in nine years out of ten.
Sydney as a whole escapes the extremes of temperature experienced by inland Australia, due to proximity to the ocean, although this influence is less pronounced in Western Sydney (Tables A1.3 and A1.4 in Appendix 1). Mean maximum temperature over most of the Plain is between 23 and 24°C, somewhat above the 21.6°C average in Sydney City, and well above the Katoomba mean of 16.6°C. Mean minima range from
30 10.5 to 12.3°C on the Plain, below the 13.7°C average in Sydney City near the Harbour, but above the Blue Mountains mean of 7.9°C.
January is the hottest month at most stations. The differential effect of cooling off- shore winds and altitude is most evident at this time of year, when average maximum temperatures range from below 26°C in the City and 23°C at Katoomba, to over 29°C on the Plain. “In January a maximum temperature greater than or equal to 35°C occurs at Richmond 16 per cent of the time, while at Observatory Hill [Sydney City] and Katoomba these very hot days are experienced on three per cent and 0.4 per cent of days respectively” (Bureau of Meteorology 1991:50).
The coolest month at all stations is July. Although average Western Sydney maxima for this month, which range between 16.7 and 17.4°C, are considerably higher than the Katoomba figure of 9.2°C, cold air drainage on to some parts of the Plain results in average minima not much above those in the Mountains. July minima in Campbelltown and Richmond average 3.1 and 3.2°C respectively, while the equivalent figure for Katoomba is 2.5°C. Winter minima on the Plain vary considerably, with Prospect averaging 6.1°C, not too far below Sydney City’s 8.0°C. This unevenness is reflected in the incidence and severity of frosts. The average number of frost days per annum at Richmond is 32.9, however this drops to 12.1 at Bankstown, and 3.1 at Prospect (Bureau of Meteorology 1991).
2.3.2 Rainfall during the project
Rainfall was relatively low throughout the project, with particularly dry patches in the cooler months in several years.
Data on monthly rainfall for the years 2001 to 2004 were sourced from the Bureau of Meteorology, for eight locations on the Cumberland Plain. Details are given in Appendix 2. Rainfall was below long-term annual averages at all weather stations for which data were available, in all four years, except in 2001 at Parramatta. For six of the eight localities, the discrepancy was greatest in 2002, while 2003 and 2004 were also relatively dry.
Seasonal rainfall patterns were similar across Cumberland Plain sites. In general, patterns during the years of the project mirrored long-term trends in seasonality, but
31 were more extreme. Winter and Spring 2002 were particularly dry, with periods of below-average rainfall also occurring in Winter/Spring 2001, Winter 2003, and Autumn/Winter 2004. Spikes of above-average rainfall occurred in most Cumberland Plain sites in Summer 2001/2, Autumn 2003 and Spring 2004. Figure 2.2, showing seasonal rainfall at Prospect during the years of the project relative to long-term averages, is typical. Only 57 mm of rain was recorded for the six months from June and November 2002 at this station.
400
350
300
250
200
Rainfall (mm) Rainfall 150
100
50
0 Autumn Winter Spring Summer Autumn Winter Spring Summer Autumn Winter Spring Summer Autumn Winter Spring 2001 2001 2001 2001/2 2002 2002 2002 2002/3 2003 2003 2003 2003/4 2004 2004 2004
Figure 2.2. Seasonal rainfall (mm) at Prospect from 2001 to 2004, relative to long-term averages. Solid red line, 2001-2004 rainfall data (source: Bureau of Meteorology pers. comm. 2005). Dashed blue line, long-term averages (source: Bureau of Meteorology 2004).
2.4 Vegetation
Differences in geology, topography and climate between the Plain and the surrounding plateaux are reflected in the composition of their respective native plant communities. The sandstone heath and woodland communities are characterised by a diverse, sclerophyllous shrub layer. While some species found on the Plain also occur in the sandstone country, many are restricted to one environment or the other. Grass and herb species play a considerably more prominent role on the Plain (Benson 1992, James et al. 1999, Tozer 2003).
32 Two broad categories of vegetation have long been distinguished on the Cumberland Plain: grassy woodlands on shale, and more shrubby woodlands on Tertiary alluvium (Phillips 1947 cited in Benson 1992, Kartzoff 1969, Robinson 1994). These broad vegetation types have been designated Cumberland Plain Woodlands (CPW), and Castlereagh Woodlands, respectively (Benson and Howell 1990, James et al. 1999). Subgroups can be distinguished within each category, and gradients exist between the two, as well as where shale meets sandstone. Areas of increased water availability, whether through higher rainfall, proximity to rivers and creeks, or topographic depressions, often support somewhat different vegetation to that found on the gentle slopes and low ridges which make up the bulk of the Cumberland Plain.
Three major studies of western Sydney’s vegetation have been conducted over the past 15 years. Benson (1992) interpreted data from remnant vegetation in the light of underlying geology and historical evidence to produce a typology and map of extant vegetation, along with detailed descriptions of community types. A few years later, NSW National Parks and Wildlife Service (NPWS) staff compiled an inventory of vegetation remnants of conservation significance, on the basis of extensive survey work (NPWS 1997). Results of this study informed the revision of the book Rare Bushland Plants of Western Sydney, which provided information on plant communities and individual species, along with a comprehensive species list (James et al. 1999). Most recently, Tozer (2003) coordinated a stratified survey of over 600 sample sites both on the Cumberland Plain, and on shale soils where these occurred on the adjacent plateaux. This project used cluster analysis to distinguish vegetation types, and modelling based on environmental variables to map community distributions past and present. The resulting typology was similar to Benson’s, but not identical.
All three studies highlight the mosaic of vegetation types which exist across the Cumberland Plain (Figure 2.1). Cluster analysis confirmed the primacy of the division between the vegetation of the sandy soils and those on clay-loam: this was the first division in the hierarchy. Benson’s conjectures with respect to relationships between substrate and vegetation units were also mostly borne out, although boundaries between communities frequently did not conform exactly to geological boundaries, and anomalies were found (Tozer 2003).
33 James et al. (1999) list 800 native plant species for the Cumberland Plain, while Tozer (2003) found 831 native taxa in his sample sites. The nomenclature in the James et al. (1999) species list will be followed throughout this document, unless otherwise stated.
2.4.1 Cumberland Plain Woodland
James et al. (1999:18) describe Cumberland Plain Woodland (Figure 2.3) as “a complex of intergrading vegetation units classified according to dominant canopy species which vary with local conditions.” Benson (1992) identified three CPW ‘vegetation units’: Grey Box Woodland, Grey Box-Ironbark Woodland and Spotted Gum Forest. Tozer’s cluster analysis failed to uphold these categories, identifying instead two CPW communities: Shale Plains Woodland and Shale Hills Woodland. Shale Plains Woodland typically occurred in flat, low rainfall areas, while Shale Hills Woodland was found at higher elevations in the southern part of the Plain (Tozer 2003).
Cumberland Plain Woodland vegetation shows considerable variation across remnants, and even within one site (French et al. 2000). Species are often patchily distributed, and many have sparse distributions (Benson 1992, Tozer 2003). The visibility of species varies with time of year (Thomas 1994, J. Howell pers. comm. 2001), rainfall (French et al. 2000, S. Clarke 2003, Hill and French 2004), and with time since fire (Hill and French 2004). French et al. (2000), writing before the Tozer study, suggested that the differences between the three Benson CPW categories might more usefully be thought of as the expression of variability within one community. Similar thoughts have been expressed in relation to the Tozer subgroups (R. Bradstock pers. comm. 2002).
Eucalyptus moluccana (Grey Box) and Eucalyptus tereticornis (Forest Red Gum) are the most common tree species in both Shale Plains and Shale Hills sites (Tozer 2003), E. moluccana tending to favour slightly higher ground than E. tereticornis (Benson 1992). Eucalyptus crebra (Narrow-leaved Ironbark) is also well-represented. A scattering of other eucalypt species also occurs: Eucalyptus eugenioides and Eucalyptus fibrosa are often mentioned (Benson and Howell 1990, Benson 1992, Tozer 2003). All these trees are tall, with straight boles, reflecting the positioning of CPW at the top end of the woodland rainfall spectrum.
34
Figure 2.3. Cumberland Plain Woodland remnants at Plumpton Park (left) and Prospect Reservoir (right). The Plumpton Park photo shows the open grassy woodland structure found in some CPW remnants. The flowering shrub in the Prospect photo is Bursaria spinosa, the shrub in the foreground is Pultenaea microphylla. Gum-barked trees in both photos are Eucalyptus tereticornis, half-barks are Eucalyptus moluccana. Plumpton Park photo by Annie Storey.
Shrub species richness in Cumberland Plain Woodland is low relative to that of the sandstone woodlands (James et al. 1999, Benson and Howell 2002). The resprouting shrub Bursaria spinosa is by far the most commonly encountered species, often forming dense thickets (Benson 1992, James et al. 1999, Benson and Howell 2002, Tozer 2003). Other shrubs recorded relatively frequently include the wattles Acacia parramattensis, A. decurrens, A. implexa, and A. falcata (the first three can grow quite tall, and are sometimes included in descriptions of the tree layer). Peas such as Dillwynia sieberi, Daviesia ulicifolia and Indigofera australis occur in some sites. Shrubs tend to be patchily distributed (Benson 1992, Tozer 2003).
Much of the plant biodiversity in Cumberland Plain Woodland is found in the ground layer, which supports a wide array of grasses and forbs (Hill and French 2003). Many forbs are long-lived perennials with thick fleshy roots which enable them to regenerate after drought, grazing or fire (Benson and Howell 2002).
Beadle (1981) considers CPW part of the Eucalyptus moluccana Alliance, a broad vegetation type which occurs in other rainshadow areas near the coast, such as the floors
35 of the Hunter and Clarence River valleys. The CPW flora also has similarities to that of grassy woodlands to the west: 74% of CPW species recorded by Benson (1992) occur on the Western Slopes and 44% on the Western Plains. Equivalent figures for sandstone flora in Ku-ring-gai Chase National Park are 34% and 10% (Benson 1992), showing that these inland floras have a much closer affinity with CPW than with Sydney’s sandstone communities.
Where shale meets Tertiary alluvium, CPW grades into Shale Gravel Transition Forest. This ecotonal community also occurs on isolated deposits of tertiary alluvium overlying shale, and in areas with a high concentration of lateritic gravels. Shale Gravel Transition Forest is usually dominated by Eucalyptus fibrosa (Broad-leaved Ironbark), with Eucalyptus moluccana occurring, and sometimes dominating, in a minority of sites. Melaleuca decora often forms a small tree stratum. A low shrub layer, “with a mixture of the hardier species from the Wianamatta Shale and Tertiary Alluvium” (Benson 1992:558) is generally present, as are a variety of grasses and herbs (Benson 1992, Tozer 2003).
2.4.2 Castlereagh Woodland
While the project focussed primarily on Cumberland Plain Woodlands, plant responses to fire in Castlereagh Woodland were also investigated.
The two most extensive subunits identified under this heading by both Benson (1992) and Tozer (2003) are Castlereagh Ironbark Forest and Castlereagh Scribbly Gum Woodland (Figure 2.4). Castlereagh Ironbark Forest generally occurs on Tertiary alluvium with relatively high clay content, while Scribbly Gum Woodland is more common on sandy loams. A third subunit, Castlereagh Swamp Woodland, is also identified by both authorities, although Benson uses a slightly different name. The following descriptions draw on both Tozer (2003) and Benson (1992).
Castlereagh Ironbark Forest is dominated by Eucalyptus fibrosa, with Melaleuca decora also generally present. This vegetation type has an understorey of sclerophyllous shrubs from several families; fabaceous taxa are particularly well represented. Somewhere in the vicinity of the boundary between Tertiary alluvium and shale, Castlereagh Ironbark Forest grades into Shale-Gravel Transition Forest. As soils become more sandy, Castlereagh Scribbly Gum Woodland takes over.
36
Figure 2.4. Castlereagh Woodland at Nutt Road, Castlereagh, showing the relatively shrubby nature of this vegetation type. Flowering shrubs are Banksia spinulosa var. spinulosa (orange flowers) and Pimelea linifolia ssp. linifolia (white flowers). Gum-barked trees are Eucalyptus sclerophylla, full-barks are Angophora bakeri.
Castlereagh Scribbly Gum Woodland is dominated by the Scribbly Gum Eucalyptus sclerophylla, and by Eucalyptus parramattensis (Parramatta Red Gum), Angophora bakeri (Narrow-leafed Apple), and Melaleuca decora. This subunit tends to occur on slightly higher ground than Castlereagh Ironbark Forest, with E. parramattensis more likely than E. sclerophylla to be found in areas of relatively poor drainage. The shrub understorey is diverse, and includes a greater representation of Myrtaceous and Proteaceous species than the subunits discussed to this point, as well as a variety of peas and wattles.
In poorly-drained depressions and along creeklines on Tertiary alluvium, the two more extensive Castlereagh vegetation types described above may grade into Castlereagh Swamp Woodland. Eucalyptus parramattensis (Parramatta Red Gum) and Melaleuca decora are the dominant overstorey species in this map unit. The shrub layer is generally less developed here than in adjacent Scribbly Gum and Ironbark areas. The herbaceous understorey contains species which respond to wet conditions.
37 2.5 Management history 2.5.1 Aboriginal management
Prior to the arrival of the First Fleet in 1788, between 5000 and 8000 people lived in the greater Sydney region. The Dharag people lived in the area from the Hawkesbury River in the north to Appin in the south, and west into the Blue Mountains – in other words, on the Cumberland Plain (Kohen 1993).
European settlement had a profound and almost immediate effect on the Aboriginal population. In April 1789, a smallpox epidemic started. Only one colonist died, but over 50 percent of the Aboriginal population is estimated to have perished (Kohen 1993, Bear 1997).
Although relations between remaining Aboriginal clans and colonists were initially cordial, conflicts inevitably arose as settlers cleared traditional hunting and gathering areas, and traditional life was increasingly disrupted. Records show that a number of bands continued to survive on the Cumberland Plain through the first half of the 19th century. Many families today can trace their ancestry back to these groups (Kohen 1993).
What did the Cumberland Plain landscape look like when it was under Aboriginal custodianship? Early European accounts of the vegetation of Western Sydney present a clear picture of open grassy woodland with localised shrubby patches associated with less fertile ground (Benson 1992). The earliest description is provided by Governor Phillip, in the first year of the colony:
“The country through which they [Phillip and his party] travelled was singularly fine, level, or rising in small hills of a very pleasing and picturesque appearance. The soil excellent, except in a few small spots where it was stony. The trees growing at a distance of from 20 to 40 feet [6-12 m] from each other, and in general entirely free from brushwood, which was confined to the stony and barren spots” (Phillip 1789, quoted in Benson and Howell 2002).
In 1827, Peter Cunningham wrote:
“…the land immediately bordering upon the coast is of a light, barren, sandy nature, thinly besprinkled with stunted bushes; while, from ten to fifteen miles interiorly, it consists of poor clayey or ironstone soils, thickly covered with our usual evergreen forest timber and underwood. Beyond this commences a fine timbered country, perfectly clear of brush, through which you might, generally
38 speaking, drive a gig in all directions, without any impediment in the shape of rocks, scrubs or close forest. This description of country commences immediately beyond Parramatta on the one hand, and Liverpool on the other; stretching in length south-easterly obliquely towards the sea, about forty miles, and varying in breadth near twenty” (quoted in Benson and Redpath 1997; italics added to emphasise the description of the Cumberland Plain).
Ethnobotanists Kohen and Downing (1992) report that Aboriginal people were known to use many plants currently found on the Cumberland Plain. These included herbaceous species with edible bulbs or tubers, collectively called ‘yams’ by both Aboriginal people and early settlers. In fact, the name ‘Dharag’ which literally means ‘tooth’ is also the word used for these tubers (Kohen and Downing 1992). Kohen and Downing (1992:4) note that “one important aspect of Aboriginal economy was the practice of burning the underbrush,” and speculate that regular burning may have maintained an environment particularly suitable for tuber-producing species.
There is no doubt that Aboriginal people all over the continent used fire for a variety of purposes. There is equally no doubt that burning was anything but random; rather it was a controlled and skilled part of traditional life designed to meet particular goals (NPWS 1997, Bowman 1998, Gott 1999, Yibarbuk et al. 2001). On the Cumberland Plain, fire appears to have been used as an aid to hunting (Benson and Howell 1990). White (1790, cited in Benson and Redpath 1997) reports its use to smoke out game from trees. Kohen (1993) reports that kangaroos were certainly hunted around Sydney, although they formed a relatively minor component of the food intake. He speculates that this was partly because of the effort needed to hunt them: 50 or 60 men would form a large circle, set fire to the grass, and spear animals as they tried to escape. This hunting technique was known as walbunga, literally ‘wallaby dead’.
Ecologists Clark and McLoughlin (1986), in a study of pre-European and recent fire regimes around Sydney, argue that shale areas were probably burnt more frequently than sandstone vegetation, as do Benson and Howell (1990). Even Benson and Redpath (1997:292), in what is generally a vigorous critique of the suggestion that Aborigines made frequent use of landscape fires, acknowledge that “Aboriginal people may have burnt grassy woodlands such as the Cumberland Plain fairly regularly in patches to attract game and to ease travel.”
39 2.5.2 European management
The relatively rich shale soils and gentle topography of the Cumberland Plain encouraged early settlement by colonists after the arrival of the First Fleet in 1788. Agricultural development was underway by 1792; domestic livestock followed shortly thereafter. Grey Box (Eucalyptus moluccana) was seen as an indicator of good grazing land, so the country supporting this species was taken up rapidly (Kartzoff 1969). By the mid 19th century, the majority of the Cumberland Plain was either under cultivation, or subject to grazing (Kartzoff 1969, Benson and Howell 2002, Tozer 2003). Agriculture was preceded and accompanied by timber-getting, which again was concentrated on shale soils (Recher et al. 1993).
More recently, farms have been replaced by suburbs and factories as Sydney reaches west. Urban expansion into the Cumberland Plain intensified from the 1960s, bringing major impacts. Apart from on-going clearing, weeds, feral animals, rubbish and four- wheel drives have increasingly been introduced to remaining remnants (Rawling 1994, James et al. 1999, Tozer 2003).
A potential turning-point for Western Sydney’s bushland occurred with the publication in 1990 of Benson and Howell’s Taken for Granted: the Bushland of Sydney and its Suburbs. This book presented residents of Western Sydney with a clear and alluring account of the bushland in their local neighbourhood, and with an impassioned plea for its preservation. By 1999, James et al. were able to report that local councils were starting to manage bushland remnants for their conservation values. A number of the larger remnants were gazetted as national parks.
There is little information on fire regimes since European settlement. We do know that the practice of widespread frequent burning for grazing ‘green pick’, still common in grassy woodlands further north today, was practiced by pastoralists in the early years of the colony. In 1926 James Atkinson wrote:
“In the unoccupied districts in the interior, and also in those tracts that are only used for the purposes of grazing, the grass in winter becomes withered by the frosts, and assumes the appearance of bad coloured hay; in this state it is refused by the cattle; and as it impedes the growth of young grass, the common practice is to set fire to it. The Natives also pursue the same system setting fire to the thick brushes and old grass every summer; the young herbage that springs up in these places, is sure to attract the kangaroos and other game; and the horned cattle are also very fond of
40 feeding upon this burnt ground, as it is termed in the colony; .... In dry seasons these periodical burnings sometimes assume a truly awful appearance, the country seems on fire in all directions, and if the weather is calm, is enveloped in dense smoke” (quoted in Benson 1992:569).
At some point after 1827, this practice ceased. As far back as 1848, Mitchell noted that the “omission of the annual periodical burning by natives, of grass and young saplings, has already produced in the open forest lands nearest to Sydney, thick forests of young trees, where, formerly, a man might gallop without impediment, and see miles before him” (quoted in Kartzoff 1969). As Benson and Howell (2002) point out, the term ‘forest land’ was used in the 19th century to describe grassy woodland; this quote thus almost certainly relates to the shale country of the Cumberland Plain.
At least in terms of planned burns, CPW now conforms to the generalisation by Hobbs (2002:310) that “Burning in woodland communities is extremely rare in eastern Australia.” However it is likely that management regimes included somewhat more frequent use of fire pre-1985 than has been the case over the past two decades. Reports from Rural Fire Service (RFS) personnel and older land managers indicate that burns for fuel reduction purposes were relatively common in CPW remnants until its conservation potential was recognized in the late 80s and early 90s, at which point burning was curtailed due to a fear that fire would damage native flora and fauna. Nowadays, remnants in highly urbanised areas tend to fall into one of two categories: either they have not been burnt for many years, or they have been subject to quite frequent fire where arsonists are active. Burning for conservation purposes is in its infancy in Western Sydney. Exceptions include a small number of experimental fires at Mt Annan Botanic Gardens, and one at Scheyville National Park.
2.6 Current status of vegetation
Western Sydney’s bush is now, to a large extent, suburban bush (Rawling 1994). Most has been cleared and what is left is fragmented and subject to a wide range of threatening processes (James et al. 1999).
Under the NSW Threatened Species Conservation (TSC) Act 1995, a species assemblage at risk of extinction can be listed as an Endangered Ecological Community. Over recent years all Western Sydney’s plant communities other than Castlereagh
41 Scribbly Gum Woodland have been listed. Estimates of pre-European extent, extent in 1997, and percentage remaining in 1997 for vegetation types addressed in this project are given in Table 2.1.
Table 2.1. Estimates of pre-European and contemporary (1997) extent of native vegetation communities on the Cumberland Plain. From Tozer (2003:15).
Assemblage name Estimated pre- Estimated extent in Percent of pre- (Tozer 2003) European extent 1997 (ha) European extent (ha) remaining in 1997 (± range) Shale Plains 87,175 6,745 7.7 (± 1.1) Woodland Shale Hills Woodland 38,274 4,309 11.3 (± 1.5) Shale Gravel 5,427 1,721 31.7 (± 3.1) Transition Forest Castlereagh Ironbark 12,211 1,012 8.3 (± 0.8) Forest Castlereagh Scribbly 5,852 3,083 52.7 (± 2.9) Gum Woodland
The proportion of pre-European vegetation remaining increases with increasing sandstone influence (Tozer 2003). Despite this, there are few conservation reserves on shale or alluvium relative to those on sandstone. The current reserve system in Western Sydney is neither comprehensive nor representative of the range of habitats, although progress has been made in recent years. Conservation strategies are still much needed (NPWS 1997, James et al. 1999).
Clearing, fragmentation and tardiness in reservation of Cumberland Plain vegetation communities has naturally had an effect on the conservation of individual plant species. Of the approximately 1200 species recorded in Western Sydney, more than 30 are listed under the NSW Threatened Species Conservation (TSC) Act 1995 as endangered or vulnerable (James et al. 1999).
On a more positive note, Benson and Howell (2002) conclude from an examination of historical records that although the relative abundance of plant species in Western Sydney vegetation may well have changed, very few species present at the time of European colonisation appear to have been totally lost. Many CPW and Castlereagh woodland remnants are still dominated by native species (Benson 1992, Tozer 2003), a testimony to the resilience of these vegetation types. Removal of threats such as
42 mowing can produce encouragingly rapid increases in native species richness (NPWS 1997).
Weeds are more problematic on the richer shale soils than in the less fertile, and less disturbed, Castlereagh Ironbark and Scribbly Gum Woodlands (Benson 1992, Tozer 2003). Weed species considered particularly challenging include African Olive Olea europaea ssp. africana (Benson 1992, Rawling 1994, Tozer 2003), Bridal Creeper Myrsiphyllum asparagoides (Tozer 2003) and Boxthorn Lycium ferocissimum (Rawling 1994). Sida rhombifolia (Paddy’s Lucerne) is a common small woody weed in many CPW remnants (Tozer 2003). A range of herbaceous species such as Senecio madagascariensis, Cirsium vulgare and Setaria gracilis are also widely encountered.
While this project focussed on the role of fire in maintaining native plant species, exotic species were inevitably encountered. The fire frequency studies reported in Chapters 4 and 8 provided data germane to the question:
Does fire frequency influence exotic weeds in CPW?
2.7 Study sites 2.7.1 Locating remnants
The initial step in determining where to study the effects of fire on the Cumberland Plain was to identify good-quality remnants. Information on remnants was available in NPWS (1997), which detailed bushland areas in each local government area. Vegetation types, condition, and in some instances indications of past fire history were given for each remnant. A spreadsheet of potential study sites was developed using this information. Additional possibilities were sought through discussion with land and fire managers in agencies including NPWS, RFS, Greening Australia, and local councils. Advertisements were placed in community bushcare publications in an effort to ensure that privately-owned remnants were also considered. Remnants were visited and assessed for condition, size, dominant tree species, and access.
43 2.7.2 Identifying fire history
For a remnant to be considered a potential study site, I needed to know its fire history. Both the time of the most recent fire, and fire frequency over recent decades were of interest.
I had hoped that written records of fires would be available from land and fire managers, however although this was sometimes the case, often little useful documentation could be located. Records often had not been kept, or had been lost, or were not in a form that provided relevant information. For example RFS records of fires did not necessarily distinguish small ‘car fires’3 from moderately extensive events. Area burnt was often not mapped. Planned burns were not recorded.
However land managers, fire managers, and other contacts were frequently able to provide verbal information about the fire history of remnants. In some cases managers had been associated with their sites for over 20 years. Field visits with informants helped ensure an accurate understanding of the location and timing of fires.
Aerial photos from 1982 to 2002 were also perused for signs of fires. While some useful information was obtained in this way, there were major limitations:
• Photos were only available for a limited number of years; • Major fires in the summer of 2001/2 were not easily identified on photos taken in March 2002 – ie the ‘signal’ appears to fade rapidly in these woodlands; • Fires in remnants are often small and/or of low intensity, making their detection difficult.
Once contact had been established with managers, I was often informed when fires occurred. I was thus able to date and map many fires accurately from late 2001 on.
Fire history information was added to the spreadsheet, as was information about remnant size, condition, access and past land management.
Criteria used to select remnants for each study are outlined in the appropriate methods sections.
3 Western Sydney remnants are not infrequently used as repositories for stolen cars, which may be burnt by those depositing them. These fires seldom spread far, but may necessitate the presence of the local urban or rural fire service (pers. comm., many fire and land managers, 2001-4).
44 CHAPTER 3 SHRUB VITAL ATTRIBUTES
3.1 Introduction 3.1.1 Vital attributes and plant functional types
In this chapter, shrub species in Cumberland Plain and Castlereagh woodlands are characterised in terms of their fire-related attributes, and inferences drawn as to their probable responses to different frequencies of fire. The approach used draws heavily on that outlined by the NSW Department of Environment and Conservation (DEC 2002), which in turn draws on Noble and Slatyer’s vital attributes model (Noble and Slatyer 1980).
The vital attributes model uses a small number of life history characteristics of plant species, termed ‘vital attributes’ because they are “vital to the role of the species in a vegetation replacement sequence,” to predict successional pathways (Noble and Slatyer 1980:6). These attributes can also be used to group species into functional types whose populations have similar methods of persistence and re-establishment after fire (Noble and Slatyer 1980, Keith et al. 2002b). The fire regime requirements of these groups can then be considered and distilled into management guidelines (Bradstock 1999, DEC 2002).
Three sets of vital attributes are used in the model (Noble and Slatyer 1980, Friend et al. 2003):
1. The method of arrival or persistence at a site after fire. Both propagule-based, and vegetative methods of persistence are recognised, as are combinations of the two. 2. The environmental conditions required for establishment, particularly whether a species can establish only in the immediate post-fire period (I species), or whether establishment can occur at any time between fires (T species). 3. The time taken to reach critical life stages. Primary juvenile period, longevity of mature individuals, and longevity of propagules are the key factors.
45 Combinations of the first two attributes define 14 plant functional types (PFTs). Species in each type share critical factors which mediate their response to various levels of fire recurrence (Keith et al. 2002b). Detailed information on vital attribute sets, plant functional types and fire interval sensitivities is provided in Appendix 3.
3.1.2 Interval sensitivities
Key fire response species are those whose vital attributes indicate that they are most likely to be eliminated by very frequent, or very infrequent, fires (Friend et al. 2003).
Functional types most sensitive to short interfire intervals (high fire frequency) contain obligate seeder species whose seed reserves are exhausted by disturbance. Populations of these species are liable to local extinction if the interval between fires is shorter than their primary juvenile period (Noble and Slatyer 1980). The minimum interfire interval to retain all species in a particular vegetation type therefore needs to accommodate the taxon in this category with the longest juvenile period (DEC 2002). Obligate seeders with seeds that can persist through more than one fire are moderately sensitive to short interfire intervals. These species will decline if fires continue to occur within the juvenile period (Bradstock and Kenny 2003).
Species whose establishment is keyed to fire (I species) are highly sensitive to long interfire intervals (infrequent fire): they are liable to local extinction if fire does not occur within the lifespan of established plants and/or seedbanks (Noble and Slatyer 1980). Maximum intervals therefore need to accommodate the taxon in this category with the shortest lifespan, seedbank included (DEC 2002, Bradstock and Kenny 2003).
3.1.3 Study aims, boundaries and questions
The broad aims of this study were: • To document fire-related life-history characteristics of Cumberland Plain shrub species; • To characterise the shrub species complements of Cumberland Plain and Castlereagh woodlands in terms of their fire-related attributes; and • To use this information to identify interfire intervals that may be incompatible with retention of shrub species currently found in each vegetation type (thresholds of concern).
46 Some of the data used in this study were sourced locally, while some were drawn from the NSW Flora Fire Response Database (DEC 2002).
Profiles of the two vegetation types explored in this study were constructed using the list of native plants of Western Sydney provided in James et al. (1999). These vegetation types are likely to cover a somewhat wider range of sites and species than Cumberland Plain Woodland and Castlereagh Woodland as defined in Section 2.4. James et al. (1999) do not include a separate category for Shale-Gravel Transition Forest, and may also have included some sites and species from shale-sandstone transition areas (James et al. 1999). In deference to these differences, the terms ‘Cumberland Plain woodlands’ and ‘Castlereagh woodlands’ (rather than Cumberland Plain Woodland – CPW – and Castlereagh Woodland) have been used in this chapter.
The decision to focus on shrubs reflects both the relative sensitivities of life-forms, and data availability. A scoping exercise using the NSW Database (DEC 2002) revealed that no tree species in the vegetation types of interest was classified as highly sensitive to frequent fire. Most were also considered insensitive to low fire frequencies, and species classified as sensitive had lifespans well above those of most shrubs. A similar exercise using common CPW herbs found that many were unclassified due to lack of data, and of those that had been allocated to a plant functional type, few were considered sensitive to frequent fire. Juvenile periods, where available, were all below two years. It therefore seemed unlikely that a herb would be the key fire response species for frequent fire. Vulnerability to infrequent fire was more common, but longevity data were too sparse for useful analysis.
Specific questions addressed in this chapter include: • What post-fire regeneration mechanisms do Cumberland Plain shrubs employ? • How long are the primary and, where relevant, the secondary juvenile periods of these species? • Do local observations of shrub species regeneration modes and juvenile periods accord with those in the NSW Flora Fire Response Database? • What regeneration modes, plant functional types and fire regime sensitivities characterise Cumberland Plain and Castlereagh woodlands, respectively? • What range of juvenile periods is found in Cumberland Plain and Castlereagh woodlands?
47 • Which shrub species in each vegetation type are likely to be most sensitive to short interfire intervals, and at what point is this vulnerability likely to become critical? • How long are species sensitive to infrequent fire, and their propagules, estimated to survive, in Cumberland Plain and Castlereagh woodlands? • Which shrub species in each vegetation type are likely to be most sensitive to long interfire intervals, and at what point is this vulnerability likely to become critical? • What do we know about the fire-related characteristics of threatened shrub species within Cumberland Plain and Castlereagh woodlands?
3.2 Methods 3.2.1 Documenting post-fire regeneration modes
Ideally, regeneration modes would have been assessed through post-fire observation of plants tagged prior to a burn, over multiple sites. Time generally precluded the use of this approach, although it was adopted at a single site for Bursaria spinosa (Chapter 6). Instead, post-fire observation across a wide range of Cumberland Plain sites was employed. Site locations are shown in Figure 3.1. A reliability scale (Kitchin 2001) reflecting the limits of this methodology, and the degree to which observations were replicated across sites, was developed (Section 3.2.3).
Where possible, seedlings were distinguished from resprouts by the presence of cotyledons. Living shoots attached to charred remains were assumed to be resprouts. Small living plants without charred remains were generally assumed to be seedlings, particularly if single-stemmed. Where categorisation was unclear, a small number of plants of the disputed species were dug up, and their roots examined for evidence of suckering, lignotubers or other storage organs, and burnt remains. Where none of these signs was found, plants were considered seedlings. Dead remains of burnt plants were also examined. Plant species present after fire only as seedlings were assumed to be obligate seeders (Bradstock et al. 1997).
3.2.2 Documenting juvenile periods
Two strategies were employed to assess shrub juvenile periods. First, seedlings were tagged at two sites approximately six months after fires, and monitored regularly for
48 flowers and fruits. Time precluded expansion of this strategy. Second, many sites across the Plain were visited at known times-since-fire, and species budding, flowering or fruiting were recorded. A reliability scale similar to that for regeneration mode was employed.
Two seeding lines were set out in June 2002. The first was in clay soil in an area dominated by Eucalyptus fibrosa at Shanes Park, the second in sandy soil at Devlin Road, Castlereagh in a patch dominated by Eucalyptus sclerophylla (see Figure 3.1 for site locations). The first site had burnt in January 2002 at fairly high intensity (scorch height to approximately 15 m), the second in November 2001 at moderate intensity (scorch height to approximately 10 m). At each site, a 30 m tape was laid on the ground, and a subset of the seedlings lying within 1.5 m of the tape, tagged. Seedlings were identified by the presence of cotyledons, by the absence of charred remains, and/or by comparison with other plants which had been dug up and their roots examined. All shrub species with seedlings in this 30 x 3 m area were included. It was not possible to tag an equal number of seedlings for each species, as numbers varied widely between species, with some represented only by one or a few seedlings. All seedlings of these poorly-represented species were tagged, while for more common species a subset was chosen. Chosen seedlings were those which appeared robust and thus likely to survive, a course of action recommended by Doug Benson (pers. comm. 2001), on the basis of his experience in Brisbane Water National Park (Benson 1985), where many tagged seedlings died in the initial post-fire years.
Seedling lines were revisited at least every three months; visits were more closely spaced in Spring as most species were known to flower at this time. Seedling deaths were recorded, and for the first four monitoring visits (to 18 and 20 months post-fire at Shanes Park and Castlereagh respectively), dead seedlings were replaced with robust living individuals of similar size. A few additional seedlings were tagged during these early visits; these seedlings were of species not found initially, probably due to delayed emergence. At each visit, living seedlings were assessed for flowers and fruits. Nearby vegetation which had burnt at the same time as the seedling lines was searched for flowering or fruiting shrubs.
Other sites where fires had occurred were visited at various times post fire (see Figure 3.1 for site locations). Effort was concentrated between August and October, as most
49
Figure 3.1. Map of the Cumberland Plain showing location of sites used to assess shrub species regeneration modes and juvenile periods. Sites designated by a street name (eg Devlin Rd, Nutt Rd) are all in the Castlereagh area. Orange, Cumberland Plain woodlands; pink, Castlereagh woodlands; green, other native vegetation (source: M. Tozer, DEC, pers. comm. 2005).
50 species were expected to flower in Spring, however some observations were also made at other times. Many fires occurred across the Plain during the summer of 2001/02, in Spring 2002 and in Summer 2002/03. Where possible, sites were visited soon after fire, and the burnt area mapped. Early post-fire observations allowed identification of areas where regeneration of certain species was limited to seedlings. Flowering observed during later visits could thus confidently be considered indicative of primary juvenile period in these species. Flowering on resprouts generally commenced while evidence of resprouting was still apparent. A species was recorded as flowering or fruiting at a site even if only a small number of plants was showing this level of maturity. While this meant observations reflected a liberal approach to defining juvenile period, later decisions were more conservative.
3.2.3 Consolidating observations
Observations of regeneration modes and juvenile periods of Cumberland Plain shrub species were consolidated into two tables (Appendices 4 and 5; to avoid duplication, these appendices cover species from both Cumberland Plain and Castlereagh woodlands). In each table, species for which observations had been made were listed, along with relevant observations. For each observation, site location, date and time since fire were recorded. Where similar observations were made at numerous sites, three or four were detailed. Species were characterised as either resprouters (R) or obligate seeders (S) (Appendix 4) Where observations of regeneration mode conflicted across sites, a primary and a secondary mode were designated (Rs or Sr). Juvenile period (Appendix 5) was defined in relation to the time of the last fire, rather than from the time of seedling emergence. The former was more easily ascertained, and more relevant for determining interval sensitivities. Juvenile period figures were rounded up to the next whole year (eg 13 months became 2 years, as did 24 months). Note these decision rules are conservative. Where observations of time to first flowering differed between sites, the summary figure was expressed as a range (eg 2 - 4 years). Site differences may reflect genuine differences in maturation times, or may have resulted from failure to note earlier flowering due to limited site visits.
A reliability rating was attached to each summary regeneration mode and juvenile period figure. Reliability ratings reflect data source and extent of replication:
51 1. tested by following the fate of individual plants through time, at more than one site 2. tested by following the fate of individual plants through time, at a single site 3. post-burn observation in three or more sites 4. post-burn observation in two sites 5. post-burn observation in one site.
3.2.4 Comparison with NSW Database
A copy of the NSW Flora Fire Response Database (DEC 2002) was obtained from DEC in August 2004. A new table listing all shrub species for which local information on regeneration mode was available, was constructed (Appendix 6). Summary information from the two tables described above was transferred to this table, and juxtaposed with the equivalent data from the NSW Database. The extent to which regeneration mode and juvenile period data from the two sources coincided, was assessed.
3.2.5 Assigning vital attributes
For each listed species, the vital attributes category assigned in the NSW Database was transposed to the table (Appendix 6). Using this information, together with local observations, each species was assigned to a Cumberland Plain vital attributes group. The following decision rules were used in this process:
• Local regeneration modes were given preference; • Where local regeneration mode was Sr, the species was classified in the appropriate seeder category; • Where local regeneration mode was Rs, the species was classified in the appropriate resprouter category; • Resprouting species were classified U if the Cumberland Plain secondary juvenile period was consistently ≤ 2 years, otherwise V (these terms are defined in Appendix 3); • Seedbank and establishment categories from the NSW Database were used, where these were available; • Where no seedbank category was given in the NSW Database, species with on- plant storage were classified C, legumes were classified S, wind-dispersed species were classified D, all others were classified G (terms defined in Appendix 3); • Where no establishment category was given in the NSW Database, species were classified I unless observation or previous research indicated establishment between fires in vegetated areas, in which case they were classified T.
52 3.2.6 Characterising shrub species in broad vegetation types
The next step was to profile the two broad vegetation types, Cumberland Plain woodlands and Castlereagh woodlands, in terms of the vital attributes and fire regime sensitivities of their shrub species. Profiles drew on the list of native plants of Western Sydney provided in James et al. (1999). This listing, which includes an estimate of abundance for each species in the two broad vegetation types of interest, is based on surveys conducted by botanists over 20-30 years (James et al. 1999). A table was constructed for each vegetation type, with species arranged in two groups (Appendix 7). The first group included all shrub species listed by James et al. (1999) as either ‘very common’, ‘common’ or ‘frequent’ in the relevant vegetation type. The second group included shrub species listed as ‘occasional’, plus ‘rare’ species which were also listed under the NSW Threatened Species Conservation (TSC) Act 1995.
Regeneration mode, primary and secondary juvenile period, and vital attributes group were listed for each species. Where local information was available or, in the case of vital attributes, where it had been assigned as above, these data were used. Gaps were filled through entering information from the NSW Database, where this was available. For a small number of species, neither local data nor Database information existed. Regeneration mode and vital attributes were allocated to these species as follows:
• all species were assumed to be seeders; • species with on-plant storage were classified C, legumes were classified S, wind- dispersed species were classified D, and others were classified G; • establishment was assumed to be limited to the immediate post-fire period, ie all species were classified I.
These rules were designed to ensure possible sensitivities would not be underestimated.
‘Lifespan + seedbank’ figures were transferred over from the Database, for all species where figures were given.
Vital attributes were used to assign each species to its appropriate plant functional type. Associated sensitivities to frequent, and to infrequent, fire were added. Conceptual relationships between vital attributes, plant functional types and sensitivities are outlined in Appendix 3. Sensitivity categories paralleled those used in the Database (DEC 2002) and in Bradstock and Kenny (2003), ie:
53 1. regime leads to local decline or extinction 2. persistent regime likely to lead to local decline or extinction 3. regime unlikely to lead to local decline or extinction.
3.2.7 Identifying key fire response species and thresholds of concern
Frequent fire
Juvenile period data were charted for species in each vegetation type. Species most likely to be sensitive to frequent fire, that is the obligate seeders in sensitivity categories 1 and 2 with the longest juvenile periods, were identified. The following decision rules were used in this process:
• Where the figure for juvenile period included a range of values (eg 3 - 5 years), the midpoint of the range was used. • Where the midpoint was not a whole number, and the data had been locally sourced, the figure was rounded down. This was considered appropriate as juvenile period figures had been previously been rounded up to whole years (Section 3.2.3). • Where the midpoint was not a whole number, and the data had been obtained from the NSW Database, the figure was rounded up. This was a conservative decision as these figures may reflect reports from inland areas where juvenile periods may be longer than on the near-coastal Cumberland Plain (Knox and Clarke 2004). • Where the figure for juvenile period was “> x”, x + 2 was used. • Where the figure for juvenile period was “< x”, x was used.
Infrequent fire
As very little local information was available, NSW Database figures for lifespan + seedbank were examined to identify species most likely to be at risk under long fire return intervals. DEC (2002) is quick to point out that these figures are often broad estimates, based on plant life form and structure, or in some cases, on post-fire age at which a species assumed or known to be part of a community, was no longer found. Seedbank longevity has also been estimated for many species in the Database, as follows: transient or canopy seedbank, 0 years; persistent seedbank, seeds without a hard seedcoat, 10 years; persistent seedbank, seeds with a hard seedcoat, 30 years (DEC 2002).
54 Again, charts were constructed for each vegetation type, and species most likely to be sensitive to infrequent fire, identified. Where the figure given in the Database for lifespan + seedbank was in the form “> x,” x +5 was used. Figures were rounded to the nearest multiple of five.
3.2.8 Threatened species
Species listed under the NSW Threatened Species Conservation (TSC) Act 1995 were identified for each vegetation type. Local observations and information from the NSW Database was used to outline the fire regime sensitivities of these taxa.
3.3 Results
In the initial paragraphs of the results section (3.3.1 to 3.3.5), the information presented refers to plants which may be found in either of the two vegetation types of interest. In later sections (3.3.6 to 3.3.9) findings pertaining to each vegetation type are distinguished.
3.3.1 Regeneration modes
Local information on post-fire regeneration mode was gathered for 95 shrub species. Appendix 4 details observations for each species. In 62 cases (65%), regeneration mode was determined from observations at at least three separate sites. Twenty-six species (27%) were classified as obligate seeders, while post-fire recruitment for a further five species usually relied on seedlings, though some resprouting was also observed. Sixty- four species (67%) were classified as resprouters.
3.3.2 Seedling lines
Ninety-four seedlings were initially tagged at Shanes Park, 80 at Castlereagh. Almost all seedlings at each site were of obligate seeder species. Exceptions were Acacia elongata and Indigofera australis at Shanes Park, and Platysace ericoides and possibly Kunzea capitata at Castlereagh. Two Banksia spinulosa var. spinulosa seedlings were tagged at Castlereagh, but they had both died by 14 months post-fire.
55 Seedlings deaths were high in the first 18 months post-fire, although timing varied between sites. Many seedlings died in the first 12 months after fire at Shanes Park, with few deaths in the second year, and none in the third (Table 3.1). Seedlings at Castlereagh were tagged at eight months post-fire, so only one monitoring visit fell within the first year after the burn; deaths at this visit were relatively few. Over the next six months, however, many seedlings died. Death rates declined sharply after 18 months post-fire, although occasional losses were recorded (Table 3.2). At each site, percentage losses are likely to have been higher in the wider population of seedlings than amongst tagged individuals, as tagged seedlings were generally chosen for their robustness; in other words, this was not a random sample of the seedling population.
Table 3.1. Percentage of tagged seedlings which died between visits at Shanes Park.
Visit number Months post-fire % deaths 1 9 25 2 12 29 3 15 2 4 18 3 5 20 1 6 23 0 7 26 0 8 29 0 9 32 0 10 34 0
Table 3.2. Percentage of tagged seedlings which died between visits at Castlereagh.
Visit number Months post-fire % deaths 1 11 6 2 14 58 3 17 24 4 20 4 5 22 3 6 25 3 7 28 8 8 31 3 9 34 0 10 36 2
56 No flowering of tagged seedlings was observed at either site in the first post-fire year, while flowering in year two was limited to five percent of seedlings at each site. Flowering in the third post-fire year was quite profuse at Shanes Park (Figure 3.2), but limited at Castlereagh. At Shanes Park, 62% of the 98 seedlings followed through to 34 months post-fire had flowered by the end of the study (Table 3.3). However at Castlereagh, only 16% of 65 seedlings followed to 36 months post-fire had flowered (Table 3.4).
Table 3.3. Post-fire flowering of seedlings at Shanes Park.
Taxon Total Number Number Number number of first first NOT seedlings flowering flowering 2- flowering at final visit 1-2 yrs 3 yrs post- by 3 yrs post-fire fire post-fire Acacia elongata 3 0 0 3 Acacia falcata 8 0 1 7 Cryptandra amara var. amara 5 0 5 0 Daviesia ulicifolia 13 0 1 12 Dillwynia tenuifolia 15 1 14 0 Grevillea juniperina ssp. 2 0 0 2 juniperina Indigofera australis 4 2 0 2 Pultenaea parviflora 36 2 30 4 Ozothamnus diosmifolius 12 0 5 7 Total 98 5 56 37
Table 3.4. Post-fire flowering of seedlings at Castlereagh.
Taxon Total Number Number Number number of first first NOT seedlings flowering flowering flowering at final visit 1-2 yrs 2-3 yrs by 3 yrs post-fire post-fire post-fire Cryptandra amara var. amara 1 0 1 0 Dillwynia rudis 20 0 0 20 Hakea sericea 12 0 0 12 Kunzea capitata? 1 0 0 1 Pimelea linifolia ssp. linifolia 26 3 7 16 Platysace ericoides 4 0 0 4 Unknown sp. 1 0 0 1 Total 65 3 8 54
57 Inter-specific differences were considerable. At Shanes Park, while all Dillwynia tenuifolia and Cryptandra seedlings had flowered by 34 months post-fire, no Acacia elongata or Grevillea juniperina seedling had done so. Some seedlings of all other species tagged at this site had flowered, ranging from a single Acacia falcata to 89% of the many Pultenaea parviflora seedlings (Table 3.3).
Of the three species represented by more than four seedlings at Castlereagh, Pimelea linifolia ssp. linifolia, Dillwynia rudis and Hakea sericea, only the Pimelea had flowered in any numbers by three years post-fire (Table 3.4). This pattern was replicated in the surrounding area. However profuse flowering in small Dillwynia rudis plants was observed in low-lying areas approximately 500 m away, just 23 months after fire. Some fruiting on Hakea sericea seedlings was also observed at a number of other Castlereagh sites which had not yet reached three years post-fire.
Figure 3.2. Shrubs flowering along the seedling line at Shanes Park, 34 months post-fire. Foreground, Ozothamnus diosmifolius; inset, Dillwynia tenuifolia, a species listed under the NSW Threatened Species Conservation (TSC) Act 1995.
58 3.3.3 Juvenile periods
When data from the seedling lines were combined with other observations, local information on primary juvenile period was available for 31 shrub species. In 42% of cases, data came from on-going monitoring, in an additional 39%, from post-fire observations in at least two separate sites.
Local information on secondary juvenile period was gathered for 58 shrub species. For 28 of these species (48%), secondary juvenile period was determined from observations in at least three separate sites. Observations with respect to an additional ten species (17%) encompassed two sites.
Appendix 5 details juvenile period observations for each species. A degree of intra- species variation in primary juvenile period between sites, and between areas within sites, was widely observed across the Plain. Seedlings appeared to flower more rapidly where resources might have been expected to be more available, for instance on lower slopes and in open patches where competition from trees and resprouting plants was relatively low.
3.3.4 Comparison with NSW Database
Primary regeneration mode from Cumberland Plain observations corresponded with that in the NSW Database for 67 (71%) of the 95 species for which local information was available, and differed for 13 (14%). In eight of these 13 species, however, the Database listing included a secondary mode, indicating some records from around the state accorded with Cumberland Plain observations. Seven species for which local observations had been made were not included in the NSW Database, while the Database listing for a further eight taxa was not definitive (ie SR, indicating regeneration mode has been reported to vary).
Data on primary juvenile period were available for 31 Cumberland Plain shrub species; the NSW Database provided information on this variable for 13 (42%) of these. In all but one case the primary juvenile period based on local observations either fell within the range given in the NSW Database, or was within one year of it. The exception was Pomax umbellata.
59 Secondary juvenile period data were gathered for 58 Cumberland Plain shrub taxa; for 31 (53%) of these, information was also available in the NSW Database. There was a discrepancy of more than one year between the two sources in only two cases. Both were myrtaceous species, Leptospermum trinervium and Melaleuca thymifolia. For two further myrtaceous taxa, the Database figure was “< x”, while local information was more specific.
Comparison data for individual species can be found in Appendix 6.
3.3.5 Assigned vital attributes
Seven species were assigned to the ‘T’ establishment category (T species are able to establish at any time between fires), while all the rest were categorised ‘I’ (able to establish only in the immediate post-fire period). For four of the T species – Allocasuarina littoralis, Banksia oblongifolia, Breynia oblongifolia and Hakea sericea – this vital attribute was transferred from the NSW Database. The remaining three species – Bursaria spinosa, Grevillea juniperina ssp. juniperina and Styphelia laeta ssp. laeta – were awarded this designation on the basis of local observation. Mixed size classes of B. spinosa were often observed in infrequently burnt areas, while G. juniperina can be seen regenerating in many parts of the large, unburnt development site at St Mary’s. S. laeta was found in sites which had not burnt for several years, but not in nearby recently burnt areas, suggesting this species establishes progressively post-fire, perhaps through animal dispersal.
3.3.6 Characterisation of broad vegetation types
More shrub species occur in Castlereagh woodlands than in Cumberland Plain woodlands, according to James et al. (1999). Twenty-three shrub species are frequent, common or very common in Cumberland Plain woodlands; 37 in Castlereagh woodlands (Table 3.5).
60 Table 3.5. Number of shrub species listed by James et al. (1999) in various frequency categories.
Frequency Cumberland Plain woodlands Castlereagh woodlands Very common 2 2 Common 7 9 Frequent 14 26 Occasional 44 53 Rare 38 49 Total 105 139
Appendix 7 details regeneration modes, vital attributes, plant functional types, sensitivities, juvenile periods, longevity, conservation status and family for shrub species in each broad vegetation type. In the table for each vegetation type, two groups of shrubs are distinguished on the basis of their relative frequency. I will use the term ‘more common species’ to refer to the first group (those listed by James et al. 1999 as ‘very common’, ‘common’ or ‘frequent’). The second group (‘occasional’ species plus ‘rare’ species listed under the TSC Act) I will call ‘less common species’.
Regeneration modes
Local information on regeneration mode was available for a majority of species in each vegetation type. This was particularly the case for the more common species: local information was available for over 90% of these taxa (Table 3.6).
Table 3.6. Number and percentage of species in two broad vegetation types for which local information on regeneration mode was available.
Local information Cumberland Plain woodlands Castlereagh woodlands available? more common less common more common less common species species species species Yes 21 (91%) 25 (57%) 35 (95%) 36 (64%) No 2 (9%) 19 (43%) 2 (5%) 20 (36%) Total 23 44 37 56
While resprouting species were in the majority in each vegetation type, the obligate seeder complement was also substantial. Forty-eight percent of more common shrub species in Cumberland Plain woodlands were entirely, or primarily, obligate seeders, as were at least 41% of the less common taxa in this vegetation type. Equivalent figures
61 for Castlereagh woodlands were 32% (more common species) and 45% (less common species). A relatively high percentage of Cumberland Plain woodland obligate seeders were occasional resprouters (Table 3.7).
Table 3.7. Number and percentage of species in two broad vegetation types exhibiting various regeneration modes. S, obligate seeder; R, resprouter; r, usually killed by fire but sometimes resprouts; s, usually resprouts but sometimes killed; SR, S and R modes equally reported.
Regeneration Cumberland Plain woodlands Castlereagh woodlands mode more common less common more common less common species species species species S 6 (26%) 13 (30%) 9 (24%) 20 (36%) Sr 5 (22%) 5 (11%) 3 (8%) 5 (9%) R 11 (48%) 20 (45%) 24 (65%) 24 (43%) Rs 1 (4%) 3 (7%) 1 (3%) 3 (5%) SR 0 (0%) 1 (2%) 0 (0%) 1 (2%) Unknown 0 (0%) 2 (5%) 0 (0%) 3 (5%) Total 23 44 37 56
Families
Table 3.8 lists the dominant shrub families in the two broad vegetation types. The leguminous families, Fabaceae and Mimosaceae, have large numbers of species in each. The Proteaceae and Myrtaceae families are both represented by considerably more species in Castlereagh than in Cumberland Plain woodlands.
Table 3.8. Number of shrub species in various families in two broad vegetation types.
Family Cumberland Plain woodlands Castlereagh woodlands more common less common more common less common species species species species Asteraceae 1 4 2 0 Dilleniaceae 1 2 1 4 Epacridaceae 3 1 3 4 Fabaceae 6 9 8 19 Mimosaceae 3 6 4 5 Myrtaceae 0 4 7 4 Proteaceae 2 2 4 8 Other 7 16 8 12 Total 23 44 37 56
62 Plant functional types
Shrub species in each vegetation type were heavily concentrated in two plant functional types, groups 4 and 12. These types encompass obligate seeders and resprouters with persistent seedbanks, which recruit only after fire (see Appendix 3 for more detail). Seventy-four percent of the more common Cumberland Plain woodlands species belonged to one of these two groups, as did 51% of less common species. Common Castlereagh woodlands shrub taxa were somewhat more diverse, although 51% still fell into these two groups, as did 72% of less common Castlereagh species. Groups 5 and 7 were the next most populous groups: each of these groups was represented by between four and 11 species in each vegetation type (Table 3.9).
Table 3.9. Number of species in two broad vegetation types belonging to various plant functional types (PFTs). See Appendix 3 for an explanation of these groups.
PFT Cumberland Plain woodlands Castlereagh woodlands more common less common more common less common species species species species 1 1 3 1 1 2 1 2 2 1 3 0 3 0 1 4 10 16 7 29 5 4 5 4 7 6 0 2 1 1 7 0 4 8 2 8 0 0 0 0 9 0 0 0 0 10a 0 1 0 0 10b 0 0 0 1 11 0 0 0 0 12 7 7 12 11 13 0 0 0 0 14 0 1 2 2 Total 23 44 37 56
63 Fire regime sensitivities
Collation of fire regime sensitivities indicated that shrub species in Cumberland Plain and Castlereagh woodlands are highly sensitive to infrequent fire. Most species are also sensitive, though to a lesser degree, to frequent burning.
About 20% of species in each broad vegetation type were categorised as highly vulnerable to frequent fire (Table 3.10), while the large majority were moderately sensitive (vulnerable to local decline or extinction under persistent frequent burning). Few species were considered invulnerable to persistent short intervals.
Over 90% of common shrub species in each vegetation type were categorised as highly vulnerable to infrequent fire (Table 3.11), as were the large majority of less common species (Cumberland Plain woodlands, 80%; Castlereagh woodlands, 93%). A small number of species in each vegetation type were coded insensitive to infrequent fire. In Cumberland Plain woodlands, two common species, Bursaria spinosa and Hakea sericea, fell into this category, as did six less common species (Table A7.1 in Appendix 7).
Table 3.10. Number of species in two broad vegetation types exhibiting various levels of sensitivity to frequent fire. 1, frequent fire leads to local decline or extinction; 2, persistent frequent fire likely to lead to local decline or extinction; 3, frequent fire unlikely to lead to local decline or extinction.
Sensitivity to Cumberland Plain woodlands Castlereagh woodlands frequent fire more common less common more common less common species species species species 1 5 9 7 9 2 17 31 29 46 3 1 4 1 1 Total 23 44 37 56
64 Table 3.11. Number of species in two broad vegetation types exhibiting various levels of sensitivity to infrequent fire. 1, infrequent fire leads to local decline or extinction; 2, persistent regime of infrequent fire likely to lead to local decline or extinction; 3, infrequent fire unlikely to lead to local decline or extinction.
Sensitivity to Cumberland Plain woodlands Castlereagh woodlands infrequent fire more common less common more common less common species species species species 1 21 35 34 52 2 0 3 0 1 3 2 6 3 3 Total 23 44 37 56
3.3.7 Key fire response species in Cumberland Plain woodlands
Fourteen Cumberland Plain woodlands shrub species were classified as highly sensitive to frequent fire. Information on primary juvenile period was available for nine (64%) of these. Forty-eight species were potentially sensitive to repeated short intervals. Primary juvenile period data were available for 15 (83%) of the 18 obligate seeders in this group. Primary juvenile period data were also available for 12 resprouters. Secondary juvenile period data were available for 34 Cumberland Plain woodlands species.
Primary juvenile periods for Cumberland Plain woodlands shrubs ranged from 1 to 5 years (Figure 3.3). Obligate seeder shrubs in the most sensitive category first flowered between 1 and 4 years post-fire, with a mode of two years. Grevillea juniperina ssp. juniperina had the longest juvenile period of this group. Species sensitive to repeated short interfire intervals first flowered between 2 and 5 years post-fire (mode also two years); those taking the longest to reach reproductive maturity were Acacia decurrens and Dodonaea triquetra.
Almost all Cumberland Plain woodlands shrub species had a secondary juvenile period of between 1 and 3 years (Figure 3.4). The secondary juvenile period for the single resprouting species charted as first flowering at five years post-fire, Exocarpos cupressiformis, was taken from the NSW Database where it was given as >3 years.
65 16
14
12
10
8
6
Number of shrub species 4
2
0 12345 Primary juvenile period (years)
Figure 3.3. Primary juvenile periods of shrubs in Cumberland Plain woodlands. Cream, obligate seeders highly sensitive to frequent fire; green, obligate seeders moderately sensitive to frequent fire; maroon, resprouters.
16
14
12
10
8
6
Number of shrub species 4
2
0 12345 Secondary juvenile period (years)
Figure 3.4. Secondary juvenile periods of shrubs in Cumberland Plain woodlands. Maroon, resprouters; green, obligate seeders which sometimes resprout.
Longevity estimates were available for 29 (52%) of the 56 shrub species categorised as highly sensitive to infrequent fire in Cumberland Plain woodlands. Estimates of lifespan + seedbank longevity for highly sensitive species ranged from 20 to 150 years, with clusters of taxa around 35, and 55 to 60, years (Figure 3.5). The species with the shortest estimated survival time were Cassinia arcuata (20 years) and Indigofera australis (30 years). The 50 to 60 year cluster included the obligate seeder legumes Acacia falcata, Daviesia ulicifolia, Dillwynia sieberi, Dillwynia tenuifolia and Pultenaea parviflora.
66
9
8
7
6
5
4
3 Number of species
2
1
0 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90+ Estimated lifespan + seedbank (years)
Figure 3.5. Estimated longevity (lifespan + seedbank) of shrubs in Cumberland Plain woodlands. Cream, species highly sensitive to infrequent fire; green, species moderately sensitive to infrequent fire; maroon, species insensitive to infrequent fire.
3.3.8 Key fire response species in Castlereagh woodlands
Sixteen species were categorised as highly sensitive to frequent fire; primary juvenile period data were available for 13 (81%) of these. Seventy-five shrub species, twenty- five of them obligate seeders, were potentially sensitive to persistent frequent burning. Primary juvenile period data were available for 18 (72%) of these obligate seeders. Secondary juvenile period data were available for 52 Castlereagh woodlands species.
Castlereagh woodlands shrubs exhibited a wider range of primary juvenile periods than those in Cumberland Plain Woodlands. High sensitivity obligate seeders had juvenile periods of between 1 and 6 years, with two and three years the modal categories (Figure 3.6). Petrophile pulchella had the longest juvenile period in this group: the figure of six years for this species reflects both local findings (one record of flowering at less than three years post-fire) and the Database listing for this species (4 to 9 years). Partially sensitive obligate seeder species had juvenile periods between 2 and 5 years, with a mode of two years. Persoonia lanceolata and Acacia decurrens were the slowest maturing species in this group. Three resprouting species had very long primary juvenile periods – Banksia oblongifolia (5-10 years), Isopogon anemonifolius (10 years), and Macrozamia spiralis (10-20 years) – these figures were all taken from the NSW Database.
67 Secondary juvenile period for all Castlereagh shrub species except one was three years or less (Figure 3.7). Again the outliner was Exocarpos cupressiformis.
16
14
12
10
8
6
Number of shrub species 4
2
0 123456789101112131415 Primary juvenile period (years)
Figure 3.6. Primary juvenile periods of shrubs in Castlereagh woodlands. Cream, obligate seeders highly sensitive to frequent fire; green, obligate seeders moderately sensitive to frequent fire; maroon, resprouters.
30
25
20
15
10 Number of shrub species
5
0 12345 Secondary juvenile period (years)
Figure 3.7. Secondary juvenile periods of shrubs in Castlereagh woodlands. Maroon, resprouters; green, obligate seeders which sometimes resprout.
Longevity estimates were available for 49 (57%) of the 86 shrub species categorised as highly sensitive to infrequent fire in Castlereagh woodlands. In this vegetation type, longevity estimates ranged from 30 to 150 years (Figure 3.8). Indigofera australis was the species with the shortest estimated lifespan (30 years); other short-lived species included Grevillea mucronulata and Pomax umbellata (each 35 years). Many leguminous species were estimated to survive, at least as seeds, for 50 to 60 years. The lifespan of the CI species Petrophile pulchella was also estimated at 60 years.
68 12
10
8
6
4 Number of species
2
0 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90+ Estimated lifespan + seedbank (years)
Figure 3.8. Estimated longevity (lifespan + seedbank) of shrubs in Castlereagh woodlands. Cream, species highly sensitive to infrequent fire; green, species moderately sensitive to infrequent fire; maroon, species insensitive to infrequent fire.
3.3.9 Threatened species
Five shrub species listed under the NSW Threatened Species Conservation (TSC) Act 1995 occurred in Cumberland Plain woodlands (James et al. 1999): Acacia pubescens, Pimelea spicata, Dillwynia tenuifolia, Pultenaea parviflora and Grevillea juniperina ssp. juniperina. The first two taxa were classified as resprouters, the others as obligate seeders. All five were considered highly sensitive to long interfire intervals, and at least moderately sensitive to frequent fire. Local information on primary juvenile period for the two leguminous obligate seeders was collected from tagged seedlings as well as from observation in other sites: flowering in these species began by two years post-fire, and was well established by three years after a burn. Flowering in Grevillea juniperina ssp. juniperina was observed in numerous plants 3.1 and 4.7 years post-fire (Appendix 5). No information on primary juvenile period was collected during the current project for the resprouters, however the NSW Database gave a figure of two years for P. spicata, and 3-5 years for A. pubescens. P. spicata flowered within the first year after fire, on resprouts (Appendix 5). The NSW Database provided longevity estimates for P. parviflora (55 years), D. tenuifolia (60 years) and A. pubescens (80 years). These figures include an estimate of 30 years for seedbank longevity (DEC 2002). Local observation, however, suggested that the lifespans of the two obligate seeders may be shorter than 25-30 years: many dead and dying D. tenuifolia plants were observed ten
69 years post-fire at Scheyville, and both species were senescent at Shanes Park in an area whose time-since-fire was unknown, but where arson was common.
Castlereagh woodlands also hosted the three legume species and the Grevillea discussed above, plus an additional four listed species: Acacia bynoeana, Allocasuarina glareicola, Micromyrtus minutiflora and Persoonia nutans. The first two species were classified as resprouters, the Micromyrtus and Persoonia as obligate seeders. All were considered highly vulnerable to long interfire intervals. The obligate seeders were also considered highly vulnerable to frequent fire, the resprouters moderately so. Persoonia nutans was listed in the Database as a GI species (but see Section 3.4.7). A single local observation placed the juvenile period of this taxon at approximately four years. Micromyrtus minutiflora was observed flowering by two years post-fire; profuse flowering was observed 4.7 years after a burn (Appendix 5). No information on primary juvenile period was available for either of the resprouters. Acacia bynoeana was observed flowering on resprouts 14 months and 2.2 years post-fire (Appendix 5). The secondary juvenile period figure for Allocasuarina glareicola in the Database was >3 years; this species was not encountered during the current study. The Database provided a longevity estimate only for A. glareicola (100 years).
3.4 Discussion 3.4.1 Reliability and availability of data
While some data on primary juvenile periods were gathered through monitoring tagged plants, time precluded extensive use of this strategy. Rather, post-fire observation in sites across the Plain was used to gather information on life history characteristics. For most species, evidence for their regeneration mode and juvenile period(s) came from observations across multiple sites. Previous research suggests post-fire observations can provide reliable data: Bradstock and Kenny (2003) found much common ground between juvenile period data derived from anecdotal sources, and published demographic studies. The finding that local observations closely agreed with records in the NSW Database (DEC 2002) provides confidence in both data sets.
70 Regeneration mode is known to vary across vegetation types and geographic regions, for a number of species (Sandercoe 1989, Williams et al. 1994, Benwell 1998). It is therefore not surprising that local observations for a small number of species did not accord with Database records. More often than not, where differences did occur, the possibility of variability in regeneration mode was reflected in the Database record, through listing of a secondary mode.
Discrepancy with respect to juvenile periods was minimal: only in a handful of cases did local records fall outside Database ranges by more than a year. Where discrepancies were identified, they were as likely to reflect differences in method as substance. The primary juvenile period for Pomax umbellata was given as <1 year in the Database. The figure of two years for this species on the Cumberland Plain was rounded up from observations of fruiting at 13, 18, 20 and 23 months post-fire: the difference in this case was therefore a function of the rounding convention. Leptospermum trinervium and Melaleuca thymifolia were listed as flowering on resprouts by 0.5 years post-fire in the NSW Database, while local observations suggested flowering did not occur in the first post-fire year: local figures were 3, and 2-3 years, respectively. Again the discrepancy owed something to the rounding conventions adopted in the current study. It may also reflect dry post-fire conditions or inadequate local observation.
While within a year of the range in the Database once it was rounded up, the one local observation for the primary juvenile period of Petrophile pulchella (observed flowering at 2.8 years post-fire versus a Database range of 4-9 years), suggests that this species may reach reproductive maturity more rapidly in the Castlereagh area than on sandstone.
While the NSW Database provided information for many species, local observations added considerably to the data available: information on primary juvenile period was gathered for an additional 18 species, while data on secondary juvenile period were added for 27 taxa. Local information on regeneration mode was available for a large majority of species in each vegetation type. When primary juvenile period data from the two sources were combined, information on this variable was available for approximately three-quarters of taxa in the groups most sensitive to frequent fire.
Overall, data needed to identify critical thresholds with respect to frequent fire were relatively available, and likely to be reliable.
71 Data needed to identify critical thresholds for infrequent fire, however, were much more limited. Local information on longevity of adult plants and seedbanks was rarely available. Database records reflected estimates rather than solid data on individual species. Even then, estimates were available for only half the species rated as highly sensitive to infrequent fire in each vegetation type. This situation reflects the relative paucity of lifespan data; ability to draw useful conclusions on upper fire frequency thresholds is correspondingly limited.
3.4.2 Post-fire seedling establishment
Shrub seedlings were common in the post-fire environment, reflecting the ‘I’ vital attribute status of most local species (I species recruit only after fire). Almost all tagged seedlings were from obligate seeder species: obligate seeders generally produce more post-fire seedlings that resprouters (Wark et al. 1987, Benwell 1998, Enright and Goldblum 1999). Resprouting shrub species were present along both seedling lines, however seedlings were only found for a few of them, and then in small numbers. Obviously resprouting species have less need to recruit than obligate seeders. Some recruitment is necessary, however, to replace adult losses, raising the possibility that some of these species may recruit between fires.
Seedling death rates were considerable in the first 12-18 months post-fire, but minimal after that, despite unusually dry weather in the Winter of 2003, and again in the Autumn and Winter of 2004 (Section 2.3.2). Other studies, both after fire and in its absence, have also found high levels of seedling mortality (Benson 1985, Clarke 2002, Moles and Westoby 2004). All obligate seeder species, however, were represented by a healthy population of surviving plants either on the seedling line, or in the adjacent area, at the end of the study approximately three years after fire. This result contrasts with the complete demise of obligate seeder seedlings sown into unburnt woodland on the New England Tablelands (Clarke 2002), suggesting that the post-fire environment is relatively conducive to seedling survival.
72 3.4.3 Primary juvenile periods of obligate seeders
Primary juvenile periods for Cumberland Plain obligate seeder shrubs ranged from 1 to 6 years. The Cumberland Plain does not support the large dominant obligate seeding species found in Sydney’s sandstone heaths and woodlands. Juvenile periods of Cumberland Plain obligate seeders were all lower than those recorded by Benson (1985) for tagged seedlings of sandstone dominants, for example 7 to 8 years for Banksia ericifolia, and 6 to 7 years for Hakea teretifolia.
The intra-species variation in time to first flowering found in this study has been found elsewhere; availability of resources, particularly water, appears to be a key factor (Bradstock and O’Connell 1988, Burrows and Wardell-Johnson 2003). As rainfall was relatively low throughout the study period (Section 2.3.2), local juvenile period data used in the analysis presented in this chapter are likely to reflect the upper end of natural variation in this parameter. Conclusions based on these data are therefore likely to be fairly conservative.
3.4.4 Characteristics of the Cumberland Plain shrub flora
Castlereagh woodlands support more shrub species than Cumberland Plain woodlands; this finding is in line with the general trend in woodlands towards greater shrubbiness with increasing sandiness (Section 1.2.2). In fact, figures for shrub species richness in strictly-defined Cumberland Plain Woodland would probably be even lower than the figures in Section 3.3.6, as the vegetation type boundaries in the study are comparatively broad.
In terms of number of species, legumes dominate both Cumberland Plain and Castlereagh woodlands. The range of species in the Proteaceae, Myrtaceae and Epacridaceae families, though greater in Castlereagh than in Cumberland Plain woodlands, does not approach that found in Sydney’s sandstone areas (Robinson 2003). Additionally, almost all Cumberland Plain species in these families resprout. Proteaceous obligate seeders, where they exist, differ from many of their sandstone counterparts in not being serotinous. The exception is Petrophile pulchella, which occurs at low frequency in sandy Castlereagh Scribbly Gum sites, but not on the finer- textured soils in Castlereagh Ironbark Forest (Tozer 2003). A similar pattern is reported
73 by Bond (1997) in South Africa, where the dominant fynbos species are often serotinous obligate seeding taxa in the Proteaceae family. Protea species on the savannas, however, are not serotinous, and post-fire regeneration in savanna trees and shrubs is mostly by resprouting.
While most shrub species resprout, the extent of obligate seeding amongst Cumberland Plain shrub taxa – around 40% in each vegetation type – is not inconsiderable. Resprouting is, however, the primary regeneration mode in both the tree and herb strata, so the proportion of obligate seeders in each vegetation type as a whole would be considerably less than 40%. It has been suggested that the percentage of obligate seeders in a vegetation type may provide a rough indication of appropriate fire frequencies. In some places, high proportions of obligate seeder species tend to occur in areas where fire is a relatively rare event, such as on rocky outcrops and in sheltered gullies, while higher proportions of resprouters are found in ecosystems where burns are more frequent (Hunter 1998, Burrows et al. 1999, Clarke and Knox 2002, though see Lloret et al. 1999). The nature of the obligate seeder species in the vegetation type may have a bearing on this question (Hunter 2003).
Cumberland Plain obligate seeders are overwhelmingly legumes in plant functional type 4. Noble and Gitay (1996:333) note that group 4 encompasses “Early successional species that regenerate from long-lived seed pools,” and add that “many understorey shrubs of recurrently burnt communities are of this species type.” Leguminous ‘fire weed’ obligate seeders are often short-lived (Bell and Koch 1980, Auld 1987, McCaw 1988). This distinguishes them from the long-lived, ‘persistence’ obligate seeders that dominate granite outcrops in the New England region (Hunter 2003).
Type 5 obligate seeders have soil-stored seedbanks exhausted by a single disturbance; they are thus particularly vulnerable to frequent fire. Nine, and 11 species were allocated to this group, in Cumberland Plain and Castlereagh woodlands respectively. However numbers in type 5 were somewhat inflated by the decision rules for species for which information was not available, and by species more commonly found in shale- sandstone transition areas (see Section 3.4.6).
Type 6 species, obligate seeders with canopy-stored seed, form a very minor component of the two vegetation types.
74 Plant functional types in Castlereagh woodlands were somewhat more diverse that those in Cumberland Plain woodlands. Castlereagh shrubs also exhibited a wider array of juvenile periods than Cumberland Plain woodlands species. Both these factors reflect the more diverse shrub flora in the Castlereagh vegetation type.
Finally, the shrub complement of both vegetation types is characterised by its extreme vulnerability to infrequent fire: only a very few species are coded as insensitive to long interfire intervals. Vulnerability to frequent fire is also considerable, although for the majority of species vulnerability at this end of the fire interval domain relates to recurrent short intervals, rather than to a single ‘double burn’.
3.4.5 Vital characteristics and fire interval domains
What implications for fire interval domains can, or should, be drawn from findings with respect to the vital attributes of interval-sensitive species?
The longest time to maturity (x) amongst taxa classified as highly sensitive to frequent fire is clearly a key variable in determining the shortest tolerable interfire interval, or lower threshold. Some authors suggest that the lower threshold should be set higher than x, for example at 2x, or x+3, on the grounds that species sensitive to frequent fire need time to accumulate a seedbank (Gill and Nicholls 1989, Bell 2001, Burrows and Wardell-Johnson 2003). However both the Victorian (Friend et al. 2003) and the NSW guidelines (NPWS 2004c) take x as the minimum interval, within the context of a clear understanding that intervals should vary. In particular, the NSW guidelines stress that communities should not be burnt repeatedly at the minimum interval, and that any intervals at or below the minimum should be followed by a longer interval, to allow seedbank reserves to accumulate. The practice of setting minimum intervals at, or just above, x, recognises that short intervals within a variable regime play an important role in maintenance of biodiversity in some ecosystems (Cowling and Gxaba 1990, Yeaton and Bond 1991, Keith and Bradstock 1994, Keith et al. 2002b, Tozer and Bradstock 2002).
While the nature and implications of the vulnerabilities of plant functional types have been clearly articulated for species categorised as highly vulnerable to particular fire regimes, this is not the case for groups considered partially sensitive. Particularly relevant in the context of the current study are PFTs 4, 7 and 12, as a large majority of
75 Cumberland Plain shrub species fall into these groups. All three PFTs were considered vulnerable only to infrequent disturbance by Noble and Slatyer (1980), whereas Bradstock and Kenny (2003) consider them also vulnerable to local extinction or severe decline if short intervals persist.
PFTs 4, 7 and 12 cover eight combinations of vital attributes, seven of which are resprouters. All combinations recruit only after fire. The single obligate seeder combination, SI, occurs in PFT 4. These species have soil-stored seedbanks which are not exhausted by a single disturbance. For example, six out of seven leguminous obligate seeder species were still surviving in reasonable numbers after three fires within three years in the sandstone site studied by Bradstock et al. (1997). However continued interfire intervals below the primary juvenile period can be expected to exhaust seedbanks. Thus a good proportion of intervals (but not all) will need to be longer than the juvenile periods of these species.
All other species in groups 4, 7 and 12 resprout. Extent of vulnerability to frequent fire will be influenced by survival rates of existing plants, whether seedbanks are exhausted by a single disturbance, whether recruitment occurs vegetatively and/or through seeds, length of secondary juvenile period, and time to fire tolerance of juveniles. Site and climatic factors will affect some of these variables. This picture is sufficiently complex to suggest that attempts at quantification are unlikely to be useful for defining fire interval domains, particularly as key fire response species for frequent fire are more likely to be drawn from other PFTs. It may be desirable to ensure that, say, no more than two consecutive interfire intervals fall within these species secondary juvenile period, and that some intervals are relatively long, in order to give seedlings time to attain fire tolerance.
The minimum lifespan, including that of propagules (y) amongst taxa sensitive to a long interfire interval, is a key variable in determining the longest tolerable interfire interval, or upper threshold. In Victoria (Friend et al. 2003) and NSW (NPWS 2004c), the upper threshold is set at y. Burrows and Abbott (2003:446), however, suggest 0.75y may be more appropriate, and I am inclined to agree. Population decline, both above and below ground, may occur over a long period prior to the point of local extinction (Bond 1980). Flowering may peak in the years following the end of the juvenile period. McFarland (1990) found flowering and seeding in south-east Queensland’s wallum
76 heath peaked at four to eight years after a burn, and dropped markedly by 11 years post- fire. Fruiting in the leguminous sandstone obligate seeder Acacia suaveolens peaks 2-5 years post-fire (Auld and Myerscough 1986). This species is short-lived: seedbank numbers peak at six years post-fire then decline (Auld 1987). Morrison et al. (1996) found a significantly reduced abundance of A. suaveolens in sites with interfire intervals greater than seven years. Upper thresholds may also need to take into account competitive interactions between species (Huston 2003). This issue will be further explored in Chapter 10.
3.4.6 Fire interval domain in Cumberland Plain woodlands
The longest time to maturity amongst shrub species highly sensitive to frequent fire, for species for which information was available, was four years. Using the NSW convention, the shortest tolerable interfire interval for retention of shrub species in this vegetation type would therefore be four years. The key fire response species here is a ‘listed’ species, Grevillea juniperina ssp. juniperina. Time to maturity varied between sites for this species, and may be related to rainfall. Managers could monitor the progress of seedlings over the initial post-fire years, and adjust minimum intervals accordingly.
No data were available for five species classified as highly sensitive to frequent fire. Investigation of these species revealed that three of them were not recorded in any Shale Hills, Shale Plains or Shale-Gravel Transition Forest sites surveyed by Tozer (2003). Leucopogon juniperinus was found in some shale sites (approximately 10% frequency), but was more common in shale-sandstone transition areas. Olearia microphylla occurred in Shale-Gravel Transition Forest, but not in Shale Hills or Shale Plains Woodland (Tozer 2003). The four year threshold is therefore considered well-founded for Cumberland Plain Woodland, with the proviso that monitoring should include Leucopogon juniperinus where this occurs. The other two ‘listed’ obligate seeders in CPW both flower profusely by three years post-fire, so a four-year lower threshold would meet their needs. Since the key fire response species in relation to fire frequency is unlikely to be a herb or a tree (Section 3.1.3), this minimum is probably appropriate for the community as a whole.
77 Two Cumberland Plain woodland shrub species with maturation times of five years were considered vulnerable to persistent short intervals. Both these species are capable of massive germination after a fire following a moderately long interfire interval (Acacia decurrens: M. Peterson pers. comm. and pers. obs. Holsworthy Military Base, 2004; Dodonaea triquetra: Floyd 1966, pers. obs Castlereagh Nature Reserve 2004). In addition, Acacia decurrens may survive low to moderate intensity fire through avoiding 100% leaf scorch (pers. obs., Lansdowne, 2004). Secondary juvenile periods for Cumberland Plain woodlands shrubs were almost all below three years, although one resprouting species was reputed to take five years to flower. Variable intervals across space and time, including some well above five years, should ensure retention of both sets of species.
The shrub taxa with the shortest estimated lifespans were Cassinia arcuata (20 years) and Indigofera australis (30 years). Cassinia arcuata was not found in shale sites in Tozer’s study (Tozer 2003). It seems reasonable to suppose, however, that the group of species for which estimates are not available may include some shrubs with similar lifespans. Using a figure of 20 years for y, and the Burrows and Abbott (2003) convention (0.75y), the longest tolerable interfire interval for retention of shrub species in Cumberland Plain woodlands would be 15 years. A precautionary approach to maximum interfire intervals for Cumberland Plain woodlands shrubs is appropriate because of the paucity of quantitative life history information, and because of the very high percentage of species at risk of local extinction in the face of long interfire intervals. In particular, all ‘listed’ species are highly vulnerable to local extinction if intervals become too long. Local observations suggest that populations of two of these species, Dillwynia tenuifolia and Pultenaea parviflora, as well as some other Cumberland Plain shrubs (eg Ozothamnus diosmifolius) may begin to senesce less than ten years post-fire. Considering these factors, it seems reasonable to conclude that the maximum threshold to conserve the shrub flora of Cumberland Plain woodlands should not exceed 15 years. The threshold for the community as a whole might need to be lower, if short-lived herb species sensitive to low fire frequency, are identified. Competition issues may also need to be considered.
78 3.4.7 Fire interval domain in Castlereagh woodlands
Petrophile pulchella was the key fire response species for frequent fire in Castlereagh woodlands. Detailed demographic work on this species in Sydney’s sandstone woodlands has been carried out (Bradstock and O’Connell 1988). In this environment, flowers did not start to emerge until four years post-fire, a year later than the single local observation. Seedling emergence was higher after more intense fires, as cones opened more quickly, releasing seeds at a time when viability was relatively high. Bradstock and O’Connell (1988) concluded that intense summer fires at 8-10 year intervals would be unlikely to send this species into permanent decline, although less intense fires might be problematic; populations would be stable under 10-15 year intervals regardless of season or intensity, and that populations would increase greatly after fires at 15-20 year intervals. Variable intervals between six and 20 years could therefore be expected to allow this species to thrive across the landscape.
No juvenile period data were available for three species rated as highly sensitive to frequent fire. One species was not found by Tozer (2003) at any relevant sites. Olearia microphylla, however, was common in Castlereagh Ironbark Forest (43% frequency), while Styphelia laeta ssp. laeta was occasionally found on tertiary alluvium (Tozer 2003). There is no reason to suppose that a minimum interval of six years would not be appropriate for these species.
Two ‘listed’ species were considered highly vulnerable to frequent fire. Micromyrtus minutiflora matures well before six years post-fire. Less is known, however, about Persoonia nutans. One local observation of fruiting was made in the current study, 4.8 years after a burn. Individuals of 30-60 cm can produce fruits (Robertson et al. 1996), suggesting juvenile periods are probably not long.
Persoonia nutans may operate in a somewhat different manner in relation to fire than do most other shrubs in Castlereagh woodlands. A survey by Robertson et al. (1996) found stands often contained individuals of different sizes. Concentrations of plants were found in heavily disturbed areas, although the species tended to occur as scattered individuals in intact bush (Robertson et al. 1996). High densities of plants have recently been reported from a weedy Cumberland Plain site after sand-mining (D. Wotherspoon pers. comm. 2004). These observations suggest P. nutans may recruit between fires from fruits dispersed across the landscape by animals, making it a DT
79 species. The ecology of obligate seeder DT species in fire-prone environments is not well studied; they are uncommon, and likely to employ somewhat different strategies to the majority of obligate seeders. They may take some time to build up population numbers after a fire, as seeds deposited on the soil surface are unlikely to survive a burn, making regeneration reliant on occasional buried seed or on seeds arriving from unburnt areas (French and Westoby 1996). Dispersal distance and timing will depend on when and how far vectors travel into, or fly over, areas of various post-fire ages. The extent of patchiness in fires may therefore influence abundance of these species.
Since they have the capacity to continue to recruit throughout the interfire period, DT species have the potential to do very well in those parts of the landscape where time- since-fire is long (McMahon et al. 1996). Their distribution is likely to be patchy, and linked to time-since-fire. Their success in keeping their place in a community would thus need to be assessed across fire boundaries. The relationship of Persoonia nutans to time-since-fire, and fire extent, requires further investigation. This is particularly important given the potential tension between the need to ensure fire-free periods/patches where P. nutans populations can build up, and the need to ensure fires are sufficiently frequent to meet the requirements of other vulnerable Castlereagh woodland species. Given the propensity of P. nutans to establish in cleared areas, fire may play a role in its lifecycle by providing gaps for seedling establishment; this may only be able to occur in the early post-fire years, and may be of increasing importance as other sources of disturbance are removed from reserves. Fire may also help keep country open for vectors.
No Castlereagh obligate seeders regarded as partially sensitive to frequent fire had juvenile periods above that of P. pulchella, and known secondary juvenile periods were all below six years.
Indigofera australis was the species with the shortest estimated lifespan (30 years). The Burrows and Abbott (2003) convention gives a maximum tolerable interfire interval of 23 years. Again, the paucity of data and the extreme sensitivity of shrub species in this vegetation type to infrequent fire (over 90% of species, including all listed species, were rating highly sensitive) argue for a precautionary approach to upper thresholds. And again, short-lived herbs and competition factors may need to be considered, if and when they come to light.
80 3.5 Conclusions
From this examination of shrub species vital attributes it can be concluded that: • The minimum interfire interval in Cumberland Plain woodlands should be no lower than four years • The minimum interfire interval in Castlereagh Scribbly Gum woodland should be no lower than six years • The maximum interval in Cumberland Plain woodlands should be no higher than 15 years • The maximum interval in Castlereagh woodlands should be no higher than 23 years • Variable interfire intervals within thresholds will be important for ensuring persistence of all shrub species.
It needs to be recognised, however, that the scheme on which these conclusions are based is a theoretical model incorporating a limited number of life-history characteristics, applied using decision rules whose validity can be questioned, using limited data. Sensitivities relate to a conceptualisation of fire frequency which makes no allowance for patchiness or intensity. Rates of seed production and mortality, effects of weather and site productivity, and interspecific interactions are not considered. As Gill and Bradstock (1995:311) point out:
“Models are a valuable guide to prediction. However, models are abstractions of reality, caricatures, and do not emulate the real world. Models focus on attributes and events that explain or predict as much variation as possible while keeping their structure as simple as possible… They provide us with a framework around which complications can be considered….”
It is therefore essential that predictions from models be verified though observation and field measurement (Gill et al. 2002). While not specifically designed to do so, the study reported in the next two chapters should throw some light on the accuracy of predictions in relation to Cumberland Plain Woodland shrub species.
81 CHAPTER 4 FIRE FREQUENCY AND WOODLAND LANDSCAPES: SHRUBS, TREES AND GRASSES
4.1 Introduction 4.1.1 Fire frequency and woodland structure
An increase in woody plant density and cover in grasslands and open woodlands has been observed over the last century in many parts of the world. In North America, for example, field sampling and analysis of aerial photos demonstrate that savannas in southern Texas are being replaced by closed thorn woodlands (Archer et al. 1988), a process which has occurred within the last 100 years (Archer 1989). In Africa, aerial photographs in combination with on-ground estimates show a three-fold increase in shrub biomass in a savanna in Eastern Botswana (van Vegten 1983). It is estimated that thornbush encroachment affects 13 million hectares in South Africa (Trollope et al. 1989, cited in Roques et al. 2001). In South America, longitudinal surveys have documented a massive increase in tree stems in Venezuelan savanna (San Jose and Farinas 1991).
Encroachment has also been documented in Australia. For example, native trees and shrubs have multiplied in the semi-arid woodlands of Eastern Australia since European settlement (Peacock 1900, Hodgkinson and Harrington 1985, Noble 1997, Russell- Smith 2002). In the Northern Territory, Melaleuca minutifolia has proliferated in seasonally-inundated wetlands in the Victoria River district (Sharp and Bowman 2004). In North Queensland, Melaleuca viridiflora is spreading into grasslands on Cape York, reducing habitat for the endangered Golden-shouldered Parrot (Crowley and Garnett 1998), while Eucalyptus grandis grassy forests are predicted to disappear within the next century due to rainforest expansion (Harrington and Sanderson 1994). In the Bunya Mountains in south-east Queensland, the area occupied by grassy ‘balds’ shrunk by 26% in the forty years to 1991, although not in one frequently burnt area (Fensham and Fairfax 1996).
82 While many examples of shrub encroachment come from tropical, subtropical, or semi- arid areas, this process can also occur in temperate ecosystems. For example Withers and Ashton (1977) and Lunt (1998a, b) have clearly demonstrated a change from very open grassy woodland to Casuarina thicket at Ocean Grove on the Victorian coast. Also in Victoria, Bennett (1994) documented a reduction from 1496 ha to 389 ha in the area of Banksia integrifolia grassy woodland on the Yanakie Isthmus near Melbourne, due to spread of Leptospermum laevigatum. Acacia sophorae is encroaching into coastal grasslands near Moruya on the New South Wales south coast (Costello et al. 2000).
Shrubs encroaching into grassy ecosystems may be species that already exist in the landscape but which have increased in density, cover or abundance (Madany and West 1983, Archer et al. 1988, Archer 1989, Noble and Grice 2002). Alternatively they may be species extending their geographic range (Gleadow and Ashton 1981, Lunt 1998a).
While the role of fire in the encroachment process is not always clear, there are reasons to suspect it may be pivotal. A reduction in fire frequency often accompanies increasing woody plant density (Bradfield 1981, San Jose and Farinas 1991, Kington 1997, Scholes and Archer 1997, Rocques et al. 2001, Woinarski et al. 2004). Cessation of indigenous burning may be involved (Crowley and Garnett 1998, Noble and Grice 2002, Russell-Smith 2002). In Australia, experimental studies of fire frequency have found that infrequently burnt sites develop a higher density of trees and/or shrubs than do frequently burnt ones (Bowman et al. 1988, Birk and Bridges 1989, Taylor 1992, House 1995, Russell-Smith et al. 2003). South African experiments report similar findings (Trapnell 1959, Westfall et al. 1983, Bond 1997).
Explanations for a link between woody plant density and fire frequency are available. Fires can limit woody plant density by killing seedlings of obligate seeder species before they have time to reach reproductive maturity, and those of resprouting species before they have time to reach fire tolerance (Keith 1996). Tree saplings within the flame zone may not be able to ‘get away’ to become adults if fire is frequent and/or intense (Scholes and Archer 1997).
While the issue of shrub encroachment has not been formally studied on the Cumberland Plain, observers have noted increases in cover of the dominant shrub, Bursaria spinosa (Benson and Howell 2002, Robinson 2003, H. Recher pers. comm.
83 2004). This study seeks to ascertain whether either Bursaria abundance, or tree density, is linked to fire frequency.
4.1.2 Fire frequency and shrub species types
Although a lack of fire may encourage shrub encroachment, often only a limited number of species expand. As Noble and Slatyer (1980) point out, in the absence of disturbance, long-lived species and those which can regenerate in the presence of their own adults (T and R species – see Appendix 3) will finally come to dominate in any community. In the last chapter, assessment of shrub species vital attributes led to the conclusion that a large majority of Cumberland Plain shrub species would be vulnerable to local extinction in the absence of fire, as they can only regenerate in the immediate post-fire period. Only one common shrub in Cumberland Plain Woodland was classified as having the ability to recruit throughout the interfire period: Bursaria spinosa (Table A7.1 in Appendix 7 – Hakea sericea was also classified as a T species, however it is found in transitional sites rather than in CPW, according to Tozer 2003). Vital attributes analysis therefore leads to the prediction that infrequently burnt sites will support larger populations of Bursaria, but smaller populations of other shrubs, than sites burnt at a moderate frequency. If the thresholds proposed in Chapter 3 are well- founded, ‘sites burnt at a moderate frequency’ should encompass remnants burnt at intervals between 4 and 15 years. The extent to which species other than Bursaria are affected by long interfire intervals will be a function of their longevity, including that of their propagules.
At the other end of the fire frequency spectrum, vital attributes analysis would predict a decline in the abundance of most shrub species where repeated interfire intervals below 4 years are the norm, relative to areas where fire frequency is moderate. Only Bursaria spinosa is predicted to be relatively unaffected by very frequent fire (Appendix 3). The extent of decline of shrub species other than Bursaria, under frequent burning, will be influenced by their regeneration mode, according to the vital attributes model. Obligate seeders are likely to be more vulnerable than resprouters (Section 3.1.2, Appendix 3), particularly where their juvenile periods are relatively long.
Because of timing constraints, the study described in this and the following chapter was not specifically designed to test these predictions. The vital attributes analysis reported
84 in Chapter 3 relied on local data gathered over the course of the study, and on access to the NSW Flora Fire Response Database, which only became available in 2003, while data collection for this study commenced in 2002. Nevertheless, the findings of the current study provide an opportunity to assess the accuracy of vital attributes-based predictions for shrubs through an examination of shrub richness and abundance in sites from a range of fire frequencies.
4.1.3 Fire frequency and grasses
Experimental studies of fire frequency in grassy woodlands have noted that while infrequent fire benefits woody vegetation, frequent burning is associated with grass dominance in the understorey (Westfall et al. 1983, Bowman et al. 1988, Birk and Bridges 1989, Taylor 1992). Unequivocal documentation of differential degrees of grass cover under different fire frequencies is, however, less easily located.
In terms of grass composition, a number of authors have noted that Themeda australis benefits from fire (Robertson 1985, Belsky 1992, Prober 1996, Morgan and Lunt 1999, S. Clarke 2003). It may therefore decline if fire frequency is low.
4.1.4 Fire frequency and exotics
Exotic weed species often have attributes which enable them to take advantage of the establishment opportunities offered by disturbance; these include generalist modes of dispersal, and the ability to germinate and grow rapidly (Hobbs 2002). Grassy urban remnants may be particularly vulnerable to weed invasion, due to ready availability of propagules (Lunt and Morgan 1999), and indeed non-native species are common on Cumberland Plain shale soils (Tozer 2003). Since fire provides opportunities for recruitment in CPW (Chapter 7), might frequent burns encourage weeds?
Concerns that frequent fire may, or will, encourage weeds have certainly been expressed (Christensen and Burrows 1986, Fensham 1992, Stuwe 1994, Asquith and Messer 1998, Nadolny et al. 2003, Stenhouse 2004). However the experimental literature is equivocal.
Few Australian studies have assessed the effects of a series of fires. In Perth, Baird (1977) reported that frequent burning had resulted in the spread of the African grass
85 Ehrharta calycina into the shrubby woodlands of Perth’s Kings Park. The relative contributions of fire frequency and soil disturbance associated with ploughed fire breaks could not, however, be disentangled. Later studies of weed invasion in south-west Western Australia after a single fire suggest the two factors may interact to encourage weed establishment (Milberg and Lamont 1995), but that fire in otherwise undisturbed vegetation does not have this negative effect (Hobbs and Atkins 1991, Hester and Hobbs 1992).
On the other hand, Grice (1997) concluded that repeated burns had the potential to help control Cryptostegia grandiflora ( Rubber Vine) in grassy woodland near Townsville, particularly if applied in the early stages of invasion. This conclusion was based on the ability of a single fire to kill plants in lower size classes. Russell and Roberts (1996) assessed the effects of four low-intensity fires in a 14 year period against an unburnt control. Lantana density increased more rapidly in the unburnt area, although this trend was not significant. Fire frequency was not associated with exotic species richness, either above or below ground, in Victorian basalt grasslands (Morgan 1998b), and a single fire in a Victorian dry sclerophyll forest did not lead to weed invasion (Adams and Simmons 1996). Morgan (1998d) considers that fire may enhance the invasive potential of some exotic species, but not others.
The current study provides an opportunity to assess the relative abundance of woody exotics in sites with differing degrees of exposure to fire. The relationship of herbaceous weeds to fire frequency will be explored in Chapter 8.
4.1.5 Study aims, approach, and questions
The broad aims of this study were: 4 • To document the structure of vegetation in a range of shale woodland remnants across Western Sydney; • To determine the extent and nature of structural differences between sites which have experienced different burning frequencies; and
4 This chapter, and the data chapters that follow (Chapters 5 to 9) focus on Western Sydney’s shale-based woodlands, which I will refer to as Cumberland Plain Woodland, or CPW (see Section 2.4). Time precluded further study of the role of fire in Castlereagh Woodland (beyond that reported in Chapter 3).
86 • To begin to investigate the relationship between fire frequency and species richness and composition, in the field (this process was continued in a second study, reported in Chapter 8).
The study compared shale-based woodland sites which had experienced low, medium and high fire frequencies over the last 20 years. Investigation focused on features related to the character of woodland landscapes at a hectare scale. Broad characteristics of the three structural layers of Cumberland Plain woodlands – trees, shrubs and grasses – were measured. Because of its obvious dominance of many remnants, Bursaria spinosa was accorded particular attention.
Study methods are outlined in Section 4.2 below. Results are reported and discussed in this and the subsequent chapter. Specific questions addressed in this chapter include:
• Does Bursaria spinosa frequency, density and/or dominance increase with decreasing fire frequency? It was hypothesised, on the basis of the literature and the vital attributes analysis in Chapter 3, that this would indeed be the case. • Does the species richness or frequency of native shrubs other than Bursaria vary with burning frequency? It is hypothesized, on the basis of the vital attributes analysis in Chapter 3, that sites burnt at moderate frequency would support more of these shrubs than more, or less, frequently burnt areas. • Do shrub species which resprout after a fire and those which rely on seeding regeneration differ in their response to fire frequency? It was hypothesised, on the basis of their vital attributes, that obligate seeders would be more sensitive to variations in interfire interval. • Does the frequency of exotic shrubs vary with burning frequency? The hypothesis inherent in the oft-expressed fear about the effect of fire on weediness was tested, ie that more frequently burnt areas would support more exotic shrubs than less frequently burnt areas. • Does tree density and/or basal area increase as interfire interval increases? It was hypothesised, on the basis of the literature linking increased woody plant density to reduced exposure to fire, that this would be the case. • Is tree recruitment affected by fire frequency? In particular, the hypothesis that frequent fire prevents suppressed seedlings from moving into and through the sapling stage was examined. • Does grass cover increase with burning frequency? The hypothesis that grass cover would be greater in frequently burnt areas was tested. • Does grass composition differ with burning frequency? The hypothesis that frequently burning favours Themeda australis was of particular interest.
87 4.2 Methods 4.2.1 Fire frequency categories and study sites
Vegetation parameters were sampled at a relatively large spatial scale in nine sites, three in each of three fire frequency categories. Fire frequency categories were based on the incidence of burns over the past 20 years. As available fire history information did not necessarily include exact dates and locations of fires (see Section 2.7.2), broad categories were used:
• High fire frequency sites had burnt very regularly over the past 20 years. Intervals were reported to be mostly between 1 and 3 years. • Moderate fire frequency sites had experienced some fire over the past 20 years, generally in the form of patchy arson or fuel reduction burns. Intervals were reported to be mostly between 4 and 10 years. • Low fire frequency sites had had at least a 20 year fire-free interval prior to their most recent fire.
Moderate fire frequency sites therefore fell roughly within the fire interval domain proposed for Cumberland Plain woodlands in Chapter 3, while sites in the other two categories fell outside these thresholds.
This study design assumes that earlier fire regimes were similar to those of recent years, or at least were random. It seems reasonable to suppose that Aborigines employed a consistent approach to fire management for Western Sydney’s grassy woodlands, and that early settlers running farming enterprises might also have taken a common approach. Divergent strategies may, however, have been employed by land managers as land uses diversified with the growth of the metropolitan area.
Remnants were only considered for inclusion in the study where information from at least two apparently-reliable sources concurred on recent fire history to a degree which made allocation to a category unambiguous. Though not ideal, this process allowed identification of a reasonable number of remnants, or parts of remnants, in each category. Given the subjective nature of the fire history information, sites in each category should be regarded as having a good probability of having experienced the particular level of fire exposure implied by the category name. Without detailed fire records, however, these probabilities cannot be regarded as certainties.
88 Sites selected for sampling in each category were those which came nearest to meeting the following criteria:
• Site not subject to broad-acre mowing or large-scale tree removal within the last 20 years; • Minimal grazing by domestic animals over the past 20 years; • Most recent burn between 9 months and 3 years prior to sampling.
In addition, all sites met these criteria: • Site visually dominated by native species; • Site not currently grazed by domestic animals; • Eucalyptus moluccana is a prominent species in the site; • Site is at least 3 ha in area.
The grazing, mowing and logging criteria were designed to minimise the confounding effects of disturbances other than fire.
Limiting sampling to areas with prominent Eucalyptus moluccana ensured all sites fell within a broad definition of Cumberland Plain woodlands. E. moluccana is a common and characteristic species of both Shale Plains and Shale Hills Woodland, and is also sometimes prominent in the related Shale-Gravel Transition Forest (Tozer 2003). While E. tereticornis tends to favour lower slopes and depressions, E. moluccana is more abundant on drier rises (Benson et al. 1996).
Time-since-fire affects vegetation composition and structure (Specht et al. 1958, Posamentier et al. 1981, McFarland 1988a, Coops and Catling 2000). For example seedlings germinating post-fire may not survive into the later post-fire years (Section 3.4.2), and sites which have not burnt for many years may harbour shrub species in the soil seed bank which are not visible above-ground (Posamentier et al. 1981, Gill and Bradstock 1995, Bond and Ladd 2001). Thus ideally, study sites would all have been of similar post-fire age. Since the frequently-burnt sites, by definition, had burnt recently, ideally all sites would have burnt in recent years. This was particularly important in infrequently burnt sites, to ensure fire-cued shrub seeds had had a chance to germinate, bringing these species into the above-ground community (Bond and Ladd 2001). All three infrequently burnt sites had, in fact, burnt about 12 months prior to sampling. High fire frequency sites were also mostly recently burnt, and one of the three moderate
89 fire frequency sites had burnt completely just over 12 months prior to sampling. Patchy burns in the other two moderate fire frequency sites meant that some transects (or parts of transects) in these sites met the time-since-fire criterion, but others did not.
Table 4.1 lists the nine sample sites and provides further information on their vegetation types, management authorities, and recent fire history. Sites are listed from most to least frequently burnt (this order will be maintained throughout the chapters pertaining to this study). Figure 4.1 shows site locations. Sites in each fire frequency category were fairly well-separated geographically, and treatments were interspersed. All sites were between 3 and 6 ha in area, and most were part of a larger remnant.
Figure 4.1. Map of the Cumberland Plain showing remnant native vegetation (green) and study sites in three fire frequency categories. ▲, high fire frequency; ■, moderate fire frequency; ▼, low fire frequency.
90 Table 4.1. Sites used in study of relationship of landscape-level vegetation parameters to fire frequency in Cumberland Plain Woodland.
Site name Fire frequency Management Vegetation type Fire history Approx. time category authority since fire Benson (1992) Tozer (2003)* when sampled Ropes Creek high Penrith Council Not mapped; SGTF Burnt approximately annually by arsonists since mid 1970s (pers. comm. John Pearson, 10 mths over Grey Box Penrith RFS, 2002; John Diamond, Bushcare group coordinator and former local most of site Woodland resident, 2002; Maria Mura, local resident and former student at adjacent high school, nearby 2004). Many fires 2001-2004 (pers. obs.). Shanes Park high Air Services Grey Box SGTF (70%) Burnt every 1-3 years since 1980 or earlier (pers. comm. Craig Sampson and Ian 12 mths over Australia Woodland SPW (30%) Lovelock, Air Service Australia, 2002; Daniel Busch, Shanes Park Rural Fire Brigade, most of site 2003). Several fires 2001-2004 (pers. obs). Holsworthy high Department of Grey Box SPW (70%) Annual patchy fuel reduction burns through 1980s and early 1990s. Wildfire 1997 (part of 12 mths over Defence Woodland SGTF (30%) site), extensive arson burn 2000, wildfire 2002 (pers. comm. Sarah Hill, University of whole site Wooloongong, 2001; Marina Peterson, Defence Dept., 2002; Ian Jackson, Penrith RFS, 2003). Prospect moderate Sydney Grey Box - SHW (80%) Patchy fuel reduction burns every 1-2 yrs 1967-1987, one or two hazard reduction burns 13 mths over 91 Catchment Ironbark SPW (20%) mid 1990s, small patchy arson burns late 1990s and 2001, wildfire 2002. (Thomas, 1994; whole site Authority Woodland pers. comm. David Warren, Greening Australia, 2001; Wayne Butcher, former Captain Blacktown RFS, 2002; Ron Hicks, former employee of Water Board, 2003; Jane MacCormick, Sydney Catchment Authority, 2003). Plumpton moderate Blacktown Grey Box SPW Patchy arson burns of varying extent through 80s and 90s. Most of reserve burnt in 1995. 2.3 yrs over Park Council Woodland About half burnt in 2000. (pers. comm. Ted Williams, Wayne Butcher and Wira Tewhare, 50% of site; 7 Blacktown RFS, 2002; Glenn White and Mark Anderson, Blacktown Council, 2002; John yrs over much Diamond, bush regeneration contractor, 2003; local residents, 2003). of the rest. Lansdowne moderate Bankstown Grey Box SPW Patchy arson burns of varying extent through 80s and 90s (pers. comm. Debra Little and 16 mths over Council Woodland Rob Corby, former Bankstown Council Bushland Managers, 2002; Carol Mara, Bankstown 50% of site, 3 Council Parks staff member, 2003; H.J. Jansen, local resident, 2003). Arson burns 2001 yrs over 20%. (pers. obs). Mt Annan low Royal Botanic Grey Box – SHW Last interfire interval probably 20 years. Lido Turrin, Campbelltown RFS, has a diary 15 mths over Gardens Ironbark record of a wildfire in 1982 which appears to include this area. Burnt for ecological almost whole Woodland** purposes by Gardens staff in 2002 (pers. obs. 2002; pers. comm. Debra Little, Royal of site Botanic Gardens, 2002). Orchard Hills Low Department of Grey Box SPW (50%) Last interfire interval at least 30 years. No fire in this area until a wildfire in December 15 mths over Defence Woodland SHW (50%) 2001. (Pers. comm. Don Himsley, former leasee, 2002; Marina Peterson, Defence whole site Department, 2002; Steve Weyman, Orchard Hills Rural Fire Brigade, 2002). Scheyville Low NPWS Grey Box - SPW Last interfire interval approximately 50 years (NPWS records based on information from 12 mths over Ironbark Peter Speet, Oakville Rural Fire Brigade, 2001). Burnt for experimental purposes by whole site Woodland NPWS staff in November 2001 (pers. obs; pers. comm Jonathan Sanders, NPWS, 2001). * SPW, Shale Plains Woodland; SHW, Shale Hills Woodland; SGTF, Shale-Gravel Transition Forest. ** Not covered by Benson (1992); pers. comm. D. Benson 2004.
4.2.2 Sampling and data collection
Belt transects
The basic sampling unit was a 100m-long belt transect, either 2 m or 6 m in width depending on what was being measured. Belt transects are long, thin quadrats (Ruyle and Frost 1993). The distribution of shrubs and grasses in Cumberland Plain Woodland is patchy, a feature which has caused difficulties in previous research comparing areas with differing fire histories (Thomas 1994). Long thin plots are more efficient at picking up the species richness of an area than square plots (Stohlgren et al. 1995), and are more likely to cover a range of patch types. Vigilante and Bowman (2004) used belt transects to minimise the effects of landscape heterogeneity when studying the effects of fire in the Kimberley.
There were 12 transects in each site, four in each of three blocks, giving 108 transects in total.
Belt transects were centred on line transects. In each site the area which met the fire history and other relevant sampling criteria was mapped. This area was divided, on paper, into 100m-wide strips. Strips were grouped into three blocks of roughly equal size. Within each block, the sides of all 100m-wide strips were numbered consecutively in metres. These numbers were used to randomly select four line transects, together with some ‘spares’. If a randomly-selected line transect was within 15 m of a transect which had already been selected, a further selection was made. This procedure ensured a reasonable degree of separation between transects, thus minimising autocorrelation.
Once on site, selected transects were surveyed unless: • The vegetation did not include sufficient E. moluccana. The rule here was: at least one adult E. moluccana tree within 3 m of each half of the line transect, ie one between 0 and 50 m, and one between 50 and 100 m. (Three transects at Mt Annan failed to meet this rule. Tree density at Mt Annan was relatively low, so there were no trees along long stretches of transects. E. moluccana was, however, the dominant species along and around the three transects.) • The transect crossed an area subject to substantial disturbance, eg a major track, an area in which a significant amount of rubbish had been dumped, an area where vegetation had been subject to a burn within the last six months.
92 If a selected transect was not used for one of these reasons, then a ‘spare’ transect was surveyed instead.
Features of the woodland landscape were recorded for each 2 m segment of transect (Figure 4.2). Grass and shrub variables were recorded in 2 x 2 m subplots, ie 1 m to either side of the central line. Variables relating to tree species were recorded in 2 x 6 m subplots, ie 3 m to either side of the central line. There were 50 subplots along a transect. Subplots falling on minor paths were omitted; substitute subplots were added to the end of the transect.
Figure 4.2. Recording data in a 2 x 2 m subplot part way along a 100 m transect. Photo by Venesa Brusic.
Vegetation categories
In this study, the terms ‘shrub’ and ‘tree’ refer to groups of particular plant species (see Appendix 8 for species lists). A liberal definition of ‘shrub’ was used. Included were:
• woody climbers/scramblers (eg Hardenbergia violacea, Clematis glycinoides), • prostrate and low-growing shrubs/subshrubs (eg Chorizema parviflorum, Eremophila debilis), and
93 • species which may become trees but which were generally represented by individuals of shrubby habit in study sites (eg Acacia implexa, Acacia paramattensis, Olea europaea spp. africana). Species descriptions in the Flora of New South Wales (Harden 1992 - 2002) guided classification of species on the boundary between shrub and forb/non-woody plant. Where the description was ambiguous (eg “shrub or herb,” or life form not mentioned), allocation was based on:
• the extent of woodiness in individuals found in survey sites; • the persistence of above-ground material. In some species, the above-ground parts die back during a dry spell. These species were generally excluded from the shrub list, as an accurate record of their presence throughout the sampling period could not be ensured.
T-square sampling
An additional strategy, the T-square technique, was employed for adult trees (defined below), and Bursaria spinosa bushes over 80 cm in height. This technique has the advantage of providing data which can be used to objectively determine spatial pattern of a species or life-form (Ludwig and Reynolds 1988, Krebs 1999). Dixon (2001:13) describes the method as follows:
“A point, A, is randomly chosen and the nearest neighbor, B, is found. Then, the study area is divided into two half planes by a line through B and perpendicular to AB (hence the name, T-square). Attention is restricted to the half plane that does not contain point A. The distance to the nearest neighbor, Z, of B in that half plane is measured.”
Eight points were randomly selected along each transect. If a randomly-selected point was within 8 m of a point which had already been selected, a further selection was made, ensuring points were at least 8 m apart.
4.2.3 Vegetation parameters
Bursaria measures
In each 2 x 2 m subplot, the presence or absence of Bursaria spinosa was noted, along with its maximum height. Height was recorded in categories: 1 = < 25 cm, 2 = 25 – 80 cm, 3 = 80 – 200 cm, 4 = > 200 cm.
94 This information was used to create two indices:
• Bursaria frequency. This variable reflects the presence or absence of Bursaria spinosa in the 50 subplots along a transect. The number of subplots with Bursaria present was multiplied by 2, giving a score from 0 to 100. • Bursaria dominance index. This index adds a dimension to the previous measure by incorporating the height of Bursaria spinosa bushes. Height category ratings in each subplot along a transect were summed, giving a number between 0 and 200. This number was divided by 2 to give the transect’s score on the index. B. spinosa height is significantly correlated with diameter, volume and number of stems (Thomas 1994), so this index provides an indication of the dominance of Bursaria in the landscape.
A third measure, Bursaria density, was estimated from the T-square data at block level, using the formula:
2 E* = N / 2 ∑ (xi)[√2 ∑ (yi)] where 2 • E* = density (Bursaria plants over 80 cm per m ) • N = sample size (32, 8 points from each of 4 transects)
• xi = distance from random point i to nearest Bursaria plant over 80 cm high
• yi = T-square distance associated with Bursaria plant i.
Byth (1982), in a simulation study designed to test a wide range of distance-based density estimators, found this formula to be robust to changes in underlying processes. It is also recommended by Ludwig and Reynolds (1988) and Krebs (1999). E* multiplied by 10,000 provided a measure of Bursaria density (in units of plants over 80 cm) per hectare.
Other shrub measures
In each 2 x 2 m subplot, all native shrub species with cover in the subplot were recorded, along with the number of individuals of each species other than Bursaria spinosa rooted in the plot. Exotic shrub species with cover in the subplot were also noted.
This information was used to create several measures:
• Other native shrub species richness, the number of native shrub species other than Bursaria spinosa recorded along the belt transect (200 m2).
95 • Other native shrub frequency, the number of subplots along a transect in which one or more native shrub species other than Bursaria was recorded, multiplied by 2 (scale 0 to 100). • Subshrub frequency, the number of subplots along a transect in which at least one subshrub species was recorded, multiplied by 2 (scale 0-100). Smaller native shrub species were assigned to this category, along with woody climbers and scramblers (see Table A8.1 in Appendix 8). Species descriptions in the Flora of New South Wales (Harden 1992-2002) guided the allocation process. • Larger shrub frequency, the number of subplots along a transect in which at least one shrub, other than a subshrub or B. spinosa, was recorded, multiplied by 2 (scale 0-100). • Obligate seeder frequency, the number of subplots along a transect in which at least one native obligate seeder shrub species was recorded, multiplied by 2 (scale 0–100). Allocation to obligate seeder and resprouter categories was based on field observations (Appendix 4, Table A8.1 in Appendix 8). • Other resprouter frequency, the number of subplots along a transect in which at least one native resprouter shrub species other than Bursaria spinosa was recorded, multiplied by 2 (scale 0-100). • Exotic shrub frequency, the number of subplots along a transect in which a shrub species not native to Australia was recorded, multiplied by 2 (scale 0-100).
Tree measures
In each 2 x 6 m subplot, the number of individuals of each tree species rooted in the plot was noted. Individuals were allocated to one of three life stage categories:
• Adult, defined as a member of a tree species with at least one live trunk whose diameter at breast height equalled or exceeded 10 cm. Species and diameter at breast height (DBH) were recorded for each adult tree. Measurement procedures followed Australian forestry conventions (Wood et al. 1999). Where a tree had more than one trunk over 10 cm DHB, each trunk was measured. • Sapling, defined as a member of a tree species over 2 m high, with stem(s) under 10 cm DBH (Figure 4.3A). Saplings were recorded as eucalypt, or non-eucalypt. • Juvenile, defined as a member of a tree species below 2 m high, with multiple stems. This definition excluded recently-germinated single-stemmed seedlings, and included lignotuberous multi-stemmed plants (Figure 4.3B) sometimes known as “suppressed seedlings” (Clarke 2002), “lignotuberous seedlings” (Henry and Florence 1966) or “gullivers” (Bond and van Wilgen 1996). Single-stemmed seedlings were excluded to minimise the influence of time-since-fire: in the early months post-fire there may be many small seedlings, most of which will not live to become established juveniles (Wellington and Noble 1985). Juveniles were recorded as eucalypt, or non-eucalypt.
96 A.
B.
Figure 4.3. A, sapling, and B, juvenile eucalpts.
DBH figures for adult trees were used to calculate basal area. Where a tree had multiple trunks, basal area was calculated for each trunk, and the resulting figures summed. Figures for basal area per transect (600 m2) were converted to m2/ha, for ease of reporting.
The percentage of non-adult trees which were saplings was also calculated for each transect, as follows: (number of saplings / (number of saplings + number of juveniles)) x 100.
Grass measures
In each 2 x 2 m subplot, an estimate of grass cover was recorded, along with the name of the dominant grass species, defined as the species with the highest cover. Grass cover estimates were based on the proportion of the subplot over which grass clumps formed a mosaic, rather than simply on projective foliage cover, as the former was less subject to change with time-since-fire or recent grazing. The rating scale was:
0 = no grass 1 = low, grass clumps found over up to 20% of the subplot 2 = medium, grass clumps found over 20-80% of the subplot 3 = high, grass clumps found over > 80% of the subplot
97 This information was used to create four indices:
• A grass cover index, constructed by summing the rating figures for grass cover in each subplot along a transect, giving a number between 0 and 150. This number, divided by 1.5, was the transect’s score on the index, which thus ranged from 0 to 100. • An index for each of the three most common grass taxa found in the study. The number of subplots along a transect in which the taxon dominated was calculated, then multiplied by 2, giving an index between 0 and 100.
4.2.4 Data analysis
Each of the variables described above was analysed through a nested analysis of variance, with fire frequency as a fixed factor (high, moderate, or low), and site and block as nested, random factors. The unit of replication was the transect.
For each variable, Cochran’s test was used to assess homogeneity of variances at block level. If variances were homogeneous (p > 0.05 when C3,27 < 0.17), the analysis proceeded, giving significance levels for differences between fire frequency categories, sites within these categories, and blocks within sites.
Where block variances were heterogeneous, the analysis was moved up a level, using the average score per block as the unit of analysis, and Cochran’s test was again applied. If variances at site level were homogeneous (p > 0.05 when C2,9 < 0.48), the analysis proceeded, giving significance levels for differences between fire frequency categories and sites within these categories.
If variances were still heterogeneous, plots of means against variances were examined to see whether a transformation was likely to be effective. Where large variances were simply a product of an unusual transect, the option of running the analysis on the untransformed data, but lowering the α level, was considered.
Where the ANOVA indicated significant differences between fire frequency categories, Student-Newman-Kuel tests were used to identify where differences lay.
98 4.3 Results 4.3.1 Bursaria
Infrequently burnt sites supported more Bursaria that sites burnt at least once a decade. Bursaria frequency, dominance and density were all considerably, and significantly, higher where fire frequency was low.
Bursaria frequency
Bursaria frequency was over twice as great in low fire frequency sites, on average, than in sites in the other fire frequency categories (Figure 4.4). Eighty-six percent of subplots in low fire frequency sites had Bursaria cover, while the equivalent figures in moderate and high frequency sites were 36% and 29% respectively (site and fire frequency category means and standard errors for all variables are given in Appendix 9). This difference was highly significant, while differences at site and block level were not (Table 4.2). While heterogeneity of variances precluded post-hoc comparisons, it was clearly the low fire frequency sites which differed from the rest.
A. Bursaria frequency by fire frequency B. Bursaria frequency by site
100.0 100.0
90.0 90.0
80.0 80.0
70.0 70.0
60.0 60.0
50.0 50.0
40.0 40.0
30.0 30.0 Bursaria frequency Bursaria Bursaria frequency frequency Bursaria
20.0 20.0
10.0 10.0
0.0 0.0 Ropes Ck Shanes Pk Holsworthy Prospect Plumpton Lansdowne Mt Annan Orchard Scheyville high moderate low Pk Hills Fire frequency Site Figure 4.4. Bursaria frequency (mean + S.E., scale 0-100) for A, three fire frequency categories and B, nine sites, three within each category. Red, high fire frequency; orange, moderate fire frequency; cream, low fire frequency.
99 Table 4.2. ANOVA of Bursaria frequency for sites in three fire frequency categories.
Source Sums of squares d.f. Mean square F P Fire frequency 68610 2 34305 43.539 < 0.001 Site (Fire frequency) 4728 6 788 1.301 0.306 Block (Site) 10901 18 606 1.778 0.042 Error 27584 81 341 Total 111823 107
Cochran’s test C3,27 = 0.18, P < 0.05, α reduced to 0.01.
Bursaria dominance index
Mean score on the Bursaria dominance index decreased with fire frequency, with low fire frequency sites scoring almost four times higher than those burnt every 1-3 years (Figure 4.5). Again this difference was highly significant, while differences at site and block level were not significant under the revised α level adopted for this analysis (Table 4.3).
Table 4.3. ANOVA of Bursaria dominance for sites in three fire frequency categories.
Source Sums of squares d.f. Mean square F P Fire frequency 51142 2 25571 38.944 < 0.001 Site (Fire frequency) 3940 6 657 1.605 0.203 Block (Site) 7366 18 409 1.965 0.021 Error 16872 81 208 Total 79319 107
Cochran’s test C3,27 = 0.25, P < 0.05, α reduced to 0.01.
A. Bursaria dominance by fire frequency B. Bursaria dominance by site
100.0 100.0
90.0 90.0
80.0 80.0
70.0 70.0
60.0 60.0
50.0 50.0
40.0 40.0
30.0 30.0
20.0
20.0 Bursaria dominance index Bursaria dominanceBursaria index
10.0 10.0
0.0 0.0 Ropes Ck Shanes Pk Holsworthy Prospect Plumpton Lansdowne Mt Annan Orchard Scheyville high moderate low Pk Hills Fire frequency Site Figure 4.5. Bursaria dominance (mean + S.E., scale 0-100) for A, three fire frequency categories and B, nine sites, three within each category. Red, high fire frequency; orange, moderate fire frequency; cream, low fire frequency.
100 Bursaria density
Differences between fire frequency categories emerged most strongly on this measure (Figure 4.6). The mean density of Bursaria plants over 80cm high was 92 per hectare on high fire frequency sites, 272 on sites subject to moderately frequent fires, and 1872 in low fire frequency areas. Densities in the low frequency sites ranged from 575 per ha at Mt Annan (probable interfire interval of 20 years), to 2009 at Orchard Hills (33+ year interfire interval), to 3033 at Scheyville (approximately 50 year interfire interval).
Cochran’s test showed site variances were heterogeneous (C2,9 = 0.63, P < 0.05), so data were log10 transformed. Low fire frequency sites had significantly more Bursaria plants per hectare than moderate or high fire frequency areas (Table 4.4). Moderate and high fire frequency sites did not differ on this variable, nor did sites within each fire frequency category.
A. Bursaria density by fire frequency B. Bursaria density by site
3000.0 4000.0
3500.0 2500.0
3000.0
2000.0 2500.0
1500.0 2000.0
1500.0
plants over 80cm high/ha over plants 1000.0 plants over 80cm high/ha over plants
1000.0 500.0
500.0 B. spinosa B. B. spinosa B. 0.0 0.0 high moderate low Ropes Ck Shanes Pk Holsworthy Prospect Plumpton Pk Lansdowne Mt Annan Orchard Hills Scheyville Fire frequency Site
C. Bursaria density (log10) by fire frequency D. Bursaria density (log10) by site
4.00 b 4.00 3.50 a a 3.50 3.00 3.00
2.50 2.50 ) ) 10 2.00 10 2.00 (log (log 1.50 1.50 plants over 80 cm 80 over high plants plants over 80 cm high/ha cm80 high/ha over plants 1.00 1.00
0.50 0.50 B. spinosa B. B. spinosa B. 0.00 0.00 high moderate low Ropes Ck Shanes Pk Holsworthy Prospect Plumpton Pk Lansdowne Mt Annan Orchard Hills Scheyville Fire frequency Site Figure 4.6. Density of Bursaria spinosa plants over 80 cm high/ha (mean + S.E.) for three fire frequency categories (A, C) and nine sites, three within each category (B, D). A, B, untransformed data; C, D, log10 transformed data. Red, high fire frequency; orange, moderate fire frequency; cream, low fire frequency. Means labelled with different letters in Chart A are significantly different at P = 0.01 (SNK).
101 Table 4.4. ANOVA of Bursaria density (plants over 80 cm/ha, log10 transformed) for sites in three fire frequency categories. Block means used as replicates.
Source Sums of squares d.f. Mean square F P Fire frequency 7.224 2 3.612 17.892 < 0.005 Site (Fire frequency) 1.211 6 0.202 2.551 > 0.05 Error 1.425 18 0.079 Total 9.860 26
Cochran’s test C2,9 = 0.396, P = NS
4.3.2 Other native shrubs
Thirty-six native shrub taxa other than B. spinosa were recorded in belt transects; 23 occurred in more than one site. No shrub taxon other than B. spinosa was recorded in every site, and the highest percent frequency for any one species across all study sites was 18.0 (Hardenbergia violacea) compared to 50.3 for Bursaria. The nine most abundant taxa, after B. spinosa, were legumes. Eleven of the 37 shrub taxa (30%) were classified as obligate seeders, including four of the five most abundant species after B. spinosa. Fourteen (38%) were categorised as subshrubs. Native shrub taxa encountered in the study are listed in Table A8.1 in Appendix 8, together with regeneration mode, life form, conservation status and overall frequency. Site-specific frequencies are given in Table A8.2, also in Appendix 8.
Other native shrub species richness
Native shrub species richness was highest in moderate fire frequency areas, with an average of 6.6 species per transect (200 m2). High fire frequency sites averaged 4.4 species, while sites which had experienced a long interfire interval before a recent fire were relatively species poor, averaging 2.5 species per transect (Figure 4.7). There were significant differences both between fire frequency categories, and between sites within those categories (Table 4.5). However only the difference between low and moderate fire frequency areas was significant on the post-hoc comparisons (P < 0.05).
102 A. Other native shrub species richness by fire frequency B. Other native shrub species richness by site
8.0 b 9.0 7.0 ab 8.0 7.0 6.0 6.0 5.0 inosa inosa 5.0 p p 4.0 a B. s B. B. s B. 4.0 of native shrubs other other shrubs native of of native shrubs other other shrubs native of 2 2 3.0 3.0 than than
2.0 2.0
1.0 1.0 SR per 200m SR 200m per 0.0 0.0 Ropes Ck Shanes Pk Holsworthy Prospect Plumpton Lansdowne Mt Annan Orchard Scheyville high moderate low Pk Hills Fire frequency Site
Figure 4.7. Species richness per 200 m2 of native shrubs other than Bursaria spinosa (mean + S.E., scale 0-100) for A, three fire frequency categories and B, nine sites, three within each category. Red, high fire frequency; orange, moderate fire frequency; cream, low fire frequency. In this and following figures, means labelled with different letters are significantly different at at least P = 0.05 (SNK).
Table 4.5. ANOVA of species richness per 200 m2 of native shrubs other than Bursaria spinosa for sites in three fire frequency categories. Block means used as replicates.
Source Sums of squares d.f. Mean square F P Fire frequency 76.2 2 38.11 5.828 0.039 Site (Fire frequency) 39.2 6 6.54 3.395 0.020 Error 34.7 18 1.93 Total 150.1 26
Cochran’s test C2,9 = 0.438, P = NS
Other native shrub frequency
The picture for this variable was similar to that for shrub species richness, with moderate frequency sites scoring higher than those with either a high or low fire frequency (Figure 4.8). Differences between fire frequency categories, and between sites within categories, were again significant (Table 4.6). This time both low and high fire frequency sites scored significantly lower than areas with a moderate fire frequency (P < 0.05), but did not differ from each other.
103 Table 4.6. ANOVA of other native shrub frequency for sites in three fire frequency categories. Block means used as replicates.
Source Sums of squares d.f. Mean square F P Fire frequency 13756 2 6878 5.503 0.044 Site (Fire frequency) 7500 6 1250 10.307 <0.001 Error 2183 18 121 Total 23439 26
Cochran’s test C2,9 = 0.294, P = NS
A. Other native shrub frequency by fire frequency B. Other native shrub frequency by site
100.0 100.0 90.0 b 90.0 80.0 80.0 70.0 a 70.0 60.0 60.0
50.0 a 50.0
40.0
40.0 B. spinosa B. spinosa spinosa B. 30.0 30.0
20.0 20.0
10.0 10.0 Frequency of native shrubs other than than other shrubs native of Frequency Frequency of native shrubs other than than other shrubs native of Frequency 0.0 0.0 Ropes Ck Shanes Pk Holsworthy Prospect Plumpton Lansdowne Mt Annan Orchard Scheyville high moderate low Pk Hills Fire frequency Site Figure 4.8. Frequency of native shrubs other than Bursaria spinosa (mean + S.E., scale 0- 100) for A, three fire frequency categories and B, nine sites, three within each category. Red, high fire frequency; orange, moderate fire frequency; cream, low fire frequency.
This pattern held, and in fact was marginally stronger, when subshrub species were excluded from the analysis. Larger shrub frequency averaged 67.3 in moderate fire frequency sites, 23.9 in high fire frequency sites, and 17.8 in low fire frequency sites (Table 4.7, Figure 4.9).
Table 4.7. ANOVA of larger shrub frequency for sites in three fire frequency categories. Block means used as replicates.
Source Sums of squares d.f. Mean square F P Fire frequency 13098 2 6549 6.347 0.033 Site (Fire frequency) 6191 6 1032 7.880 <0.001 Error 2357 18 131 Total 21647 26
Cochran’s test C2,9 = 0.302, P = NS
104 A. Frequency of larger shrubs by fire frequency B. Frequency of larger native shrubs by site 100.0 b 100.0 90.0 90.0
80.0 80.0
70.0 70.0
60.0 60.0 B. spinosa B. B. spinosa B. 50.0 a 50.0 40.0 a 40.0 30.0 30.0 other than than other other than than other 20.0 20.0
10.0
10.0 shrubs native larger of Frequency Frequency of larger native shrubs shrubs native larger of Frequency
0.0 0.0 Ropes Ck Shanes Pk Holsworthy Prospect Plumpton Lansdowne Mt Annan Orchard Scheyville high moderate low Pk Hills Fire frequency Site
Figure 4.9. Frequency of larger native shrub species other than Bursaria spinosa (mean + S.E., scale 0-100) for A, three fire frequency categories and B, nine sites, three within each category. Red, high fire frequency; orange, moderate fire frequency; cream, low fire frequency.
The average frequency of subshrubs was also highest in moderate fire frequency sites (44.9), next highest in very frequently burnt sites (22.3), and lowest in infrequently burnt sites (9.3). However there was considerable variability between sites, and these differences were not significant (F(2,6) = 1.412, P = 0.314, NS).
When post-fire regeneration mode was considered, differences in frequency were significant for obligate seeders, but not for resprouters (Bursaria spinosa was excluded from the list of resprouters for this analysis; Tables 4.8 and 4.9, Figure 4.10). Obligate seeder frequency averaged 58.8 in sites with a moderate fire frequency, significantly more than in either high or low fire frequency sites, which averaged 21.7 (P < 0.05) and 10.9 (P < 0.01) respectively.
Table 4.8. ANOVA of resprouter frequency (shrub species other than B. spinosa) for sites in three fire frequency categories.
Source Sums of squares d.f. Mean square F P Fire frequency 25147 2 12573 1.646 0.269 Site (Fire frequency) 45835 6 7639 15.383 <0.001 Block (Site) 8939 18 497 2.242 0.008 Error 17939 81 222 Total 97860 107
Cochran’s test C3.27 = 0.137, P = NS
105 Table 4.9. ANOVA of obligate seeder shrub frequency for sites in three fire frequency categories.
Source Sums of squares d.f. Mean square F P Fire frequency 45328 2 22664 8.312 0.019 Site (Fire frequency) 16359 6 2727 5.469 0.002 Block (Site) 8974 18 499 2.000 0.019 Error 20194 81 249 Total 90855 107
Cochran’s test C3.27 = 0.168, P = NS
A. Resprouter frequency by fire frequency B. Resprouter frequency by site
100.0 100.0
90.0 90.0
80.0 80.0
70.0 70.0 B. spinosa B. B. spinosa B. 60.0 60.0
50.0 50.0
40.0 40.0
30.0 30.0
20.0 20.0 shrubs other than than other shrubs shrubs other than than other shrubs Frequency of resprouting native native resprouting of Frequency Frequency of resprouting native native resprouting of Frequency 10.0 10.0
0.0 0.0 Ropes Ck Shanes Pk Holsworthy Prospect Plumpton Lansdowne Mt Annan Orchard Scheyville high moderate low Pk Hills Fire frequency Site
C. Obligate seeder frequency by fire frequency D. Obligate seeder frequency by site
100.0 100.0
90.0 90.0 80.0 b 80.0 70.0 70.0
60.0 a 60.0
50.0 50.0 40.0 a 40.0 30.0 30.0
20.0 20.0
10.0 10.0 Frequency of obligate seeder shrubs seeder obligate of Frequency Frequency of obligate seeder shrubs seeder obligate of Frequency 0.0 0.0 Ropes Ck Shanes Pk Holsworthy Prospect Plumpton Lansdowne Mt Annan Orchard Scheyville high moderate low Pk Hills Fire frequency Site
Figure 4.10. Frequency of native shrubs exhibiting two post-fire regeneration modes (mean + S.E., scale 0-100) in three fire frequency categories (A, C) and nine sites, three within each category (B, D). A, B, resprouters; C, D, obligate seeders. Red, high fire frequency; orange, moderate fire frequency; cream, low fire frequency.
Summary
Shrubs other than Bursaria spinosa were consistently better represented in sites subject to moderately frequent fires, than in either high or low fire frequency areas. Obligate seeder shrubs were particularly sensitive to fire frequency. High fire frequency sites had more shrubs other than Bursaria, on average, than low fire frequency sites, although these differences were not statistically significant.
106 4.3.3 Exotic shrubs
Ten exotic shrub species were recorded in study sites (Table A8.3, Appendix 8): Sida rhombifolia was the most common, occurring in five sites, followed by Olea europaea ssp. africana, Lycium ferocissimum and Araujia sericiflora (three sites each). There were many more exotic shrubs in sites which had experienced a long interfire interval, than in more frequently burnt sites. No exotic shrubs at all were found in two high fire frequency sites, and in the third, only four subplots contained exotics, giving high fire frequency areas a mean rating on the exotic shrub index of 0.1 (scale 0 -100). Moderate fire frequency sites averaged 5.1 on this index, while low fire frequency sites averaged 30.2.
It was necessary both to move the analysis to block level, and to log-transform the data, to homogenise variances.
When this was done, the analysis of variance returned significant results for fire frequency, and for site (Table 4.10, Figure 4.11). Low fire frequency sites had significantly more exotic shrubs than either high (P < 0.01) or moderate fire frequency sites (P < 0.05).
Table 4.10. ANOVA of exotic shrub frequency (ln(x+1) transformed) for sites in three fire frequency categories.
Source Sums of squares d.f. Mean square F P Fire frequency 34.5 2 17.27 9.561 < 0.05 Site (Fire frequency) 10.8 6 1.81 3.490 < 0.05 Error 9.3 18 0.52 Total 54.7 26
Cochran’s test C2,9 = 0.330, P = NS
107
A. Exotic shrub frequency by fire frequency B. Exotic shrub frequency by site
100.0 100.0
90.0 90.0
80.0 80.0
70.0 70.0
60.0 60.0
50.0 50.0
40.0 40.0
30.0 30.0 Exotic shrub frequency frequency shrub Exotic Exotic shrub frequency shrub Exotic 20.0 20.0
10.0 10.0
0.0 0.0 Ropes Ck Shanes Pk Holsworthy Prospect Plumpton Lansdowne Mt Annan Orchard Scheyville high moderate low Pk Hills Fire frequency Site
C. Exotic shrub frequency (ln(x + 1)) by fire frequency D. Exotic shrub frequency (ln(x + 1)) by site 4.50 b 4.50 4.00 4.00
3.50 3.50 3.00 a 3.00 2.50 2.50
2.00 a 2.00 1.50 1.50
1.00 1.00
0.50
0.50 + 1)) (ln(x frequency shrub Exotic Exotic shrub frequency (ln(x + 1)) (ln(x frequency shrub Exotic 0.00 0.00 Ropes Ck Shanes Pk Holsworthy Prospect Plumpton Pk Lansdowne Mt Annan Orchard Scheyville high moderate low Hills Fire frequency Site
Figure 4.11. Frequency of exotic shrubs (mean + S.E.) for three fire frequency categories (A, C) and nine sites, three within each category (B, D). A, B, untransformed data; C, D, ln(x + 1) transformed data. Red, high fire frequency; orange, moderate fire frequency; cream, low fire frequency.
4.3.4 Trees
Tree density
Nine tree taxa were recorded in the study, five of them eucalypts (Table A8.4, Appendix 8). Eucalyptus moluccana was the most common species (62% of adults), followed by E. tereticornis (16%) and E. crebra (9%). Ninety-five percent of adult trees were eucalypts, with E. eugenioides and E. fibrosa each contributing around 4% to this total. Melaleuca decora made up almost all the remaining 5%. Findings reported in detail below are for all tree species together.
Adult tree densities ranged from 129 trees/ha at Mt Annan, to 450 trees/ha at Shanes Park. Tree densities were higher, on average, in frequently burnt sites (337 trees/ha) than in moderate and low fire frequency areas (237 and 219 trees/ha respectively). These overall differences masked large differences between sites within fire frequency
108 categories, and between blocks within sites, and were not statistically significant (Table 4.11, Figure 4.12).
Table 4.11. ANOVA of tree density (trees over 10 cm DHB) for sites in three fire frequency categories.
Source Sums of squares d.f. Mean square F P Fire frequency 1040 2 520 0.845 0.475 Site (Fire frequency) 3692 6 615 6.834 < 0.001 Block (Site) 1621 18 90 3.649 < 0.001 Error 1999 81 25 Total 8351 107
Cochran’s test C3.27 = 0.115, P = NS
A. Adult trees by fire frequency B. Adult trees by site
600.0 600.0
500.0 500.0
400.0 400.0
300.0 300.0
200.0 200.0
100.0 100.0 Number >10cm trees of DBH ha / Number of trees >10cmNumber trees DBH of ha /
0.0 0.0 Ropes Ck Shanes Pk Holsworthy Prospect Plumpton Lansdowne Mt Annan Orchard Scheyville high moderate low Pk Hills Fire frequency Site Figure 4.12. Density of trees over 10cm DBH per hectare (mean + S.E.) for A, three fire frequency categories and B, nine sites, three within each category. Red, high fire frequency; orange, moderate fire frequency; cream, low fire frequency.
When eucalypts alone were considered, the pattern was similar. Average density was still highest in frequently burnt sites (297 trees/ha), with moderate and low fire frequency areas supporting 236 and 219 trees/ha respectively. These differences were not significant (F(2,6) = 0.260, P = 0.779, NS).
Basal area
Basal area figures for the nine sites all fell between 10 and 16 m2/ha. Average basal area was only marginally higher in frequently burnt sites (14.1 m2/ha) than in moderate or low fire frequency areas (each 12.2 m2/ha). Differences between fire frequency
109 categories were not significant, although there were significant differences between sites within categories (Table 4.12, Figure 4.13).
Table 4.12. ANOVA of tree basal area (m2/ha) for sites in three fire frequency categories.
Source Sums of squares d.f. Mean square F P Fire frequency 87 2 43.4 0.528 0.615 Site (Fire frequency) 493 6 82.2 5.141 0.003 Block (Site) 288 18 16.0 0.921 0.557 Error 1406 81 17.4 Total 2274 107
Cochran’s test C3,27 = 0.24, P < 0.05, α reduced to 0.01.
A. Adult tree basal area B. Adult tree basal area by site
18.0 18.0
16.0 16.0
14.0 14.0
12.0 12.0
10.0 10.0 /ha) /ha) 2 2 8.0
8.0 (m (m
6.0 6.0
4.0 4.0
Basal area of trees >10cm diam trees of area Basal 2.0 Basal area of trees >10cm diam trees of area Basal 2.0
0.0 0.0 Ropes Ck Shanes Pk Holsworthy Prospect Plumpton Lansdowne Mt Annan Orchard Scheyville high moderate low Pk Hills Fire frequency Site Figure 4.13. Basal area (m2/ha) of trees over 10cm DBH (mean + S.E.) for A, three fire frequency categories and B, nine sites, three within each category. Red, high fire frequency; orange, moderate fire frequency; cream, low fire frequency.
When trees other than eucalypts were removed from the analysis, the similarity between fire frequency categories became even more marked: basal area averaged 12.6 m2/ha in high fire frequency sites, and 12.2 m2/ha in each of the other two fire frequency categories (F(2,6) = 0.018, P = 0.982, NS).
Juveniles
The density of tree species juveniles varied widely between sites, ranging from an average of 206/ha at Mt Annan, to 1231 at Lansdowne. Juvenile densities were higher, on average, in high and moderate fire frequency sites (684 and 625/ha respectively) than in low fire frequency areas (393/ha). However again these differences masked variation
110 between sites within fire frequency categories, and between blocks within sites, and were not statistically significant (Table 4.13, Figure 4.14).
Table 4.13. ANOVA of density of juveniles (tree species individuals under 2m in height) for sites in three fire frequency categories.
Source Sums of squares d.f. Mean square F P Fire frequency 6160 2 3080 0.607 0.576 Site (Fire frequency) 30464 6 5077 5.268 0.003 Block (Site) 17348 18 964 2.088 0.014 Error 37396 81 462 Total 91368 107
Cochran’s test C3.27 = 0.162, P = NS
A. Juvenile trees by fire frequency B. Juvenile trees by site
1400.0 1400.0
1200.0 1200.0
1000.0 1000.0
800.0 800.0
600.0 600.0
400.0 400.0 Number of juveniles / ha / Number juveniles of Number of juveniles / ha / juveniles Number of 200.0 200.0
0.0 0.0 Ropes Ck Shanes Pk Holsworthy Prospect Plumpton Lansdowne Mt Annan Orchard Scheyville high moderate low Pk Hills Fire frequency Site
Figure 4.14. Density of juveniles (under 2 m in height) of all tree species per hectare (mean + S.E.) for A, three fire frequency categories and B, nine sites, three within each category. Red, high fire frequency; orange, moderate fire frequency; cream, low fire frequency.
Again, removing non-eucalypt species from the analysis blunted the differences between fire frequency categories. Moderate fire frequency sites averaged a slightly larger number of juveniles than high frequency areas (559 versus 511/ha), while juvenile numbers in infrequently burnt sites, at 391/ha, were relatively low. However again sites within fire frequency categories differed considerably on this variable, and the differences between fire frequencies categories were not significant (F(2,6) = 0.286, P = 0.761, NS).
111 Saplings
Sapling density also varied considerably between sites, ranging from an average of 29/ha at Holsworthy (a high fire frequency site) and Orchard Hills (a low fire frequency site), to 240/ha at Lansdowne. Sapling densities were lower, on average, in frequently burnt sites (75/ha) than in sites subject to moderate and low fire frequencies (134 and 137 saplings/ha respectively). However differences between fire frequency categories were not significant, and nor were differences between sites within categories (Table 4.14, Figure 4.15). Eucalypts alone presented a similar picture, and differences between fire frequency categories were again not significant (F(2,6) = 1.460, P = 0.304, NS).
Table 4.14. ANOVA of density of saplings (tree species individuals over 2m in height but less than 10cm DBH) for sites in three fire frequency categories. Block means used as replicates.
Source Sums of squares d.f. Mean square F P Fire frequency 80.2 2 40.11 0.562 0.598 Site (Fire frequency) 428.5 6 71.41 2.488 0.063 Error 516.7 18 28.71 Total 1025.4 26
Cochran’s test C2,9 = 0.347, P = NS.
A. Sapling trees by fire frequency B. Sapling trees by site
300.0 300.0
250.0 250.0
200.0 200.0
150.0 150.0
100.0 100.0 Number of saplings / haNumber / saplings of Number of saplings / ha / saplings Number of 50.0 50.0
0.0 0.0 Ropes Ck Shanes Pk Holsworthy Prospect Plumpton Lansdowne Mt Annan Orchard Scheyville high moderate low Pk Hills Fire frequency Site
Figure 4.15. Density of saplings (over 2 m in height and under 10 cm DBH) of all tree species, per hectare (mean + S.E.) for A, three fire frequency categories and B, nine sites, three within each category. Red, high fire frequency; orange, moderate fire frequency; cream, low fire frequency.
Given the considerable variation between sites in numbers of both juveniles and saplings, might a clearer pattern emerge if the ratio of saplings to all non-adult individuals of tree species, was considered? The average percentage of young trees (juveniles + saplings) which were saplings extended from 5.4% at Holsworthy, and
112 5.9% at Orchard Hills, to 43.6% at Mt Annan. Overall, saplings made up a greater proportion of the young tree population as fire frequency decreased. Per transect averages ranged from 11.7% in frequently burnt sites, to 17% in sites subject to a moderate fire frequency, to 27% in low fire frequency areas. At site level, however, these trends were not consistent, and the difference between fire frequency categories was not significant (Table 4.15, Figure 4.16). Eucalypts alone presented a similar picture, with differences between fire frequency categories again failing to reach significance (F(2,6) = 2.099, P = 0.204, NS).
Table 4.15. ANOVA of saplings as a percentage of all young trees (juveniles + saplings) for sites in three fire frequency categories. Ratios calculated at transect level.
Source Sums of squares d.f. Mean square F P Fire frequency 4419 2 2210 1.354 0.327 Site (Fire frequency) 9795 6 1632 8.658 < 0.001 Block (Site) 3394 18 189 0.950 0.524 Error 16084 81 199 Total 33691 107
Cochran’s test C3.27 = 0.131, P = NS.
A. Percentage of non-adult trees which were saplings, B. Percentage of non-adult trees which were saplings, by by fire frequency site
100.0 100.0
90.0 90.0
80.0 80.0
70.0 70.0
60.0 60.0
50.0 50.0
40.0 40.0 trees, x 100 trees, trees, x 100 trees, 30.0 30.0
20.0 20.0
10.0 10.0 Ratio of saplings to all non-adult non-adult all to saplings of Ratio Ratio of saplings to all non adult nonadult all to saplings of Ratio
0.0 0.0 high moderate low Ropes Ck Shanes Pk Holsworthy Prospect Plumpton Lansdowne Mt Annan Orchard Scheyville Pk Hills Fire frequency Site
Figure 4.16. Saplings as a percentage of all young trees (juveniles + saplings) for A, three fire frequency categories and B, nine sites, three within each category. Mean + S.E.; ratios calculated at transect level. Red, high fire frequency; orange, moderate fire frequency; cream, low fire frequency.
Summary
Although the average density of adult trees was highest on frequently burnt sites, neither tree density nor basal area differed significantly between fire frequency categories, while sites within categories were significantly different from one another.
113 Low fire frequency sites had the lowest number of juveniles and the highest number of saplings, on average, however again there was considerable variability between sites within categories, and differences between fire frequency levels were not significant. A trend towards a higher ratio of saplings to all young trees in long interfire interval sites also did not reach significance, perhaps due to the low level of saplings, and of saplings relative to juveniles, at Orchard Hills.
4.3.5 Grasses
Grass cover index
Grass cover was relatively unaffected by fire frequency, which was not significant in the analysis (Table 4.16). Fire frequency means all fell between 60 and 70 (scale 0 – 100), while site means varied from 54.1 to just below 78.8 (Figure 4.17).
Table 4.16. ANOVA of grass cover index for sites in three fire frequency categories; block means used as replicates.
Source Sums of squares d.f. Mean square F P Fire frequency 240.8 2 120.4 0.583 0.587 Site (Fire frequency) 1238.9 6 206.5 2.355 0.074 Error 1578.3 18 87.7 Total 3058.0 26
Cochran’s test C2,9 = 0.312, P = NS
A. Grass cover by fire frequency B. Grass cover by site
100.0 100.0
90.0 90.0
80.0 80.0
70.0 70.0
60.0 60.0 50.0 50.0 40.0 40.0 30.0 Grass index cover
Grass cover index 30.0 20.0 20.0 10.0 10.0 0.0 high moderate low 0.0 Ropes Ck Shanes Pk Holsworthy Prospect Plumpton Pk Lansdowne Mt Annan Orchard Hills Scheyville Fire frequency Site Figure 4.17. Grass cover index (mean + S.E., scale 0-100) for A, three fire frequency categories and B, nine sites, three within each category. Red, high fire frequency; orange, moderate fire frequency; cream, low fire frequency.
114 Grass species composition
The three most common grass taxa were Themeda australis, Microlaena stipoides and Aristida spp.
Themeda australis
Themeda dominated over 50%, and up to 83%, of subplots in sites with a high or moderate fire frequency (Figure 4.18). In contrast, on all low fire frequency sites the Themeda index was less than 50, dropping as low as 0.5 at Scheyville, where an estimated interval of 50 years pertained. The index differed significantly between fire frequency categories ((high = moderate) > low), but not between sites within categories (Table 4.17).
A. Themeda australis dominance by fire frequency B. Themeda australis dominance by site
100.0 100.0
90.0 a a 90.0
80.0 80.0
70.0 70.0 60.0 b 60.0 50.0 50.0 dominance index dominance
dominance index 40.0 40.0
30.0 30.0
20.0 20.0 Themeda Themeda 10.0 10.0
0.0 0.0 Ropes Ck Shanes Pk Holsworthy Prospect Plumpton Lansdowne Mt Annan Orchard Scheyville high moderate low Pk Hills Fire frequency Site Figure 4.18. Themeda index (mean + S.E., scale 0-100) for A, three fire frequency categories and B, nine sites, three within each category. Red, high fire frequency; orange, moderate fire frequency; cream, low fire frequency. In chart A, means labelled with different letters are significantly different at at least P = 0.05 (SNK).
Table 4.17. ANOVA of Themeda index for sites in three fire frequency categories. Block means used as replicates.
Source Sums of squares d.f. Mean square F P Fire frequency 14282 2 7141 9.662 0.013 Site (Fire frequency) 4434 6 739 2.195 0.092 Error 6059 18 337 Total 24776 26
Cochran’s test C2,9 = 0.399, P = NS
115 Microlaena stipoides
Although the Microlaena index was over four times higher, on average, on low fire frequency sites than where fire frequency was high or moderate, these differences were not statistically significant (Table 4.18, Figure 4.19). There were significant differences between sites within fire frequency categories, and between blocks within sites Microlaena stipoides dominated over 90% of subplots in one low fire frequency site, 28% in another site in this category, but less than 15% of subplots in all other sites.
Table 4.18. ANOVA of Microlaena index for sites in three fire frequency categories.
Source Sums of squares d.f. Mean square F P Fire frequency 26415 2 13207 1.636 0.271 Site (Fire frequency) 48428 6 8071 23.451 < 0.001 Block (Site) 6195 18 344 3.902 < 0.001 Error 7144 81 88 Total 88182 107
Cochran’s test C3,27 = 0.157, P = NS
A. Microlaena stipoides dominance by fire frequency B. Microlaena stipoides dominance by site
100.0 100.0
90.0 90.0
80.0 80.0
70.0 70.0
60.0 60.0
50.0 50.0 dominance index dominance dominance index dominance 40.0 40.0
30.0 30.0 20.0 20.0 Microlaena Microlaena Microlaena 10.0 10.0 0.0 0.0 Ropes Ck Shanes Pk Holsworthy Prospect Plumpton Lansdowne Mt Annan Orchard Scheyville high moderate low Pk Hills Site Fire frequency Figure 4.19. Microlaena index (mean + S.E., scale 0-100) for A, three fire frequency categories and B, nine sites, three within each category. Red, high fire frequency; orange, moderate fire frequency; cream, low fire frequency.
Aristida ssp.
Aristida ssp. were found in all sites, but did not dominate more than 25% of subplots in any site. The Aristida index did not differ between fire frequency categories, although
116 there were significant differences between sites within categories, and between blocks within sites (Table 4.19, Figure 4.20).
Table 4.19. ANOVA of Aristida index for sites in three fire frequency categories.
Source Sums of squares d.f. Mean square F P Fire frequency 2737 2 1368 1.133 0.382 Site (Fire frequency) 7246 6 1208 4.847 0.004 Block (Site) 4485 18 249 3.096 < 0.001 Error 6519 81 81 Total 20988 107
Cochran’s test C3,27 = 0.151, P = NS
A. Aristida ssp. dominance by fire frequency B. Aristida ssp. dominance by site
100.0 100.0
90.0 90.0
80.0 80.0
70.0 70.0
60.0 60.0
50.0 50.0
40.0 dominance index dominance
dominance index dominance 40.0
30.0 30.0
20.0
20.0 Aristida Aristida
10.0 10.0
0.0 0.0 Ropes Ck Shanes Pk Holsworthy Prospect Plumpton Lansdowne Mt Annan Orchard Scheyville high moderate low Pk Hills Fire frequency Site Figure 4.20. Aristida index (mean + S.E., scale 0-100) for A, three fire frequency categories and B, nine sites, three within each category. Red, high fire frequency; orange, moderate fire frequency; cream, low fire frequency.
Summary
There were no significant differences between different levels of fire frequency for grass cover. In terms of grass species composition, there was significantly more Themeda in sites with a high and moderate fire frequency, than in sites with a long interfire interval. Microlaena showed the opposite trend, though this was not consistent or significant.
117 4.4 Discussion 4.4.1 Bursaria
It was hypothesised, on the basis of the literature on shrub encroachment together with vital attributes analysis, that the frequency, density and dominance of Bursaria would all increase with decreasing fire frequency. This hypothesis was supported by the data.
The effect of fire frequency on Bursaria was large and consistent, across both sites, and measures. The differences were readily apparent to the eye: the character of sites from which fire had been excluded for many years was fundamentally different to that of those subject to regular burning (Figures 4.21 and 4.22). Bursaria was not eliminated from frequently burnt remnants, again in line with predictions based on the vital attributes model – it was in fact the most common shrub in the two most frequently burnt sites. However in moderate and high fire frequency sites its distribution was limited, and often patchy. Data on spatial pattern of Bursaria is presented in Chapter 5.
In a yet-to-be published study in Scheyville National Park, Bursaria cover-abundance decreased after a burn (C. Morris pers. comm. 2004). As all infrequently burnt sites in the current study were also recently burnt, Bursaria cover in some long unburnt Cumberland Plain Woodland remnants may be even higher than that reported here. The same may be true for survey sites as time-since-fire increases: many Bursaria resprouts at Orchard Hills and Scheyville had grown to over 2 m in height when these sites were sampled for fuel load 2.5 years post-fire.
Although the current study was not longitudinal, documented observations from past years at two of the low fire frequency study sites, Mt Annan and Scheyville, confirm that Bursaria has expanded in recent decades. Benson and Howell (2002:635) provide photographs from 1988 and 2002 at Mt Annan, showing a change from open grassy understorey to Bursaria patches. Prof Harry Recher (pers. comm. 2004) reports that the Bursaria expansion during the 1970s and 80s at Scheyville “reached a stage when I was doing my bird work that I needed to cut paths through the thicket.”
The finding of shrub ‘thickening’ in the absence of fire is consistent with findings in other parts of Australia and the world (Section 4.1.1).
118
Figure 4.21. Frequently burnt woodland at Ropes Creek, showing Bursaria patches in a predominantly open grassy understorey.
Figure 4.22. Infrequently burnt woodland at Orchard Hills, showing dominance of Bursaria in the understorey. This photo was taken 2.5 years after a wildfire following a 30+ year interfire interval.
119 It has been suggested that shrub thickening may be a result of the increased CO2 in the atmosphere (Berry and Roderick 2002, Bond et al. 2003). This factor may have contributed to the build-up of Bursaria in infrequently burnt remnants. If so, regular burning seems to be keeping this effect at bay. However Bursaria encroachment in Cumberland Plain Woodland may not be new. In 1817, after a visit to a farm near Liverpool, Allan Cunningham wrote: “Like other farms in the neighbourhood it is overrun with the Bursaria spinosa, now in fruit” (quoted in Benson and Howell 2002).
Might variations in vegetation type, or grazing history, account for the observed differences in Bursaria abundance between remnants? This question will be explored in the next chapter.
4.4.2 Other native shrubs
Species richness of native shrubs in study sites was relatively low: in no site did it average above eight species per 200 m2. By contrast, average shrub species richness in shrubby woodland on sandstone was 51.3 per 1000 m2 (Rice and Westoby 1983). The importance of leguminous species in CPW (Section 3.4.4) was reflected in the finding that the nine most abundant shrub taxa, after Bursaria, were legumes.
Species richness and frequency of native shrubs other than Bursaria varied with fire frequency: both high and low fire frequencies were associated with a lesser abundance of these shrubs. The hypothesis, based on the vital attributes analysis in Chapter 3, that sites burnt at moderate frequency would support more of these shrubs than more, or less, frequently burnt areas was therefore supported. So too was the hypothesis that obligate seeder shrubs would be more sensitive to variations in interfire interval than resprouters.
The lack of obligate seeders in very frequently burnt areas is consistent with the life history finding that many of these species need several years to reach reproductive maturity. Very frequent burning at Ropes Creek and Shanes Park means seedlings of these species would often have been subject to a second fire before they were sufficiently developed to flower and set seed.
The relative abundance of obligate seeders at Holsworthy and Prospect, both sites whose recent fire history has included a series of very short interfire intervals, requires
120 some explanation. The ability of soil-stored seed of plant functional type 4 obligate seeders to persist through more than one fire (Section 3.4.5) undoubtedly helped these species to survive frequent burning in these sites. In addition, frequent burns may have been patchy. Thus fire frequency may have been lower, at a small scale, than reports of fire at a landscape scale might suggest, allowing sensitive species to reach maturity at certain points across the landscape. Prospect is of particular interest, as annual to biannual fuel reduction burning was carried out there for at least 20 years prior to 1985 (Thomas 1994). Most intervals in the sampling area since this regime ceased have been several years longer than juvenile periods: the result appears to be that these shrubs have recovered extremely well, giving this location the highest frequency of obligate seeders of any study site. This outcome demonstrates the effectiveness of variable intervals in maintaining populations of hard-seeded and quick-maturing legumes, even where some intervals are very short.
Obligate seeder shrubs were also more sensitive to low fire frequency than resprouters: long periods of fire exclusion were associated with a lower abundance of these shrubs, even relative to frequently burnt areas. The low frequency of obligate seeder shrubs in infrequently burnt areas suggests these species may be adopting a ‘live fast, die young’ strategy. This supposition is consistent with observations of deaths of obligate seeder species in remnants which had not burnt for 8 to 10 years (Sections 3.3.9, 3.4.6). Soil- stored seed might be expected to persist for some years after adult plants die, ready to germinate when cued by fire. The longest interfire interval in the study was approximately 50 years at Scheyville. Here, very few obligate seeder legumes were found along transects, although several species occur on the site. One explanation for these findings is that few soil-stored seeds survived through this protracted period.
Little is known about seed longevity. Experimental studies generally involve seed burial for periods of a year or two, with survivorship curves used to extrapolate ability to persist long-term. Investigation of a range of non-leguminous sandstone species showed a wide range of responses: some species showed virtually no loss of viability over two years; estimated half-lived for those that did ranged from 0.4 to 7.6 years (Auld et al. 2000). The half-life of the seeds of Acacia suaveolens, a short-lived sandstone obligate seeder legume has been estimated, on the basis of seed losses after two years of burial, at 10.7 years. This equates to three percent of seeds remaining after 50 years (Auld 1986), suggesting that although this species may be able to survive very
121 long interfire intervals, its abundance might well be reduced. Seed longevity estimates of 30-40 years are considered reasonable for hard-seeded species (DEC 2002, Bradstock and Kenny 2003), however there may well be variability between species. Half-lives of fynbos legumes in South Africa varied from 3.3 to 13.6 years (Holmes and Newton 2004). Floyd (1976) found no seed of a common quick-growing wattle in the soil of wet sclerophyll sites near Coffs Harbour with a time-since-fire of 30 years, although it was abundant in the seedbank of nearby sites which had burnt 14 years previously. There may also be differences between habitats. Clay soils may be less conducive to seed persistence than sandy soils. Cumberland Plain ‘fire weed’ obligate seeder legumes produce seed rapidly post-fire (Chapter 3); seed durability may be a trade-off. Moles et al. (2003) found bigger seeds survived longer in the soil, on average, than smaller ones. CPW obligate seeder legumes Dillwynia sieberi and Dillwynia tenuifolia have seeds weighing approximately 4.0 mg (B. Lomov, unpub data 2004) and 3.0 mg (Rymer 1999) respectively, towards the bottom of the range reported for sandstone legumes by Auld and O’Connell (1991). Grevillea juniperina seeds were the lightest of seven Sydney Grevillea species studied by Morris (2000). This is an area for further research.
Might obligate seeders have been present in low fire frequency sites as soil-stored seed which was not visible above ground during sampling? While this possibility cannot be discounted, it is hard to see why ungerminated seed should persist in low frequency sites to a greater extent than in those subject to a moderate or high fire frequency. In fact, the opposite is more probable, as fire-related germination cues were likely to have been stronger in low fire frequency sites. These sites were all burnt approximately 12 months prior to sampling, whereas two of the three moderate fire frequency sites had only been partially burnt in recent years. The recent fires in low fire frequency sites were all of moderate to high intensity, whereas those in low and moderate fire frequency sites, with the exception of Prospect, were low intensity arson fires. Study of the effects of different levels of heat on legume seeds from Sydney sandstone habitats has led to the conclusion that moderate intensity fires should promote more germination than fires of low intensity (Auld 1987, Auld and O’Connell 1991).
Another possible explanation for the relative dearth of obligate seeder shrubs in low fire frequency sites is that these species have suffered disproportionately from competition
122 with Bursaria. This question will be explored in the next chapter, along with the effects of grazing and vegetation type on shrub species abundance and composition.
The finding that shrub abundance was highest in sites with a history of moderately frequent burning accords with those of Kitchin (2001), who found highest woody plant densities at Guy Fawkes National Park in sites where fire had been moderately frequent.
4.4.3 Exotic shrubs
The hypothesis that frequent burning would be associated with weediness was not supported by the data, in fact the opposite was found. Frequently burnt remnants were virtually free of woody weeds. This finding provides much-needed information on the response of woody exotics to fire cycles, and should help allaying the fears, noted in the Introduction, that frequent burning will encourage weeds.
It may be that a number of woody weed species common in CPW are T or R species (Noble and Slatyer 1980), able to recruit between fires: this trait would give them an advantage, in the absence of fire, over all other shrub species except Bursaria.
The significantly higher frequency of exotic shrubs in areas which had not burnt for many years suggests that these areas may be at increased risk of invasion by the ever- growing number of naturalised exotic shrub species which pose a particular threat in urban remnants. In the semi-arid rangelands, exotic shrubs are increasingly invading grassy areas from which fire has been excluded (Noble and Grice 2002).
4.4.4 Trees
Tree density findings did not support the hypothesis that frequent fire is detrimental to trees, in fact the trend was in the opposite direction, with frequently burnt remnants supporting more adult trees than sites subject to a moderate or low fire frequency. This finding is contrary to the predictions of woody thickening outlined in the Introduction. Fire-related woody plant encroachment in CPW thus appears to be limited to the shrub layer.
Canopy eucalypts in grassy ecosystems in other parts of Australia also seem to be unaffected by frequent fire. In the Northern Territory, savanna eucalypts were not significantly disadvantaged by annual burns at Solar Village near Darwin, in fact the
123 trend in this site was towards a higher eucalypt density in regularly burnt areas (Woinarksi et al. 2004). In eucalypt woodland at Munmarlary, canopy stem density remained stable over the course of a 25 year study, irrespective of fire regime; the limiting factor here appeared to be root competition between canopy individuals (Russell-Smith et al. 2003). A model parameterised for savanna country at Kidman Springs, also in the Northern Territory, found tree cover was not affected by mild fires regardless of frequency, although intense burns reduced tree cover even if these fires were only occasional (Liedloff et al. 2001).
The large and significant differences in tree density between sites within fire frequency categories probably reflect different management histories: all sites have almost certainly been logged at some time during the two centuries of European settlement.
There was less divergence between sites, and between fire frequency categories, in basal area than in tree density. In other words, sites tended to have many relatively small trees, or a smaller number of larger ones. Again, however, there was no indication that frequent fire was disadvantaging trees.
Using Governor Phillip’s description of distances between trees on the Cumberland Plain in 1789 (Section 2.5.1) Benson (1992) calculated that tree density prior to settlement would have been somewhere between 68 and 272 trees/ha. Five sample sites from the current study fell within this range, while tree density was greater than 272 in four sites. Benson’s pre-settlement tree density estimate is well above that of Benson and Redpath (1997), who suggest that grassy woodlands may have supported approximately 30 trees/ha. This estimate is based on Cunningham’s description of woodlands in the upper Hunter Valley in 1827, soon after settlers arrived, and on the average density on the lowland Gippsland Plains in Victoria calculated from historical maps (Lunt 1997d). The lowest tree density recorded in this study is 129 trees/ha at Mt Annan, while the average for all sites is 279. It seems likely that current densities, at least at some sites, and probably at all, are well above those which pertained prior to European colonisation.
Still, density may be decreasing. Average tree density per hectare in 100 CPW remnants surveyed during the 1970s was 378, while basal area averaged 14.4 m2/ha (Benson 1992; these figures may be slightly inflated through inclusion of Acacia species not counted as trees in the current study), giving an average DBH of 22.0 cm.
124 Equivalent averages from the nine sites in the current study are 279 and 12.9 m2/ha, an average DBH of 24.3 cm. There is a slight trend over the last three decades towards smaller numbers of larger trees.
Trees are long-lived organisms, so population trends may take many years to become apparent. The non-adult population is therefore of interest, as trends in the adult population of the future may emerge there.
In terms of juveniles, the story was similar to that for adult trees: not only was there no evidence that frequent fire was disadvantaging tree species, the trend was in the opposite direction. This trend was also found in the Kapalga experiment (Williams 2000). While fire may kill some juveniles (Noble 1984), it may also encourage seedling recruitment (Wellington and Noble 1985, Hill and French 2004). Thus recruitment into the juvenile population may more than offset mortality in frequently burnt areas.
The trend reversed, however, for saplings. Although again not significant, there were fewer saplings in frequently burnt areas. This trend was even more pronounced when the ratio of saplings to all non-adult trees was considered. Overall, more young eucalypts were ‘getting away’ on their journey towards the canopy in less frequently burnt sites than in more frequently burnt ones.
Neither trend, however, was significant: the hypothesis that frequent fire prevents suppressed seedlings from moving into and through the sapling stage was therefore rejected. While the relative abundance of saplings was high in two of the three low fire frequency sites, it was very low at the third site, Orchard Hills. This site had burnt in a severe summer wildfire, whereas the fires at the other two low fire frequency sites, Scheyville and Mt Annan, were moderate intensity planned burns. The other site with a low number of saplings, and a low sapling to non-adult ratio, was Holsworthy, the only other site which had recently burnt in a summer wildfire. Perhaps infrequent burning is associated with increased recruitment of eucalypts into the adult population, but this process can be severely disrupted by intense summer wildfires.
The Liedloff et al. (2001) findings in semi-arid savanna suggest that fire intensity may have a greater affect on canopy cover than fire frequency. However even in the face of severe fire, adult eucalypts may persist for many years. At Kapalga, tree stem survival was greatest in the unburnt and early dry season (ie low intensity) treatments, whereas late dry season fires, and a wildfire which followed five years of fire exclusion, both led
125 to a one third reduction in tree stems. At whole tree level, however differences were small (Williams et al. 1999). Modelling indicates that even with repeated late season fires, a stable population of mid-sized, fire-tolerant trees would persist (Andersen et al. 2003).
Overall, the lack of a significant relationship between the various tree parameters and fire frequency suggests that Cumberland Plain Woodland tree species have recovery and recruitment strategies which allow them to cope with a wide range of fire regimes. This accords with the VCT vital attribute status accorded to the three CPW dominants (E. moluccana, E. tereticornis and E. crebra) in the NSW Flora Fire Response Database (DEC 2002). Species in these categories are considered unlikely to decline under any fire regime. The strategies adopted by CPW tree species may assist them to persist in the face of various aspects of the Cumberland Plain environment in addition to fire: episodic drought, harsh temperatures, and soils which can be rock hard, or waterlogged (Sections 2.2 and 2.3).
4.4.5 Grasses
The hypothesis that grass cover would increase with frequency of burning was not supported by the data: although there was a slight trend in the predicted direction, it was not significant. Although grasses visually dominated the understorey in frequently burnt areas – a trend which has been noted elsewhere (Westfall et al. 1983, Bowman et al. 1988, Birk and Bridges 1989, Taylor 1992) – this may reflect absence of shrubs, rather than enhanced grass cover.
Grass species composition, however, did differ with fire frequency, in that the dominance of Themeda australis was significantly lower in low fire frequency sites than in those subject to a high and moderate fire frequency. The hypothesis that frequent burning favours Themeda was therefore supported.
A number of studies have documented a decline in Themeda abundance and vigour with increasing time-since-fire, however so far as I am aware its response to a series of fires has not previously been studied in Australia. Locally, S. Clarke (2003) found cover- abundance of Themeda at Holsworthy was higher in recently-burnt than in unburnt sites. At Gellibrand Hill in Victoria, Themeda abundance decreased in unburnt woodland areas, while Microlaena stipoides increased (Robertson 1985). Prober (1996) observed
126 that Themeda appeared to be favoured over Poa sieberiana in White Box woodland sites in NSW which were regularly mown or burnt. Perhaps most tellingly, a study of Themeda at various times-since-fire near Melbourne found a decline in numbers of tussocks, in numbers of tillers per tussock, and in number of inflorescences with time since fire. Significant declines were first observed at five years post-fire. By 11 years without disturbance, almost all vegetative matter in tussocks was dead, and tussock numbers per unit area were half those in recently-burnt areas. Long-unburnt tussocks were significantly slower to recover when a fire did finally occur, and had fewer tillers (Morgan and Lunt 1999). Similar responses have also been reported from South Africa, where some forms of Themeda triandra “become moribund in the absence of fire.” Periodic defoliation is needed to prevent self-shading which suppresses tiller production (Bond 1997:434). See Figure 4.23.
Figure 4.23. Themeda understorey at Lansdowne. Left, long unburnt Themeda australis. Right, patchiness resulting from different times-since-fire. The green patch to the left of this picture is rejuvenated Themeda. It was burnt approximately 18 months before the photos were taken.
As well as contributing to vigour of existing tussocks, fire may play a role in seedling establishment. Wood (2001) recorded a strong germination response in Themeda to smoke, although S. Clarke (2003) found germination in Themeda was not affected by either heat or smoke. Studies elsewhere have also obtained mixed results (Baxter et al. 1994, Clarke et al. 2000).
Summer-growing (C4) grasses such as Themeda use water more efficiently and have lower nutrient requirements than all-season and winter-growing (C3) grasses like Microlaena stipoides and Austrodanthonia species (Nadolny et al. 2003). These
127 characteristics may give these species a competitive advantage in a frequently-burnt environment.
Recent work in White Box woodlands suggests that Themeda may play a key role in ecosystem function, regulating nitrogen to the advantage of native perennials over exotic annuals (Prober et al. 2002b, Prober et al. 2004). Morgan (1998d) found a significant negative correlation between Themeda canopy cover and exotic species richness in a Victorian grassland.
4.5 Conclusion
This study found a strong association between interfire interval and the structure, native shrub species richness and abundance, and dominant grass species of Cumberland Plain Woodland remnants. Differences between sites with a 20+ year interfire interval, and more frequently burnt sites, were most pronounced. Low fire frequency sites were dominated by Bursaria spinosa, had relatively few other native shrubs but more woody exotics, and did not have a Themeda-dominated ground layer. Sites with mostly 1-3 year interfire intervals also had a low frequency of native shrubs relative to sites with predominantly 4-10 year intervals, although sites subject to high and moderate fire frequencies were both dominated by Themeda. Frequently burnt sites had virtually no woody exotics.
Tree density varied widely between sites but fire frequency was not a factor in these differences, nor in the more subdued variations in basal area. Juvenile and sapling numbers also fluctuated between sites, but not with fire frequency. Interfire interval did not affect grass cover.
The consistent differences between sites in the three fire frequency categories suggest that Cumberland Plain Woodland can exist in different states. A state and transition model for CPW is presented in Chapter 10.
128 CHAPTER 5 FIRE FREQUENCY AND WOODLAND LANDSCAPES: SPATIAL PATTERN, SHRUB SPECIES COMPOSITION, ASSOCIATIONS AND GRAZING
5.1 Introduction
Investigation of the effects of fire frequency in Cumberland Plain Woodland landscapes continues in this chapter, through further analysis of the data collected in the nine study sites described in Chapter 4. Factors other than fire which may influence CPW structure and floristics are considered, and explanations for the findings reported in the previous chapter are canvassed.
5.1.1 Spatial pattern
An understanding of vegetation patterns across a landscape can help illuminate ecological processes (Turner et al. 2001).
Bursaria spinosa, the dominant shrub in Cumberland Plain Woodland, was more prominent in low fire frequency sites than in areas which had been burnt more regularly (Chapter 4). The tendency for this species to form thickets has been noted by a number of authors (Benson 1992, James et al. 1999, Benson and Howell 2002). Does clumping occur consistently across CPW sites? Does it vary with fire frequency? The question of whether trees occur in clumps in CPW is also investigated, as is the influence of fire frequency on tree spatial pattern.
5.1.2 Species composition
Shrub species richness and frequency in CPW varied with fire frequency (Chapter 4), but what about species composition? Studies from elsewhere in south-eastern Australia have found significant differences in floristics between areas with different burning
129 histories (Cary and Morrison 1995, Morrison et al. 1995a, Kitchin 2001, Watson and Wardell-Johnson 2004).
The relationship between vegetation type and shrub species composition is also investigated in this chapter. The vegetation of study sites spanned three of Tozer’s vegetation categories (Tozer 2003): Shale Plains Woodland, Shale Hills Woodland, and Shale Gravel Transition Forest (Table 4.1). If shrub species composition does vary between study sites, might this reflect differences between vegetation types, rather than differences between fire frequency categories?
5.1.3 Associations
Bursaria and other shrubs
Understorey species richness can be greatly reduced through competition with resprouting dominants in heath environments (Bond and Ladd 2001). Bursaria density was high in long unburnt sites, while the abundance of other shrubs, particularly obligate seeders, was low (Chapter 4). What role might competition between Bursaria and other shrubs have played in producing this outcome? This question is explored through analysis of the association between Bursaria and other shrubs in moderate fire frequency sites, where both elements are prominent in the landscape. If shrubs other than Bursaria are less abundant in Bursaria thickets than elsewhere, then the hypothesis that Bursaria has the potential to out-compete other shrub species remains tenable – although other explanations would also need to be considered (Kershaw 1973). On the other hand, if other shrubs are more abundant where Bursaria is most dominant, the competition hypothesis can be rejected.
Adult trees, saplings and juveniles
Early reports of CPW suggest these woodlands were dominated by a relatively small number of large trees (Benson 1992). It seems likely that the trees in most if not all study sites are regrowth following logging since European settlement. Are the trees in CPW remnants ‘in transition’ towards a landscape dominated by small numbers of large trees?
130 The finding that tree basal area was considerably less variable across sites than tree density (Chapter 4) suggests that the CPW environment may have a ‘carrying capacity’ in terms of tree basal area. If tree basal area is indeed limited by environmental resources, then as some trees grow bigger, others will be out-competed and either die or remain stunted, and a smaller number of large trees may indeed become dominant. In this chapter the effect of tree basal area on the abundance of juveniles and saplings at subplot and site level, is investigated.
5.1.4 Grazing pressure
Studies have shown that grazing can cause an increase in woody plant density. Grazing reduces grass biomass and can thus directly reduce the competition faced by woody seedlings, favouring shrubs and trees (Madany and West 1983, Harrington 1991). Grazing can also indirectly encourage tree and shrub establishment by limiting grass fuel, and thus the frequency and intensity of fires that might otherwise have killed young trees and shrubs, or prevented juveniles reaching adult size (Hodgkinson and Harrington 1985, Scholes and Archer 1997). However grazing – or perhaps more correctly browsing – may also be associated with a decreased abundance of shrubs (Leigh and Holgate 1979, Henderson and Keith 2002, P.J. Clarke 2003).
Sites in the current study have almost certainly all been grazed by domestic animals at some time during the last 200 years, although this is no longer the case. Grazing by native and/or feral animals, particularly macropods, rabbits and hares, still occurs to varying degrees. Given the limited number of remnants available for study, it was not possible to control for grazing. Might differences in grazing pressure explain differences in shrub abundance between fire frequency categories?
5.1.5 Study questions
Specific questions addressed in this chapter include:
• Does the spatial pattern of Bursaria spinosa plants differ in areas subject to differing fire frequencies? • Does the spatial pattern of trees differ in areas subject to differing fire frequencies?
131 • Does shrub species composition vary with fire frequency? It was hypothesised, on the basis of findings from elsewhere in south-eastern Australia, that it would indeed differ. • Do variations in shrub species composition correspond to variations in vegetation type? • Does the abundance of shrubs other than Bursaria spinosa decrease as Bursaria dominance increases? • Does tree basal area affect the density of juveniles and saplings? • Might variations in grazing pressure have influenced the abundance of Bursaria or other shrubs in study sites?
5.2 Methods
The analyses reported in this chapter use the data generated in the study described in Chapter 4.
5.2.1 Spatial pattern
Spatial pattern was assessed statistically for Bursaria spinosa (plants over 80 cm in height), and for adults trees (members of tree species over 10 cm in diameter), using data from the T-square sampling procedure described in Section 4.2.2. Ludwig and Reynolds (1988:56-7) present an index of spatial pattern, C, which is a ratio of squared point-to-individual distances and squared individual-to-neighbour distances:
2 2 2 C = ∑ [xi /( xi + ½ yi )] n where
• xi = distance from random point i to nearest individual i
• yi = T-square distance from individual i to neighbour • n = sample size
This index will be approximately 0.5 for random patterns, significantly less than 0.5 for uniform patterns, and significantly greater than 0.5 for clumped patterns. Ludwig and Diggle found this index powerful in detecting both clumped and uniform patterns (unpublished study cited in Ludwig and Reynolds 1988).
132 Departures from randomness can be tested for significance, using the formula: z = (C – 0.5)/√(1/12n). (Ludwig and Reynolds 1988:58). z will be zero when C = 0.5 and the pattern is random, increasingly positive as the pattern tends towards clumping, and increasingly negative as the pattern tends towards evenness. The statistical significance of z can be obtained from a probability table for the standard normal distribution. Using a two- tailed test, z = 1.96 at P = 0.05.
A nested analysis of variance was used to test for differences in spatial pattern between fire frequency categories, and sites within categories.
5.2.2 Shrub species composition
The spatial analysis program PRIMER was used to explore floristic patterns in the shrub data. Frequency scores were used because this measure was available for all shrub species, including Bursaria and exotics. Data from the four transects in each block were combined for this analysis, as some individual transects contained either no shrubs, or very few species.
Similarity between plots was quantified using the Bray-Curtis metric (Bray and Curtis 1957), a method that has performed consistently well in a variety of tests and simulations on different types of data (Faith et al. 1987). ‘Joint absences’ (shared zeros) have no effect on the Bray-Curtis metric, and abundance is considered in weighing species contributions (Clarke and Warwick 2001).
Data were clustered through group averaging, and visually presented through a multi- dimensional scaling (MDS) ordination. MDS uses rank correlations, and seeks to provide an accurate visual representation of the similarity between a set of samples (in this case, blocks), on the basis of their attribute (species) profiles. Representations may be in two or more dimensions. The stress value associated with a particular representation provides a measure of its accuracy: for two-dimensional representations, stress values above 0.2 indicate limited ability to portray sample relationships (Clarke and Warwick 2001).
Relationships between fire frequency and shrub species composition were further assessed through multi-variate analysis of similarities, using the ANOSIM option in PRIMER. This process uses ranks to generate a ‘Global R’ summarising the differences
133 between samples with differing values on a specified variable (in this case, fire frequency). The distribution-free nature of the test was appropriate given the prevalence of zeros in the dataset.
Finally, similarity and dissimilarity percentages were calculated for the three fire frequency categories using the SIMPER option in PRIMER. SIMPER breaks down the Bray-Curtis similarity/dissimilarity matrix (dissimilarity is simply 1 – similarity) into contributions from each species. By looking at the percent contribution a species makes to the dissimilarity between two groups of samples (for example, blocks exposed to a high frequency of fire and blocks exposed to moderately frequent burning), species can be listed in order of importance in terms of discriminating between groups (Clarke and Gorley, 2001).
The influence of vegetation type was assessed through examination of the ordination pattern, and through analysis of similarities (ANOSIM). Final vegetation maps for Western Sydney (Tozer 2003) were published by NPWS during the course of the study (NPWS 2004a). These maps were used to allocate each block to a vegetation type.
5.2.3 Associations
Associations were assessed through analysis of data from a random sample of subplots along the randomly located transects described in Section 4.2.2.
Bursaria and other shrubs
To assess the association between Bursaria and other shrubs, a sample of subplots with a reasonable representation of Bursaria, Bursaria-free patches, and other shrubs, was needed. The analysis was therefore limited to the three moderate fire frequency sites. A separate analysis was conducted for each site. Shrub density at Prospect was considerably higher than that at the other two sites: Prospect had burnt approximately 13 months prior to sampling, leaving most species represented by many small seedlings and/or root resprouts.
Six hundred 2 m x 2 m subplots were available for sampling at each site, 50 from each of the 12 randomly-located transects. A random sample of 200 subplots was drawn from each site. Univariate permutation tests, using the program Resampling Stats
134 (Blank et al. 2001), were used to assess the significance of differences in shrub density between:
• subplots in which Bursaria was present, and subplots in which Bursaria was absent; and
• subplots in which Bursaria cover was present but under 80 cm in height, and subplots in which Bursaria was over 80 cm in height.
Obligate seeder species and resprouting species (other than Bursaria) were analysed separately.
Permutation tests are distribution-free and involve permuting a data set many times (Manly 1991). They are able to accommodate samples of different sizes, and multiple zeros in the data set. Data from the relevant subplots were listed in three columns in an Excel spreadsheet. For each comparison, the difference between the means of the appropriate groups was calculated. Subplot data were then permuted 5000 times, each time the difference between means was again calculated. The significance level for the actual difference was assessed by finding its place in the distribution developed over the multiple permutations.
Adult trees, saplings and juveniles
Sites chosen for the sub-plot level analysis were those with a good representation of both large trees and saplings: Ropes Creek, Prospect, Lansdowne and Scheyville. Again, a separate analysis was conducted for each site, using data from 200 randomly- chosen 2 m x 6 m subplots. Permutation tests were used to assess the significance of differences in number of a) saplings and b) juveniles between:
• subplots with adult trees present, and subplots with no adult trees; and 2 • subplots in which adult tree basal area was under 800 cm and subplots in which adult tree basal area was over 800 cm2.
(The terms ‘adult’, ‘juvenile’ and ‘sapling’ are defined in Section 4.2.3.)
At site level, the ratio of non-adult trees (juveniles + saplings) to adult trees was calculated. The relationship between this ratio and adult tree basal area was assessed through a Pearson correlation coefficient, as was that between basal area and number of non-adult trees.
135 5.2.4 Grazing pressure
To get some idea of the level of grazing pressure in study sites, a grazing index was constructed. This index was the sum of two ratings:
• Recency of grazing by domestic stock. Information from managers, local residents, and documentation was used to allocate each site to one of the following categories. 1 = domestic grazing ceased over 20 years prior to sampling. 2 = domestic grazing ceased between 10 and 20 years prior to sampling. 3 = domestic grazing ceased between 5 and 10 years prior to sampling. 4 = domestic grazing ceased less than 5 years prior to sampling.
• Current level of grazing. Mammals which currently graze CPW remnants include macropods, rabbits and hares. On-site observation of animals and scats, along with information from managers, local residents, and documentation, were employed to allocate each site to one of the following categories: 1 = Current level of grazing is estimated to be low 2 = Current level of grazing is estimated to be moderate 3 = Current level of grazing is estimated to be high
Clearly these scales are rough, and site allocations are based on subjective information. Still, the index provides some clues as to the relative status of study sites with respect to grazing. Ratings for study sites are given in Table 5.1.
Table 5.1 Ratings of study sites with respect to past and present grazing.
Site Recency of grazing Current grazing level Score on grazing index Ropes Creek 1 2 3 Shanes Park 2 2 4 Holsworthy 1 2 3 Prospect 1 1 2 Plumpton Park 1 2 3 Lansdowne 1 1 2 Mt Annan 2 3 5 Orchard Hills 4 2 6 Scheyville 3 1 4
136 The relationship between grazing and the various shrub frequency measures was assessed through Spearman rank correlation coefficients, corrected for ties, using the nine study sites as nine data points.
5.3 Results 5.3.1 Spatial pattern
Bursaria
Bursaria spinosa plants over 80 cm high exhibited substantial clumping on high and moderate fire frequency sites, with C values calculated at site level all highly significant (P < 0.001). Overall C values for low fire frequency sites were well below those for sites in the other two fire frequency categories. Clumping was still significant at Mt Annan (P < 0.05), but not at Orchard Hills or Scheyville. C values for the Bursarias at Scheyville, with its very long interfire interval, were below 0.5, indicating the pattern in that site tended towards evenness, rather than clumping (Table 5.2, Figure 5.1).
Spatial pattern of Bursaria spinosa varied significantly between fire frequency categories (F(2,6) = 8.808, P = 0.016), with post-hoc tests confirming a significantly greater tendency towards clumping in moderate (P < 0.01) and high (P < 0.05) fire frequency sites than in areas which had not burnt for many years before the most recent fire.
137
Spatial pattern, Bursaria spinosa
1 C
0.5 Index of spatial pattern,
0 Ropes ASA Holsworthy Prospect Plumpton Lansdowne Mt Annan Orchard Hills Scheyville Creek Park Site
Figure 5.1. Index of spatial pattern, C (mean + S.E.), for Bursaria spinosa plants over 80 cm high at nine sites, three in each of three fire frequency categories. A C value of 0.5 indicates random patterning, values over 0.5 indicate a tendency towards clumping, while values below 0.5 indicate a tendency towards uniformity. Red, high fire frequency; orange, moderate fire frequency, cream, low fire frequency.
Trees
At site level trees were randomly distributed in all sites except Shanes Park, where the C value indicated evenness (P <0.05). Overall C values were below 0.5 in most sites, indicating a tendency towards evenness rather than clumping (Table 5.3). Patterning did not vary significantly with fire frequency (F(2,6) = 3.594, P = 0.094, NS).
138 Table 5.2. Spatial pattern of Bursaria spinosa plants over 80 cm high. C = index of spatial pattern calculated from T-square sampling distances.
Site Number of Fire C z P† C value data points Frequency indicates: Ropes Creek Block 1 32 high 0.69 3.77 *** clumping Block 2 32 high 0.64 2.65 ** clumping Block 3 32 high 0.79 5.63 *** clumping Site 96 high 0.71 6.96 *** clumping Shanes Park Block 1 32 high 0.76 5.00 *** clumping Block 2 32 high 0.65 2.88 ** clumping Block 3 32 high 0.61 2.07 * clumping Site 96 high 0.67 5.74 *** clumping Holsworthy Block 1 32 high 0.59 1.71 NS randomness Block 2 32 high 0.70 3.96 *** clumping Block 3 32 high 0.62 2.41 * clumping Site 96 high 0.64 4.68 *** clumping Prospect Block 1 32 moderate 0.69 3.71 *** clumping Block 2 32 moderate 0.56 1.19 NS randomness Block 3 32 moderate 0.77 5.26 *** clumping Site 96 moderate 0.67 5.87 *** clumping Plumpton Park Block 1 32 moderate 0.76 5.06 *** clumping Block 2 32 moderate 0.80 5.96 *** clumping Block 3 32 moderate 0.78 5.51 *** clumping Site 96 moderate 0.78 9.54 *** clumping Lansdowne Block 1 32 moderate 0.68 3.46 *** clumping Block 2 32 moderate 0.63 2.58 ** clumping Block 3 32 moderate 0.60 1.88 NS randomness Site 96 moderate 0.64 4.58 *** clumping Mt Annan Block 1 32 low 0.61 2.15 * clumping Block 2 32 low 0.52 0.36 NS randomness Block 3 32 low 0.59 1.72 NS randomness Site 96 low 0.57 2.44 * clumping Orchard Hills Block 1 32 low 0.55 0.89 NS randomness Block 2 32 low 0.51 0.10 NS randomness Block 3 32 low 0.47 -0.61 NS randomness Site 96 low 0.51 0.51 NS randomness Scheyville Block 1 32 low 0.45 -0.94 NS randomness Block 2 32 low 0.45 -0.95 NS randomness Block 3 32 low 0.48 -0.41 NS randomness Site 96 low 0.46 -1.32 NS randomness † *, P < 0.05; **, P < 0.01; ***, P < 0.001; NS, not significant
139 Table 5.3. Spatial pattern of trees over 10 cm DBH. C = index of spatial pattern calculated from T-square sampling distances.
Site Number of Fire C z P† C value data points Frequency indicates: Ropes Creek Block 1 32 high 0.49 -0.14 NS randomness Block 2 32 high 0.51 0.13 NS randomness Block 3 32 high 0.46 -0.87 NS randomness Site 96 high 0.49 -0.51 NS randomness Shanes Park Block 1 32 high 0.48 -0.32 NS randomness Block 2 32 high 0.41 -1.84 NS randomness Block 3 32 high 0.44 -1.26 NS randomness Site 96 high 0.44 -1.98 * evenness Holsworthy Block 1 32 high 0.46 -0.76 NS randomness Block 2 32 high 0.43 -1.40 NS randomness Block 3 32 high 0.45 -1.04 NS randomness Site 96 high 0.45 -1.85 NS randomness Prospect Block 1 32 moderate 0.48 -0.43 NS randomness Block 2 32 moderate 0.45 -1.03 NS randomness Block 3 32 moderate 0.46 -0.82 NS randomness Site 96 moderate 0.46 -1.32 NS randomness Plumpton Park Block 1 32 moderate 0.49 -0.26 NS randomness Block 2 32 moderate 0.49 -0.17 NS randomness Block 3 32 moderate 0.53 0.65 NS randomness Site 96 moderate 0.50 0.13 NS randomness Lansdowne Block 1 32 moderate 0.52 0.34 NS randomness Block 2 32 moderate 0.41 -1.75 NS randomness Block 3 32 moderate 0.52 0.34 NS randomness Site 96 moderate 0.48 -0.62 NS randomness Mt Annan Block 1 32 low 0.48 -0.34 NS ramdomness Block 2 32 low 0.56 1.11 NS ramdomness Block 3 32 low 0.46 -0.84 NS randomness Site 96 low 0.50 -0.04 NS randomness Orchard Hills Block 1 32 low 0.46 -0.75 NS randomness Block 2 32 low 0.52 0.32 NS randomness Block 3 32 low 0.62 2.27 * clumping Site 96 low 0.53 1.06 NSrandomness Scheyville Block 1 32 low 0.49 -0.25 NS randomness Block 2 32 low 0.49 -0.23 NS randomness Block 3 32 low 0.49 -0.17 NS randomness Site 96 low 0.49 -0.38 NS randomness † *, P < 0.05; **, P < 0.01; ***, P < 0.001; NS, not significant
140 5.3.2 Shrub species composition
Cluster analysis
Cluster analysis separated the plots cleanly into five groups, each with between three and nine members (Figure 5.2). The largest group (Group 5) was made up exclusively, and exhaustively, of blocks from low fire frequency sites. The three blocks at Prospect formed a group (Group 1), as did the three from Plumpton Park (Group 4): these were both moderate fire frequency sites. The three blocks from the other moderate fire frequency site, Lansdowne, were grouped with four high fire frequency blocks, three from Holsworthy, and one from Shanes Park (Group 2). The remaining group consisted of high fire frequency sites from Ropes Creek and Shanes Park (Group 3).
Figure 5.2. Dendrogram showing classification of 27 blocks, three in each of nine sites, based on shrub floristic composition. Black, high fire frequency; green, moderate fire frequency, white, low fire frequency.
141 Ordination
The groupings in the dendrogram were generally reflected in the ordination, which attained an acceptable stress level of 0.17 when presented on two dimensions (Figure 5.3). The nine low fire frequency blocks were clumped together at some distance from the other blocks. Blocks from Lansdowne (moderate fire frequency) and Holsworthy (high fire frequency) were also closely grouped in their own area of the ordination space. A third grouping, consisting of the high fire frequency sites in Group 3, was again quite clearly defined. The third Shanes Park (high fire frequency) block was located in the middle of the ordination space reasonably close to both the three Prospect, and the three Plumpton Park blocks. Overall, while the ordination clearly separated out the low fire frequency blocks, complete separation between high and moderate fire frequency blocks was not achieved.
Figure 5.3. Two-dimensional ordination (MDS) of 27 blocks, three in each of nine sites, based on shrub floristic composition. Black, high fire frequency; green, moderate fire frequency, white, low fire frequency.
Analysis of similarities
Shrub community floristics varied significantly with fire frequency. Initial analysis through a two-way nested ANOSIM - site nested in fire frequency – returned a significant result for each factor (Table 5.4). As there were insufficient permutations in this analysis to test the significance of differences between levels of fire frequency, a one-way ANOSIM was also conducted. All pairwise tests were significant, with
142 differences between low fire frequency blocks and those from each of the other two fire frequency categories reaching 0.001 level (Table 5.5).
Table 5.4. Association between shrub community floristics and blocks in nine sites, three within each of three fire frequency categories: two-way nested ANOSIM, site nested in fire frequency.
Factor Global R P Fire frequency (using sites as samples) 0.630 0.011 Site 0.813 < 0.001
Table 5.5. Association between shrub community floristics and blocks subject to high, moderate and low fire frequencies: one-way ANOSIM.
Analysis Global R P Fire frequency (overall analysis) 0.646 < 0.001 High vs. moderate fire frequency (pairwise test) 0.311 0.008 High vs. low fire frequency (pairwise test) 0.827 < 0.001 Moderate vs. low fire frequency (pairwise test) 0.729 < 0.001
Similarity percentages and species contributions
Low fire frequency blocks showed considerably greater within-group floristic similarity (68.0%) than did blocks in the other two fire frequency groupings. Blocks in moderate fire frequency sites were the slightly more dissimilar (41.7% similarity) than those from sites subject to a high frequency of fire (46.9% similarity). Moderate and high fire frequency blocks were more akin floristically than were blocks in each of these categories and those from low fire frequency sites (Table 5.6).
Table 5.6. Average similarity (percent) in shrub community floristics within and between areas subject to high, moderate and low fire frequencies.
High fire frequency Moderate fire frequency Low fire frequency High fire frequency 46.9 34.8 28.4 Moderate fire frequency - 41.7 26.9 Low fire frequency - - 68.0
Bursaria spinosa contributed most towards the dissimilarity between blocks in low fire frequency areas and those in the other two fire frequency categories. The exotic Sida rhombifolia also featured in both these lists. Both Bursaria and S. rhombifolia were more abundant in low fire frequency sites. Hardenbergia violacea was a major
143 contributor to dissimilarity in all three pair-wise comparisons: this species was not common on infrequently burnt sites. Three obligate seeder species contributed more than 4% to the dissimilarity between low and moderate frequency blocks; all were more abundant where there had been some fire over the last 20 years. Obligate seeder species also contributed to the dissimilarity between high and moderate fire frequency blocks, with four out of five more common in blocks where fire frequency had been moderate (Table 5.7).
Table 5.7. Species contributing more than 4% to the average dissimilarity in shrub community floristics between blocks subject to high, moderate and low fire frequencies.
Fire frequency categories Regeneration More abundant when % contribution Cumula Species mechanism** fire frequency is†: to dissimilarity -tive % High vs. low Bursaria spinosa resprouter low 36.13 36.13 Sida rhombifolia* n/a low 15.83 51.96 Hardenbergia violacea resprouter high 7.12 59.08 Acacia decurrens seeder high 4.86 63.95 Eremophila debilis resprouter low 4.15 68.10 Dillwynia sieberi seeder low 4.09 72.19
Moderate vs. low Bursaria spinosa resprouter low 20.84 20.84 Hardenbergia violacea resprouter moderate 13.58 34.42 Pultenaea microphylla seeder moderate 11.16 45.58 Sida rhombifolia* n/a low 9.95 55.53 Dillwynia sieberi seeder moderate 7.60 63.13 Daviesia ulicifolia seeder moderate 7.12 70.26
High vs. moderate Hardenbergia violacea resprouter moderate 17.62 17.62 Pultenaea microphylla seeder moderate 14.33 31.95 Dillwynia sieberi seeder moderate 10.72 42.68 Daviesia ulicifolia seeder moderate 9.26 51.94 Bursaria spinosa resprouter moderate 7.28 59.21 Acacia decurrens seeder high 7.27 66.48 Grevillea juniperina seeder moderate 5.24 71.72 * exotic sp. ** regeneration mechanism given for native species only. †Frequencies of native shrub species are detailed in Table A8.2 (Appendix 8).
144 Vegetation type and shrub community floristics
When points on the ordination shown in Figure 5.3 were categorised according to Tozer’s vegetation types, vegetation types did not cluster, and were scattered across fire frequencies (Figure 5.4). Shrub community floristics were not significantly associated with vegetation type in the one-way ANOSIM analysis for this variable (Global R = 0.100, P = 0.10).
Stress: 0.17
SGTF
SPW
SHW
Figure 5.4. Ordination shown in Figure 5.3, showing vegetation type of 27 blocks. SGTF, Shale Gravel Transition Forest; SPW, Shale Plains Woodland; SHW, Shale Hills Woodland (from NPWS 2004a).
5.3.3 Associations
Bursaria and other shrubs
In all three moderate fire frequency sites, obligate seeders were more abundant, on average, when Bursaria was absent than when it was present; these differences, however, only reached significance at Plumpton Park. Similarly, obligate seeders in each site were more abundant when co-occurring Bursaria plants were under 80 cm, than when they were above this height, although these differences were not significant. Patterns for resprouters were less consistent, and no significant differences were found (Table 5.8).
145 Table 5.8. Association between Bursaria spinosa and other native shrubs in three sites. P values determined through permutation tests.
Site Native shrubs Mean (SE) number of individuals per subplot (4 m2) other than Bursaria Bursaria Bursaria P Bursaria < 80 Bursaria > 80 P absent present cm cm Prospect Obl. seeders 27.5 (4.2) 20.4 (5.3) 0.289 26.2 (9.1) 18.6 (6.3) 0.524 Resprouters 21.4 (1.8) 18.9 (1.8) 0.379 24.7 (4.3) 17.2 (1.9) 0.122 Plumpton Obl. seeders 2.18 (0.27) 1.21 (0.28) 0.034 1.31 (0.83) 1.17 (0.26) 0.881 Park Resprouters 1.01 (0.32) 0.77 (0.25) 0.699 0.19 (0.10) 0.98 (0.33) 0.388 Lansdowne Obl. seeders 2.94 (0.67) 1.91 (0.40) 0.197 3.05 (0.99) 1.57 (0.42) 0.268 Resprouters 1.22 (0.27) 1.99 (0.52) 0.207 2.73 (1.24) 1.77 (0.56) 0.311
Adult trees, saplings and juveniles
There were significantly more saplings in subplots where trees were present at Ropes Creek and Prospect, than in subplots without adult trees. This pattern was not repeated, however, in the other two sites investigated. Subplots with adult trees whose basal area was above 800 cm2 had consistently fewer saplings than subplots with smaller trees; these differences were significant at Ropes Creek and Lansdowne (Table 5.9).
Patterns for juveniles were generally similar in individual sites as those for saplings, however differences were not so pronounced and none reached significance. Again, subplots with larger trees had consistently fewer juveniles than those containing adult trees with a basal area below 800 cm2 (Table 5.9).
Table 5.9. Association between adult and non-adult trees in three sites. P values determined through permutation tests.
Site Non-adult Mean (SE) number of individuals per subplot (12 m2) trees adult trees adult trees P adult tree < adult tree > P absent present 800 cm2 800 cm2 dbh dbh Ropes Creek Juveniles 1.03 (0.18) 1.61 (0.41) 0.155 1.93 (0.52) 0.93 (0.63) 0.175 Saplings 0.08 (0.03) 0.25 (0.08) 0.005 0.33 (0.11) 0.07 (0.07) 0.046 Prospect Juveniles 0.51 (0.09) 0.67 (0.23) 0.464 0.85 (0.34) 0.38 (0.22) 0.201 Saplings 0.11 (0.03) 0.26 (0.08) 0.033 0.35 (0.11) 0.13 (0.09) 0.081 Lansdowne Juveniles 1.58 (0.22) 1.13 (0.29) 0.338 1.26 (0.53) 1.00 (0.27) 0.676 Saplings 0.32 (0.06) 0.29 (0.09) 0.879 0.53 (0.16) 0.05 (0.05) 0.013 Scheyville Juveniles 0.50 (0.08) 0.45 (0.08) 0.649 0.52 (0.10) 0.21 (0.11) 0.230 Saplings 0.25 (0.05) 0.21 (0.06) 0.699 0.23 (0.07) 0.14 (0.10) 0.650
146 The ratio of non-adult to adult trees varied widely between sites, from 0.86 at Plumpton Park to 8.54 at Lansdowne. The correlation between the number of young trees in a site and mean adult tree basal area was not significant (r = -0.278, P > 0.05). There was, however, a significant negative correlation between adult tree basal area and the ratio of non-adult to adult trees (r = -0.629, P < 0.05). This ratio was over 3.0 in the five sites with a basal area of less than 13 m2/ha, but less than 2.0 in the three sites where it was over 15 m2/ha.
5.3.4 Grazing pressure
Correlation coefficients between shrub variables, the density of non-adult trees, and the grazing index are given in Table 5.10. With n = 9, rs0.05(2) = 0.700 (Zar, 1999). Thus none of these correlations is significant, though several come close.
Table 5.10. Association between grazing index and shrub variables in nine study sites.
Variable Spearman Rank Correlation coefficient Significance Bursaria frequency 0.479 NS Other native shrub species richness - 0.581 NS Other native shrub frequency - 0.641 NS Obligate seeder frequency - 0.650 NS Other resprouter frequency - 0.573 NS Exotic shrub frequency 0.296 NS Young trees (juveniles + saplings) -0.248 NS
147 5.4 Discussion 5.4.1 Spatial pattern
In high and moderate fire frequency sites, Bursaria spinosa occurred in clumps separated by open grassy areas. Fire can play a role in creating and maintaining clumps: in a simulation study, Green (1989) found clumped distributions initiated by local seed dispersal were intensified by fire, which acted as a space-creating mechanism. In infrequently burnt sites, however, this pattern weakened and finally disappeared. The tendency towards an even pattern found at Scheyville, where fire had reportedly been excluded for 50 years, is indicative of the close packing of Bursaria plants in that site. This finding suggests that Bursaria clumps are expanding in the absence of fire, until they coalesce to cover the landscape.
Clumping in shrubs and trees can occur because seeds accumulate in the neighbourhood of adult plants, and because reduced grass cover under trees and shrubs reduces fire- induced mortality (Jeltsch et al. 1998). Patchiness may also reflect uneven distribution of nutrients, for example areas where logs have decomposed may be nutrient-enriched, while patches subject to water flows may be depleted of nutrients (Nadolny et al. 2003).
Spatial pattern will be further discussed in Section 5.4.4.
5.4.2 Shrub species composition
The hypothesis that shrub species composition would differ with fire frequency was supported by the data. Cluster, ordination, ANOSIM and SIMPER analyses all highlighted differences between low fire frequency blocks and those in the other two fire frequency categories. Low fire frequency sites resembled each other in terms of shrub floristics, while sites in the other two fire frequency categories supported a much more diverse shrub flora. Although not as pronounced, differences between areas subject to high and moderate fire frequencies could also be discerned.
The high fire frequency blocks subject to the most extreme fire regimes, ie those at Ropes Creek and Shanes Park, were located to the same side of the ordination space as the nine blocks from low fire frequency sites. Extreme fire regimes can be viewed as constraining the shrub floristics in both these sets of sites, while the floristics of sites
148 which have experienced a moderate level of burning are less constrained, and thus more clearly reflect the effects of environmental variables other than fire.
SIMPER analysis highlighted the central role played by Bursaria in distinguishing the floristics of low fire frequency sites from those of sites with shorter interfire intervals. It also confirmed the part played by obligate seeder shrubs in distinguishing moderate fire frequency sites from those with both shorter and longer interfire intervals. Shrubs characterising moderate fire frequency sites were overwhelmingly legumes: most noteworthy were Pultenaea microphylla, Dillwynia sieberi and Daviesia ulicifolia. While it could be argued that the limited distribution of Pultenaea microphylla on the Cumberland Plain may have been a factor in its absence from some survey sites, the other two species are widespread (James et al. 1999).
The limited association between shrub floristics and vegetation type vis-a-vis fire frequency highlights the strength of the fire frequency effect. Soil type is closely associated with vegetation patterns on the Cumberland Plain (Tozer 2003). Thus the lack of association with vegetation type implies that the findings of the study are unlikely to reflect edaphic differences.
5.4.3 Bursaria in relation to other shrubs
Native shrubs in moderate fire frequency sites were found both in open areas and under Bursaria. The lack of a consistent reduction in native resprouter abundance with increasing Bursaria dominance indicates that these species are not generally at risk of being out-competed, or otherwise disadvantaged, by Bursaria. However the finding that Bossiaea prostrata is not tolerant of overtopping by Acacia sophorae (Costello et al. 2000) suggests that this conclusion may not apply to every individual resprouting species.
On the other hand, the consistently lower abundance of obligate seeders with increasing Bursaria dominance suggests these species may be disadvantaged by association with Bursaria. The fact that this trend was only reflected in significant findings at one site may be partly a result of the imprecise nature of the Bursaria dominance measure, which was based on cover anywhere in the 2 x 2 m subplot. Dominant resprouters have been found to have a detrimental effect on understorey species elsewhere (Bond and Ladd 2001).
149 There are a number of mechanisms by which a negative association may be mediated. The pattern of association between Bursaria and obligate seeders could be the result of direct competition for resources. Huston’s dynamic equilibrium model considers that competitive exclusion will result in biodiversity reduction under low disturbance frequencies in productive environments (Huston 1979, 2003). However the current findings could also be the result of allelopathy, or could reflect differences in environmental requirements of different shrub species. Indirect effects, such as disruption of ant-plant interactions due to increased shade (Cowling and Gxaba 1990, Yeaton and Bond 1991), may also be involved.
Although their abundance was reduced, some obligate seeder plants did occur in plots dominated by Bursaria. Interspecific interactions are thus unlikely to be the sole factor driving their relatively low abundance in low fire frequency sites, although they could play a part.
5.4.4 Tree density and recruitment
The tendency towards evenness in the distribution of trees supports the hypothesis that adult tree density is nearing the level where it is limited by environmental resources (Section 5.1.3).
The finding that young trees are more likely to occur with adult trees of limited basal area than either in subplots without trees, or with adult trees with a relatively high basal area, reflects findings elsewhere. Curtis (1990) found that eucalypt recruitment in New England was concentrated in the ‘halo’ just beyond the tree canopy, but not directly beneath it. Large adult trees appear to be suppressing recruitment in their immediate neighbourhood – particularly recruitment into the canopy. This is consistent with the hypothesis that as individual trees move into the higher size classes, they limit the development of other trees (Henry and Florence 1966; Florence 1996).
On the other hand, saplings are most likely to occur where small adult trees already exist. This may be because these trees are providing a seed source for recruitment, or because these saplings established at the same time as the nearby trees, but are subdominant.
150 If it is the case that most CPW remnants are in post-logging transition towards a landscape with a relatively small number of large trees, the death of saplings or small trees as a result of frequent or intense fire may be part of the transition process. An implication of the trees in transition hypothesis is that tree recruitment is not a particularly vital process in the recovery of study remnants, at this time. The significant negative correlation between adult tree basal area and the ratio of non-adult to adult trees suggests a degree of self-regulation of recruitment as adult tree basal area approaches carrying capacity.
5.4.5 Grazing pressure
The grazing index was lower in moderate fire frequency sites than in low fire frequency sites, with high fire frequency sites somewhere in between. While a more random distribution of grazing pressure vis-a-vis fire frequency would have been desirable, the small number of CPW remnants available for study made this impossible.
Correlations between the grazing index and the various shrub frequency variables were all non significant, although those for other native shrubs overall, and for obligate seeders, approached significance. Given the coarseness of the index, this result suggests that the effect of grazing on shrub species is worthy of further consideration.
The idea that stock grazing is associated with a high density of Bursaria was expressed by several people during the course of the project. A substantial increase in Bursaria density was observed during the course of research into avifauna at Scheyville during the 1970s and 80s, while horse and cattle grazing was still extant (H. Recher pers. comm. 2004). That grazing can encourage shrub growth in some circumstances is also attested in the literature (Wimbush and Costin 1979, Madany and West 1983, Noble 1997).
As the relationship between fire frequency and grazing is not linear, it is relatively easy to disentangle the effects of these variables on Bursaria density. Although low fire frequency sites had both high Bursaria densities and relatively high exposure to grazing, this pattern was not encountered elsewhere in the data set. High fire frequency sites scored higher on the grazing index than moderate fire frequency sites, but had lower densities of Bursaria. Prospect, where grazing has been absent for many years due to its status as a water catchment, had the highest Bursaria density of the three intermediate
151 sites. Scheyville rated below the other two low fire frequency sites on the grazing index, but had by far the highest density of Bursaria, as well as the longest reported fire free period. It seems unlikely that the very significant findings with respect to fire frequency and Bursaria density are an artifact of grazing pressure.
A similar argument can be advanced for exotic shrubs. Higher grazing pressure in high fire frequency sites is not associated with an increase in weediness. Thus the greater frequency of exotic shrubs in low fire frequency areas is unlikely to have resulted solely from grazing pressure. Alternatively, one could hypothesise that grazing may indeed play a role in helping exotic shrubs establish in Cumberland Plain Woodland, but that frequent fire counteracts that trend.
The effects of fire and grazing on the frequency of native shrubs other than Bursaria are not so readily disentangled. The correlation between grazing and these shrubs was negative, and substantial though not significant. Might grazing, rather than fire, be responsible for the relatively low abundance of shrubs other than Bursaria in high and low fire frequency sites? While a definitive answer cannot be garnered from the available data, several lines of evidence suggest that fire frequency is the more important factor. Considerations include:
• The frequency and species richness of native shrubs other than Bursaria overall, and the frequency of obligate seeder shrubs in particular, rises sharply and smoothly as frequency of burning increases in high fire frequency sites. Ropes Creek and Holsworthy have equivalent scores on the grazing index, however Ropes Creek with its relentless arson burning has virtually no shrubs, while Holsworthy, with intervals between approximately 2 and 5 years, boasts a shrub frequency almost as great as some moderate fire frequency sites.
• In the same way, other native shrub species richness and frequency falls smoothly as the last interfire interval increases in low fire frequency sites. Scheyville, the site with the longest reported interfire interval, had very few native shrubs and virtually no obligate seeders, despite the occurrence of several species in the area. Current grazing levels at Scheyville are well below those at Mt Annan and Orchard Hills, and domestic grazing prior to gazettal in the mid 1990s was confined to the more open areas (CALM 1993, H. Recher pers. comm. 2004).
• While regional studies have come to differing conclusions with respect to the effect of grazing on shrubs (eg Leigh and Holgate 1979 versus Wimbush and Costin 1979), local research at Holsworthy found an association between shrub abundance and fire, but not between shrub abundance and grazing by macropods and rabbits (Hill and French 2004).
152 • Windsor Downs, a Cumberland Plain nature reserve which experiences fairly frequent fires (it would fall into this study’s moderate fire frequency category) supports thriving populations of many obligate seeder shrubs in Shale-Gravel Transition Forest (pers. obs. 2001-4). Prior to becoming a reserve in 1990, this site was owned and managed by the Riverstone meat works; animal watering points are still to be found in the bush (NPWS 1999).
• The density of young trees in study sites is not related to grazing pressure, suggesting that grazing past and present in the high quality remnants chosen for study has not been a major pressure on vegetation.
• Similarly, grazing-intolerant herbs, while somewhat less dominant in the three low fire frequency sites than in three other sites in this study, were still an important component of the ground flora (Section 8.3.2). This finding again suggests that grazing has not been a major pressure on the vegetation of study sites.
These points do not, however, allow total dismissal of the role of grazing in reducing shrub abundance. Further research is recommended.
5.5 Conclusion
Further analysis of the landscape study data again highlights differences between sites with a recent long interfire interval and those burnt more often. Bursaria, which formed clumps in sites with 1-10 year interfire intervals, tended towards evenness as fire frequency decreased. Shrub floristics in the three low fire frequency sites were similar due to ample Bursaria, a relative dearth of other native shrubs, and the presence of woody exotics. The species composition of these sites differed significantly from that in more frequently burnt sites. Shrub floristics also differed between high and moderate fire frequency sites. A higher abundance of obligate seeder legumes distinguished sites with mostly 4 to 10 years between fires. While competition from Bursaria may have played a role in the reduced abundance of obligate seeders in low fire frequency sites, the direct impact of a lack of fire on obligate seeders (Chapter 3) is likely to have been at least as important.
Relative to fire frequency, vegetation type had a weak association with shrub floristics, making it unlikely that differences between study sites can be attributed to vegetation type or edaphic variation. Grazing pressure is also unlikely to have had a significant
153 impact on the abundance of Bursaria or exotic shrubs. The possibility that grazing may have played a role in reducing obligate seeder abundance in high and low fire frequency sites cannot be discounted; however a number of factors suggest that fire frequency was the major influence.
Tree density in most study sites is probably higher today than it was prior to the logging that almost certainly took place soon after European settlement. Several considerations suggest that as sites recover and some trees grow larger, other trees will die out, allowing basal area to remain relatively constant. If this hypothesis is correct, the pool of tree species juveniles would appear to be adequate in all sites, as recruitment will not be a major need for some years to come.
154 CHAPTER 6 POST-FIRE REGENERATION OF BURSARIA SPINOSA
6.1 Introduction 6.1.1 Bursaria spinosa and fire
“Bursaria spinosa Canavilles. Blackthorn, Kurwan... An erect shrub usually 2-3m, occasionally to 5m, tall, with light foliage and side-branches ending in thorns. It is opportunistic and highly successful on the Cumberland Plain where it now dominates the forest understorey over extensive areas. It also occurs near streams in sandstone country. A rather dreary plant most of the year but it puts on a magnificent display of fragrant white flowers in late summer.” From the popular Field Guide to the Native Plants of Sydney by Les Robinson (1994:196).
Bursaria spinosa (Figure 6.1) is clearly a key component of Cumberland Plain Woodland. It is far and away the most abundant shrub species in this vegetation type (James et al. 1999, Tozer 2003, Chapter 4), forming dense thickets, particularly in sites with a low fire frequency. It influences the composition of ground layer species (Chapter 8), and may also affect fire-cued shrubs (Chapter 5). However its fire ecology has not previously been investigated.
Figure 6.1. Bursaria spinosa in flower.
155 There is no doubt that Bursaria resprouts successfully post-fire: its ability to do so was evident in every study site. Even resprouting species, however, may lose some individuals to fire (Morrison 1995, Hodgkinson 1998, Morrison and Renwick 2000). The fact that Bursaria is much less abundant where fires occur once or twice a decade, than in areas where intervals exceed 20 years (Chapter 4), suggests that fire-induced mortality must occur under some circumstances.
This chapter reports a small, unreplicated study which sought to provide some albeit limited insight into the regeneration of Bursaria spinosa after a single fire.5
6.1.2 Study aims and limitations
The study followed a cohort of Bursaria plants tagged prior to an ecological burn in a single site through their initial years of recovery. Questions addressed in this chapter include:
• What proportion of plants resprouted after the fire? • Did smaller plants experience greater mortality than larger plants? • Did isolated plants experience greater mortality than plants crowded together? • Was survival lower in patches with large plants than in patches without large plants? • How quickly did resprouting plants grow? • Did seedlings establish after the fire? • Did density and/or size of existing plants affect seedling establishment?
Time prevented extension of this study to other sites, or other fires. It is therefore unreplicated. Similar studies, across space and time, will be needed to assess how generally findings apply.
5 Staff at the Royal Botanic Gardens kindly contributed to this study in a number of ways – see Acknowledgements section for details.
156 6.2 Methods 6.2.1 Study site and burn
The study was conducted in one of several Cumberland Plain Woodland conservation areas maintained as part of the Royal Botanic Gardens (RBG) at Mt Annan. The study site, which was approximately 350 m x 100 m in area, was burnt for ecological purposes on 23 September 2002 (Figure 6.2). Fire intensity varied across the site but overall was low to moderate. The site came under RBG management in 1985, and had not been exposed to fire in that time (D. Benson, RBG, pers. comm. 2001) although a wildfire may have burnt across it in 1982 (L. Turrin, RFS, pers. comm. 2002). Grazing by domestic animals had also ceased by 1985, although rabbit and macropod grazing remains.
Figure 6.2. The study area at Mt Annan at the start (left), and finish (right) of the planned burn. These photos and those in subsequent figures in this chapter courtesy of Lotte von Richter, RBG.
6.2.2 Tagging plants in plots
In the week before the burn, sixteen 2 m x 2 m plots were identified in the woodland area. In order to ensure sampling encompassed site variability, the area scheduled for burning was divided, on paper, into two blocks of approximately equal size, and eight plots selected in each block. Plots were of three types:
a. Plots containing a large Bursaria plant with smaller plants nearby. (Definition: at least one Bursaria plant over 3 m in height with at least six other plants in a 2 x 2 m area.)
157 b. Plots containing a high density of small Bursaria plants, relatively isolated from established specimens. (Definition: at least ten Bursaria plants under 1 m high in a 2 x 2 m area, with no Bursaria bushes over 2 m in height within 2 m of plot boundaries.) c. Plots containing a low density of small Bursaria plants, relatively isolated from established specimens. (Definition: two to four plants under 1 m high in a 2 x 2 m area, with no Bursaria bushes over 2 m in height within 2 m of plot boundaries.)
Two ‘a’ plots, two ‘b’ plots, and four ‘c’ plots were identified within each block. A higher number of ‘c’ plots were identified because by definition there were fewer plants in these patches. Plots were located by randomly picking a location, and then finding the nearest patch which met the relevant criteria.
There were thus 16 plots overall, 4 plots with large plants, 4 plots with many small plants, and 8 plots with a few small plants.
Plots were laid out on a NSEW grid (Figure 6.3), and the south-east corner of each plot marked with a metal stake. Stakes were labelled with thick aluminium tags attached with copper wire. In each plot, all Bursaria plants were tagged with copper wire (circles of wire were placed on the ground, away from plant stems). The position of each plant in the plot was mapped, and its height in centimetres recorded. Each plant was numbered, so its individual fate could be tracked.
Figure 6.3. Two study plots prior to the fire: left, plot with many small Bursaria plants; right, plot with one large and several small Bursarias.
158 6.2.3 Assessing post-fire recovery
Approximately three weeks after the fire (on 11 October 2002), each plot was visited to assess the extent to which the fire had succeeded in burning the tagged plants. The position of tags was checked. Where necessary, tags were renewed (Figure 6.4).
Figure 6.4. Study plots shown in Figure 6.3, three weeks post-fire.
Lotte von Richter, a researcher located at the Mt Annan Gardens, monitored plots for resprouting each week after the burn. Arrangements were made to reassess plots two weeks after resprouting was first observed. After this, plots were assessed at three monthly intervals.
At each post-fire assessment, compass bearings from the metal stake at the corner of each plot were used to re-establish plot boundaries. All visible live Bursaria plants within a plot were flagged. Flagged plants were then compared with mapped plants. Where flagged plants corresponded with mapped plants, they were ticked off. Where plants were mapped but not flagged, a search was conducted to determine whether the plant was resprouting, but had been missed (plants were sometime difficult to locate amongst grass and litter). Where plants were flagged but not mapped, they were examined to determine whether they were resprouts or seedlings. If cotyledons were present, the plant was classified as a seedling. If cotyledons were absent, plants with a single thin stem were classified as seedlings, while individuals with multiple thick stems were considered resprouts. Additional plants were mapped and numbered. At every assessment, at least two observers helped locate plants.
159 Three-monthly assessments continued until 25 months after the fire (Figure 6.5). At 13 and 25 months post-fire, the height of each plant, from the ground to the highest living shoot; was measured.
Figure 6.5. Study plots shown in Figure 6.3, 25 months post-fire.
Rainfall data for the Mt Annan Gardens were provided by RBG staff. Monthly totals for the study period were plotted against long-term averages for nearby Campbelltown, and the relationship of the resulting pattern to the pattern of post-fire Bursaria recovery, explored.
6.2.4 Bursaria excavation
In an attempt to discover whether Bursaria spinosa spreads by suckering, approximately 20 plants under 20 cm tall were dug up. Plants were in an unburnt area of the Mt Annan Gardens not designated for conservation. Soil was scraped away carefully in an attempt to ensure any underground connections between plants were not disrupted.
6.2.5 Data analysis
The significance of differences in mortality between plot types (Section 6.2.2), and between size classes, was assessed through chi-square tests incorporating Yates correction for continuity (Zar 1999). Chi-square tests for goodness of fit were used evaluate differences in seedling numbers between plot types.
160 6.3 Results
One hundred and fifty-three plants were tagged prior to the burn. All tagged plants burnt in the fire, except one. This plant, located in a plot where fuel was sparse and patchy, was excluded from monitoring and analysis, leaving a sample of 152 burnt plants.
6.3.1 Resprouts
Resprouting commenced 3.5 months after the fire, on 7 January 2003; two weeks later, on January 22, 43 (28%) of the 152 tagged plants had green shoots (Figure 6.6). By seven months post-fire, 87% of tagged plants were resprouting. Numbers peaked ten months after the burn, when live shoots were found on 138 (91%) of pre-fire plants, then dropped slightly, settling at a constant level of 132 plants, 87% of the pre-fire total (Table 6.1, column 5). These figures mask a degree of fluctuation: 24 tagged plants resprouted but then were not visible at at least one subsequent assessment; of these 17 subsequently reappeared. Only 11 plants, 7% of the pre-fire total, did not resprout at all.
Figure 6.6. Tagged Bursaria resprouting four months post-fire.
Table 6.1. Resprouting of Bursaria spinosa in relation to time-since-fire.
Assess- Date Time-since- Number of pre- Percentage of Additional Total Percentage of ment fire (mths) fire plants pre-fire plants resprouts number of pre-fire total number resprouting resprouting found resprouts (n = 152) (n = 152) 1 Oct 02 1 0 0% 0 0 0% 2 Jan 03 4 43 28% 0 43 28% 3 April 03 7 132 87% 11 143 94% 4 July 03 10 138 91% 12 150 99% 5 Oct 03 13 137 90% 16 153 101% 6 Jan 04 16 128 84% 14 142 93% 7 April 04 19 132 87% 12 144 95% 8 July 04 22 132 87% 12 144 95% 9 Oct. 04 25 132 87% 14 146 96%
161 As well as the tagged plants, an additional eighteen resprouts appeared during the course of the study. These plants were tagged as they appeared. Fourteen of them were visible at the last visit (Table 6.1). Three disappeared, then reappeared, over the course of the study.
By the end of the study, resprouting had replaced 96% of plant numbers prior to the fire.
Small plants were less likely to resprout than larger ones: 13% of plants under 25 cm high prior to the fire did not resprout at all, compared with 4% of plants above this height. Small plants were also twice as likely to ‘come and go’ (Table 6.2). No plant over 50 cm failed to resprout. Differences between height categories were significant (χ2 = 8.57, df = 2, P < 0.05 – ‘variable resprouts’ included in analysis; χ2 = 4.25, df = 1, P < 0.05 – ‘variable resprouts’ omitted from analysis).
Table 6.2. Resprouting success of Bursaria spinosa by size class (one plant omitted due to missing data). Constant resprout, resprouted and remained visible at every subsequent assessment; variable resprout, resprouted but was not visible at at least one subsequent assessment.
Number of plants (percent of column totals) ≤ 25 cm in height pre- > 25 cm in height pre- Total fire fire Constant resprout 33 (63%) 83 (84%) 116 (77%) Variable resprout 12 (23%) 12 (12%) 24 (16%) Did not resprout 7 (13%) 4 (4%) 11 (7%) Total 52 (86%) 99 (14%) 151
Small plants in plots with a large plant were somewhat less likely to resprout than plants in plots where large plants were absent (Table 6.3). However the difference between plots with a large plant, and other plots, was not significant (χ2 = 1.39, df = 1, NS; the two ‘small plant’ categories were combined for this analysis).
Table 6.3. Extent of resprouting in tagged Bursaria spinosa plants, 25 months post-fire. Large plants (over 3 m high prior to fire, n = 4) not included.
Plots containing: Number (percent) of Number (percent) of tagged Total tagged plants resprouting plants not resprouting at 25 at 25 mths post-fire mths post-fire A large plant plus some small 26 (79%) 7 (21%) 33 plants A high density of small plants 79 (88%) 11 (12%) 90 A low density of small plants 23 (92%) 2 (8%) 25 Total 128 20 148
162
6.3.2 Seedlings
Seedlings first appeared in the Winter after the fire. Sixty-three seedlings were identified during the July 2003 visit ten months post-fire, with a further batch found in October of that year. A small number of seedlings also appeared in the second post-fire year. Most of the seedlings found in 2003 were identified by their cotyledons. Cotyledons were not, in general, apparent on seedlings found in 2004. The majority of seedlings died within months of germination: by 16 months post-fire only 14 of the initial cohort remained. However over the next nine months numbers slowly built up, and at the final assessment 25 months after the burn, 21 live seedlings were counted (Table 6.4).
Almost all seedlings (97%) were found in plots containing a plant over 3 m in height (Table 6.5). All four plots of this type had seedlings; numbers per plot ranged from 6 to 45. In contrast, only three seedlings were found in all the ‘small plant’ plots together. This finding was highly significant (χ2 = 250.8, df = 2, P < 0.001). Differences in seedling numbers between plot types at the end of the study were also significant (χ2 = 37.4, df = 2, P < 0.001).
Table 6.4. Fluctations in seedling numbers with time-since-fire. na, not applicable.
Assessment Date Time-since- Number of Number of Number of live number fire (mths) new seedlings seedling seedlings deaths 1 Oct 02 1 0 na 0 2 Jan 03 4 0 na 0 3 April 03 7 0 na 0 4 July 03 10 63 na 63 5 Oct 03 13 18 21 60 6 Jan 04 16 0 46 14 7 April 04 19 4 2 16 8 July 04 22 3 1 18 9 Oct. 04 25 5 2 21
163 Table 6.5. Bursaria spinosa seedlings in plots with different configurations of existing plants: emergence, and survival at 25 mths post-fire.
Plots containing: Number Number of seedlings Number of live seedlings 25 of plots emerging months post-fire Total Mean (S.E.) / Total Mean (S.E.) / plot plot A large plant plus 4 90 22.50 (8.21) 18 4.50 (1.32) some small plants A high density of 4 1 0.25 (0.25) 1 0.25 (0.25) small plants A low density of 8 2 0.25 (0.25) 2 0.25 (0.25) small plants Total 16 93 5.81 (3.09) 21 1.31 (0.58)
6.3.3 Flowering
Very little flowering was observed during the course of the study. Two plants were flowering or budding in January 2004: one of the four plants which was over 3 m before the burn, and another smaller plant in one of the high density plots.
6.3.4 Overall plant numbers
Considering seedlings and resprouts together, numbers of plants peaked in July and October 2003, during the initial flush of seedling emergence, then dropped when most of these seedlings died (Table 6.6). However through 2004 the total number of plants grew steadily. At the end of the study, 25 months after the fire, 167 Bursaria plants were located in plots, 110% of the pre-fire number (Table 6.6).
Table 6.6. Overall numbers of Bursaria spinosa plants relative to time-since-fire.
Assessment Date Time-since- Number of Number of Total number number fire (mths) resprouts live seedlings of plants 1 Oct 02 1 0 0 0 2 Jan 03 4 43 0 43 3 April 03 7 143 0 143 4 July 03 10 150 63 213 5 Oct 03 13 153 60 213 6 Jan 04 16 142 14 156 7 April 04 19 144 16 160 8 July 04 22 144 18 162 9 Oct. 04 25 146 21 167
164 6.3.5 Growth
After the fire, all Bursaria plants in the study either completely disappeared, or were reduced to dead, burnt sticks. All resprouting was basal; no epicormic resprouting occurred.
The average height of tagged Bursarias prior to the fire was 47.7 cm. Thirteen months after the burn, the average height of live shoots was 16.4 cm. A year later, 25 months post-fire, little progress had been made: the average height of tagged plants was 17.6 cm. Of the 143 plants which were measured in both years, 51% had grown, 7% had remained the same size, and 42% were smaller in 2004 than they had been in 2003. A greater percentage of resprouts grew in height in plots with a large Bursaria plant than in plots with small plants only (Table 6.7), however this difference was not significant (χ2 = 2.65, df = 1, NS; the two ‘small plant’ categories were combined for this analysis).
Table 6.7. Relative size of resprouting Bursaria spinosa plants in October 2004 (25 months post-fire) and October 2003 (13 months post-fire). Both tagged plants and additional resprouts included.
Plots containing: Number (percent) of Number (percent) of Total resprouting plants resprouting plants which were larger in which were smaller or Oct 2004 than in Oct the same in Oct 2004 2003 as in Oct 2003 A large plant plus 22 (65%) 12 (35%) 34 some small plants A high density of 40 (48%) 44 (52%) 84 small plants A low density of small 11 (44%) 14 (56%) 25 plants Total 73 (51%) 70 (49%) 143
The average height of surviving seedlings increased appreciably between October 2003 and October 2004, from 1.9 cm to 6.3 cm. Of the nine seedlings present at both times, five had grown, one was the same height as in 2003, and three were smaller. The tallest seedlings were 15 cm high in October 2004.
165 6.3.6 Rainfall
Rainfall during the study period was well below average. Annual totals at Mt Annan were 436 mm, 607 mm and 625 mm for 2002, 2003 and 2004 respectively, relative to a long-term yearly average of 830 mm at nearby Campbelltown. Monthly totals were also almost all below average (Figure 6.7).
The weather leading up to the burn was very dry, with just 15 mm of rainfall recorded for the preceding three months. Only 11 mm fell in the two months after the burn, however slightly better falls in December and January preceded and accompanied the first appearance of resprouts. The initial flush of seedlings occurred between April and July 2003: rainfall was relatively good in these months, with May in particular recording well above average rainfall (155 mm). In the year between October 2003, when plant heights were first measured, and October 2004 when they were measured again, rainfall was below average in almost every month. Good rain did occur in October 2004, but most of this fell only days before plants were re-measured.
180 planned burn 160 first seedling resprouts flush 140
120
100
80 Rainfall (mm) Rainfall
60
40
20
0 Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct
2002 2003 2004
Figure 6.7. Monthly rainfall (mm) at Mt Annan from July 2002 to October 2004 (solid red line; source, L. von Richter, RBG, pers. comm 2004, 2005), relative to long-term averages at nearby Campbelltown (dashed blue line; source, Bureau of Meteorology, 2004).
166 6.3.7 Excavation
All Bursaria plants excavated, even very small ones, had swollen, solid, tap roots (Figure 6.8). Roots began to swell several centimetres below ground level, and could be longer than above-ground plant height. In only one instance did there appear to be a root connection between plants, and even in that case there was some doubt as to whether the root, which was thin and broken, had in fact joined two individuals. In all other cases, no connection was found.
Figure 6.8. Excavated Bursaria spinosa plant showing swollen tap root.
6.4 Discussion 6.4.1 Status of findings
It could be legitimately argued that the results reported in this chapter relate only to the particular study site (Mt Annan), fire (in spring, of low to moderate intensity, following a long interfire interval), and weather conditions (mostly very dry). They can, however, suggest hypotheses for testing when other planned burns occur in Cumberland Plain Woodland.
167 6.4.2 Does fire kill Bursaria?
Seven percent of plants tagged prior to the fire did not resprout at all, while a further six percent were not visible at the final assessment. It seems reasonable to assume that plants in the first group were killed by the fire. Plants in the second group may or may not be dead: quite a few plants ‘came and went’ during the course of the study. Deaths in this latter group, if they exist, may be due to a combination of fire and grazing. Although no formal assessment of grazing pressure was carried out during the study, its effects were readily apparent at every visit, as were rabbit, and occasionally macropod, scats. The many plants which decreased in height between October 2003 and October 2004 attest to the impact of grazing animals.
Of course, the dry spell following the fire might also have contributed to mortality in both groups of plants, as might the dry weather before the burn. On the other hand, fire intensity was relatively low. Mortality in resprouting shrubs has been found to increase with fire intensity (Morrison and Renwick 2000), so more plants might have died if the fire had been more intense. A smaller number of plants may have died, however, if rainfall had not been so low. The long fire-free period prior to the experimental fire may also have contributed to the low mortality rate. The ability of Bursaria plants to ‘bounce back’ might be reduced if resources such as underground starch storages have not had time to build up. Noble (1982) found that although few mallee eucalypts died after a single burn, a second fire one year later, if fires were in autumn, caused considerable mortality.
Smaller plants were more vulnerable to fire than larger ones. This finding accords with other studies in the Sydney region (Morrison 1995, Morrison and Renwick 2000) and elsewhere (Hodgkinson 1998), and implies that individual plants must attain a certain minimum size before they become fire-tolerant. Most of the 11 plants which did not resprout were below 25 cm in height, while all plants over 50 cm resprouted. Vulnerability in a number of Poplar Box woodland shrub species was also concentrated in plants under around 25 cm in height (Hodgkinson 1998). Root rather than shoot development may be the critical factor for young Bursaria plants. Hodgkinson (1998) found that failure to resprout in semi-arid woodland shrubs was not due to lack of meristems, and speculated, on the basis of prior research into these species, that the ability of roots to supply nutrients, carbohydrates and water to meristems might be the
168 key factor. Resprouter seedlings are known to devote energy, in their early years, to developing the capacity to store starch (Bell 2001). Even small plants exhumed during the current study had impressive tap roots.
Although some tagged plants did not resprout, other resprouts appeared, as did seedlings. Over the second post-fire year, plant numbers gradually increased to 110% of the pre-fire total. This occurred despite unusually dry weather and grazing. So although a single fire may kill some individual Bursaria plants, this does not necessarily equate to a reduction in population density.
6.4.3 How does Bursaria reproduce?
The appearance of untagged post-fire resprouts suggested that plants might be reproducing through root suckering. A number of plants were excavated to test this idea, however no conclusive evidence of suckering was found. At this point, therefore, it seems reasonable to assume that Bursaria at Mt Annan reproduces by seed alone.
What, then, of the ‘new resprouts’? Presumably these plants existed prior to the fire, but were not visible above the ground. Given the propensity of some tagged plants to be visible at some assessments but not others, this supposition is not unreasonable.
Bursaria at Mt Annan certainly appears to reproduce from seed. Although many post- fire seedlings died, reasonable numbers were still alive and growing at the end of the study, despite low rainfall and grazing. Seedlings were overwhelmingly found in plots containing a large Bursaria plant.
6.4.4 When does Bursaria reproduce?
As this study covered only the first 25 months post-fire, the extent to which findings bear on this issue is limited. Certainly reproduction occurred in the post-fire period: many seedlings germinated in the first year after the fire, and some were also found in the second year.
The picture with respect to germination cues for this species is not particularly clear. Locally, Wood (2001) found significantly increased germination in response to heat, however Thomas (1994) did not. Roche et al. (1997), using seed from an undisclosed source, found a negative response to smoke; Clarke et al. (2000), using seed from the
169 New England Tablelands, did not. Germination on the Plain is promoted by cool temperatures (Warren 2003, L. von Richter pers. comm. 2003); this finding fits with the timing of germination in the current study. According to local Greening Australia staff, Bursaria seed is viable only when fresh (Warren 2003). Studies elsewhere support this observation: Roche et al. (1997) found 87% viability in fresh seed decreased to 8% after a year of soil storage, and Grant and MacGregor (2001) failed to find Bursaria seed in New England forest soil, despite the fact that B. spinosa was the most common shrub in the standing vegetation. However seedlings germinating at Mt Annan ten to 13 months after the fire must have come from seeds set prior to the burn, as resprouting plants certainly had not flowered by that time. Seed must therefore have remained viable for at least 18 months, and must also have remained ungerminated through the dry winter preceding the fire. Some seeds germinating in the second post-fire year could conceivably have come from plants which flowered the previous summer, as a small amount of flowering was observed in January 2004. (It is also possible that these seedlings germinated the previous year but were missed during earlier assessments.) Research at Mt Annan has found that local Bursarias, unlike their inland counterparts, need light to germinate (L. von Richter, pers. comm., 2004). The implication is that seeds do not germinate readily when buried, again suggesting that recruitment tends to occur from fresh seed. The extent of on-plant storage is unknown.
Hopefully monitoring at Mt Annan can continue over later post-fire years, thus providing data on the extent to which Bursaria recruits between fires. At present the assessment that it is able to do so is based on documented but casual observations of researchers at Mt Annan and Scheyville (Section 4.4.1), on widespread observations of increasing abundance across the Plain (eg Robinson 2003), and on variation in size classes in low fire frequency sites such as Mt Annan and Orchard Hills. Shrub species with a similar ability to build up numbers in the absence of fire occur in Australia’s semi-arid woodlands. Here, germination is episodic, occurring only after certain rainfall events (Hodginson 1979). Episodic recruitment of Melaleuca minutifolia, a savanna species whose abundance is also on the increase, has been demonstrated (Sharp and Bowman 2004).
In vital attributes terms Bursaria spinosa appears to be a VGT species (Appendix 3). Noble and Gitay (1996:333) include this combination of attributes in their species type 9. They describe members of this group – able to reproduce in the presence of their
170 own adults, resprout and reproduce by soil-stored seed – as ‘superspecies’, and note that taxa with such a wide range of advantages are probably uncommon.
6.4.5 When does Bursaria grow?
Resprouting plants grew between January and October 2003, which included a month of well above average rainfall (Figure 6.7). Growth over the next year of below average rainfall, however, was minimal, with many plants going backwards. Rainfall may play a vital role at a number of points in Bursaria’s life-cycle. In the current study, seedlings germinated after rain and died after dry weather. Drought-affected Bursaria plants across the Plain lose most of their leaves, though stems are often still alive (pers. obs. 2002, 2004).
6.4.6 Why is Bursaria more abundant in sites where fire is relatively infrequent: towards a model
Outlined below are some thoughts as to how recurrent fire, rainfall and Bursaria life history characteristics may interact to produce the patterns seen in CPW sites, particularly those reported in Chapters 4 and 5. While beyond the scope of the current project, in the future these thoughts could be refined on the basis of further research, and developed into a model.
• Large Bursaria plants may flower more abundantly than small plants, and set larger quantities of seed. Seed presumably falls mostly near fruiting plants. Germination in the first post-fire year may be higher under large Bursaria plants (Section 6.3.2) because these plants had a larger pre-fire seed pool, and also perhaps because higher fire intensity, due to the larger amount of shrub fuel in these patches, cues seeds to germinate (Section 6.4.4). • Bursaria seedlings germinate in winter following good rainfall (Section 6.3.6). Germination occurs in the first post-fire year, if rainfall is adequate. Many seedlings die but some survive (Section 6.3.2). Seedlings grow slowly (Section 6.3.5). The priority is development of a tap root (Section 6.3.7) that will enable the new plant to survive fire, grazing and drought. Acquiring this safeguard, however, takes time. Small plants are vulnerable to fire (Section 6.3.1). If a fire occurs before this crop of young plants becomes fire-tolerant, they will die. Existing adults will resprout, although perhaps at reduced levels (Section 6.4.2). The Bursaria population will stay steady or decline. Thus in sites with a high fire frequency, Bursaria populations could be expected to remain relatively low, with plants tending to occur in clumps. This was found to be the case in the three high fire frequency sites in the landscape study (Sections 4.3.1 and 5.3.1).
171 • However if a fire does not occur for several years, the post-fire seedlings, most of which established near large plants, will gain fire-tolerance. Existing thickets will expand slightly. Open areas away from thickets, however, will be relatively unaffected, as few post-fire seedlings occur in these patches. • Bursaria seeds germinate or otherwise lose viability fairly rapidly (Section 6.4.4). Thus although a small amount of germination may occur in the second post-fire year, large-scale germination will not occur again until resprouting plants mature, flower and set seed. Adequate rainfall will be required for good germination to occur; it may also be needed for flowering and seedset. As rainfall fluctuates year to year, it may be some time before a second wave of seedlings establishes near large bushes and starts to make its way towards fire tolerance. It may be even longer before small Bursaria plants burnt in the original fire grow large enough to flower and produce off-spring. If fire occurs during this period – perhaps 4 to 10 years post-fire – seedlings away from established thickets will not have reached fire tolerance. Thus Bursaria populations will remain roughly stable, and the existing patch structure will be reinforced. This pattern fits with findings in Chapters 4 and 5 with respect to moderate fire frequency sites. • However if a second fire still does not occur, the ability of Bursaria to produce seed and establish seedlings may gather pace, although recruitment is still likely to be episodic due to rainfall requirements. More plants flower, more seedlings attain fire tolerance. Thickets expand. Some seedlings manage to establish in open areas, and if no fire occurs for a number of years, attain both fire tolerance and sexual maturity. Now these plants too are in a position to father a thicket. • More than a decade has now passed since the original fire. Themeda clumps may be starting to senesce (Morgan and Lunt 1999); competition from the grass layer may therefore be reduced. Short-lived shrubs may also be dying, again reducing competition. The potential for Bursaria seedlings to establish may be further enhanced by these changes. • If a fire occurs after this time, the expanded Bursaria population may not be greatly affected, and may in fact be enhanced. This could be the scenario in the study site described in this chapter, where fire followed a probable 20 year interval. It could also explain the density and distribution of Bursaria in the other two low fire frequency sites in the landscape study.
While some of the ideas in the embryonic model presented above have a basis in research (albeit limited), others are currently pure speculation, for example the supposition that large plants flower more readily and produce more seeds than smaller plants, and that rainfall plays a key role in that process. Further research into the life history of Bursaria, at multiple sites across the Plain, will be needed before the interactions suggested here can be confidently deemed to resemble reality. Revisions and refinements could be made as knowledge increases. Research could aim to parameterise developmental stages and interactions.
The role of grazing could also be considered. Grazing may:
172 • Create gaps in the grass layer thus assisting Bursaria seedlings to establish. • Kill young seedlings (they could be eaten or trampled), particularly those not protected by large plants. • Restrict growth of Bursaria resprouts post-fire, particularly if grass fodder is limited by drought, and thus delay flowering and seed-set. • Limit flowering of small Bursaria bushes, but not large ones which have grown out of browsing range.
If correct, points 1 and 3 could help explain the observation that Bursaria abundance can increase rapidly when grazing is removed from a CPW site (S. Burgin pers. comm. 2001, Benson and Howell 2002). At this point, any existing browsed resprouts would be released, while grazing-induced gaps might still exist. Observation suggests, however, that cattle avoid Bursaria (N. Carlile, H. Recher pers. comm. 2004), so resprouts might not suffer as much from cattle grazing as they do where rabbits and macropods occupy a site. Points 2 and 4 suggest additional mechanisms whereby the patchiness of Bursaria may be maintained.
A future project could produce a simulation based on a refined version of these ideas, and compare results with the findings in Chapters 4 and 5.
6.5 Conclusion
Fire, rainfall and grazing probably all affect the ability of Bursaria spinosa to grow and reproduce. Some thoughts as to how these factors may interact are presented. The success of Bursaria where fire frequency is low may hinge on its ability to recruit between fires, to outlive its competitors, and to resprout and survive once a deep tap root has been developed. Vulnerabilities relate to the slow growth rate of seedlings, and of resprouts when rainfall is low and grazing pressure is high. While recruitment is probably episodic, long interfire intervals may allow existing plants to produce large quantities of seed, young plants to reach fire tolerance, and a number of recruitment events to occur. When a fire does occur after a period of Bursaria expansion, little impact on population numbers can be expected. In the current study, plant numbers at 25 months post-fire, after a probable 20 year interfire interval, were higher than they were prior to the burn. This finding underlines the need for periodic fire if open patches, as well as Bursaria thickets, are to be retained in the CPW landscape.
173 CHAPTER 7 FIRE AND HERB REPRODUCTION
7.1 Introduction
Soon after the project commenced in October 2001, the 2001/2 fire season began, with arson burns in several remnants. On Christmas Day 2001 a wildfire jumped the Nepean River and swept east through Cumberland Plain Woodland in Mulgoa Nature Reserve. Six weeks later, forbs and grasses had resprouted and were beginning to flower. This observation raised questions about the role of fire in recruitment of ground layer species in Cumberland Plain Woodland.
In Victoria’s lowland grasslands, facets of reproduction are associated with fire. Almost all species flower within 12 months of a burn (Lunt 1990, Morgan 1996), and Lunt (1994) found that seven of nine forb species investigated in this vegetation type flowered more abundantly in the first year post-fire than in the second. Seedling establishment appears to be focussed on the early post fire years, as expanding Themeda clumps limit opportunities by three years post-fire (Morgan 1998a).
Little is known, however, about the fire-related dynamics of reproduction in ground layer species in Australia’s grassy woodlands. This chapter reports a small study which sought to explore whether similar processes to those found in Victoria’s grasslands might pertain in Cumberland Plain Woodland. Areas burnt within the last six months were compared to nearby areas which had not experienced a recent fire, with respect to:
• the extent of flowering and fruiting in ground layer species; • the extent of bare ground in which seedlings could potentially establish; • numbers of seedlings; and • species composition of flowering and fruiting plants.
174 7.2 Methods 7.2.1 Study sites and fires
Sampling was carried out at two sites, Nurragingy and Glenmore Park, in Autumn 2002. These sites were similar in that they contained recently burnt areas in reasonable proximity to areas which had not been burnt for at least five years, however they also differed in a number of ways. They are therefore best conceptualised as two case studies.
Nurragingy Nature Reserve (Nurragingy) is a popular recreation area managed by Blacktown Council. It contains a small patch of Shale Plains Woodland adjoining a strip of Alluvial Woodland along Eastern Creek (NPWS 2004a). Between late November 2001 and early January 2002, several arson fires occurred in the woodland, burning about half the CPW area. These fires were of moderate intensity, causing some canopy scorch. The previous fire history at Nurragingy was not clear in detail, however a large fire reportedly burnt right through the area in the mid 1980s. Another major fire occurred in the mid 1990s, although it did not burn the entire CPW section of the Reserve (pers. comm. Ted Williams, Wayne Butcher and Wira Tewhare, Blacktown RFS, 2002; Blacktown Council staff via Glenn White and Clare Nuttgens, 2002). At the time of sampling, therefore, time-since-fire was between 3 and 5 months in the burnt area, and between 6 and 17 years in the unburnt area. The burnt and unburnt areas were either contiguous, or separated only by a walking track.
Glenmore Park is a housing estate just south of Penrith. It is bounded to the west and south by Mulgoa Nature Reserve (Mulgoa), an extensive patch of bushland managed by the NSW National Parks and Wildlife Service. The entire Reserve burnt on 25 December 2001 in an intense fire which swept across the Nepean River from the Blue Mountains National Park. Forest Red Gum Reserve (FRG) is a small bushland park located near to, but not contiguous with, Mulgoa. Surrounded by housing, it remained unburnt. This area, which is managed by Penrith Council, had not burnt for at least ten years prior to sampling, and probably longer (pers. comm. Will Robertson, Penrith Council, 2002). Mulgoa covers a large area relative to FRG, and includes a number of different vegetation types. Shale Hills Woodland predominates (NPWS 2004a). FRG is mapped as Shale Plains Woodland. The area of Mulgoa selected for study was that judged most likely to carry similar vegetation to FRG, on the basis of its topography
175 (ridge-top and adjacent gentle slopes) and the fact that it contained vegetation mapped as Shale Plains Woodland in the version of the NPWS survey maps available at the time (NPWS 2000). FRG and the area sampled at Mulgoa are approximately 2 km from each other. Updated maps now categorise this part of Mulgoa as Shale Hills Woodland. The past fire history at Mulgoa is unclear, but the last interfire interval was at least ten years, and probably longer (pers. comm. John Foster, Rural Fire Brigade, Regentville, 2002). Crown scorch from the Christmas wildfire was extensive through most of the study area; some trees were burnt to the ground. At the time of sampling, time-since- fire was approximately four months in the burnt area, and over ten years in the unburnt area.
7.2.2 Sampling and data collection
Each treatment area (burnt, unburnt) was divided into three blocks of approximately equal size. Two 25 m x 25 m plots were randomly selected within each block. At Mulgoa, plots crossing two drainage lines in the study area were omitted from the draw, as it was considered they might differ from the well-drained areas in the rest of this site, and at FRG.
Within each plot, six one metre square subplots were randomly selected, along with some ‘spares’. In the field, a ‘spare’ subplot was surveyed if one of the original six proved to be on a path, to contain rubbish, or to have been recently mowed (Council mowed strips along paths in FRG during the course of the study). A ‘spare’ was substituted for one originally-selected subplot at Nurragingy where the corner of an unburnt plot overlapped the burn edge.
Sampling was conducted between 5 and 19 April 2002 at Nurragingy, and between 26 April and 9 May at Glenmore Park, after a peak of above-average rainfall the preceding summer (Section 2.3.2). It was recognised that flowers and fruit would form and fall as the season progressed towards winter, so burnt and unburnt plots were sampled alternately, to minimise any bias this might cause.
Within each 1 m2 subplot, reproductive structures, seedlings, and bare ground were assessed.
176 Reproductive structures of all ground layer species, both native and exotic, were counted.
• For most forb species, the unit counted was the individual flower or fruit. Lily- and iris-like monocots were included in this category. • For Asteraceous forbs, flowering or fruiting heads were counted. • For grasses and graminoids (‘grass-like’ plants – in this study all relevant species fell within the family Cyperaceae), the number of inflorescence-bearing culms was tallied.
Most taxa were identified to species level. However native Glycine species were grouped, as were native Oxalis, Arthropodium, Panicum and Sporobolus, and exotic Solanum, species. This was because:
• individual species could not always be reliably distinguished in the field, and • Herbarium staff familiar with the Cumberland Plain advised that taxonomic distinctions in existing keys were not clear-cut (D. Benson and J. Howell, pers. comm. 2003).
Taxa were classified as either native to the Cumberland Plain, or exotic. Digitaria didactyla and Bidens pilosa were classified as exotics.
Seedling numbers in each subplot were assessed on a five point scale: 1, one seedling; 2, 2-9 seedlings; 3, 10-49 seedlings; 4, 50-99 seedlings; 5, 100 or more seedlings. Numbers of dicots and monocots were counted separately, but no attempt was made to identify individual species. Where cotyledons were not apparent, seedlings were distinguished from resprouts by their size and single-stemmed habit.
The proportion of exposed bare ground in each subplot was estimated to the nearest ten percent.
7.2.3 Univariate data analysis
For most variables, a nested analysis of variance, with fire recency as a fixed factor, and block and plot as nested, random factors, was used to assess differences between the burnt and unburnt areas at each site. The unit of replication was the subplot.
For each variable, Cochran’s test was used to assess homogeneity of variances at plot level. If variances were homogeneous (P > 0.05 when C6,12 < 0.244), the analysis
177 proceeded, giving significance levels for differences between burnt and unburnt areas, blocks within these categories, and plots within blocks.
Where plot variances were heterogeneous, the analysis was moved up a level, using the average score per plot as the unit of analysis, and Cochran’s test was again applied. If variances at block level were homogeneous (P > 0.05 when C4,6 < 0.480), the analysis proceeded, giving significance levels for differences between fire recency categories and blocks within these categories.
Where variance was heterogeneous at both plot and block level, univariate permutation tests were employed to test for differences between plot means in burnt and unburnt areas, using 5000 permutations in the program Resampling Stats (Blank et al. 2001).
Count data were transformed for analysis, using the formula ln(x+1).
Ordinal-scale seedling data from the 72 subplots in each site were grouped into 2 x 3 contingency tables, with categories for subplots containing ≤ 1, 2 - 9, and ≥ 10 seedlings. Chi-square tests were used to assess differences between burnt and unburnt areas.
7.2.4 Multivariate data analysis
The spatial analysis program PRIMER was used to explore patterns in the flowering and fruiting of native species, at each site. Plot level data, log (x + 1) transformed, were used for these analyses.
A similarity matrix was constructed using the Bray-Curtis metric. Data were clustered through group averaging, and visually presented through a multi-dimensional scaling ordination. Analysis of similarities was employed to further assess the effect of burning on reproductive patterns. Similarity percentages and species contributions were calculated using the SIMPER option in PRIMER. (See Section 5.2.2 for a more detailed explanation of these analyses).
178 7.3 Results 7.3.1 Nurragingy forbs
Forb flowers and fruits
At Nurragingy, forb fruits far outnumbered flowers, and forb reproductive structures were considerably more abundant in recently burnt plots.
Initial evaluation of the forb data revealed the presence of an outlier in the unburnt area. In one subplot a single species, Veronica plebeia, a native which grows as a weed in lawns (Harden 1992), produced 298 fruits, 75% of the total for all 36 unburnt subplots. This species was not found in any other subplot, in either treatment area. Grubbs’ test (Sokal and Rohlf 1995) clearly identified the V. plebeia subplot as an outlier (z = 3.330, P < 0.01).
The analysis in this section will focus on the situation with V. plebeia excluded from the data. However figures including this species are also provided, in brackets.
Forb fruits far outnumbered flowers, making up 90.1% (92.8%) of all forb reproductive structures across the site (Figure 7.1). Flowers were therefore combined with fruits for some analyses. Native flowers and fruits far outnumbered exotics: 97.8% (98.3%) of reproductive structures were attributable to native species.
20
18
16 2 14
12
10
8
6 Mean number per 1m per number Mean 4
2
0 Native flowers Native fruits Exotic flowers Exotic fruits Type of reproductive structure
Figure 7.1. Mean number of reproductive structures per 1 m2 in recently burnt (red) and unburnt (blue) areas at Nurragingy. Veronica plebeia excluded.
179 Native forbs in recently burnt plots produced many more flowers and fruits than those in the unburnt area. Eighty-six point three percent (60.6%) of fruits occurred in the burnt area, as did 94.4% of flowers. Excluding Veronica plebeia, almost seven times more reproductive effort was occurring in recently burnt plots. Differences were significant for native fruits alone (Table 7.1) and for flowers and fruits together (Figure 7.2, Table 7.2). Even with Veronica plebeia included, differences were still significant
(fruits only, F(1,4) = 17.191, P = 0.014; flowers and fruits together, F(1,4) = 19.770, P = 0.011).
3.00
2.50 (ln(x+1)) 2 2.00
1.50
1.00
0.50 Mean no. flowers and fruits/1m 0.00 Burnt Unburnt Figure 7.2. Mean number of native forb reproductive structures per 1 m2 (log transformed data) in six blocks at Nurragingy, three recently burnt, three unburnt. Veronica plebeia excluded.
Table 7.1. ANOVA comparing forb fruits in burnt and unburnt areas at Nurragingy. Veronica plebeia excluded from outlier plot; ln(x+1) transformed.
Source Sums of squares d.f. Mean square F P Fire recency 37.633 1 37.633 52.850 0.002 Block (Fire recency) 2.848 4 0.712 0.364 0.827 Plot (Block) 11.753 6 1.959 1.471 0.203 Error 79.874 60 1.331 Total 132.109 71
Cochran’s test C6,12 = 0.236, P = NS
180 Table 7.2. ANOVA comparing forb reproductive structures (flowers and fruits combined) in burnt and unburnt areas at Nurragingy. Veronica plebeia excluded from outlier plot; ln(x+1) transformed.
Source Sums of squares d.f. Mean square F P Fire recency 44.619 1 44.619 56.588 0.002 Block (Fire recency) 3.154 4 0.788 0.436 0.779 Plot (Block) 10.842 6 1.807 1.312 0.266 Error 82.666 60 1.378 Total 141.281 71
Cochran’s test C6,12 = 0.229, P = NS
The few exotic forb flowers and fruits found at Nurragingy were all in the recently burnt area (Figure 7.1). However they occurred in only three subplots, and the difference between burnt and unburnt plots was not significant (P = 0.453).
Native forb species
The tendency for increased reproduction in the recently-burnt area could also be discerned at the species level.
Nineteen native forb species were flowering or fruiting in plots at Nurragingy. Of these, 15 were found in the burnt area, and ten in the unburnt.
The average number of species flowering or fruiting per square metre was significantly higher in the burnt area (1.86 species) than in the unburnt (0.69 species; F(1,4) = 24.845, P = 0.008).
181 7.3.2 Nurragingy grasses and graminoids
Culms
Native grasses and graminoids produced many more flowering stalks than did exotic species: exotic culms made up only 1.9% of the 5506 culms counted across the Nurragingy site (Figure 7.3).
140.0
120.0 2 100.0
80.0
60.0
40.0 Mean number per 1m per number Mean
20.0
0.0 Native culms Exotic culms Type of reproductive structure
Figure 7.3. Mean number of grass and graminoid culms per 1 m2 in recently burnt (red) and unburnt (blue) areas at Nurragingy.
There were significantly more native culms in the recently burnt area (Figure 7.4, Table 7.3); 73.3% of native culms were found here (Figure 7.3).
6.00
5.00 (ln (x+1)) 2 4.00
3.00
2.00
1.00 Mean numberculms per of 1m 0.00 Burnt Unburnt Figure 7.4. Mean number of native grass and graminoid culms per 1 m2 (log transformed data) in six blocks at Nurragingy, three recently burnt, three unburnt.
182
Table 7.3. ANOVA comparing number of grass and graminoid culms (ln(x+1) transformed) in burnt and unburnt areas at Nurragingy.
Source Sums of squares d.f. Mean square F P Fire recency 17.037 1 17.037 14.253 0.020 Block (Fire recency) 4.781 4 1.195 0.272 0.886 Plot (Block) 26.415 6 4.403 4.172 0.001 Error 63.310 60 1.055 Total 111.544 71
Cochran’s test C6,12 = 0.214, P = NS
Flowering of exotic grasses and graminoids was not extensive anywhere at Nurragingy, with only 102 culms recorded across the site. Although a large majority of these (89 culms) occurred in the burnt area, the difference between treatments was not significant (P = 0.509).
Native grass and graminoid species
At species level, differences between burnt and unburnt areas at Nurragingy were minor. Of the 20 native grass and graminoid species found flowering across the site, 15 were in the burnt area, and 13 in the unburnt. The average number of species flowering in the burnt area was 3.00 per square metre, relative to 2.89 in the unburnt plots, an insignificant difference (F(1,4) = 0.050, P = 0.835).
183 7.3.3 Bare ground and seedlings at Nurragingy
Bare ground
Bare ground was definitely a feature of the burnt area, but was almost non-existent in the unburnt area. Three-quarters of the unburnt plots were scored as having no bare ground, while none of the remaining plots scored above 10%. In contrast, bare ground in the burnt area averaged 43%. The difference between burnt and unburnt areas was significant at 0.001 level (Figure 7.5, Table 7.4).
100.00
90.00
80.00
70.00
60.00
50.00
40.00
30.00
20.00
Mean percent exposed bare ground bare exposed percent Mean 10.00
0.00 Burnt Unburnt
Figure 7.5. Average percent exposed bare ground in six blocks at Nurragingy, three recently burnt, three unburnt.
Table 7.4. ANOVA comparing percent exposed bare ground in burnt and unburnt areas at Nurragingy. Plot means used as replicates.
Source Sums of squares d.f. Mean square F P Fire recency 33368 1 33368 75.313 0.001 Block (Fire recency) 1772 4 443 0.958 0.493 Error 2775 6 463 Total 37915 11
Cochran’s test: C4,6 = 0.382, P = NS
184 Seedlings
There were significantly more seedlings in recently burnt areas at Nurragingy.
Thirteen subplots in the burnt area (36%) had 50 or more dicot seedlings, whereas no subplots in the unburnt area had this level of recruitment. On the other hand, 19 unburnt plots (53%) had either no dicot seedlings, or only one, while only one burnt subplot had less than two seedlings (Table 7.5). The difference was highly significant (χ2 = 43.57, df = 2, P < 0.001).
Table 7.5. Number of 1 m2 subplots containing various levels of dicot seedlings in burnt and unburnt areas at Nurragingy.
Number of subplots with no one 2 - 9 10 - 49 50 - 99 100 + Total seedlings seedling seedlings seedlings seedlings seedlings Burnt 0 1 6 16 7 6 36 Unburnt 14 5 15 2 0 0 36
No unburnt subplot had more than nine monocot seedlings, whereas 26 burnt plots (72%) had at least this level of recruitment. Eight recently burnt subplots (22%) had 100 or more monocot seedlings (Table 7.6). Again, the difference between treatments was highly significant (χ2 = 46.26, df = 2, P < 0.001).
Table 7.6. Number of 1 m2 subplots containing various levels of monocot seedlings in burnt and unburnt areas at Nurragingy.
Number of subplots with no one 2 - 9 10 - 49 50 - 99 100 + Total seedlings seedling seedlings seedlings seedlings seedlings Burnt 1 0 9 13 5 8 36 Unburnt 10 12 14 0 0 0 36
185 7.3.4 Composition of species reproducing at Nurragingy
Ordination cleanly separated the burnt and unburnt plots at Nurragingy (Figure 7.6). Analysis of similarities confirmed the existence of a significant difference between the complement of species flowering and fruiting at the two different times after fire (Global R = 0.641, P = 0.002).
Burnt plots were more similar to each other (average similarity 57.1%) than were unburnt plots (44.9%). The average dissimilarity between burnt and unburnt plots was 60.0%.
Stress: 0.13
Burnt
Unburnt
Figure 7.6. Two dimensional ordination (MDS) of 12 plots at Nurragingy, six burnt, six unburnt, based on extent of ground layer flowering and fruiting.
Fifteen species contributed more than three percent to the dissimilarity between the two treatment areas. Twelve of these species were flowering and/or fruiting more abundantly in the burnt area: six forbs, five grasses and one graminoid. The three discriminating species reproducing more abundantly in the unburnt area were all grasses (Table 7.7).
186 Table 7.7. Species whose flowering and/or fruiting contributing more than 3% to the average dissimilarity between recently burnt and unburnt plots at Nurragingy.
Species Life-form More abundant in % contribution Cumulative burnt or unburnt to dissimilarity % area? Aristida vagans grass burnt 6.84 6.84 Hypoxis hygrometrica forb burnt 6.70 13.54 Entolasia marginata grass unburnt 6.25 19.79 Brunoniella australis forb burnt 5.96 25.76 Digitaria diffusa grass burnt 5.90 31.65 Microlaena stipoides grass burnt 5.39 37.04 Themeda australis grass unburnt 5.04 42.09 Fimbristylis dichotoma graminoid burnt 4.33 46.42 Stackhousia viminea forb burnt 4.24 50.66 Eragrostis leptostachya grass unburnt 4.17 54.83 Glycine tabacina forb burnt 3.69 58.51 Tricoryne elatior forb burnt 3.63 62.15 Paspilidium distans grass burnt 3.51 65.66 Sporobolus creber grass burnt 3.47 69.12 Opercularia diphylla forb burnt 3.12 72.24
7.3.5 Glenmore Park forbs
Forb flowers and fruits
There were many more fruits than flowers at Glenmore Park: fruits made up 95.3% of all reproductive structures across the two treatment areas. Natives again predominated, although exotics were more in evidence than at Nurragingy, making up 19.8% of all forb reproductive structures (Figure 7.7).
187 60
50 2
40
30
20 Mean number per 1m per number Mean 10
0 Native flowers Native fruits Exotic flowers Exotic fruits Type of reproductive structure
Figure 7.7. Mean number of reproductive structures per 1 m2 in recently burnt (red) and unburnt (blue) areas at Glenmore Park.
A large majority of native forb flowers and fruits, 87.1%, occurred in recently burnt plots (Figure 7.7). For fruits alone, the equivalent figure was 86.9%. The difference between fire recency categories was significant both for fruits alone (Table 7.8) and for all reproductive structures together (Figure 7.8, Table 7.9).
5
4.5
4 (ln(x+1)) 2 3.5
3
2.5
2
1.5
1
0.5 Mean no. flowers and fruits/1m 0 Burnt Unburnt
Figure 7.8. Mean number of native forb reproductive structures per 1 m2 (log transformed data) in six blocks at Glenmore Park, three recently burnt, three unburnt.
188 Table 7.8. ANOVA comparing forb fruits (ln(x+1) transformed) in burnt and unburnt areas at Glenmore Park.
Source Sums of squares d.f. Mean square F P Fire recency 93.911 1 93.911 14.527 0.019 Block (Fire recency) 25.858 4 6.465 5.861 0.029 Plot (Block) 6.617 6 1.103 0.742 0.618 Error 89.239 60 1.487 Total 215.626 71
Cochran’s test C6,12 = 0.191, P = NS
Table 7.9. ANOVA comparing forb reproductive structures (flowers and fruits combined, ln(x+1) transformed) in burnt and unburnt areas at Glenmore Park.
Source Sums of squares d.f. Mean square F P Fire recency 122.066 1 122.066 30.360 0.005 Block (Fire recency) 16.083 4 4.021 5.636 0.031 Plot (Block) 4.280 6 0.713 0.502 0.805 Error 85.287 60 1.421 Total 227.715 71
Cochran’s test C6,12 = 0.203, P = NS
Exotic flowers and fruits were much more prevalent in the burnt area: 96% of them were found here. However they were found in only a small number of subplots, and the difference between fire recency categories failed to reach significance (P = 0.085).
Native forb species
Twenty-one native forb species were flowering or fruiting at Glenmore Park, 18 in the burnt area, and 12 in the unburnt. The average number of species reproducing per square metre was considerably higher in the burnt plots (3.42) than in the unburnt ones
(0.89). This difference was highly significant (F(1,4) = 31.607, P = 0.005).
189 7.3.6 Glenmore Park grasses and graminoids
Culms
Although native grass and graminoid culms predominated, many exotic culms were counted at Glenmore Park, particularly in the unburnt FRG plots (Figure 7.9). Exotic culms made up 27.1% of the total across the Glenmore Park site.
90.0
80.0
70.0 2
60.0
50.0
40.0
30.0
Mean number per 1m per number Mean 20.0
10.0
0.0 Native culms Exotic culms Type of reproductive structure
Figure 7.9. Mean number of grass and graminoid culms per 1 m2 in recently burnt (red) and unburnt (blue) areas at Glenmore Park.
Over half the native culms (54.4%) occurred in the recently burnt area. However the difference between the two treatments was not significant (Table 7.10, Figure 7.10).
Table 7.10. ANOVA comparing number of grass and graminoid culms (ln(x+1) transformed) in burnt and unburnt areas at Glenmore Park. Plot means used as replicates.
Source Sums of squares d.f. Mean square F P Fire recency 0.118 1 0.118 0.045 0.842 Block (Fire recency) 10.401 4 2.600 0.701 0.619 Error 22.272 6 3.712 Total 32.731 11
Cochran’s test C4,6 = 0.193, P = NS
190 4.50
4.00
3.50 (ln(x+1)) 2 3.00
2.50
2.00
1.50
1.00
0.50 Mean number of culms per 1m of culms Mean number 0.00 Burnt Unburnt
Figure 7.10. Mean number of native grass and graminoid culms per 1 m2 (log transformed data) in six blocks at Glenmore Park, three recently burnt, three unburnt.
In a reversal of trends to date, significantly more exotic grass culms were found in the unburnt area at Glenmore Park (F(1,4) = 17.435, P = 0.014). Ninety-two percent (1495) of the exotic culms counted across the site occurred in unburnt plots. A large majority of this total – 1410 culms – was attributable to one species, Digitaria didactyla. D. didactyla was not recorded in the burnt area. If this species is excluded, there were in fact more exotic culms in the burnt area (125) than in the unburnt (85), although this difference was not significant.
Native grass and graminoid species
There were few differences, at a species level, between burnt and unburnt plots with respect to native grasses and graminoids. Of the 24 native species flowering at Glenmore Park, 19 were found in burnt plots and 18 in unburnt. The average number of native species with culms in the burnt and unburnt areas was 3.3 and 4.1 per m2 respectively, a difference that was not significant (F(1,4) = 0.656, P = 0.464).
191 7.3.7 Bare ground and seedlings at Glenmore Park
Bare ground
Again, bare ground was found throughout the burnt area, but was rare where a fire had not occurred. All subplots in the burnt area had at least some patches which were free of litter, with exposed bare ground averaging 32%. In the unburnt reserve, however, all but three subplots were scored as having zero bare ground. The difference between burnt and unburnt areas was significant at greater than 0.001 level (Figure 7.11, Table 7.11).
100.00
90.00
80.00
70.00
60.00
50.00
40.00
30.00
20.00
10.00 Mean percent exposed bare ground bare exposed percent Mean 0.00 Burnt Unburnt
Figure 7.11. Average percent exposed bare ground in six blocks at Glenmore Pak, three recently burnt, three unburnt.
Table 7.11. ANOVA comparing percent exposed bare ground in burnt and unburnt areas at Glenmore Park. Plot means used as replicates.
Source Sums of squares d.f. Mean square F P Fire recency 17422 1 17422 179.2 < 0.001 Block (Fire recency) 389 4 97 0.232 0.911 Error 2517 6 419 Total 11
Cochran’s test: C4,6 = 0.356, P = NS
192 Seedlings
Seedlings in the recently burnt areas at Mulgoa far outnumbered those in the unburnt FRG plots.
Twenty-seven subplots in the burnt area (75%) had ten or more dicot seedlings, whereas this applied to only two subplots (6%) in the unburnt area. The number of subplots with less than two seedlings was 16 (44%) in the unburnt area, but zero where fire had occurred four months previously (Table 7.12). This difference was highly significant (χ2 = 40.55, df = 2, P < 0.001).
Table 7.12. Number of 1 m2 subplots containing various levels of dicot seedlings in burnt and unburnt areas at Glenmore Park.
Number of subplots with no one 2 - 9 10 - 49 50 - 99 100 + Total seedlings seedling seedlings seedlings seedlings seedlings Burnt 0 0 9 24 2 1 36 Unburnt 10 6 18 2 0 0 36
No unburnt subplot had more than nine monocot seedlings, whereas 13 burnt plots (36%) had at least this level of recruitment. Twenty-four unburnt subplots (67%), but only two burnt ones (6%) had less than two seedlings (Table 7.13). Again, the difference between categories was highly significant (χ2 = 34.07, df = 2, P < 0.001).
Table 7.13. Number of 1 m2 subplots containing various levels of monocot seedlings in burnt and unburnt areas at Glenmore Park.
Number of subplots with no one 2 - 9 10 - 49 50 - 99 100 + Total seedlings seedling seedlings seedlings seedlings seedlings Burnt 0 2 21 12 1 0 36 Unburnt 13 11 12 0 0 0 36
193 7.3.8 Composition of species reproducing at Glenmore Park
The burnt and unburnt plots at Glenmore Park were clearly separated in the MDS (Figure 7.12). The analysis of similarities Global R was 0.807 (P = 0.002), indicating a very significant difference between the complement of species reproducing in the two different treatment areas.
Burnt plots were again more similar to each other (average similarity 52.3%) than were unburnt plots (43.6%). The average dissimilarity between burnt and unburnt plots was 76.4%.
Stress: 0.07
Burnt
Unburnt
Figure 7.12. Two dimensional ordination (MDS) of 12 plots at Glenmore Park, six burnt, six unburnt, based on extent of ground layer flowering and fruiting.
The majority of the thirteen species contributing over three percent to the dissimilarity between the burnt and unburnt areas were found in the burnt area: six of these eight species were forbs. All five discriminating species flowering or fruiting more abundantly in the unburnt area were grasses (Table 7.14).
194 Table 7.14. Species whose flowering and/or fruiting contributing more than 3% to the average dissimilarity between recently burnt and unburnt plots at Nurragingy.
Species Life-form More abundant in % contribution Cumulative burnt or unburnt to dissimilarity % area? Paspilidium distans grass burnt 5.77 5.77 Brunoniella australis forb burnt 5.23 11.00 Microlaena stipoides grass unburnt 5.10 16.10 Scleria mackaviensis graminoid burnt 4.73 20.83 Arthropodium milleflorum forb burnt 4.43 25.27 Phyllanthus virgatus forb burnt 4.38 29.65 Hypoxis pratensis forb burnt 4.23 33.88 Digitaria diffusa grass unburnt 3.62 37.51 Stackhousia viminea forb burnt 3.35 40.86 Eragrostis brownii grass unburnt 3.27 44.13 Glycine tabacina forb burnt 3.15 47.28 Aristida ramosa grass unburnt 3.10 50.37 Aristida warburgii grass unburnt 3.03 53.40
7.4 Discussion 7.4.1 Status of findings
The study reported in this chapter provides some limited information about the extent of ground layer reproduction in recently burnt Cumberland Plain Woodland, relative to that in nearby patches which had not had a fire for some time. Its primary weakness is limited replication: only two sites were surveyed, each somewhat different in terms of vegetation type, fire history, intensity of the recent fire, and proximity of the treatment areas. Further sampling across a wider range of CPW sites and fires will be needed before findings can be considered to apply across the Plain.
The study was also limited in that it did not measure recruitment into the adult population directly. Rather, a number of variables which may be associated with successful recruitment were assessed. It would be instructive to follow the fate of seeds from fruits, and of seedlings, over time. It is also important to note that measures of recruitment are not the same as measures of population increase; mortality rates also
195 need to be considered. This study thus contributes information in relation to a subset of the processes which determine populations over time.
Within these limitations, some interesting findings emerged. Both the moderate intensity burn at Nurragingy and the high intensity fire at Mulgoa were rapidly followed by a pulse of flowering and fruiting, particularly amongst forbs. Bare ground was significantly more prevalent shortly after fire, and seedlings were abundant.
7.4.2 Flowering and fruiting
Although the two case study sites differed in a number of respects, the patterns of flowering and fruiting in native species were remarkably similar. In each site, forb reproduction was greatly enhanced in recently burnt areas. Significant findings were recorded both at species level, and in number of reproductive structures. When Veronica plebeia was omitted from an outlier subplot at Nurragingy, there were almost seven times more native forb flowers and fruits in burnt plots, at each site. The SIMPER analyses reinforced these findings. At both sites, all forb species contributing more than three percent to the dissimilarity between burnt and unburnt plots were more abundant in the burnt area. These forbs were all resprouters, a number of which – for example Hypoxis spp., Brunoniella australis and Arthropodium milleflorum – have underground storage organs (Harden 1992, 1993). Rapid post-fire flowering by forbs resprouting from subterranean storage organs has also been documented in Victoria (Morgan 1996). In fact, flowering in CPW may be even more rapid than in these grasslands, where the majority of species failed to flower in the first six months after a late November burn, despite high rainfall (Morgan 1996).
A number of the forb species which discriminated the burnt from the unburnt plots formed part of Aboriginal diets in south-eastern Australia (Kohen and Downing 1992), including Arthropodium milleflorum, Glycine tabacina and Hypoxis hygrometrica. Gott (1999) repeats a quote from Robinson in 1841: “they [the Aborigines] burn the grass, the better to see these roots, but this burning is a fault charged against them by the squatters.” Certainly Hypoxis species, in particular, are difficult to locate when not in flower (pers. obs. 2002-4).
196 The difference in flowering and fruiting between burnt and unburnt areas was much less dramatic for native grass and graminoid species than for forbs. While more native culms occurred in the burnt area at each site, the difference only reached significance at Nurragingy. There were no significant differences between the treatments in numbers of flowering grass and graminoid species. The SIMPER analyses reflected the ambiguity of these findings. At Nurragingy, five of the eight grass species contributing more than three percent to the dissimilarity between treatments were more abundant in the burnt area, whereas five of the six doing so at Glenmore Park occurred in the unburnt area.
The findings of this study in relation to forbs accord with those of Lunt (1994): seven of nine grassland forb and subshrub species in a Victorian Themeda grassland flowered more abundantly at six months post-fire than they did two years after a burn. This grassland was once a Eucalyptus tereticornis woodland, and its forb complement has much in common with that in CPW. Fire-stimulated flowering is also reportedly common amongst South African monocots (Bond 1997), including orchids and lilies (Kruger 1983 and Front 1984, cited in Bell et al. 1993). Herb flowering is particularly profuse in the first post-fire year in shrubby Western Australian woodlands (Baird 1977), and amongst geophytes in Californian coastal sage scrub (Keeley and Keeley 1984).
A number of factors may have contributed to the differential response of forbs and grasses.
• Forbs are more likely than grasses to be interstitial species, growing in the gaps between grass clumps (Figure 7.13). As time-since-fire progresses, these gaps close. At 3 to 5 months post-fire at Nurragingy, bare ground averaged 43%, at four months post-fire at Glenmore Park, 32%. There was therefore much more room for interstitial species soon after fire than in the later post-fire years, when gaps were rare. In a Victorian grassland, gaps between Themeda clumps were common in the first two years after fire, rare at three years, and had virtually disappeared by four years post-fire (Morgan 1998a). In woodlands, gaps are filled both through expansion of grass clumps, and by eucalypt litter (Figure 7.14, Chapter 9). The initial post-fire years therefore offer interstitial species access to light, and possibly other resources, to a degree which they are unlikely to encounter in later years. Grass species, which are often relatively large in stature and may themselves constitute the matrix, are more likely to be able to access resources in the later post-fire years. It therefore makes ecological sense that forbs are more likely than grasses to concentrate their growth and flowering in the early post-fire period.
197 • Themeda australis is an important grass species in CPW, dominating many sites (Chapter 4). Very few culms of this species were encountered in burnt plots during this study. Observation over the course of the project indicated that abundant flowering in Themeda occurs in the second, or sometimes in the later months of the first, post-fire year, but not in the initial post-fire months. Flowering in this species has been shown to decline sharply by five years post-fire in a Victorian grassland (Morgan and Lunt 1999). This may explain the fact that although a large majority of culms of this species were encountered in the unburnt plots both at Nurragingy and Glenmore Park, culm numbers were still not high. If the study had taken place a few months later, once Themeda flowering had commenced in the burnt plots, the difference in numbers of grass culms between burnt and unburnt areas may well have been significant at Glenmore Park as well as Nurragingy. • The lack of a significant difference in the number of grass culms at Glenmore Park may also owe something to the severity of the fire there, relative to that at Nurragingy. Grass species may have recovered and flowered more rapidly after the more moderate burn at the latter site. Adams and Simmons (1996) found higher grass cover in a Victorian dry sclerophyll forest 2.5 years post-fire where fire had been less intense.
Figure 7.13. Native herbs flowering two months after fire, in gaps between grass clumps.
Exotic flowers and fruits were generally encountered in small numbers, making it difficult to draw meaningful conclusions. However there is no evidence that exotic forbs are exhibiting a slower post-fire flowering response than native forbs. In each site, flowering and fruiting in exotic forbs was much greater in the recently burnt areas, though differences did not reach significance.
198 A. B.
. . C.
Figure 7.14. A: Grass clumps and litter three months post-fire at Nurragingy, showing exposed bare ground. B. Litter building up between grass clumps 16 months post-fire at Orchard Hills. C. Grass clumps and litter four years post-fire at Windsor Downs.
The significantly greater degree of flowering amongst grass and graminoid species in the unburnt area at Glenmore Park is attributable to Digitaria didactyla, Queensland Blue Couch. The status of this species on the Cumberland Plain (exotic or native) is not clear (pers. comm., Royal Botanic Gardens Herbarium staff, 2002). It has been classified as an exotic in Forest Red Gum Reserve because it was not encountered in other CPW sites surveyed for this project, and because it occurred in situations where it could easily have established from garden waste from the surrounding housing estate. If this species is not considered, again there is no evidence that exotic grasses are slower than natives to flower post-fire.
7.4.3 Seedling emergence
In each site, seedling numbers were considerably greater in recently burnt plots. This finding applied equally to dicots and monocots.
199 The presence of gaps in the burnt area probably contributed to the high rates of seedling emergence (Hitchmough et al. 1996, Odgers 1996, Morgan 1998a, but see Clarke and Davison 2004).
It is sometimes suggested that drought creates bare ground for seedling establishment in CPW, so fire is not necessary. The winter preceding the current study was sufficiently dry to produce a destructive fire season (see Section 2.3.2 for information on rainfall during the project), however this did not result in either substantial areas of bare ground in unburnt plots, nor in any great degree of seedling establishment. Litter in CPW sites is more likely to come from eucalypts than grasses (Chapter 9), and eucalypt litter is more likely to increase than decrease during a drought, due to both greater accession, and reduced decomposition (Simmons and Adams 1986).
Seedling numbers in the burnt area may have been enhanced by fire cued germination. While many CPW species do not respond to fire-related cues (heat and smoke), some do (Section 1.3.4). Heat and/or smoke-responsive species include several legumes, the dicot forbs Dichondra repens, Wahlenbergia gracilis and Poranthera microphylla (Warcup 1980, Wood 2001, Hill and French 2003), and the grasses Eragrostis leptostachya, Panicum effusum, Panicum simile (S. Clarke 2003), Eragrostis brownii and Paspalidium distans (Read et al. 2000, Wood 2001). CPW populations of Themeda australis have been found to have a negative response to heat (Wood 2001, Hill and French 2003), but a positive response to smoke (Wood 2001).
Might some seedlings in burnt plots have germinated from seeds resulting from the post-fire flowering pulse observed in this study? While this is possible, the timing of seed release in the months leading up to the relatively dry, frosty, winter suggests any seedlings from this source would have little time to establish. Perhaps these seeds join the soil seedbank, ready to germinate with spring or summer rains around nine to eighteen months post-fire. As noted above, Morgan (1998a) found interstitial gaps were still relatively common in the second post-fire year. Native seedling recruitment occurred two and three years post-fire in grasslands burnt periodically, though not in long-unburnt grassland, in western Victoria (Morgan 2001). Observation in CPW suggests seedling germination in this vegetation type may also be higher at this time- since-fire than in later post-fire years (Figure 7.15)
200
Figure 7.15. Bare ground with seedlings 19 months post-fire at Orchard Hills. Photo by Venesa Brusic.
7.4.4 Towards an understanding of ground layer recruitment
The hypothesis that rapid post-fire flowering results in seedling recruitment in the second post-fire growing season is consistent with some common characteristics of ground layer plants in grassy vegetation.
• Germination in many of these species is not fire cued (Clarke et al. 2000, Read et al. 2000, Grant and Macgregor 2001, Morgan 2001, Wood 2001, S. Clarke 2003, Hill and French 2003), despite their presence in fire-prone vegetation. If seeds are to germinate a year or so after fire, fire cuing is unnecessary. Bell et al. (1987) note that two West Australian Xanthorrhoea species with fire-stimulated flowering, X. gracilis and X. preissii, respond negatively to heat. • Many species do not appear to form persistent soil-stored seedbanks. In CPW, for example, Hill and French (2003) found many species in the standing vegetation at Holsworthy did not emerge from the seedbank. Many Victorian species commonly found in grassy standing vegetation also failed to appear in seedbank studies (Lunt 1997c, Morgan 1998b). This included species known to flower rapidly post-fire in CPW, such as Tricoryne elatior and Plantago gaudichaudii. If the aim is germination when gaps are available, and gaps are maximally available early in the post-fire sequence, then long-term storage would not be required. Bond (1997) reports that seeds of fire-flowering species in South Africa typically germinate readily. California coastal sage scrub plants that flower in the first year after fire also have readily germinable seeds. Post-fire seedlings are rare, but establish abundantly in the second growing season (Keeley and Keeley 1984). Germination and dormancy in species with fire-stimulated flowering would more usefully relate to temperature or moisture requirements. On the Cumberland Plain,
201 seedbanks of perennial species might be geared to survive ungerminated through the dry frosty winter months when conditions are unlikely to be conducive to survival and growth. Odgers (1999) found most of the native grass species she studied in grassy woodland near Brisbane had seasonal, short-lived seedbanks. Hagon (1976) in Canberra found germination in two local C4 grasses, Themeda australis and Bothriochloa macra, was delayed by low temperatures. Lodge and Walley (1981) found both temperature and after-ripening effects which would tend to limit winter germination in seeds of a number of C4 species collected near Tamworth, particularly Aristida ramosa, Bothriochloa macra and Sporobolus elongatus, all species which also occur in CPW.
Forb species differentiating burnt from unburnt areas were often resprouters with storage organs such as corms or tubers. These plants are probably long-lived (Benson and Howell 2002), and thus able to maintain population numbers through limited episodic recruitment. The best opportunities for this may arise quite infrequently, when fire and good rainfall coincide.
Fire and rainfall may interact to facilitate herb recruitment in grassy vegetation (Morgan 1998a, 2001):
• fire creates gaps (Morgan 1996, Odgers 1996, Morgan 1998a, this study); • fire (Lunt 1994, this study) and rain (pers. obs.2002-4) encourage flowering and fruiting, and thus seed production; • rain (Harrington and Johns 1990, Morgan 2001, Clarke and Davison 2004), and in some species fire (Read et al. 2000, Wood 2001, S. Clarke 2003, Hill and French 2003) stimulate germination; and • rain supports seedling survival and growth (Morgan 2001). Fire may have a role here too, through enhancing nutrient availability (Leigh et al. 1991), and reducing competition.
Gaps can also be created by agents other than fire, such as humans, animals, and perhaps drought. However as noted above, in productive woodlands like CPW, grass and eucalypt litter would limit the gap-creating potential of drought in later post-fire years.
202 It is likely that ground layer plants in grassy vegetation, including CPW, employ a range of life history strategies. Broad groups might include:
• The long-lived resprouting pyrogenic-flowering forbs discussed above. Many of these species appear to germinate readily, fail to develop a long-lived seedbank, and are not fire cued (eg Grant and Macgregor 2001). Population increase may be episodic, and tied to fire and rain. Survival between fires would be in adult form, at times as below-ground corms and tubers. • Short-lived herbs, which may include species such as Wahlenbergia gracilis, Poranthera microphylla, Cyperus gracilis, and native Senecio species. Seedling establishment would need to occur more frequently than it would in long-lived forbs. Faced with a fire, these species, which probably allocate less energy to below-ground storage, may well act as obligate seeders. They would therefore need some mechanism to establish from seed post-fire. Soil-stored, fire-cued seeds would be one option, wind- or animal-dispersal from unburnt sites would be another. Wahlenbergia gracilis and Poranthera microphylla are known to have soil-stored seed (Wood 2001), and to respond to fire-related cues (Warcup 1980, Wood 2001, Hill and French 2003). Grant and Macgregor (2001) noted that species responsive to smoke in their Tablelands study tended to be annuals, while Williams et al. (2003) observed a post-fire pulse in the abundance of ephemeral forbs in grassy woodland near Townville. Some members of this group may ‘bet hedge’, with some seeds able to germinate readily in the occasional gap between fires, and others with dormancy mechanisms able to respond to fire. Survival between fires could be as a mix of soil-stored seeds, and a moving population of adults in areas subject to disturbances other than fire. • Themeda australis may be in a class of its own, or may share characteristics with other dominant tussock-forming grasses, such as Poa sieberiana, Sorghum leiocladum and Cymbopogon refractus. Themeda regenerates slowly relative to many other ground layer species, taking somewhat longer to flower post-fire (this study), and thus creating gaps in which these other plants can grow and reproduce. Themeda also flowers less abundantly in later post-fire years than many other grass species (Morgan and Lunt 1999), but stands out amongst native grasses in possessing a persistent, rather than a transient, soil seedbank (Odgers 1999). Germination may be smoke-cued (Section 4.4.5). • Subdominant perennial grasses. These species may flower relatively abundantly all through the interfire interval, and thus have seeds ready to go when a fire occurs, without the need for long-term dormancy (Odgers 1996, 1999). A number do appear to have fire-cued seeds (Read et al. 2000, Wood 2001, S. Clarke 2003). Some may also ‘bet hedge’, establishing between fires when opportunities arise. • Species able to propagate vegetatively between fires, such as grasses with rhizomes, and stoloniferous herbs. These species may have little need for reproduction via flowers, fruits and seedlings. Clarke and Davison (2004) found species in this group were the least successful in a germination experiment on the New England Tablelands.
203 It would be instructive to survey both the seedbank, and seedlings in the field, pre-fire and in each season over the first few post-fire years. Further research could also include:
• monitoring of flowering and fruiting over the initial post-fire years (ie beyond the first six months); • comparison of seedling establishment after drought without fire to establishment after drought with fire; and • investigation of the longevity of forbs and grasses, and the extent of post-fire mortality in individual species and species groups.
7.5 Conclusion
Reproductive activity in ground layer species, particularly forbs, was greater at two CPW sites in the year immediately following a fire than some years after fire. This finding suggests that fire plays a role in regenerating the CPW understorey; in other words, some fire is likely to be desirable in this vegetation type. It does not, however, tell us how frequently fire might need to occur to provide sufficient opportunities for ground layer recruitment.
An understanding of the role of pyrogenic flowering in fire-prone grassy ecosystems is important. Without it, the lack of fire-cued germination amongst herbaceous species could be read as a sign that fire is unimportant in, or even detrimental to, reproduction in these taxa.
204 CHAPTER 8 FIRE FREQUENCY AND WOODLAND MICROHABITATS
8.1 Introduction 8.1.1 Microhabitats and herbs
As studies from Australia and elsewhere attest, trees and shrubs often provide a somewhat different microhabitat for ground layer species than open areas. Productivity, species richness and species composition may all be affected. However the influence of microhabitat on Cumberland Plain Woodland herbs has yet to be assessed.
Many studies have found that total herbaceous productivity is lower under trees and/or shrubs (Engle et al. 1987, Archer 1990, Scanlan and Burrows 1990), or that productivity increases when trees and shrubs are removed (Walker et al. 1986, Harrington and Johns 1990). Some studies, however, have documented increased productivity under trees (Weltzin and Coughenour 1990, Hobbs and Atkins 1991, Belsky et al. 1993). It appears the overstorey can have both stimulatory and competitive influences on the understorey, and that the net effect on herbage biomass may be positive or negative.
Differences in species richness and composition have also been documented, both in Australia and overseas (Pieper 1990, Hobbs and Atkins 1991, Facelli and Temby 2002). Several studies provide insight into microhabitat differences in temperate grassy woodlands in Australia’s south-east. Prober et al. (2002a) found patches under trees in Box woodlands on the Western slopes of NSW had a significantly higher richness of native understorey species than did gaps. Gap and canopy samples were clearly distinguished through ordination. Few species were exclusive to either habitat, but some were more abundant under trees (eg Dianella ssp, Austrodanthonia ssp), where the C3 grass Poa sieberiana dominated. The C4 species Themeda australis dominated in gaps. C3 grasses and herbs also dominated under eucalypt canopies at Gellibrand Hill in Victoria, while Themeda again dominated open areas (Robertson 1985). Similarly, shade-tolerant grasses such as Microlaena stipoides dominated under trees on the New England Tablelands, while Aristida ramosa – again, a C4 species – dominated
205 open spaces (Chilcott et al. 1997, Gibbs et al. 1999). Microlaena stipoides was more abundant in paddocks with >30% tree cover than in paddocks without trees across the Northern Tablelands of NSW (Magcale-Macandog and Whalley 1994).
Various explanations have been advanced to explain differences in productivity and species composition between microhabitats.
• Nutrient levels tend to be higher under trees or large bushes than in open areas (Robertson 1985, Tongway and Ludwig 1990, Hobbs and Atkins 1991, Chilcott et al.1997, Jackson and Ash 1998, Prober et al. 2002a,b). Proposed mechanisms for nutrient build-up include long-term accrual from litter (Gleadow and Ashton 1981), interception of nutrient-rich dust by the canopy, concentration of nutrients through tree root systems, and accumulation of dung of birds and animals that congregate in and under trees (Scholes and Archer 1997, Prober et al. 2002a). • Soil water levels may be higher or lower (Robertson 1985, Engle et al. 1987) under trees than in adjacent open areas. Many processes operate and net values will vary over time. Transpiration would tend to decrease water availability, while shading and the mulching effects of leaf litter would limit evaporation and thus act in the opposite direction. Hydraulic lift may increase water availability near trees (Richards and Caldwell 1987, Caldwell and Richards 1989, Dawson 1993), although the benefits may be variable (Ludwig et al. 2003), or marginal (Moreira et al. 2003, Ludwig et al. 2004). • Trees may suppress dominant grasses such as Themeda, thus allowing more gaps for other grasses and forbs. Warm season grasses may not respond well to the limited light available under trees (Prober et al. 2002a). Woody leaf litter may act as a physical barrier for large tufted grasses, favouring smaller grasses and herbs (Chilcott et al. 1997).
Higher levels of fertility under trees may increase vulnerability to weed invasion (Hobbs and Atkins 1991, Prober et al. 2002a).
Prober et al. (2002a) suggest that the influence of trees on grass biomass may reduce the need for burning, slashing or grazing in woodlands, as ‘under tree’ habitats may provide a haven for forbs which might otherwise be outcompeted by dominant grasses.
It has also been suggested that some palatable CPW forbs may only survive under Bursaria, where they will be protected from grazing pressure (D. Little, P. Cuneo, Mt Annan Botanic Gardens, pers. comm. 2002).
206 8.1.2 Fire frequency and herbs
The literature on this topic is reviewed in Section 1.7. While few Australian studies have assessed the effect of fire frequency on woodland herbs directly, studies in grasslands suggest frequent burning may be important to maintain the balance between matrix grasses and interstitial forbs (Lunt 1995a, Morgan 1998a). Annually burnt savanna woodland near Darwin supported more native herbaceous species than areas where fire had been excluded (Fensham 1990, Woinarski et al. 2004). However Lunt (1997a) found a higher species richness in unburnt woodlands than in frequently burnt grasslands, although some forb species were more abundant in frequently burnt areas.
The two case studies in Chapter 7 suggest that reproductive activity in the CPW ground layer is higher in the months after a fire than in later post-fire years. Forbs, in particular, flowered much more vigorously in recently burnt areas. Seedlings of both dicots and monocots were sparse in unburnt plots, but common where a burn had occurred. These findings suggest that periodic fire plays an important role in the maintenance of populations of these ground layer species. Extrapolating to fire cycles, might a reasonably high fire frequency be needed to allow sufficient replenishment of herb species populations?
The possibility that frequent fire may enhance opportunities for establishment of exotic species has already been noted (Section 4.1.4). However in the landscape study described in Chapter 4, woody weed abundance was significantly greater in low fire frequency sites, while sites subject to frequent burning were virtually free of exotic shrubs. The effect of fire frequency on herbaceous weeds is explored in this chapter.
A further finding from the landscape study was that Themeda australis dominated high and moderate fire frequency sites, but not those with a long recent interfire interval. The possibility that Themeda might play a role in excluding weed species (Morgan 1998d, Prober et al. 2002a, Prober et al. 2004) was noted in Section 4.4.5. This study provided an opportunity to explore the association between Themeda and herbaceous weeds in CPW.
Vigorous Themeda stands promoted by relatively frequent burning might also have negative consequences for smaller native herbs: research in Victoria’s basalt grasslands has indicated that Themeda clumps close by three or four years post-fire, leaving little
207 room for interstitial forbs (Section 1.7.1). The relationship between Themeda abundance and forbs in CPW is also addressed here.
Finally, both positive and negative effects on biodiversity have been associated with grazing (Tremont 1994, McIntyre et al. 1995, Fensham 1998, Proulx and Mazumder 1998, McIntyre and Martin 2001, Section 5.1.4). The data collected for this study provided an opportunity both to partially validate the grazing index presented in Section 5.2.4, and to make a rough assessment of the consequences of grazing on the CPW ground layer.
8.1.3 Study aims and questions
This study compared ground layer species richness and composition in three microhabitats: around trees, under Bursaria thickets, and in open areas. The comparison between Bursaria thickets and open areas was of particular interest, given the differences in the proportion of the landscape occupied by these two microhabitats in sites with different fire histories. If open areas support a different suite of ground layer species than Bursaria patches, fire frequency will influence ground layer species composition indirectly through its influence on the extent of open and shrubby patches.
Sampling was carried out at shale-based woodland sites spanning a range of fire frequencies, within two years of a fire.
The broad aims of the study were: • To determine whether different microhabitats support different herbaceous species; and • To investigate whether a direct effect of fire frequency on ground layer vegetation could be discerned.
Specific questions addressed in this chapter include:
• Does the species richness of native herbs vary with microhabitat? It was hypothesised, on the basis of the Prober et al. (2002a) findings, that patches around trees would support more native herb species than open patches. • Does the species richness of exotic herbs vary with microhabitat? • Does the species richness of native herbs vary with fire frequency? It was hypothesised, on the basis of the findings in Chapter 7 and those in the grassland
208 and savanna literature, that infrequently burnt sites would support fewer native herb species than frequently burnt areas. • Does the species richness of exotic herbs vary with fire frequency? The hypothesis that frequently burnt areas would support more exotic herb species than infrequently burnt areas was again put to the test. • Are groups of species within the native herb flora (forbs, grasses and graminoids, grasses with different photosynthetic pathways, grazing-intolerant species) differentially affected by either microhabitat or fire frequency? Of particular interest were: o The hypothesis, suggested by findings elsewhere in Australian temperate woodlands, that C3 grasses would be more abundant under trees, while C4 grasses would be favoured in open areas; o The hypothesis that grazing-intolerant herbs would be more abundant under Bursaria than in open areas, due to the protective effect of Bursaria. • Does overall floristic composition of the ground layer vary with fire frequency and/or microhabitat? It was hypothesised, on the basis of the literature, that the floristics of microhabitats would indeed vary. • Which species, if any, characterise each microhabitat? • Is Themeda australis associated with decreased richness or abundance of herbaceous weeds? • Is Themeda australis associated with a decreased richness or abundance of native forbs? • Does grazing have an association with the richness of native or exotic herbs?
8.2 Methods 8.2.1 Study sites
Ground layer vegetation was sampled in six sites, two in each of the three fire frequency categories described in Chapter 4.
Species richness and abundance in Cumberland Plain Woodland fluctuates with recency of burning and rainfall, with higher numbers reported 12-15 months post-fire, and after drought-breaking rain (unpub. data, C. Morris, UWS, 2004; unpub. data, D. Benson and J. Howell, Royal Botanic Gardens, 2004). The study was therefore limited to areas which had last burnt between 9 and 20 months prior to sampling. Field work was conducted between May and July 2003. Above-average rainfall had occurred in the early months of 2003 following a particularly dry winter and spring in 2002 (Section
209 2.3.2). Ground layer plants were thus likely to have been relatively abundant at the time of sampling. It is probable, however, that some cryptic species, particularly geophytes which die back during autumn and reappear to flower in spring (Morgan 1999), and spring-flowering annuals, were missed during the current survey, particularly in the earlier sites sampled. The initial intention was to resurvey all sites in spring, however this proved impossible in the time available.
Fire frequency categories for this study paralleled those used in the landscape study reported in Chapter 4 (high fire frequency, intervals mostly 1 – 3 years; moderate fire frequency, intervals mostly 4 – 10 years; low fire frequency, at least 20 years unburnt prior to the most recent fire). With the exception of the more stringent approach to time-since-fire, criteria for site selection were also the same (see Section 4.2.1). In fact, five of the six sites in this study were also used in the landscape study:
• Ropes Creek, a frequently burnt site; • Shanes Park, also a frequently burnt site; • Lansdowne, a moderate fire frequency site (the current study was restricted to those parts of the site burnt approximately 18 months prior to sampling, while the previous study covered a wider area); • Orchard Hills, a low fire frequency site; • Scheyville, a low fire frequency site.
The second moderate fire frequency site in the current study, Mt Annan 2, was located adjacent to the low fire frequency area sampled in the landscape study. The two adjacent areas had different fire histories. Mt Annan 2 was burnt for experimental and ecological purposes in 1991, and again in 2001, 19 months prior to sampling (D. Benson and J. Howell, Royal Botanic Gardens, pers. comm. 2001).
Table 8.1. Sites used in a study of ground layer vegetation in Cumberland Plain Woodland microhabitats. Table 4.1 gives additional detail.
Site name Fire frequency category Time since fire when sampled Ropes Creek high 9 months Shanes Park high 18-20 months Lansdowne moderate 18-19 months Mt Annan 2 moderate 19 months Orchard Hills low 18-19 months Scheyville low 18-19 months
210 The six sites are listed, from most to least frequently burnt, in Table 8.1: this order will be maintained throughout the chapter. Figure 8.1 shows site locations.
Figure 8.1. Map of the Cumberland Plain showing remnant native vegetation (green) and study sites in three fire frequency categories. ▲, high fire frequency; ■, moderate fire frequency; ▼, low fire frequency.
8.2.2 Sampling and data collection
Species sampled
The study was limited to ground layer species: herbs, grasses, graminoids, ferns, scramblers and subshrubs were included, shrub species of larger habit were not. For ease of communication, I will refer to these ground layer species collectively as “herbs.”
Most taxa were identified to species level. In some cases, related taxa were grouped. In other cases, recorded taxa may have included other closely-related taxa. In particular, native Glycine species were categorised as Glycine tabacina, native Sporobolus as Sporobolus creber, native Oxalis were grouped (Oxalis spp.), native Arthropodium species were recorded as Arthropodium milleflorum, exotic Solanum species were categorised as Solanum nigrum, and Austrodanthonia species were grouped. This was because:
211 • Some taxa could not be reliably distinguished in the field, either because only vegetative material was present, or because reproductive material was either just developing, or was decomposing; • Herbarium staff familiar with Cumberland Plain plant species advised that taxonomic distinctions in existing keys were not definitive (D. Benson and J. Howell pers. comm. 2003); • It was not considered desirable or practical to remove large numbers of specimens for taxonomic analysis, given the threatened status of the ecosystem.
Microhabitat definitions
Microhabitats were defined as follows:
• Around trees. Around the base of E. moluccana trees with a basal area, based on DBH measurements, of ≥ 1000 sq cm. This equates to a diameter of > 36 cm, for a tree with a single trunk. As some Bursaria was unavoidable around trees in infrequently burnt sites, absence of Bursaria was not a criterion for ‘around tree’ plots. • Under Bursaria. Under patches of Bursaria with shoots ≥ 1.5 m in height, sufficiently dense to severely impede access. No tree saplings of over 2 m high in the plot, and no trees over 10 cm DBH either in the plot or within 2 m of the plot boundary. • In open patches. Areas containing no Bursaria or other shrubs over 50 cm high, no tree saplings over 2 m high, and no trees over 10 cm DBH. No trees over 10 cm DBH within 2 m of the plot boundary. While ideally open patches would not have included even small Bursaria bushes, this was not realistic in infrequently burnt areas. Microhabitat types are illustrated in Figures 8.2 to 8.4.
Figure 8.2. ‘Around tree’ plots were located at the base of large E. moluccana trees like this.
212
Figure 8.3. ‘Under Bursaria’ plots were located in thickets like this.
Figure 8.4. An area similar to those in which ‘open patch’ plots were located.
213 Plots
Plots were 3.6 m in diameter, and consisted of a series of circular, concentric, ‘nested’ subplots (Figure 8.5). The inner circle, with a radius of 30 cm, was not sampled, as in ‘around tree’ microhabitats this area was occupied by the tree trunk. Subplot characteristics are outlined in the first three columns of Table 8.2. The total area sampled per plot was 9.89 m2. Occasionally, trees were over 60 cm in diameter at the base. In these cases a modified sampling frame which adjusted the distance of subplots from the centre, while retaining subplot areas, was used.
Table 8.2. Characteristics of concentric nested subplots used in a study of ground layer vegetation in Cumberland Plain Woodland microhabitats.
Subplot Distance from Subplot Importance Contribution number centre of plot (cm) area (m2) score to frequency score 1 30 – 40 0.22 5 1 2 40 – 60 0.63 4 1 3 60 – 90 1.41 3 1 4 90 – 130 2.77 2 1 5 130 - 180 4.87 1 1
Figure 8.5. Recording herb species in nested subplots within a 3.6 m diameter plot at Orchard Hills, a low fire frequency site.
214 Measures
Herb species presence was recorded for each subplot. Shoot cover, rather than root location was used to indicate presence, eliminating the need to disturb scrambling herbs.
This method of sampling species abundance is similar to that recommended by Outhred (1984) and Morrison et al. (1995b), who used nested square subplots. These authors outlined two methods for scoring species found in subplots:
• “Importance-score” allocates the highest score to species occurring in the smallest subplot, with successively lower scores for those first encountered in subsequent, larger, subplots. • “Frequency-score” is the number of subplots in which the species occurs. Simulations and field comparisons found frequency-score was more able to detect subtle community patterns than importance-score. However both methods provided a good approximation to density. Each was less affected by choice of subplot size and distribution of organisms (eg clumping) than traditional methods of assessing frequency (Outhred 1984, Morrison et al.1995b).
Both an importance score and a frequency score were calculated for each species in each plot (Table 8.2, columns 4 and 5). As noted above, frequency score may be better able to detect subtle differences. On the other hand, importance scores emphasise the core of each microhabitat, eg the area nearest a tree, or in the middle of an open patch, where the influence of a microhabitat type may be greatest. Belsky et al. (1993) found the greatest effects of tree canopies were near tree trunks.
Locating plots
In order to ensure sampling effort covered the range of variability present, each site was divided, on paper, into three approximately equal blocks. Six random points, two for each microhabitat type, were marked on the map of each block. Plots were located as near as possible to their designated point, given the sampling criteria for the relevant microhabitat type. Eighteen plots, six of each microhabitat type, were sampled in each site, giving 108 plots in all.
215 8.2.3 Data analysis
The relationship between vegetation parameters and the study factors was assessed through a mixed model analysis of variance, with fire frequency and microhabitat as fixed, orthogonal factors. Site was a random factor nested in fire frequency. Plots were replicates. Cochran’s test was used to assess homogeneity of variances, and transformations applied where necessary. Student-Newman-Kuel tests were used to evaluate differences between fire frequency and microhabitat categories, where F ratios were significant.
Vegetation parameters assessed in this way included per plot native and exotic species richness. ANOVAs of the above form were also used to evaluate the relationship between the study factors and groupings of native species:
• forbs (including lily and iris-like monocots) • grasses (family Poaceae) and graminoids (families Cyperaceae, Juncaceae and Lomandraceae) • grasses employing the C3 photosynthetic pathway (sometimes called ‘cool season’ grasses) • grasses employing the C4 photosynthetic pathway (sometimes called ‘warm season’ grasses).
In each of these groups, both species richness and an index of relative importance within the ground layer, were assessed. The index was similar to that used by Hunter (2003), and took the same form for each group. For example:
sum of frequency scores for all native forb species native forb index = x 100 sum of frequency scores for all native species
Thus if all native species in a plot were forbs, the index was 100. Ratings reflected both species richness and abundance relative to the full complement of native ground layer species, and were standardised for species richness.
Grass photosynthetic pathways were determined from the literature (Watson and Dallwitz 1992 onwards, Williams 1996).
A measure of grazing impact was also developed, again using a ratio formula. Included in this measure were species identified as grazing-intolerant by Prober and Thiele
216 (1995) and/or by P.J. Clarke (2003). These studies, conducted in White Box woodlands and on the New England Tablelands respectively, were used in lieu of local information, as this was not available. The impact of domestic grazing on CPW ground layer species has yet to be assessed, while work on low levels of grazing by macropods at Holsworthy uncovered few differences (S. Clarke 2003). Species used in the grazing- intolerant herb index are listed in Appendix 10. Lower values on this index indicate that exposure to grazing is likely to have been relatively high.
The spatial analysis program PRIMER was employed to further explore floristic patterns in the herb data from the 108 plots. Two similarity matrices were constructed using the Bray-Curtis metric, one using frequency scores, the other using importance scores. Data were visually presented through multi-dimensional scaling ordination. The process was repeated using combined data from pairs of plots (see Section 8.2.2), giving 54 points rather than 108. These points were clustered through group averaging.
Analysis of similarities was employed to further assess effects of fire frequency, site, microhabitat and vegetation type on species composition. The ANOSIM option in PRIMER can work with two variables at a time; these can be either nested, or crossed. The effect of fire frequency in relation to site was examined first, through a nested ANOSIM. The role of microhabitat was then explored through a two-way crossed ANOSIM with site. One-way ANOSIM tests for differences between microhabitats were also carried out for each individual site, using importance scores.
Species contributions to dissimilarity between microhabitats were calculated through PRIMER’s SIMPER option. Plot-level importance-score data were used in these analyses.
Section 5.2.2 gives a more detailed explanation of multivariate pattern analysis using PRIMER.
To further illuminate the question of which taxa were differentially associated with the various microhabitat types, univariate permutation tests using the program Resampling Stats (Blank et al. 2001) were applied. For each species found in five or six sites, a 6 x 3 matrix of the 18 site x microhabitat importance-score means was constructed, and the average difference between microhabitat pairs calculated. The figures in the matrix were then permuted 5000 times, each time the mean difference between the relevant
217 columns was calculated. The significance level for the actual difference was assessed by finding its place in the distribution developed over the multiple permutations.
A similar procedure was employed using importance scores in individual sites, wherever a taxon was found in more than five of the 18 plots.
Finally, the relationship between Themeda australis and two groups of ground layer species – exotics, and native forbs – was examined through computing, at plot level, the Pearson product-moment correlation coefficient between the Themeda frequency score and:
• the species richness of the group,
• the sum of the frequency scores of species in the group.
All plots were initially included in these calculations (n = 108), then correlations were recalculated with the omission of Scheyville (revised n = 90): Scheyville was omitted because virtually no Themeda was found in this site.
The relationship between the grazing-intolerant herb index and mean per plot species richness (native and exotic) was also assessed through Pearson product-moment correlations, this time at site level (n = 6). A Spearman rank correlation coefficient, corrected for ties, was used to determine whether the grazing index presented in Chapter 5, and the grazing-intolerant herb index, were related in the six sites surveyed for both the landscape and the microhabitat studies.
8.3 Results 8.3.1 Native and exotic species richness
One hundred and thirty-seven taxa were recorded at least once in the 108 plots. Of these, 100 were indigenous to the area, while 37 were exotics.
Native species richness
Mean native species richness (per plot, 9.89 m2) was remarkably similar across all fire frequency categories, and all microhabitats, ranging from 16.0 to 17.5 (Table 8.3). Neither fire frequency nor microhabitat significantly affected native species richness
218 (Table 8.4). Native species richness did, however, differ between sites. There was no significant interaction between microhabitat and either fire frequency, or site.
Table 8.3. Mean (S.E.) species richness of native herbs, per plot (9.89 m2), for three microhabitats x six sites, two sites in each of three fire frequency categories.
Fire frequency high moderate low Micro- habitat Site Ropes Shanes Lans- Mt Annan Orchard Schey- means Creek Park downe 2 Hills ville Around trees 12.5 (1.5) 19.0 (2.1) 14.7 (1.8) 20.2 (2.3) 19.2 (2.3) 13.5 (1.3) 16.5 (1.4) Under Bursaria 15.8 (1.7) 16.8 (1.8) 11.7 (1.1) 21.8 (2.1) 21.3 (1.9) 15.5 (2.5) 17.2 (1.6) In open patch 13.2 (0.5) 18.8 (0.9) 15.0 (1.5) 20.2 (2.2) 20.0 (2.0) 15.5 (2.7) 17.1 (1.2) Site means 13.8 (0.6) 18.2 (1.8) 13.8 (1.1) 20.2 (0.9) 20.7 (2.0) 14.8 (1.5) Fire frequency 16.0 (2.2) 17.3 (3.5) 17.5 (2.7) means
Table 8.4. ANOVA of species richness of native herbs, per plot (9.89 m2), for sites in three fire frequency categories and three microhabitats.
Source Sums of d.f. Mean F-test F P squares square against Fire frequency 44.7 2 22.3 Site 0.078 0.927 Site (Fire frequency) 863.4 3 287.8 Error 13.700 < 0.001 Microhabitat 9.9 2 4.9 Site x mic 0.288 0.760 Fire freq x microhabitat 24.0 4 6.0 Site x mic 0.351 0.835 Site x microhabitat 102.8 6 17.1 Error 0.815 0.561 Error 1890.7 90 21.0 Total 2935.4 107
Cochran’s test C6,18 = 0.117, P = NS
Exotic species richness
Mean exotic species richness varied with both fire frequency and microhabitat (Table 8.5). High fire frequency sites averaged 1.3 exotic species per plot, while moderate and low fire frequency sites averaged 4.6 and 4.1 respectively. The two moderate frequency sites differed considerably on this variable. Microhabitat means ranged from 2.9 species per plot under Bursaria, to 3.3 in open patches and 3.9 around trees.
219 Table 8.5. Mean (S.E.) species richness of exotic herbs, per plot (9.9 m2), for three microhabitats x six sites, two sites in each of three fire frequency categories.
Fire frequency high moderate low Micro- habitat Site Ropes Shanes Lans- Mt Annan Orchard Schey- means Creek Park downe 2 Hills ville Around trees 2.2 (0.7) 0.8 (0.2) 1.8 (0.6) 8.0 (0.9) 5.5 (0.9) 5.2 (0.4) 3.9 (1.1) Under Bursaria 1.8 (0.4) 1.2 (0.7) 3.3 (1.0) 5.7 (1.7) 2.3 (0.4) 3.2 (0.9) 2.9 (0.6) In open patch 1.0 (0.5) 1.0 (0.3) 3.2 (0.3) 5.8 (0.4) 3.2 (0.4) 5.3 (0.3) 3.3 (0.8) Site means 1.7 (0.4) 1.0 (0.3) 2.8 (0.3) 6.5 (0.5) 3.7 (0.3) 4.6 (0.2) Fire frequency 1.3 (0.3) 4.6 (1.9) 4.1 (0.4) means
As Cochran’s test was marginally significant when raw species richness figures were used, data were log transformed (ln(x+1)). Species richness did not differ significantly between fire frequency categories, or microhabitats (Table 8.6). Variance was high due to the large difference in exotic species richness between the two moderate fire frequency sites. When these sites were omitted from the analysis, the difference in exotic species richness between high and low fire frequency sites was statistically significant (Table 8.7). Interaction terms were not significant in either analysis.
Table 8.6. ANOVA of species richness per plot (9.89 m2) of exotic herbs (ln(x+1) transformed) for sites in three microhabitats and three fire frequency categories.
Source Sums of d.f. Mean F-test F P squares square against Fire frequency 17.25 2 8.62 Site 4.370 0.129 Site (Fire frequency) 5.92 3 1.97 Error 10.606 <0.001 Microhabitat 0.45 2 0.22 Site x mic 0.680 0.542 Fire freq x microhabitat 1.57 4 0.39 Site x mic 1.191 0.403 Site x microhabitat 1.97 6 0.33 Error 1.767 0.115 Error 16.75 90 0.19 Total 43.90 107
Cochran’s test C6,18 = 0.131, P = NS
220 Table 8.7. ANOVA of species richness per plot (9.89 m2) of exotic herbs (ln(x+1) transformed) for sites in three microhabitats and two fire frequency categories, high and low.
Source Sums of d.f. Mean F-test F P squares square against Fire frequency 12.44 1 12.44 Site 23.377 0.040 Site (Fire frequency) 1.06 2 0.53 Error 2.893 0.063 Microhabitat 0.77 2 0.38 Site x mic 2.090 0.239 Fire freq x microhabitat 1.22 2 0.61 Site x mic 3.316 0.142 Site x microhabitat 0.74 4 0.18 Error 1.001 0.414 Error 11.03 60 0.18 Total 27.26 71
Cochran’s test C4,12 = 0.199, P = NS.
8.3.2 Native species groupings
Native forbs
Although forb species richness was slightly higher in open patches (10.3 species per plot) than under Bursaria (9.7 species) or around trees (9.3), this difference was not significant (Table 8.8). This variable varied with site, but was unaffected by fire frequency.
Table 8.8. ANOVA of species richness per plot (9.89 m2) of native forbs for sites in three microhabitats and two fire frequency categories, high and low.
Source Sums of d.f. Mean F-test F P squares square against Fire frequency 70.9 2 35.5 Site 0.362 0.723 Site (Fire frequency) 294.2 3 98.1 Error 10.918 < 0.001 Microhabitat 15.4 2 7.7 Site x mic 1.071 0.400 Fire freq x microhabitat 28.0 4 7.0 Site x mic 0.976 0.485 Site x microhabitat 43.0 6 7.2 Error 0.798 0.574 Error 808.3 90 9.0 Total 1259.7
Cochran’s test C6,18 = 0.117, P = NS.
221 The native forb index, however, increased significantly as fire frequency decreased (Figure 8.6, Table 8.9, high fire frequency < moderate = low). This index ranged from 48.4 at Ropes Creek to 64.3 at Scheyville. Microhabitat types did not differ significantly on the native forb index, averaging 57.6, 57.1 and 54.6 around trees, in open patches, and under Bursaria respectively. Site, too, was not significant, reflecting the standardisation for native species richness inherent in this measure.
100.0
90.0
80.0
70.0
60.0
50.0
40.0
Value on index on Value 30.0
20.0
10.0
0.0 Ropes Ck Shanes Pk Lansdowne Mt Annan 2 Orchard Hills Scheyville (high) (high) (moderate) (moderate) (low) (low) Site (fire frequency)
Figure 8.6. Ratings on native forb (purple) and native grass and graminoid (green) indices (mean per plot + S.E.) for six sites, two in each of three fire frequency categories.
Table 8.9. ANOVA of native forb index per plot (9.89 m2) for sites in three microhabitats and two fire frequency categories, high and low.
Source Sums of d.f. Mean F-test F P squares square against Fire frequency 3106.5 2 1553.3 Site 14.173 0.030 Site (Fire frequency) 328.8 3 109.6 Error 1.206 0.312 Microhabitat 187.8 2 93.9 Site x mic 0.564 0.597 Fire freq x microhabitat 526.1 4 131.5 Site x mic 0.789 0.572 Site x microhabitat 999.7 6 166.6 Error 1.833 0.101 Error 8180.6 90 90.9 Total 13329.6
Cochran’s test C6,18 = 0.130, P = NS.
222 Native grasses and graminoids
Grass and graminoid species richness was similar in all three microhabitats, averaging 6.8 species per plot around trees and under Bursaria, and 6.2 in open patches. This difference was not significant (Table 8.10). Differences between fire frequency categories were also minor and insignificant, ranging from a mean of 6.3 in low fire frequency sites to 6.9 where fire occurred often.
Table 8.10. ANOVA of species richness per plot (9.89 m2) of native grasses and graminoids for sites in three microhabitats and two fire frequency categories, high and low.
Source Sums of d.f. Mean F-test F P squares square against Fire frequency 5.1 2 2.5 Site 0.040 0.962 Site (Fire frequency) 191.3 3 63.8 Error 12.956 < 0.001 Microhabitat 8.2 2 4.1 Site x mic 0.850 0.473 Fire freq x microhabitat 10.1 4 2.5 Site x mic 0.526 0.722 Site x microhabitat 28.8 6 4.8 Error 0.977 0.446 Error 442.8 90 4.9 Total 686.3
Cochran’s test C6,18 = 0.125, P = NS.
Again, however, significant differences between fire frequency categories were apparent on the native grass and graminoid index (Figure 8.6, Table 8.11, high fire frequency > moderate = low). This index showed the opposite trend to that for native forbs, moving from a high of 50.4 at Ropes Creek to a low of 32.8 at Scheyville. While patches under Bursaria scored slightly higher on this measure than patches around trees or in open areas, this difference was not significant.
223
Table 8.11. ANOVA of native grass and graminoid index per plot (9.89 m2) for sites in three microhabitats and two fire frequency categories, high and low.
Source Sums of d.f. Mean F-test F P squares square against Fire frequency 3140.1 2 1570.0 Site 12.178 0.036 Site (Fire frequency) 386.8 3 128.9 Error 1.288 0.284 Microhabitat 116.5 2 58.2 Site x mic 0.384 0.697 Fire freq x microhabitat 797.8 4 199.5 Site x mic 1.316 0.363 Site x microhabitat 909.5 6 151.6 Error 1.514 0.183 Error 9011.0 90 100.1 Total 14361.6
Cochran’s test C6,18 = 0.178, P = NS.
One trend for grass photosynthetic pathway reached significance: the C3 grass index was higher under Bursaria (11.9) than around trees (8.6) or in open areas (5.6; F2,6 = 6.395, P = 0.033; Bursaria > open). Conversely, the C4 grass index was highest in open patches (23.5). However patches under Bursaria and around trees, which rated 21.2 and
18.9 respectively, did not differ significantly from open plots on this variable (F2,6 = 1.422, P = 0.312).
Grazing-intolerant species
Twenty-one native species identified either by Prober and Thiele (1995) or P.J. Clarke (2003) as sensitive to grazing occurred in study plots, 21% of all native species. The grazing-intolerant herb index varied from 27.0 at Mt Annan 2, to 49.5 at Lansdowne (Figure 8.7), and site was a very significant factor on this variable (Table 8.12). Fire frequency categories, however, did not differ significantly. Microhabitat, too, made no difference: microhabitat means were 40.5 for open areas, 38.3 under Bursaria and 37.0 around trees.
224 100.0
90.0
80.0
70.0
60.0
50.0
40.0
30.0
20.0
10.0 Value on the grazing impact index impact grazing the on Value
0.0 Ropes Ck Shanes Pk Lansdowne Mt Annan 2 Orchard Hills Scheyville (high) (high) (moderate) (moderate) (low) (low) Site (fire frequency)
Figure 8.7. Ratings on grazing-intolerant herb index (mean + S.E.) for six sites, two in each of three fire frequency categories.
Table 8.12. ANOVA of grazing-intolerant herb index per plot (9.89 m2) for sites in three microhabitats and two fire frequency categories, high and low.
Source Sums of d.f. Mean F-test F P squares square against Fire frequency 791.4 2 395.7 Site 0.247 0.795 Site (Fire frequency) 4799.4 3 1599.8 Error 15.831 < 0.001 Microhabitat 221.3 2 110.6 Site x mic 1.043 0.408 Fire freq x microhabitat 432.2 4 108.0 Site x mic 1.019 0.468 Site x microhabitat 636.4 6 106.1 Error 1.050 0.399 Error 9095.0 90 101.1 Total 15975.7
Cochran’s test C6,18 = 0.127, P = NS.
8.3.3 Species composition
Ordination
Ordination of all 108 plots produced unacceptable stress levels when presented in two dimensions (stress = 0.23 for frequency score data, 0.24 for importance scores), although three-dimensional stress levels were lower (0.16 for frequency score data, 0.17 for importance scores). When the data from pairs of sites were combined into 54 samples, stress on the two-dimensional representation reduced to 0.20 when frequency scores were used, and to 0.19 for importance scores.
225 When data points were tagged by fire frequency and site, the patterns in these two ordinations were similar though not identical (Figures 8.8 and 8.9). Samples from particular sites tended to cluster, generally with some overlap with other sites, not necessarily in the same fire frequency category.
2
Figure 8.8. Two-dimensional ordination (MDS) of 54 samples, nine in each of six sites, based on frequency scores for herb species, showing site and fire frequency. Red, high fire frequency; green, moderate fire frequency; white, low fire frequency.
Figure 8.9. Two-dimensional ordination (MDS) of 54 samples, nine in each of six sites, based on importance scores for herb species, showing site and fire frequency. Red, high fire frequency; green, moderate fire frequency; white, low fire frequency.
226 As was to be expected given the site-based clumping, samples did not separate cleanly into microhabitat groups when tagged for this variable. There was, however, a discernable tendency for points to the left of clusters to represent open areas, while points to the right predominantly denoted patches around trees (Figures 8.10 and 8.11).
Figure 8.10. Two-dimensional ordination (MDS) of 54 samples, nine in each of six sites, based on frequency scores for herb species, showing microhabitat type.
Figure 8.11. Two-dimensional ordination (MDS) of 54 samples, nine in each of six sites, based on importance scores for herb species, showing microhabitat type.
227 Cluster analysis
The 54 samples were classified into two dendrograms, one based on frequency, and one on importance scores. Both were cut at approximately 45% similarity, giving seven (importance score) or eight (frequency score) groups. The two dendrograms showed many similarities:
• Each moderate fire frequency site formed a group of its own. • Eight of the nine samples from each low fire frequency site also grouped together. • The remaining low fire frequency samples, together with most of the samples from the high fire frequency sites, formed one large group.
The remaining plots were all from high fire frequency sites. In the frequency score dendrogram, the five residual points were classified into three groups (of three, one, and one) while the importance score dendrogram was completed by two one-sample groups. The frequency score dendrogram is presented in Figure 8.12.
Within some of the groups composed of samples from a single site, notably Lansdowne, Mt Annan 2 and Orchard Hills, samples from the same microhabitat type tended to group together.
228
Figure 8.12. Dendrogram showing classification of 54 samples, nine in each of six sites, based on frequency scores for herb species. H, high fire frequency; M, moderate fire frequency; L, low fire frequency. Initial letters of plot labels indicate site: Lans, Lansdowne; MtA, Mt Annan 2; OH, Orchard Hills; RC, Ropes Creek; Sch, Scheyville; SP, Shanes Park. Final letter of plot labels indicates microhabitat type: T, plots around trees; B, plots under Bursaria bushes; O, plots in open areas.
229 Analysis of similarities
While the effect of site on herb floristics was highly significant, fire frequency per se was not (Table 8.13).
Table 8.13. Analysis of similarities (two-way nested ANOSIM) amongst three fire frequency categories and six sites, two in each fire frequency category; site nested in fire frequency.
Factor Measure Global R P Fire frequency (using sites as samples) frequency score -0.278 0.867 Site frequency score 0.500 0.001 Fire frequency (using sites as samples) importance score -0.167 0.733 Site importance score 0.482 0.001
However herb floristics did differ significantly between microhabitats (Table 8.14). Pairwise comparisons between microhabitat types were all highly significant. Global R values were highest when open areas were compared with patches around trees, and lowest when patches around trees were compared with patches under Bursaria. All site pairs also differed significantly at 0.001 level, with the exception of the two high fire frequency sites, Shanes Park and Ropes Creek, although the difference here was still significant at P < 0.05.
Table 8.14. Analysis of similarities (two way crossed ANOSIM) amongst three microhabitats and six sites; site x microhabitat.
Factor Measure Global R P Site frequency score 0.649 0.001 Microhabitat (overall analysis) frequency score 0.246 0.001 Open area vs. around tree frequency score 0.350 0.001 Open area vs. under Bursaria frequency score 0.223 0.001 Under Bursaria vs. around tree frequency score 0.174 0.001 Site importance score 0.635 0.001 Microhabitat (overall analysis) importance score 0.252 0.001 Open area vs. around tree importance score 0.387 0.001 Open area vs. under Bursaria importance score 0.193 0.001 Under Bursaria vs. around tree importance score 0.185 0.001
230 Floristics differed significantly between microhabitats in four of the six sites. Differences were most pronounced in the two moderate fire frequency sites, and at Orchard Hills, a low fire frequency site. The most frequently burnt site, Ropes Creek, also returned a significant result (Table 8.15).
Table 8.15. Analysis of similarities (ANOSIM) amongst microhabitats in six individual sites, two in each of three fire frequency categories. Analysis based on importance scores.
Site Fire frequency Global R P Ropes Creek high 0.187 0.025 Shanes Park high 0.046 0.246 Lansdowne moderate 0.423 0.001 Mt Annan 2 moderate 0.336 0.002 Orchard Hills high 0.480 0.002 Scheyville high 0.038 0.264
8.3.4 Taxa characterising microhabitats
SIMPER analysis indicated that the dissimilarity between microhabitat types was the result of small contributions from many taxa. Those contributing over 2% are noted in Table 8.16.
Thirty-nine species were found in either five or six sites. Of these, nine species differed significantly when site means for open patches were compared with those for areas around trees. Six of these species were more abundant in open patches: Cheilanthes sieberi (P < 0.001), Zornia dyctiocarpa (P = 0.002), Aristida vagans (P = 0.007), Stackhousia viminea (P = 0.011), Tricoryne elatior (P = 0.017) and Hypericum gramineum (P = 0.023). Three – Glycine tabacina (P = 0.006), Dichondra repens (P = 0.012) and the exotic Hypochaeris radicata (P = 0.016) – were more abundant around trees. Two species showed a significance difference when patches around trees and under Bursaria were compared: Cheilanthes sieberi was more abundant under Bursaria (P = 0.025), while Hypochaeris radicata scored more highly under trees (P = 0.003). Zornia dyctiocarpa was significantly more abundant in open patches than under Bursaria (P < 0.001). Table 8.16 lists mean importance scores in each microhabitat for all taxa found in 5 or 6 sites.
231 Table 8.16. Mean importance score in each of three microhabitats, for all taxa recorded in either 5 or 6 sites. Exotic taxa are starred. Column 5 highlights taxa contributing >2% to the dissimilarity between two microhabitats (SIMPER analysis); the microhabitat in which the mean importance score was higher, is listed. Column 6 gives results of permutation tests on site x microhabitat means. T, around trees; B, under Bursaria; O, in open patches; *, P < 0.05; **, P < 0.01; ***, P < 0.001; NS, not significant. Taxa with a statistically significant habitat preference are bolded.
Taxon Mean importance score: Microhabitat Statistical preference differences around under in open from Simper between trees Bursaria patches analysis microhabitats Aristida ramosa 1.11 1.47 1.47 Burs NS Aristida vagans 0.50 1.56 2.14 Open O>T ** Arthropodium milleflorum. 2.00 1.64 1.08 Tree, Burs NS Asperula conferta 0.25 0.22 0.36 NS Austrodanthonia spp. 1.22 1.14 0.64 Tree NS Brunoniella australis 3.94 3.47 3.22 Tree, Burs NS Carex inversa 0.72 0.53 0.14 NS Cheilanthes sieberi 0.19 1.78 2.61 Open, Burs O>T *** O>B * Cymbopogon refractus 0.25 0.39 0.69 NS Cyperus gracilis 0.86 0.64 0.44 NS Desmodium varians 1.50 2.14 1.67 Open, Burs NS Dianella longifolia 1.42 0.94 1.36 Tree NS Dichondra repens 3.44 2.36 1.17 Tree, Burs T>O * Eragrostis leptostachya 0.69 0.44 0.78 NS Eremophila debilis 0.56 0.11 0.36 NS Fimbristylis dichotoma 0.61 0.69 1.14 NS Glycine tabacina 4.22 3.78 2.94 Tree, Burs T>O ** Goodenia hederacea 0.19 0.50 0.72 NS Hardenbergia violacea 0.22 0.67 0.36 NS Hypericum gramineum 0.03 0.33 0.67 O>T * Hypoxis hygrometrica 0.36 0.19 0.50 NS Hypochaeris radicata* 0.83 0.25 0.33 T>B ** T>O * Lomandra filiformis 2.86 3.33 3.56 Open, Burs NS Lomandra multiflora 1.03 0.64 1.19 NS Microleana stipoides 2.72 3.64 2.58 Tree, Burs NS Opercularia diphylla 1.17 1.39 1.44 NS Oxalis sp. 2.14 1.39 1.36 Tree, Burs NS Paspilidium distans 1.78 2.33 1.61 Tree, Burs NS Phyllanthus virgatus 1.31 0.69 0.67 NS Plantago gaudichaudii 0.53 0.39 0.53 Tree NS Senecio madagascariensis* 1.69 1.47 1.81 Open, Tree NS Setaria gracilis* 1.64 0.44 0.50 NS Solanum nigrum* 0.94 0.56 0.31 NS Sonchus oleraceus* 1.42 1.14 1.06 Tree NS Sporobolus creber 1.33 1.11 1.06 Tree NS Stackhousia viminea 0.56 0.92 1.72 Open O>T * Themeda australis 2.08 3.00 3.58 Open, Burs NS Tricoryne elatior 0.56 1.22 2.17 Open O>T * Zornia dyctiocarpa 0.08 0.06 1.14 O>T *** O>B **
232 In addition, several relatively uncommon species showed a significant preference for a particular microhabitat in sites where they were sufficiently abundant to allow statistical testing. These species and their preferences are listed in Table 8.17. Species with significant findings in a minority of tested sites are not listed.
Figure 8.13 provides a visual representation of the differential abundance of selected native species in the three microhabitats.
Table 8.17. Less common herb taxa showing a significant difference in abundance between microhabitats. T, around trees; B, under Bursaria; O, in open patches.
Taxon Number Number Mean importance score: Microhabitat of sites of sites preference with tested around under in open trees Bursaria patches taxon Bidens pilosa* 2 1 0.36 0.14 0.14 T>B * Glossogyne tannensis 2 1 0.17 0.08 0.42 O>B Goodenia hederacea 5 2 0.19 0.50 0.72 O>T=B (1 site) Hypoxis hygrometrica 6 1 0.36 0.19 0.50 O>T=B Plantago lanceolata* 3 1 0.64 0.25 0.14 T>O Scleria machaviensis 2 2 0.94 0.25 0.06 T>O (2 sites), T>B (1 site) Setaria gracilis* 5 3 1.64 0.44 0.50 T>B=O (2 sites) Solanum prinophyllum 3 3 1.58 0.50 0.50 T>B=O (2 sites) Wurmbea dioica 3 1 0.08 0.64 0.64 O>T
233
100%
80%
60%
40%
20%
0% Percent summed importance scores
r i is s p. a s s s is ia is a a o is a p ilis tus ns lla l l er s llum s cin an ide y osa m o ra lor tric um ac t reber t f e acea gans iens epen hy r ena c arian for r a elati sieb ine v r g r dis tipo ustralisv ram i ngif ulti hotoma v iocarp a onia s illeflorum taba s a diph o aus m om a ne r s virgae per a d y tannenam cha h u a ida la l a dic gr ti thes gr a nt m in bolus t l r is y prinop m c o is rmbea hdioica ris yne m u ly alis r u u ichond m oda Cyperui palidiumolaen oniella A iane styl is A Tricor og rnia dyct tr G Ox r smodium D i W ox heilan s o leria m D hyllanth ic Spor un e Themeda odenia hede eric Z c us r Operculari Lomandra fil p o C os opod P Pas M B D Lomandmbr G Stackhousia vimineal p S Solanu A hr Fi Hy G rt Hy A Taxon
Figure 8.13. Relative abundance of selected native herb taxa in three microhabitats based on mean importance score, expressed as a percentage. Selected taxa are those found in all sites, plus species found in less than six sites which exhibited a significant different in abundance between microhabitats in one or more sites. Cream, open areas; pink, under Bursaria; dark green, around trees. Taxa to the left of the chart were relatively abundant around trees, while those to the right were more often found in open areas.
8.3.5 Correlations with Themeda and grazing
The frequency score for Themeda australis was negatively correlated with all four variables assessed, however while correlations with native forb variables were not significant, those with exotic species were, at < 0.001 level (Table 8.18; r0.001 (1), 108 =
0.301; r0.001 (1), 90 = 0.338, Zar 1999).
234 Table 8.18. Pearson product-moment correlations between abundance of Themeda australis (frequency score) and selected measures for herbaceous weeds and native forbs, in plots.
Correlation between Scheyville included Scheyville excluded Themeda frequency score (n = 108) (n = 90) and: Correlation P Correlation P coefficient level† coefficient level† Exotic species richness - 0.432 *** - 0.378 *** Sum exotic frequency - 0.486 *** - 0.403 *** scores Native forb species - 0.011 NS - 0.122 NS richness Sum native forb frequency - 0.106 NS -0.157 NS scores † *, P < 0.05; **, P < 0.01; ***, P < 0.001; NS, not significant
The grazing-intolerant herb index was significantly, and negatively, correlated with both native (r = - 0.761) and exotic (r = - 0.760) species richness at 0.05 level (r0.05 (2), 6 = 0.707). As a lower abundance of grazing-intolerant herbs is indicative of a higher exposure to grazing, these negative correlations imply a positive relationship between grazing exposure and species richness. Native and exotic species richness were positively correlated with each other, but not significantly (r = 0.451).
The grazing index developed for the landscape study, and the grazing-intolerant herb index, were also significantly correlated (rs = - 0.841) at 0.05 level (rs0.05(1),6 = 0.829, Zar 1999). Again, this negative correlation is indicative of a positive concordance between the two measures of grazing pressure.
235 8.4 Discussion 8.4.1 Herbs and microhabitat
Native herbs
The hypothesis that the species richness of native herbs would be greater under trees than in open patches was not supported by the data. Differences between microhabitats were small and not significant, with mean species richness under trees slightly lower than that in the two other habitats. This finding contrasts sharply with that of Prober et al. (2002a), who recorded a very significant increase in native species richness under trees in Western Slopes woodlands.
Similarly, no significant differences between microhabitats were detected when the species richness of the two major subgroups of ground layer plants – native forbs, and native grasses and graminoids – was examined. In particular, patches round trees showed no trend towards either a higher richness of forbs, or a lower richness of grasses and graminoids. Mean forb species richness was in fact highest in open patches. Indices for these groups of plants might have been expected to provide a more sensitive measure of disparities, as they took account of abundance and standardised for species richness. However again differences were small, and not significant. There is little support here for the idea that ‘under tree’ habitats provide a haven for forbs (Prober et al. 2002a) in CPW.
One trend for a group of plants did reach significance: the C3 grass index was significantly higher under Bursaria than in open areas. However patches around trees were not significantly differentiated from the other two microhabitats on this variable, and the C4 index was not significantly affected by microhabitat. Thus the hypothesis, suggested by findings elsewhere in Australian temperate woodlands (Robertson 1985, Chilcott et al. 1997, Gibbs et al. 1999, Prober et al. 2002a), that C3 grasses would be more abundant under trees, while C4 grasses would predominate in open areas, was not supported by the data. Nevertheless, the finding of greater C3 dominance under Bursaria, and the trend in the C4 grass index, which was highest in open areas and lowest around trees, suggest the distribution of grasses in CPW may have something in common with those in other temperate woodlands.
236 The findings of the current study with respect to species richness and groups of herbs may have been affected by time-since-fire. Bond and Ladd (2001) found differences in species richness between microhabitats were accentuated in unburnt sites, whereas my sites were all sampled within two years of fire.
Exotic herbs
Although microhabitat did not significantly affect exotic species richness, there were more exotic herb species around trees than in the other two microhabitat types, and all four exotic species with a statistically significant association with a microhabitat favoured trees. The tendency for nutrient levels to be higher under trees (Section 8.1.1) may encourage exotic herbs. Birds perching in trees may drop seeds of species with fleshy fruits (Fensham and Butler 2004), such as Solanum nigrum.
Species composition
Although microhabitats were not clearly differentiated on the variables discussed so far, the hypothesis that the floristics of CPW microhabitats would vary, was strongly supported by the data. Analysis of similarities showed that microhabitat type clearly affected ground layer species composition. In particular, the floristics of open patches differed significantly from those under Bursaria thickets. Ordination and analysis of similarities both suggested that Bursaria thickets sat somewhere between open patches and patches around trees in terms of floristics, with species composition under Bursaria having more in common with patches around trees, than with open areas.
Floristic differences between microhabitats have profound implications for the dynamics of woodlands and fire. Findings from the landscape study reported in Chapters 4 and 5 strongly suggest that fire frequency influences Bursaria density, frequency and dominance in Cumberland Plain Woodland sites, and in particular, that sites from which fire is excluded for extended periods may lose their open patches to Bursaria. The findings of the current study imply that this change in the shrub layer will be accompanied by changes in ground layer floristics. In woodland sites with an open grassy structure, areas around trees may provide habitat for species which occur more frequently under Bursaria than in the open (Figure 8.14). In woodland sites with a
237 continuous Bursaria layer, species associated with open patches are liable to decline, as there is no substitute for their preferred habitat.
Figure 8.14. Dichondra repens and Glycine tabacina growing at the base of a tree. These species were both significantly more abundant around trees than in open patches.
Open patches were the favoured habitat of ten of the 14 native species with a statistically significant association with microhabitat. This included several species of lily with underground storage organs: Tricoryne elatior, Wurmbea dioica, and Hypoxis hygrometrica (Figure 8.15). The roots of these plants were all used by Aboriginal people as a source of food (Gott 1991, Kohen and Downing 1992). Collectively called ‘yams’, herbaceous plants with underground corms and tubers were so important in the diet of Western Sydney’s aboriginal population that they called themselves the Darug, or yam, people (Kohen and Downing 1992). Gott (2004) contends that an important Aboriginal use of fire was to keep country open, as yam species were known to grow best in open patches. Other ecologists have also recorded this habitat preference amongst forbs with underground tubers. Dichopogon strictus, Bulbine bulbosa and other geophytes were more abundant in open areas in woodlands near Melbourne (Gleadow and Ashton 1981), while abundance of Tricoryne elatior was reduced under shrubs in coastal grasslands (Costello et al. 2000).
238
Figure 8.15. Wumbea dioica (top), Tricoryne elatior (bottom left) and Hypoxis hygrometrica (bottom right) all have underground storage organs which formed part of traditional Aboriginal diets. All are significantly associated with open patches.
Although photosynthetic pathway did not distinguish open habitats from patches around trees when grasses were grouped, findings for individual taxa were in line with previous research. The only native grass species with a statistically significant association with a microhabitat, the C4 species Aristida vagans, was more common in open areas than under trees. The C4 species Themeda australis contributed to distinguishing open patches on the SIMPER analysis, while Microlaena stipoides and Austrodanthonia spp., both C3 taxa, were associated with trees. These findings accord with those from
239 temperate grassy woodlands in other parts of New South Wales and Victoria (Robertson 1985, Chilcott et al. 1997, Gibbs et al. 1999, Prober et al. 2002a,b).
It should be noted that individual native species showing a significant preference for a particular microhabitat were still found to some extent in the other two microhabitats. This suggests a degree of flexibility and resilience in the CPW herb layer. I believe it would be unwise, however, to rely on the ability of species to ‘hang on’ in non- preferential habitats, given the highly significant differences between microhabitats on the ANOSIM. Many species contributed to this outcome. Included may be less common species which could not be assessed individually for habitat preference, due to their limited occurrence in study sites and plots. Ideally, management should aim to retain a mix of microhabitats in CPW: both shrubby and open areas have a role to play in maintaining ground layer biodiversity. Open areas are particularly important for native herbs, as they provide a unique environment in a woodland setting.
Mt Annan 2, Lansdowne and Orchard Hills, the three sites in which microhabitats were particularly floristically distinct, all had a clear patch structure. Bursaria patches were well-developed, but open areas were also readily found (at Orchard Hills these areas sometimes contained small Bursaria plants, however these plants were not structurally dominant). On the other hand, in the two sites in which herb composition did not differ between microhabitats, patch structure was compromised: at Shanes Park, Bursaria patches were few and generally sparse; at Scheyville, open patches were hard to find as Bursaria dominated the landscape (Chapters 4 and 5). The significant difference between open patches and other microhabitats at the most frequently burnt site, Ropes Creek, may be explained by the patchiness of the very frequent arson fires which occur there. Bursaria thickets and areas around trees may burn less frequently than open areas, as a proportion of fires burning through newly-regenerated grass fuels may not be intense enough to penetrate these habitats, which may be less grassy and/or more moist than open Themeda-dominated areas.
8.4.2 Herbs and fire frequency
Native herbs and species composition
The hypothesis that infrequently burnt sites would support fewer herbaceous species than more frequently burnt sites was rejected. Mean native species richness was
240 remarkably consistent across all three fire frequency categories. Fire frequency per se also failed to influence species composition. These findings contrast sharply with the considerable differences in shrub species richness, abundance and composition between fire frequency categories (Chapters 4 and 5), and suggest that ground layer vegetation is considerably more resilient in the face of varied fire regimes than are most CPW shrubs.
The lack of a direct fire frequency effect on these measures accords with findings in Victorian grasslands. Morgan (1998b) was unable to relate differences in species richness and floristics between sites to previous fire history. They contrast, however, with results from savanna near Darwin, where herb species richness decreased in an area from which fire had been excluded (Fensham 1990, Woinarski et al. 2004).
However the finding that fire frequency per se did not affect ground layer floristics needs to be viewed in the light of the sampling strategy. As the primary aim of this survey was to assess the effect of microhabitat, I did not use the randomly-located plot design necessary to assess typical species composition across each site. Rather, sampling was stratified by microhabitat, giving greater weight to ‘rare’ microhabitats than a random sample would have done. A random sample in long unburnt sites would be likely to include many patches under Bursaria, while plots in frequently burnt sites would tend to fall in open areas (Chapter 4).
Although neither species richness nor overall floristics varied with fire frequency, the indices for native forbs and grasses showed significant movement across fire frequency categories. While the two indices were approximately equal in high fire frequency sites, forbs scored progressively higher than grasses as fire frequency decreased. Given the lack of variation in species richness, this trend must have resulted from changes in abundance in the two broad groups of native herbs. Thus
• grasses and graminoids were relatively more dominant where burning was frequent, while • forbs were relatively more dominant where burning was infrequent.
Note, however, that the forb index was higher than the grass and graminoid index in all sites except Ropes Creek (the most frequently burnt site), where the two indices were approximately equal. Thus forbs remained a central component of the ground flora under all fire regimes.
241 This finding about the relative dominance of grasses and forbs does not support the hypothesis that native forbs will decline if fire is too infrequent, a hypothesis that seemed tenable given the increased reproductive activity in this group after fire (Chapter 7). Clearly it is important to check suppositions about fire frequency generated from observations of species responses at a point, or points, in the fire cycle, through observations in areas where fire frequency itself has differed. The finding does, however, accord with previous observations that grass growth is encouraged by frequent fire (Birk and Bridges 1989), while a lack of fire may lead to a decline in this element of the ground flora (Morgan and Lunt 1999). In a series of experiments in grassy vegetation in South Africa, forbs tended to tolerate a wider range of fire regimes than dominant tussock grasses; these grasses declined as burning frequency decreased (Uys et al. 2004).
The good showing by forbs 18 months post-fire in infrequently burnt sites may be indicative of the effectiveness of their post-fire recovery strategies. The higher intensity fire generated in older fuels (Morgan 1999) may encourage increased flowering and seeding in these sites relative to sites burnt at a lower intensity. Differences in numbers of forb flowers and fruits in the burnt sections of the two case study sites investigated in Chapter 7 are in line with this hypothesis: there were three times more reproductive structures where fire had been more intense. Post-fire conditions after higher-intensity fire may also be more conducive to seedling establishment. Gaps between grass clumps may be larger, due both to decline before the fire (Section 4.4.5), and to greater mortality in the fire. More resources (light, water, nutrients) may be released, due to both greater post-fire mortality, and greater pre-fire accumulation of organic matter. Thus forbs may be able to ‘catch up’ after a fire in a long-unburnt site, even if their abundance had been reduced before the fire. Alternatively, forb species may be more able than grasses to recruit between fires, despite the relatively low levels at which this appears to occur in both groups (Chapter 7).
Competition between grasses and forbs in frequently burnt sites may also help explain the reduced forb dominance under short interfire intervals. Native forb species richness and abundance were negatively correlated with Themeda frequency in plots. Although these correlations were not significant, their sign is consistent with the concept articulated by Victorian grassland researchers (Stuwe 1994, Lunt 1995a, 1997a, Morgan 1998a, Lunt and Morgan 2002), that the competitive dynamic between dominant clump
242 grasses and interstitial species is the main driver of grassland floristics, and the primary reason disturbance is important in maintaining biodiversity. While fire may be of consequence for forb reproduction, it may provide an even more favourable environment for grasses.
Study of the relative abundance of forbs and grasses at different times-since-fire would help illuminate this issue.
Exotic herbs
Fire frequency directly affected the species richness of exotic herbs: there were significantly fewer exotic species per plot in frequently burnt sites than in infrequently burnt ones. The hypothesis that frequently burnt areas would support more exotic species than infrequently burnt areas was thus rejected once again, as it was for exotic shrubs (Section 4.4.3). Although contrary to expectations (Section 4.1.4), current findings are consistent with previous work in CPW and other woodlands: ten of eleven exotic species found by Thomas (1994) in a single ‘treatment’ in CPW at Prospect occurred in infrequently burnt plots. Woinarski et al. (2004) did not encounter any exotic species in annually burnt savanna woodland in the Northern Territory, although some weed species had established where fire had been excluded.
The increased presence of weeds in infrequently burnt sites may owe something to the decline of Themeda in these areas (Section 4.4.5). The finding that Themeda frequency had a highly significant negative correlation with both exotic species richness and exotic abundance does not prove a causal link between Themeda and a lack of weeds. It is, however, consistent with the hypothesis that exotics establish less readily in a vigorous Themeda sward. It is also consistent with the recent finding that Themeda establishment can render the soil environment less conducive to weeds (Prober et al. 2004).
8.4.3 Diversity across sites
Native species
A comparison between the species list for this study and that from the landscape-scale survey reported in Chapter 4 clearly demonstrates the crucial contribution of ground
243 layer vegetation to the biodiversity of Cumberland Plain Woodland. In this study 100 native herb species were recorded in 1068 m2 sampled over six sites. By contrast, 21,600 m2 – over 20 times the area – were sampled over nine sites for the landscape study, but only 37 native shrub taxa were encountered. These findings accord with those of Hill and French (2003), who recorded many more herbaceous species than shrubs in CPW at Holsworthy.
There were very significant differences in ground layer native species richness between sites within fire frequency categories in this study. Factors which may have contributed to these differences include:
• Recency of burning. The relatively low species richness at Ropes Creek might reflect the shorter post-fire recovery time in this site, which was sampled in the first, rather than the second, year post-fire. • Grazing, past and present. The negative correlation between the grazing- intolerant herb index and native species richness suggests that if grazing is a factor, it is operating to increase, rather than decrease, overall native species richness (see below). • Time of sampling. As sampling progressed into winter, shoots of some geophyte species (eg Caesia parviflora, Bulbine bulbosa) began to appear above ground, as did germinants of some short-lived species (eg Wahlenbergia gracilis). Abundance of geophytes and short-lived species may therefore have been underestimated in some study sites. Ideally, sites would have been re-surveyed in Spring, however time precluded this option.
Exotic species
Exotic species richness also varied significantly between sites. This result reflected major differences between the two moderate fire frequency sites, Lansdowne and Mt Annan 2 (site was not a significant factor when Lansdowne and Mt Annan 2 were omitted from the ANOVA). This disparity may have owed something to time of sampling, however as these two sites were assessed within the same five-week period, this explanation is unlikely to account for the major differences observed. Variation in past and present grazing provides a more likely explanation. Exotic species richness was significantly negatively correlated with the grazing-intolerant herb index, implying that grazing is associated with increased weediness. Lansdowne and Mt Annan 2 had the highest, and lowest, scores on this index respectively. Other researchers have also documented an increase in exotics with grazing (Prober and Thiele 1995).
244 Grazing effects
The finding that both native and exotic species richness were associated with a measure of grazing pressure highlights the possibility that grazing may have both positive and negative effects on the CPW ground layer. In particular, the current conservation management practice of excluding stock grazing from remnants may not have universally positive consequences for native biodiversity. Further investigation should seek to disentangle the effects of grazing by domestic animals from those of native species. Stocking level, nature of grazing (continuous or pulse), and recovery time may be important factors.
The mean grazing-intolerant herb index in study sites ranged, on a scale from 0 to 100, from 27 to 49. Grazing-intolerant herbs were thus an important component of the ground flora in every site. Again this suggests that the high-quality remnants selected for study in this project had not been severely impacted by grazing (see Section 5.4.5). The two indices of grazing used in this project correlate significantly, adding confidence in the validity of each of them.
The grazing-intolerant herb index was associated with neither microhabitat nor fire frequency. Significant findings on these variables are thus unlikely to be an artefact of grazing. This finding also suggests that the hypothesized protective effect of Bursaria on grazing-sensitive forbs mentioned in the introduction to this chapter either does not exist, or is offset by a preference by at least some palatable species for open patches. The mean value on the index was in fact higher in open areas than under Bursaria.
8.5 Conclusion
Species richness of native ground layer plants in CPW did not differ between patches around trees, under Bursaria bushes, or in open areas. Fire frequency also failed to significantly affect the species richness of native herbs, although sites with mostly 1-3 year interfire intervals had significantly less exotic herbs than sites which had had over 20 years between the last two fires. Species composition, however, did differ significantly between microhabitats. Thus while fire frequency per se may not affect species composition, structural changes in rarely burnt sites are likely to result in an overall shift in ground layer species composition due to a reduction in open patches.
245 CHAPTER 9 FUEL ACCUMULATION
9.1 Introduction 9.1.1 Vegetation as fuel
While the dynamics of bushland following fire may be conceived in terms of floristics and biodiversity by ecologists, they can also be considered from the point of view of those charged with protecting the human community from the destructive effects of bushfire. From this standpoint, vegetation is fuel, and as time-since-fire progresses, fuel accumulates.
The dynamics of fuel accumulation are of vital importance to fire managers because of the relationship between fuel load, fire intensity and potential for fire control (Raison et al. 1983, Cheney 1996). To simplify a complex situation, the likelihood of controlling any particular wildfire through human intervention is inversely related to its intensity. Fire intensity is affected by the weight of combustible fuel per unit area (fuel load), the heat yield of that fuel, and the rate of forward spread of the fire front. Rate of spread is influenced by factors outside human control, particularly ambient weather conditions and topography – and, in some circumstances, by fuel load (McAlpine 1995, Gould et al. 2004). Heat yield of fuel is determined by vegetation characteristics, and while there may be limited potential to manipulate this variable, for example by planting ‘fire retardant plants’ (Australian National Botanic Gardens 2003), this is unlikely to be either a cost-effective, or conservation-friendly, option in bushland reserves. Thus the only way to realistically manage fire intensity is through understanding, and if necessary manipulating, the amount of fuel available for a fire to consume (Simmons and Adams 1986, Morrison et al. 1996). The hazard level compatible with fire control in sclerophyll vegetation of southern Australia has been calculated to lie between 8 to 15 tonnes per hectare (Gill et al. 1987), although figures at the lower end of this range (eg 8 to 10, or 8 to 12 tonnes per hectare) are often given in the literature (Raison et al. 1983, Simmons and Adams 1986, Fensham 1992, Tolhurst 1996a).
246 Fuel dynamics relate to the relative rates of plant growth and senescence, litter deposition and breakdown of organic matter (Walker 1981, Simmons and Adams 1986). Fuel accumulates rapidly after burning (Tolhurst 1996a), particularly after fires of relatively low intensity (Birk and Bridges 1989, Adams and Simmons 1996). Grasses grow quickly in the post-fire environment, while rates of decomposition in the litter layer are low in the initial post-fire years (Fox et al. 1979, Raison et al. 1983, Woods et al. 1983, Neumann and Tolhurst 1991, Tolhurst 1996a).
9.1.2 Modelling fuel accumulation
It is generally assumed that at some point, fuel loads stabilise as decomposition rates match growth and deposition. Although this assumption has been challenged (Turner and Lambert 2002), accumulation of fuel over time is commonly modelled using an exponential saturation equation of the form:
-kt Wt = Limit (1 – e ) (1)
where Wt is fuel load (in tonnes per hectare, or t/ha) t years post-fire, Limit is the steady-state fuel load (t/ha), and k is a constant which reflects the rate of decomposition (Fox et al. 1979, Walker 1981, Raison et al. 1983, Simmons & Adams 1986, O’Connell 1991). Higher values of k are associated with more rapid progress towards maximum fuel accumulation than are lower values (Walker 1981). In recent years this equation has been modified to take into account the finding that fire generally fails to consume all available fuel (Birk and Bridges 1989, Fensham 1992). Adding the parameter Initial, the fuel load immediately after fire, equation (1) becomes:
-k t Wt = Initial + [(Limit – Initial) x (1 – e )] (2)
(Fensham 1992, Morrison et al. 1996). Initial tends to be higher after a low intensity burn than following one of high intensity (Birk and Bridges 1989).
247 9.1.3 Variation between vegetation types
Fuel accumulation parameters vary between vegetation types. There are suggestions in the literature that fuel may accumulate more rapidly in grassy than in shrubby sites, but peak at a lower level.
Rapid fuel accumulation has been documented in grasslands and grassy woodlands. Walker (1981) found that grass litter in a Heteropogon-Themeda grassland near Rockhampton in Queensland reached peak values after approximately four years, then declined, as the rate of litter accumulation was slower than the rate of decomposition. Williams et al. (2002) report that equilibrium fuel loads can be reached as early as three years post-fire in mesic tropical grassy savanna, while taking 5-10 years to do so in the more heath-like vegetation of the adjacent stone country. In Sydney’s shrubby sandstone woodlands maximum fuel loads are not reached until approximately 20 years after fire, while fuel accumulation in nearby shrubland communities continues for at least 30 years (Morrison et al. 1996). No information on fuel accumulation rates in Sydney’s grassy woodlands is currently available; this constitutes an important gap in knowledge.
On the other hand, Hobbs (2002) contends that litter accumulation in West Australian woodland communities is slower than that in forests, due both to the more open woodland canopy, and to harvesting by ants.
Fuel loads in grassy woodland sites may peak at a lower level than those in shrubby woodlands in similar climatic zones. Fuel loads in Northern Australian mesic grassy savanna reach equilibrium at around 10 t/ha, while in the more shrubby vegetation in the stone country, loads can reach 20 t/ha (Williams et al. 2002). Simmons and Adams (1999) found fuel loads in two Victorian grassy dry sclerophyll forests, last burnt approximately 13 years previously, were 18.1 and 20.2 t/ha respectively. These sites had lower fuel loads than wet sclerophyll sites, and much lower loads than a heath site, which carried 48.9 t/ha. In Sydney’s sandstone vegetation, maximum fuel loads reach at least 30 t/ha (Morrison et al. 1996). Again, data on equilibrium fuel loads for CPW are lacking.
248 9.1.4 Fuel components and patchiness
The contribution to fuel load of different elements of the flora – for example trees, shrubs and grasses – can be separated out. This practice assists understanding of both the ecology, and the fuel dynamics, of a vegetation type (Walker 1981). While total weight of combustible fine fuel is generally considered the most important fuel parameter affecting fire behaviour, attributes related to the make-up of the fuel bed are also influential. These include the degree to which the fuel is compacted or aerated (fuel density), and its vertical and horizontal continuity, or patchiness (Walker 1981, McCaw 1991, Cheney 2003). Aerated fuel is more combustible than compacted fuel, while the extent and pattern of discontinuity influences the thresholds at which fires will spread (McCaw 1991, Cheney 1996).
In grassy woodlands, available fuel may be mostly confined to ground level (Williams et al. 2002). Simmons and Adams (1999) found this to be the case in the two grassy forest sites surveyed in their comparative study of vegetation types. These sites had less bark than sites in other vegetation types, and more grass fuel, however twigs and leaves from trees still made up about half the fuel load. In shrubby heaths and forests, elevated fuel, in the form of live shrubs, and tree litter caught in shrub branches, was more abundant.
It is sometimes argued that frequent fuel-reduction burning increases the proportion of the fuel load attributable to highly flammable elements, such as grasses, thus increasing the very hazard it was designed to prevent (Asquith and Messer 1998). Conversely, infrequent burning may be associated with increased shrub fuel. We saw, in Chapters 4 and 5, that Bursaria abundance was high in sites with a low fire frequency, and that open grassy patches declined as fire frequency decreased. Birk and Bridges (1989) found a similar shift from grasses to shrubs in infrequently burnt areas, and comment that this trend has undesirable features from a fire control point of view, as fire is more likely to travel into tree crowns. Ignition thresholds, however, may be higher.
Patchiness in woodland fuels is likely to be influenced by the distribution of trees. Bark, twigs and leaves are presumed to accumulate to a greater extent under eucalypt canopy than in open areas (Walker 1981), although field measures verifying this assumption do not appear to have been published. Fire is thus likely to vary in intensity
249 across the woodland landscape. Hobbs and Atkins (1988) found considerable variability in fire intensity in experimental fires in Western Australian woodlands, and comment that these patterns will affect configuration of post-fire regeneration.
9.1.5 Study aims
The broad aim of this study was to develop an understanding of fuel dynamics in shale- based Cumberland Plain woodland habitats. Fine fuel loads were surveyed in woodland sites where the time of the last fire was known, and models of fuel accumulation developed.6
Questions associated with the models include:
• How rapidly does fine fuel accumulate in CPW?
• What maximum fuel loads are likely to be encountered in CPW?
• How much fuel remains after the passage of a fire?
• How long after fire are hazardous fuel loads likely to be achieved?
• How does fuel load accumulation in shale-based Cumberland Plain sites compare with that in Sydney’s shrubby sandstone woodlands? It is hypothesised, on the basis of the literature, that accumulation will be more rapid, but maximum loads lower, in CPW. Other questions addressed in this chapter include:
• How does the composition of fuel in CPW vary with time-since-fire?
• Does fuel composition differ with fire frequency? Two propositions were of particular interest: o The hypothesis that frequent fire leads to disproportionate growth of grass fuel, and o the hypothesis that higher levels of shrub fuel will be encountered where fire frequency has been low, than where it has been high or moderate.
• How does canopy cover influence fuel load?
6 Jikske de Bruin, an exchange student from Wageningen University in the Netherlands, contributed to this study in a number of ways – see Acknowledgements section for details.
250 9.2 Methods 9.2.1 Study sites and times-since-fire
The primary aim in selecting study sites was to maximise the number of data points representing different times after fire. Clearly, sites could only be included where the date of the last fire could be ascertained with reasonable certainty. This information was sought through land and fire managers. However because fires on the Cumberland Plain prior to the project’s commencement were not generally recorded (Section 2.7.2), options were limited. Data points were therefore obtained through surveying both different sites, and the same sites at different times.
Fourteen data points7 were obtained in total (Table 9.1). Time-since-fire ranged from two weeks, to over 50 years. Some samples were drawn from areas within the same remnant or reserve, but with differing burn histories. The year of the last fire at Richmond could be determined, but not the month (Table 9.1, point 13). The year of the last fire in the area of Scheyville where sample point 14 was obtained was not known, but time-since-fire was reportedly around 50 years. Time-since-fire for this sample point was set at 50 years.
All sites but one were classified as Shale Hills Woodland, Shale Plains Woodland, or Shale Gravel Transition Forest (Tozer 2003). These sites thus fell within a broad definition of Cumberland Plain woodlands (Section 2.4). The exception, Windsor Downs (point 10), is now classified as Castlereagh Ironbark Forest (NPWS 2004a), although listed as Shale Gravel Transition Forest on the version of the map available to the time the study was planned. Blocks sampled for this data point were located in areas where the CPW dominant Eucalyptus moluccana was a prominent species. No site was grazed by domestic animals at the time of sampling, although grazing by macropods, rabbits and hares occurred in some places.
7 The term ‘data point’ has been used throughout this chapter to describe each of the 14 site x time-since-fire combinations sampled. The term ‘site’ was not appropriate because some sites were sampled more than once.
251 Table 9.1. Location and time-since-fire of 14 fuel load data points.
Data point Site location Date Time-since- Fire recency Fire frequency number sampled fire (years) category category 1 Ropes Creek May 2004 0.0 - frequently burnt 2 Shanes Park July 2002 0.5 very recent frequently burnt 3 Orchard Hills July 2002 0.6 very recent infrequently burnt 4 Scheyville July 2002 0.7 very recent infrequently burnt 5 Ropes Creek July 2002 0.8 very recent frequently burnt 6 Shanes Park August 2003 1.7 recent frequently burnt 7 Orchard Hills May 2004 2.4 recent infrequently burnt 8 Scheyville May 2004 2.5 recent infrequently burnt 9 Plumpton August 2003 2.9 recent frequently burnt 10 Windsor Downs Sept 2003 5.8 - frequently burnt 11 Pitt Town Sept 2003 9.8 not recent frequently burnt 12 Mt Annan May 2004 21.6 not recent infrequently burnt 13 Richmond May 2004 22 not recent infrequently burnt 14 Scheyville May 2004 50 not recent infrequently burnt
9.2.2 Sampling and data collection
Sampling took place between July 2002 and May 2004, and was limited to areas of woodland vegetation visually dominated by native species.
A nested sampling design was used to obtain each data point. The portion of the site known to have burnt at the desired time was first divided, on paper, into 50 x 50 m blocks. Two blocks were randomly selected for further attention. Within each block, two 5 x 5 m plots were randomly chosen. In each plot, a ten-point interval scale, from 0 = 0-10% to 9 = 90-100%, was used to record estimates of:
• Overstorey foliage cover above the plot; • Shrub foliage cover; • Percent bare ground: ground not obscured by litter, grasses or other herbs, or low shrubs.
Five randomly-selected 25 x 25 cm subplots were located in each plot, giving a total of 20 subplots per data point (2 blocks x 2 plots x 5 subplots). In each subplot, litter depth was measured at four random points. All organic material in or above the subplot, including sticks up to 6mm in diameter, was then collected, and bagged. Six mm is the
252 limit used to define ‘fine fuel’ ─ fuel likely to be available in the event of a fire ─ by the NSW Rural Fire Service (G. Douglas, RFS, pers. comm. 2002), and by many, but not all, Australian agencies and researchers (Simmons and Adams 1986, McCaw 1991, Fensham 1992, Morrison et al. 1996, Cheney 2003, but see Fox et al. 1979, Birk and Bridges 1989). Shrubs up to 2 m in height were included. The area from which each fuel load sample was taken can thus be conceived as a three-dimensional, box-shaped space measuring 25 x 25 x 200 cm.
9.2.3 Fuel load determination
Fuel load samples were returned to the laboratory, where they were oven-dried at 105°C for 24 hours. These are the standard parameters used by the NSW Rural Fire Service when drying fuel (G. Douglas, RFS, pers. comm. 2002). Oven dried samples were weighed. Each sample was then sorted into four components, and the total weight apportioned between them. The four components were:
• ‘From trees’: material originating from tree species, mostly eucalypt bark, twigs and leaves;
• ‘From shrubs’: material originating from shrub species, mostly Bursaria twigs;
• ‘From grasses’: material originating from herbaceous species, mostly grasses, although this category also included material from other ground layer components including graminoid, forb and lichen species;
• The “comminuted fraction” (Simmons and Adams 1986, 1999): decomposing material not easily allocated to any other category, either because its origin could no longer be identified, or because it was so small as to make sorting impractical, eg pieces of decomposing eucalyptus leaf under 1 cm2.
9.2.4 Curve fitting
Exponential curves were fitted using the least squares method, with asymptotic errors calculated by matrix inversion (Johnson and Faunt 1992), using a customised Excel plug-in provided by Dr Ray Richie. The curve-fitting process generated best-fit solutions for the three fuel load accumulation parameters in equation (2) (Section 9.1.2):
• Initial, the fuel load immediately after fire, • Limit, the steady-state fuel load, and • k, the fuel accumulation rate constant.
253 Subplot data were entered rather than data point means: standard errors and confidence intervals for these parameters thus reflect the information contained in the entire data set.
Two sets of data points were used to generate two slightly different models: • All points (14 data points, 280 subplots); • All points except the three from Scheyville (11 data points, 220 subplots).
The Scheyville points were omitted from the second model because they represent an extreme situation, in ecological terms. The area of Scheyville sampled for these three points had reputedly been unburnt for 50 years until a partial burn in 2001 (point 14 was collected in the unburnt area, and points 4 and 8 in the burnt area at two different times after fire). Sampling for the landscape study took place 12 months after the 2001 fire. Results from that study indicated that Scheyville differed from other CPW sites in the almost complete absence of Themeda, and almost complete dominance of Bursaria (Chapter 4). As retention of Themeda and open patches is recommended from an ecological perspective (Chapter 10), it was considered important to model fuel accumulation in sites where these features were relatively intact.
The CPW models were compared with that generated by Morrison et al. (1996) from data collected in shrubby sandstone woodlands after low intensity burns at Ku-ring-gai Chase National Park:
(-0.08 x t) Wt = 6.8 + [(43.1 – 6.8) x (1 – e )] (3)
9.2.5 Contribution of components
Fuel load component data were analysed to ascertain whether each component contributed equally to the total dry weight of fine fuel. Percentage contributions were similarly assessed. All 14 data points were included in this analysis.
9.2.6 Time-since-fire comparisons
To examine the relationship between time-since-fire and the fuel load components, 12 of the 14 data points were allocated to one of three fire recency categories (Table 9.1, column 5). Data point 1 was omitted because regrowth had not commenced at the time
254 of sampling, two weeks post-fire. Data point 10 was omitted because it differed somewhat from the other data points in terms of vegetation type (Section 9.2.1), and because it did not group neatly with other data points in terms of time-since-fire. There were four data points in each fire recency category. The time-since-fire of the data points in each category spanned the following ranges:
• Very recent: 6 – 9 months post-fire • Recent: 1.5 to 3.0 years post-fire • Not recent: over 9 years post-fire.
Fire recency categories were compared for: • Total dry weight of fine fuel • Litter depth • Percent bare ground (using category mid-points) • The absolute dry weight of each fuel component (from trees, from shrubs, from grasses, comminuted fraction), and • The percentage of each fuel component relative to the total dry weight of fuel.
9.2.7 Fire frequency comparisons
Each data point was allocated to one of two fire frequency categories, frequently burnt, or infrequently burnt, on the basis of the fire history of the area in which it was located (Table 9.1, column 6). The frequently burnt category in this study encompassed both the high and moderate fire frequency groupings used in the landscape and microhabitat studies (Section 4.2.1). Data points in this category came from locations which had reportedly burnt at least once a decade over the last 20 years. Low fire frequency data points were located in areas that either had not burnt for over 20 years, or where the most recent interfire interval was at least 20 years. Data points 1 and 10 were retained in this analysis, giving a balanced design with seven data points in each fire frequency category. As the earlier times-since-fire were inevitably over-represented in the ‘frequently burnt’ category, comparisons between fire frequency categories focussed on the relative weight of each fuel component. Differences in shrub foliage cover were also assessed, again using category midpoints.
255 9.2.8 Effects of overstorey cover
The effect of overstorey cover on fuel load variables was investigated through comparing plots with differing degrees of overstorey cover.
Preliminary analysis revealed no significant differences, on any of the variables of interest, between plots with differing levels of overstorey cover above ten percent. Analyses therefore compared plots in the lowest canopy cover category, nominally 0- 10%, with plots in all other categories. In fact, most if not all plots scored 0-10% for overstorey cover had no canopy cover at all (pers. obs. 2002-2004). This category was therefore designated ‘away from trees’, while the category containing plots scored 1 (10-20%) or higher was labelled ‘under trees’.
Preliminary analysis also ascertained that time-since-fire did not differ significantly between the two overstorey categories (P = 0.852). The effects of overstorey cover can therefore be considered independent of those of fuel accumulation over time.
9.2.9 Statistical tests for comparisons
The initial intention was to use analysis of variance to test for differences between fire recency categories, between high and low fire frequencies, between canopy cover classes and between the absolute and relative weight of the four fuel load components. However it soon became apparent that variances, on all variables, were far from homogeneous. Attempts to ameliorate the situation through transformation, or raising the level of the analysis, were ineffective. Examination of the data showed that this situation was not a product of ‘outliers’, but merely reflected the naturally patchy nature of fuel in the CPW landscape. Resampling tests were therefore used; these tests are explained in detail in Section 5.2.3. As resampling tests are more conservative at higher levels of analysis (pers. obs. 2005), data point means were used as the unit of analysis for all comparisons. Where comparisons involved more than two categories, the sum of the absolute deviations of category means from the grand mean was first tested for significance. Where the overall deviation was found to be significant, pairwise tests were conducted to assess differences between categories. In every instance, data point means were permuted 5000 times.
256 9.3 Results 9.3.1 Fuel accumulation curves
The two sets of data points did not differ greatly in the parameter estimates they produced (Figure 9.1, Table 9.2, data in Appendix 11). Both models estimate that less than 1.5 tonnes per hectare of fuel remain after a fire in Cumberland Plain woodland, and that a steady state is achieved around or below nine tonnes per hectare. The curve generated when all 14 data points were used was:
(-0.46 x t) Wt = 1.39 + [(9.08 – 1.39) x (1 – e )], (4) while omission of the three Scheyville points produced slightly lower values on all parameters:
(-0.40 x t) Wt = 1.23 + [(8.67 – 1.23) x (1 – e )] (5)
The fit between each curve and its respective data points was highly significant (Pearson’s r, Table 9.2).
12
10
8
6
4
Fine fuel load (tonnes/ha) 2
0 0 102030405060 Time since fire (years)
Figure 9.1. Pattern of accumulation of fine fuel in Cumberland Plain woodland through time. ■, data point means; error bars give standard errors from the two replicate blocks sampled for each data point. Blue line: curve fitted to data from 14 points (all points shown). Red line: curve fitted to data from 11 points (red points only, blue points omitted; blue points come from Scheyville, a site with a very low fire frequency).
257 Table 9.2. Parameters of fuel load accumulation curves for a. Cumberland Plain woodland, based on 14 data points, b. Cumberland Plain woodland based on 11 data points (points from Scheyville, a very low fire frequency site, omitted), c. sandstone woodland in Ku-ring-gai Chase National Park (from Morrison et al. 1996). Initial, fuel load immediately after fire; Limit, steady state fuel load; k, fuel accumulation rate constant. S.E., confidence intervals, and Pearson’s r for a. and b. calculated using data from 20 subplots per data point; for c., these figures are not known.
a. CPW, 14 data b. CPW, 11 data c. Hawkesbury points points sandstone woodland Initial 1.39 1.23 6.8 S.E.(Initial) 0.46 0.43 ± 95% conf. 0.90 0.84 Limit 9.08 8.67 43.1 S.E.(Limit) 0.31 0.34 ± 95% conf. 0.61 0.67 K -0.46 -0.40 -0.08 S.E.(k) 0.07 0.07 ± 95% conf. 0.14 0.13 Pearson’s r 0.403 0.591 0.97 P 1.24 E – 242 1.19 E -226 Years post-fire when load 4.3 6.1 0.4 of 8 t/ha achieved Years post-fire when load not achieved not achieved 1.2 of 10 t/ha achieved Years post-fire when load not achieved not achieved 1.9 of 12 t/ha achieved
While the two CPW curves did not differ greatly, the slightly lower parameter values when the very low fire frequency points were omitted translated into a difference of almost two years in the estimated time to achieve a fuel load of eight tonnes per hectare. This level of fuel would accumulate in 4.3 years, according to the 14-point model, but would not be reached until 6.1 years post-fire, under the 11-point model (Table 9.2, Figure 9.2).
258 10
9
8
7
6
5
4
3
Finefuel load (tonnes/ha) 2
1
0 012345678 Time since fire (years)
Figure 9.2. Pattern of accumulation of fine fuel in Cumberland Plain woodland over the first eight years post-fire, comparing time taken to achieve a fuel load of 8 t/ha under two models. Blue line: curve fitted to all CPW data points. Red line: curve fitted to all data points other than those from Scheyville, a site with a very low fire frequency.
The two CPW models and the model generated by Morrison et al. (1996) from data collected in woodlands on sandstone, present a stark contrast (Figure 9.3, Table 9.2). Although the fuel load accumulation constant is lower, the other sandstone parameters are considerably higher than those for the shale-based woodlands. Fuel remaining immediately after a fire is estimated at 6.8 t/ha on sandstone, relative to the figure of less than 1.5 t/ha for CPW. While the steady state fuel load for CPW is around 9 t/ha, Morrison et al.’s model reaches asymptote at 43.1 t/ha, almost five times the CPW figure. These differences result in very different predictions with respect to time taken to reach potentially hazardous fuel loads. On sandstone, 8 t/ha is achieved just 0.4 years post-fire, according to the Morrison et al. model, while the equivalent estimates in the two CPW models are 4.3 and 6.1 years. CPW fuel load fails to reach 10 or 12 tonnes per hectare irrespective of time-since-fire, while these levels are achieved in 1.2 and 1.9 years respectively, in the sandstone woodland model.
259 45
40
35
30
25
20
15
10 Fine fuel load (tonnes/ha)
5
0 0 102030405060 Time since fire (years)
Figure 9.3. Pattern of accumulation of fine fuel in Cumberland Plain woodland (blue and red curves) and Hawkesbury sandstone woodland (green curve; model from Morrison et al. 1996).
9.3.2 Contribution of components
Material from trees made the greatest contribution to fuel loads, both in absolute and relative terms (Figure 9.4). Differences between components, which were significant for both average dry weight and percent contribution to fuel loads (P < 0.001), were attributable to the difference between material from trees, and each other component.
A. B.
4 100
90 3.5 80 3 70
2.5 60
2 50
40 1.5 30 1 20 Mean dry weight (t/ha) 0.5 10
0 Mean percent total fine fuel load 0 from trees from shrubs from grasses comminuted from trees from shrubs from grasses comminuted fraction fraction Fuel load component Fuel load component Figure 9.4. Contribution of four components of CPW fuels on the basis of A. mean dry weight (± S.E.) over 14 data points; B. mean percent contribution to total dry weight (± S.E.), over 14 data points.
260 Charts showing the contribution of fuel components at each data point can be found in Appendix 12.
9.3.3 Time-since-fire comparisons
Total dry weight of fuel, litter depth, and percent bare ground all varied significantly with recency of fire (Table 9.3). Total dry weight increased steadily across fire recency categories as time-since-fire increased, as did litter depth. Litter was over three times as deep, on average, where fire had not occurred for over nine years, as it was 6 - 9 months post-fire. Conversely, the proportion of bare ground decreased as time-since-fire increased. Values for individual data points on these three variables are given in Appendix 11.
Table 9.3. Variation with time-since-fire in Cumberland Plain woodland. Mean (S.E.) of total weight of oven-dry fine fuel, litter depth, and percentage of bare ground in three fire recency categories, each containing four data points.
Very recent Recent Not recent P (VR, 6 – 9 (1.5 to 3.0 (NR, over 9 mths post- years post- years post- fire) fire) fire) Total dry 3.66 (0.54) 6.28 (1.07) 9.09 (0.61) 0.018 VR < NR weight (t/ha) Litter depth 0.74 (0.08) 1.58 (0.31) 2.42 (0.60) 0.048 VR < NR (cm) Bare ground 18.1 (1.9) 11.3 (2.4) 6.3 (1.3) 0.019 VR > NR (percent)
Although overall fuel load increased significantly with time-since-fire, this was not the case for all components (Table 9.4). Dry weight attributable to tree species followed the pattern observed in the fuel load as a whole, as did the weight of the comminuted fraction. There was over five times more ‘comminuted’ material where fire had not occurred for nine years, on average, than in sites with a time-since-fire of less than a year. Dry weight attributable to grasses and herbs, however, was higher soon after fire than in subsequent time periods, although this difference was not significant. Shrub dry weight increased with time, but not significantly so.
261 Table 9.4. Variation in fuel components with time-since-fire in Cumberland Plain woodland. Mean (S.E.) weight of oven-dry fine fuel in t/ha attributable to trees, shrubs, grasses and the comminuted fraction in three fire recency categories, each containing four data points.
Fuel load Very recent Recent Not recent P component (VR, 6 – 9 (1.5 to 3.0 (NR, over 9 mths post- years post- years post- fire) fire) fire) From trees 1.64 (0.25) 3.57 (0.43) 4.05 (0.78) 0.017 VR < NR From shrubs 0.55 (0.34) 0.95 (0.54) 2.06 (1.05) 0.324 From grasses 1.08 (0.25) 0.67 (0.10) 0.75 (0.16) 0.245 Comminuted 0.40 (0.09) 1.10 (0.36) 2.23 (0.36) 0.022 VR < NR fraction
When the proportion of fuel attributable to the various components was assessed, significant findings pertained only to grasses and the comminuted fraction, while trees and shrubs contributed to much the same extent across all categories (Table 9.5). Grasses contributed a significantly higher percentage to the total fuel load where fire had occurred 6 - 9 months ago, averaging 31%, relative to their contribution at later times-since-fire (10.8% at 1.5-3.0 years post-fire, 8.3% at over 9 years post-fire). The comminuted fraction increased in importance with time-since-fire, making up 24.7% of the fine fuel load in the long unburnt sites, but only 10.8% where fire had occurred in the last year.
Table 9.5. Variation in fuel components with time-since-fire in Cumberland Plain woodland. Mean (S.E.) percentage of oven-dry fine fuel attributable to trees, shrubs, grasses and the comminuted fraction in three fire recency categories, each containing four data points.
Fuel load Very recent Recent Not recent P component (VR, 6 – 9 (R, 1.5 to 3.0 (NR, over 9 mths post- years post- years post- fire) fire) fire) From trees 45.9 (6.9) 59.6 (7.9) 44.8 (9.0) 0.373 From shrubs 12.2 (6.5) 13.4 (7.8) 22.2 (10.6) 0.641 From grasses 31.0 (8.2) 10.8 (0.7) 8.3 (1.7) 0.021 VR > R = NR Comminuted 10.8 (1.4) 16.2 (2.9) 24.7 (4.0) 0.024 NR > VR fraction
262 9.3.4 Fire frequency comparisons
The proportion of fuel load attributable to shrubs was very much lower, on average, where burns had occurred frequently (2.9%) than where they had not been common (26.2%; Figure 9.5). This difference was highly significant (P = 0.001). The percentage attributable to grasses was higher in frequently burnt sites, as was that attributable to trees, however these differences were not significant.
70
60
50
40
30
20 Percent totalPercent fine fuel load
10
0 from trees from shrubs from grasses comminuted fraction Fuel load component
Figure 9.5. Mean (S.E.) percentage of oven-dry fine fuel attributable to trees, shrubs, grasses and the comminuted fraction in two fire frequency categories. Red, frequently burnt; blue, infrequently burnt.
The relatively high contribution of shrubs in infrequently burnt sites was also apparent in the figures for shrub foliage cover. This measure averaged 7.5% where fires had occurred frequently, and 25.4% where they had not (P < 0.001).
Charts in Appendix 12 show the contribution of fuel components to individual data points in each fire frequency category.
263 9.3.5 Effects of overstorey cover
Thirty-nine (70%) of the 56 plots surveyed had over ten percent canopy cover, while 17 (30%) were classified as ‘away from trees’.
Both total dry weight of fuel and litter depth were significantly greater under trees. The total weight of fuel away from trees was 66% of that found under canopy (P = 0.013), while litter depth in open areas averaged 62% of its value where tree cover was present (P = 0.029). On the other hand, bare ground was more common away from trees, averaging 176% of that found under canopy (P = 0.009).
Not surprisingly, material from trees was a significantly greater contributor to fuel load where tree canopy was present (Figure 9.6). The dry weight of this component was over twice as great, on average, under trees than in open areas (P < 0.001, Figure 9.6A). However material from trees remained the greatest contributor to total fuel load even away from trees, averaging 42.8% in these patches relative to 55.8% under trees (Figure 9.6B). This difference in the relative importance of material from trees was also significant (P = 0.011).
A. B.
4.50 70.0 4.00 60.0 3.50
3.00 50.0
2.50 40.0 2.00 30.0 1.50
Fine fuel load1.00 (t/ha) 20.0 Percent fine total fuel load 0.50 10.0 0.00 from trees from shrubs from grasses comminuted fraction 0.0 from trees from shrubs from grasses comminuted fraction Fuel load component Fuel load component
Figure 9.6. Fuel load components in plots under trees (green) and away from trees (cream). A. Mean (± S.E.) dry weight of each component. B. Mean (± S.E.) percentage of oven-dry fine fuel attributable to each component.
The dry weight contributed by grasses was almost identical in the two habitats. The relative size of the contribution made by this component was, however, significantly greater in open areas (P = 0.031).
264 Material from shrubs did not differ significantly between the two canopy classes (P = 0.778 and 0.526 for dry weight and percent contribution respectively). And while the dry weight of the comminuted fraction tended to be higher under trees, this difference did not reach significance (P = 0.129), and the percentage contribution of this component was almost identical in the two environments (P = 0.979).
9.4 Discussion 9.4.1 Fuel accumulation in CPW
The models of fuel accumulation developed from the data collected for this study predict that little fuel – less than 1.5 t/ha – will remain after a fire in CPW. The rate of fuel build up in this vegetation type appears to be quite rapid, with values approaching asymptote by eight years post-fire. However models predict a steady state value of around or below 9 t/ha. While this figure is greater than the lower limit cited as compatible with fire control, 8 t/ha, it is below the figures of 10 and 12 t/ha also quoted in the literature (Raison et al. 1983, Simmons and Adams 1986, Fensham 1992, Tolhurst 1996a).
The rate of fuel accumulation, and the asymptotic value, were somewhat lower when points from areas with a very low fire frequency were omitted from modelling inputs. The purpose of producing this second model was to generate a curve for CPW sites with two characteristics of ecological significance: presence of Themeda australis (Chapter 4), and presence of open areas free from Bursaria (Chapters 4, 5 and 8). Of particular interest was the point at which this model would predict achievement of a fuel load of 8 t/ha. This occurred at 6.1 years post-fire. This figure is helpful in understanding the degree to which fire management regimes designed to conserve biodiversity may overlap those designed to protect life and property. This issue is addressed in Section 10.7.
265 9.4.2 CPW and sandstone woodlands
Fuel load parameters in CPW are very different to those calculated by Morrison et al. (1996) from data collected in shrubby woodlands on sandstone. Two features of Morrison et al.’s model need to be considered, however. First, their data points were taken from areas which had last burnt in low to moderate intensity fuel-reduction fires. A model presented by these authors for fuel accumulation in sandstone woodlands after high intensity fire uses a greatly-reduced post-fire fuel load (Initial), while the steady state fuel load (Limit) and accumulation rate (k) are also somewhat reduced. The data on which these changes are based are not provided. Second, Morrison et al. (1996) note that two data points from long-unburnt shrubby woodland sites fell well below the asymptote predicted by the model, and that these points were not included when the curve was fitted. Morrison et al. (1996) consider shrubby woodland fuel loads may decline somewhat from their peak value, and that the peak itself may be nearer to 30-35 t/ha than to the 43 tonnes predicted by their initial model.
Although these factors may decrease the disparity between Sydney’s shrubby and grassy woodlands, considerable differences remain. In particular, while peak fuel loads fall below 10 t/ha in CPW, they are at least three times as great on sandstone, even by the most conservative reckoning. This difference is in the direction predicted from existing literature, adding evidence for the proposition that maximum fuel loads in grassy woodlands can be quite low relative to those in nearby shrubby vegetation (Section 9.1.3).
The hypothesis that fuel would accumulate more rapidly in CPW than in shrubby woodland was also supported by the findings of the current study. The fuel accumulation rate k was considerably higher in CPW than the values used by Morrison et al. (1996) in any of their equations for shrubby sandstone woodland. However values of k calculated by other researchers for dry sclerophyll forest habitats are closer to the CPW values of 0.40 and 0.46. Simmons and Adams (1986) summarise five studies, including their own, in which k values range from 0.10 to 0.46. Fuel accumulation rates in CPW are towards the top of, but not outside, this range.
A third major difference between CPW and shrubby sandstone woodland concerns Initial, the fuel load immediately after the passage of fire. According to Morrison et al. (1996), low intensity burns are associated with much higher values of this parameter
266 than are fires of higher intensity. Data points in the current study were gathered after fires of a range of intensities. The four points from Ropes Creek and Shanes Park, both frequently burnt sites, followed low intensity arson burns. These points fell below the CPW curve, and therefore provide no evidence that residual fuel loads tend to be higher after low intensity fire. The disparity between CPW and sandstone woodland on this parameter may be because the grass and litter which make up the fuel in frequently burnt CPW sites combusts more readily, and thoroughly, than do the shrubs which dominate sandstone woodlands.
9.4.3 Fuel components in CPW
This section reviews findings related to each fuel component – trees, shrubs, grasses and the comminuted fraction – in turn. The influence of time-since-fire and fire frequency on components is addressed in this section.
Material from trees makes up the bulk of fine fuel in CPW. This may be somewhat surprising given the open woodland structure, as canopy cover is, by definition, low and patchy. But while fuel load attributable to trees was significantly higher where canopy cover occurred, eucalypt leaves and twigs were also common in open patches away from trees. Tree litter also forms the majority of the fuel load in grassy tropic Eucalptus miniata savanna (Williams et al. 2002). The absolute dry weight of material from trees in CPW increased with time-since-fire. Thus leaf litter begins to fall soon after fire, and continues to do so as time-since-fire progresses. However the increase in the mean weight of fuel from trees between 1.5-3 and 9+ years post-fire was small relative to that between 6-9 months and 1.5-3 years. The relative size of these differences probably reflects slow rates of decomposition in eucalypt litter over the early post-fire years (see discussion of comminuted fraction below).
The extent to which material from shrubs contributed to fuel load was primarily a function of fire frequency. Shrubs were a negligible component where fire had occurred at least once a decade, but a major contributor where fire had been infrequent. This was the case both in areas recently burnt after a long interfire interval, and in long- unburnt areas. These findings reflect those in Chapters 4 and 5 which showed Bursaria dominance increasing with reducing fire frequency. Almost all ‘material from shrubs’
267 was elevated fuel on Bursaria bushes. Other native shrubs were generally much smaller in stature than Bursaria and therefore contributed comparatively little to fuel loads. Shrub litter was also of little consequence (pers. obs. while collecting and sorting fuel load samples 2002-4).
A parallel to this situation is reported by Fensham (1992) for grassy woodlands growing on fertile soils with moderate rainfall in Tasmania. When regularly burnt, these woodlands, which have a ground layer dominated by Themeda australis, remain grassy, with fuel loads not exceeding 8 t/ha. When left unburnt, however, encroachment produces a shrubland dominated by Allocasuarina verticillata, Bursaria spinosa and Dodonaea viscosa, and fuel loads rise.
In Chapters 4 and 5, the patchiness of Bursaria in sites with differing fire histories was documented. In high and moderate fire frequency sites (frequently burnt sites in this study), Bursaria patches were limited in extent, grassy areas were common and tree canopies were generally well separated from the grassy understorey: eucalypts in CPW tend to have straight trunks which are leafless for the first five or so metres. In low fire frequency sites Bursaria thickets were larger and where fire had been excluded for a long time, the cover of this shrub was almost continuous. Bursaria provides elevated, aerated fuel which assists fire to travel into tree canopies (J. Pearson, Fire Control Officer, Penrith, pers. comm. 2002). This elevated fuel will be discontinuous in high and moderate fire frequency sites where Bursaria and other shrubs are patchy. Where Bursaria covers most of the landscape, as it does at Orchard Hills and Scheyville, shrub fuel will be more continuous, increasing the chances of canopy fire.
The percentage contribution of grasses and other herbaceous species to fuel loads differed significantly with time-since-fire. Grasses were an important component of fuel loads in the initial post-fire year, when on average they contributed 31% of total dry fine fuel. They then declined in importance, making up around 10% of fuel loads thereafter. This finding was consistent across fire frequency classes (Figure A12.2 in Appendix 12). Absolute dry weight of grasses was also highest in the first year after fire, although the difference between this time-since-fire and later categories was not significant. It appears that fire is followed by an initial flush of grass growth, irrespective of fire history.
268 The relatively small contribution of grasses to fuel loads after the first post-fire year echoes the findings of Simmons and Adams (1999) in grassy dry sclerophyll forest sites in Victoria. Grass fuel made up about 10% of total fuel loads; eucalypt leaves and twigs were the major contributor.
The ‘from grasses’ category included dead grass which had not yet decomposed. These findings therefore imply that dead grass decomposes rapidly relative to eucalypt litter. Walker (1981) reports that grass litter has a comparatively short half-life.
The rapid development of grassy fuel after fire suggests that Cumberland Plain woodland may be able to carry a fire as soon as twelve months after the previous burn, or perhaps even sooner, if grasses are cured by frost or drought. Both personal observation and reports from fire and land managers indicate that this is indeed the case. Bond (1997) comments that tussock grasses burn readily, because of their high surface- area to volume ratio and low moisture content when cured. Fires at yearly intervals in CPW are unlikely to be intense, as fuel loads would only have reached approximately 4 t/ha, according to the models. Spread might also be limited, as gaps would still be relatively common (Section 9.3.3).
Although the percentage of grass fuel was higher in frequently burnt areas than in those with a low fire frequency, this difference was not significant (the large disparity between the two fire frequency classes with respect to shrubs automatically meant that the proportion of other fuel components would tend to be higher in frequently burnt sites). Apart from the upswing in the first post-fire year, grasses were a small and fairly constant component of fuel loads. There is thus no evidence that frequent burning is associated with disproportionate grass growth on the Cumberland Plain. This conclusion is consistent with the finding in Chapter 4 that grass cover did not differ significantly with fire frequency.
The comminuted fraction, which consists mostly of decomposing material, increased with time-since-fire in both absolute and relative terms, making up an average of 25% of the fuel load in sites unburnt for nine years or more. Simmons and Adams (1986) also recorded an increase in this fuel component with time-since-fire. This finding accords with the literature on post-fire decomposition rates: decomposition activity is low in the early post-fire years (Woods et al. 1983), developing as litter builds up, providing an environment conducive to decomposer organisms.
269 9.4.4 Effects of overstorey cover
Fuel load was significantly greater under trees than in areas without canopy cover, due to increased input of ‘material from trees’, mostly eucalypt bark, twigs and leaves. However this component was still the major contributor to fuel load in open patches.
Although the relative contribution of grasses was higher in open patches, mean dry weight attributable to this component did not differ between the two canopy classes. This finding is at odds with studies which have demonstrated decreased herbaceous productivity under trees, or where trees have been removed (eg Walker et al. 1986, Harrington and Johns 1990, Scanlan and Burrows 1990). As noted in Section 8.1.1, trees can have both stimulatory and competitive influences on understorey vegetation. In CPW, those influences appear to balance out.
Patchiness in fuel loads across the CPW landscape will no doubt be reflected in patchiness in fire intensity, and thus in patterns of regeneration (Hobbs and Atkins 1988).
9.4.5 Litter depth and bare ground
Litter depth increased with time-since-fire, while the percentage of bare ground decreased. The figure of 6.3% bare ground for areas where fire had not occurred for over nine years is probably somewhat overstated, due to the use of category mid-points when analysing this variable. A score of 0 (0 – 10%) became 5%, making this the lowest possible value. In fact, many plots which scored zero – a score achieved by 14 of the 16 plots in the ‘not recently burnt’ category – had no bare ground (pers. obs.).
While plots burnt within the last year averaged 18.1% bare ground, those with a time- since-fire of 1.5 to 3.0 years averaged 11.3%. This finding supports the proposition in Chapter 7 that herbaceous species which flower in the months after a fire may recruit in the second post-fire year while gaps are still available. Clearly, however, gaps close quite rapidly in CPW, as litter depth builds up.
Litter depth was significantly greater under trees than in patches without canopy cover, whereas bare ground was more common beyond canopies. As noted in Chapter 1, it is sometimes suggested that herb recruitment will be less constrained by time-since-fire in
270 woodlands than in grasslands, as grass growth will be suppressed under trees, reducing competition and providing gaps in this microhabitat that would not be available in open areas. The data presented here indicate that this is unlikely to be the case: in fact open areas are more likely to provide gaps for forb recruitment than areas under trees. Also, as noted above, there is no evidence that grass growth in CPW is suppressed under trees.
9.5 Conclusion
Fuel accumulation in CPW starts from a low base as even low intensity fires consume most available fuel. Although accumulation rates are rapid relative to shrubby woodlands, maximum fuel loads are low, peaking at less than the 10 tonnes per hectare often cited as the maximum compatible with fire control. This finding suggests CPW land and fire managers are facing a very different scenario when seeking to balance fire regimes for protection of biodiversity and property, than their counterparts managing shrubby woodlands such as those on Sydney’s sandstone. This issue will be explored in Section 10.7.
271 CHAPTER 10 SYNTHESIS AND IMPLICATIONS
10.1 Introduction
The way in which disturbance, and the successional processes that follow it, shape the natural world has long been the subject of ecological theory.
Clements outlined what is now called ‘classical succession’ in 1916. Implicit in this model is the idea that only the final, ‘climax’ community is in equilibrium with the prevailing environment. The cultural metaphor for this ‘equilibrium paradigm’ is ‘the balance of nature’. Conservation practice aligned with this model focuses on objects rather than processes, concentrates on removing the natural world from human influence, and believes that desirable features will be maintained if nature is left to take its course (Pickett et al. 1992). The ‘balance of nature’ approach influenced attitudes to fire for many years. For example C.E. Lane-Poole told the Royal Commission following the 1939 fires in Victoria that “total exclusion of fire would enable natural succession to proceed, resulting in less undergrowth and a less flammable forest” (Griffiths, 2002:385).
Over recent decades, however, a paradigm shift has been underway. Drivers include the realisation that multiple states are possible within the one community (Westoby et al. 1989, Bond and Archibald 2003), as are multiple successional pathways (Connell and Slatyer 1977). Most importantly from a conservation perspective, it has increasingly been recognised that periodic disturbance is often essential to maintain diversity, allowing species which might otherwise have been displaced to continue to occur in a community (Connell 1978).
The ‘non-equilibrium paradigm’ can be encapsulated by the phrase ‘the flux of nature’. Scale is important in this paradigm: equilibrium at a landscape scale may be the product of a distribution of states or patches in flux. Implications include a legitimate role for people in ecosystem management, and a focus on the conservation of processes rather
272 than objects (Pickett et al. 1992). Fire fits more comfortably into the new paradigm, where it can be seen as a vital ecosystem process which mediates biodiversity.
Outgrowths of the new paradigm include:
• The vital attributes model of Noble and Slatyer (1980). This model provided the theoretical framework for the analysis of shrub vital attributes presented in Chapter 3, and can also be used to predict successional pathways.
• The state-and-transition model proposed by Westoby et al. (1989), which uses a catalogue of alternate states, and transitions between them, to identify land management options.
• The intermediate disturbance hypothesis, which suggests that biodiversity will be maximised where levels of disturbance are moderate (Connell 1978, Hobbs and Huenneke 1992).
• The dynamic equilibrium model (Huston 1979, 2003), which considers the interaction of productivity and disturbance in mediating species diversity.
The overall project goal was to develop a greater understanding of the relationship between fire regimes and plant diversity in the woodlands of Western Sydney’s Cumberland Plain, and to draw out implications for conservation management. This chapter brings together findings from the six project studies and explores how they can be used in management. Non-equilibrium paradigm concepts both inform this exploration, and gain validity through it.
The chapter begins with a synopsis of the effects of fire on Cumberland Plain Woodland, organised around the study questions introduced in the context of the literature review in Chapter 1.
A ‘state and transition’ model of the effects of fire in CPW is then proposed (Section 10.3). Benefits of this framework include the ability to separate ideas about what can (or may) happen in an ecosystem, from values as to what should happen (Westoby et al. 1989). Values issues are discussed in the subsequent section (Section 10.4).
Weed invasion is a major issue in CPW (Tozer 2003). Implications of project findings for management of exotic species are discussed in Section 10.5.
Although not the primary focus of the project, guidance as to appropriate fire interval domains for conserving plant diversity in Castlereagh woodlands can be gleaned from the vital attribute analysis in Chapter 3. This material is summarised in Section 10.6.
273 Section 10.7 broadens the management focus to include bushfire safety issues: how can CPW managers integrate fire management for biodiversity with protection of life and property?
Finally, the implications of project findings to processes and ecosystems beyond the Cumberland Plain are considered. Section 10.8 provides suggestions on goals and methods of determining fire frequency thresholds, particularly for grassy ecosystems. Section 10.9 draws on Huston’s dynamic equilibrium model to explore the relevance of the CPW findings for grassy woodlands elsewhere.
10.2 Key findings in Cumberland Plain Woodland
Fire exerts a powerful influence on Cumberland Plain Woodland. Remnants varied in species composition: site was a significant factor in all multivariate analyses. Nevertheless, the effects of fire frequency could be clearly discerned. These effects were most apparent in the shrub layer.
10.2.1 CPW shrubs and fire
While the shrub flora of Cumberland Plain Woodland is not as diverse as that found in the Castlereagh and sandstone vegetation types, CPW still supports many shrub species. Legumes feature large: nine of the ten most abundant species across sites surveyed for the landscape study (Chapter 4) were legumes. However the most abundant shrub species, by far, was Bursaria spinosa: it was the only one found in all study sites, and its average frequency at subplot (2 x 2 m) scale was over 50 percent.
What fire-related attributes do shrub species in Cumberland Plain and Castlereagh woodlands possess, and what does this imply for fire interval domains?
While most CPW shrub species resprout, about 40 percent are obligate seeders. In terms of Noble and Slatyer’s vital attributes scheme, a high proportion of CPW shrubs fall into plant functional types 4 and 12. Species in these groups are highly sensitive to infrequent fire, as recruitment is fire-cued. They are also moderately sensitive to high fire frequency, as although their seedbanks persist through more than one disturbance, repeated short intervals will exhaust them. A small group of non-leguminous species
274 highly sensitive to frequent fire can also be distinguished. Juvenile periods of Cumberland Plain obligate seeder shrubs are relatively short. Grevillea juniperina ssp. juniperina, with a juvenile period of four years, is the key fire response species with respect to frequent fire. Longevity data are rarely available, however analysis of estimates suggests that intervals above 15 years could cause populations of some species to decline (Chapter 3).
Does shrub species richness or composition differ with fire frequency?
Does Bursaria spinosa frequency, density and/or dominance increase with decreasing fire frequency?
Analysis of survey data from Cumberland Plain woodland sites with different fire histories produced findings consistent with predictions from the vital attributes model. The only relatively common shrub in this vegetation type able to recruit in the absence of fire, Bursaria spinosa, was found in increasing abundance as fire frequency decreased: this species dominated the landscape in sites from which fire had been excluded for 20+ years. Other shrubs, however, were most abundant in sites burnt approximately every 4 to 10 years, a finding consistent with Huston’s dynamic equilibrium model prediction for environments of “intermediate” productivity (Huston 2004). Obligate seeders were more sensitive to both high and low fire frequencies, than resprouters. Bursaria, which formed clumps in sites with 1-10 year interfire intervals, tended towards evenness as fire frequency decreased. Shrub floristics in the three low fire frequency sites were similar due to ample Bursaria, a relative dearth of other native shrubs, and the presence of woody exotics. The species composition of these sites differed significantly from that in more frequently burnt areas. High and moderate fire frequency sites also differed with respect to shrub floristics. A higher abundance of obligate seeder legumes distinguished sites with mostly 4 to 10 year intervals between fires from those which had experienced shorter or longer intervals. While competition from Bursaria probably played a role in the low abundance of obligate seeders in low fire frequency sites, the direct impact of a lack of fire is likely to have been at least as important (Chapter 3).
275 How does Bursaria spinosa regenerate after a fire?
In the single site where post-fire recovery of Bursaria was studied, a large majority of pre-fire plants resprouted, although smaller plants, particularly those below 25 cm, were less likely to recover than larger ones. Winter rain brought a flush of seedling establishment in the vicinity of plants over 3 m high. Although many seedlings died, some survived, and others established over the second post-fire year. There were more Bursaria plants 25 months post-fire than there were prior to the burn. Growth in Bursaria appears to be linked to rainfall. Even small plants have swollen tap roots, although no evidence of suckering was found. These findings suggest mechanisms to explain the distribution of Bursaria in sites with different fire frequencies (Chapter 6).
10.2.2 CPW trees and fire
The current mean density of adult trees in survey remnants, 279 per hectare, is almost certainly greater than pre-settlement values. There is evidence that density in today’s remnants is resource limited, and may be decreasing as post-logging regrowth self-thins. There is a high degree of variability between remnants in tree density, but a much smaller range in average basal area, indicating that remnants tend to have either a large number of small trees, or a smaller number of larger trees (Chapter 4). Large adult trees appear to be suppressing recruitment, particularly recruitment into the canopy (Chapter 5).
Does tree density or recruitment into the canopy vary with fire frequency?
Fire frequency did not significantly affect either adult tree density, adult tree basal area, or the density of juveniles or saplings, a finding which accords with the vital attributes status of the three dominant CPW eucalypts. Trends suggest frequent fire may be associated with an increased density of juveniles, but also with a decrease in the number of saplings ‘getting away’ into the canopy. Summer wildfire may also limit sapling numbers (Chapter 4). Both these points are moot, however, if adult tree numbers are decreasing as remnants recover from logging, and dominant trees become larger.
276 10.2.3 CPW herbs and fire
Does grass cover or dominance differ with burning frequency?
The primary impact of fire frequency on the ground layer appears to be on the tussock grass Themeda australis. This species dominated over 50%, and up to 83%, of subplots in sites with a high or moderate fire frequency, but less than 50% in all low fire frequency sites. At Scheyville, where fire had reputedly been excluded for 50 years, Themeda dominated only one subplot in 200 (Chapter 4). Themeda abundance is significantly associated with a reduction in herbaceous exotics (Chapter 8). This finding, along with findings from recent experiments which indicate that Themeda may play a role in regulating weeds through keeping nitrate levels low (Prober et al. 2002b, 2004) suggests that loss of Themeda may be associated with the loss of important ecosystem functions.
Is flowering and fruiting in CPW herbs greater in the post-fire period than some years after a fire?
Fire appears to play a role in the reproduction of many CPW ground layer species. Flowering and fruiting, particularly of forbs, was much greater in two sites burnt three to five months prior to survey, than in nearby unburnt areas. Bare ground was considerably more common, as were seedlings (Chapter 7). These findings are consistent with those from studies of seedbanks in grassy vegetation: although some ground layer species respond to fire-related cues, many do not, and persistent soil seedbanks are the exception rather than the rule (Section 1.7). Rather than enhancing seed germination, fire may enhance seed production: seeds resulting from a pulse of post-fire flowering and fruiting might be expected to germinate in the second growing season after fire, making long-term seed storage and fire cues unnecessary.
Does ground layer species richness, abundance or composition differ with fire frequency?
Despite enhanced post-fire flowering, when sites which had experienced different fire frequencies were compared, little evidence of differential effects on ground layer species was found (Chapter 8). Neither species richness of native herbs, nor species composition, differed significantly between fire frequency categories. In fact, forbs increased in importance relative to grasses as fire frequency decreased, though
277 remaining a central component of the ground flora under all fire regimes. Enhanced growth of grasses, especially Themeda, in more frequently burnt areas may explain these findings.
Do different microhabitats support different herbaceous species?
Though evidence for a direct effect of fire frequency on ground layer composition was slight, findings in relation to the influence of microhabitat on ground layer species composition point to a strong indirect impact. Open patches, patches around trees and patches under Bursaria hosted significantly different arrays of ground layer species. Ten of 14 species with a statistically significant connection with a particular microhabitat favoured open patches, including several lilies with underground storage organs. Thus the effect of fire frequency on ground layer species composition will be mediated by the changes in the shrub layer described above. If Bursaria expansion under infrequent burning leads to a loss of open patches, a decline in herbaceous ‘open patch’ species can be expected (Chapter 8). As patches under Bursaria are more similar floristically to patches around trees than to open areas, open patches provide a unique habitat in CPW.
10.2.4 CPW exotics and fire
Does fire frequency influence exotic weeds in CPW?
Woody exotics were more abundant in low fire frequency sites than in areas which had burnt at least once a decade. Very frequently burnt sites had virtually no woody exotics (Chapter 4). Similarly, significantly less herbaceous weed species were found in very frequently burnt microhabitat plots than in plots where fire frequency had been low (Chapter 8). There was a significant negative association, at a small scale, between the abundance of Themeda australis and the species richness and abundance of exotic herbs (Chapter 8).
278 10.2.5 CPW vegetation as fuel
How does fuel accumulate over time in CPW?
Fuel accumulation in CPW starts from a low base as even low intensity fires consume most available fuel. Modelling indicates that equilibrium fuel loads will be attained within ten years, however peak loads of around nine tonnes per hectare are low relative to those in nearby sandstone woodlands. Eucalypt litter – leaves, sticks and bark – was the major contributor to fuel loads in areas sampled. Grasses were an important component in the first post-fire year, but a minor contributor thereafter, while decomposing material contributed more in both absolute and relative terms as time- since-fire progressed. The contribution of shrub fuel, mostly Bursaria twigs, was many times higher in infrequently burnt sites than where fire had occurred at least once a decade.
10.2.6 Ecology informing management
The themes of the non-equilibrium paradigm resonate through the findings of this project. Vegetation on the Cumberland Plain is ‘in flux’ – it changes with time-since- fire, and it changes with fire frequency. Disturbance mediates coexistence of shrubs with different life histories, and that of grasses and herbs. Patchiness in shrub distributions, itself a product of disturbance, provides diverse habitat for ground layer species. Fire truncates the succession from grassy to shrubby woodland (and perhaps beyond), a role it seems to play across the world (Bond et al. 2005).
The management implications of the non-equilibrium paradigm also apply. If we want to conserve the diversity of Western Sydney’s grassy woodlands, we must consciously manage the processes which sustain it, including fire. The findings of this project, and the insights into ecosystem processes they engender, contain much that may be of assistance to managers. The remainder of this chapter addresses the question:
What guidance for managers can be drawn from this research?
279 10.3 A state and transition model for CPW
Project findings were used to develop a model of the effects of fire on Cumberland Plain Woodland (Figure 10.1). The model uses the state and transition framework outlined by Westoby et al. (1989). Four ‘states’ in which CPW can be found are identified. Three – States 1, 2 and 3 – are stable under certain conditions; S4 is transitory.
The model allows exploration of conditions which might maintain States 1 to 3, and of transitions between states, without imposing value judgements as to the desirability of either particular states, or particular transitions. It therefore has the potential to inform a variety of management aims, for a variety of situations. Values will be discussed subsequently.
S2 Grassy Themeda woodland with shrubs T3 T1
T2 S1 S3 T6 Grassy Themeda Bursaria- woodland dominated woodland T5
T4 S4 Recently burnt Bursaria- dominated woodland
Figure 10.1. State (S) and transition (T) model describing four states in which Cumberland Plain Woodland can be found, and transitions which may occur between these states. See text for detail of each state and transition.
280 10.3.1 Description of states
State 1 is grassy Themeda woodland with occasional Bursaria patches, but few other shrubs (Figure 10.2). The ground layer is diverse, although its composition is slanted towards open patch species. State 1 has virtually no woody exotics, and relatively few herbaceous ones.
State 2 is grassy Themeda woodland with some Bursaria patches, as well as patches of other shrubs, including obligate seeders (Figure 10.3). It has a diverse ground layer with good representation of species that favour both open and shady patches. It has few woody weeds.
In State 3 the woodland is dominated by Bursaria spinosa to the point where there are few open patches (Figure 10.4). Other native shrubs, particularly obligate seeders are in relatively low abundance. Exotic shrubs such as Olea europaea ssp. africana and Sida rhombifolia are in evidence. The ground layer is diverse, though representation of open patch species is relatively low, and herbaceous weeds are not uncommon.
State 4 is State 3 woodland which has recently been burnt (Figure 10.5). Bursaria spinosa is resprouting, as are grasses and forbs.
10.3.2 Maintaining states
What management actions might maintain states 1 to 3 (state 4 is by definition transitory)? As project studies were observational rather than manipulative, the descriptions in this section should be considered tentative. They should be used to generate hypotheses for experimental testing when resources for a long-term manipulative study are available, or to guide long-term targeted monitoring.
The results of the landscape study (Chapter 4) strongly suggest that frequent burning would be needed to maintain State 1. A fire every 2 - 4 years would allow regeneration of Themeda and other existing ground layer species, and keep the abundance of Bursaria, other native shrubs, and weeds low.
Various findings suggest a fire interval domain of approximately 4 to 12 years would be appropriate for maintaining State 2, open Themeda woodland with obligate seeder shrubs. Variable intervals between these thresholds should allow both Bursaria thickets and open grassy areas to co-exist in the landscape.
281
Figure 10.2. State 1: grassy Themeda woodland with some Bursaria patches, few other shrubs, and virtually no woody weeds.
Figure 10.3. State 2: grassy Themeda woodland with shrubs, including obligate seeders, some Bursaria thickets, and few woody weeds. Photo by Annie Storey.
282
Figure 10.4. State 3: Bursaria-dominated woodland with few other shrubs, and some woody weeds.
Figure 10.5. State 4: recently-burnt Bursaria-dominated woodland.
Interfire intervals need to be long enough to allow regeneration of the obligate seeder shrubs which characterise State 2. The key fire response species, Grevillea juniperina ssp. juniperina has a primary juvenile period of four years: the argument for pegging lower thresholds at this level is detailed in Chapter 3. The legumes amongst the CPW
283 shrub contingent are buffered from negative effects of occasional very short interfire intervals through retention of ungerminated seed in the soil. This was demonstrated for several species, including Daviesia ulicifolia and Acacia decurrens, at Holsworthy, where sampling followed a two-year interfire interval. Both species were relatively abundant, although it is unlikely that seedlings germinating after the previous fire would have had time to produce fruits. Twenty years of short interfire intervals at Prospect had also failed to eliminate obligate seeder shrubs, which were more prevalent along landscape study transects at Prospect than in any other site; this was 15 years after annual fuel reduction burning gave way to a more moderate regime (Chapter 4). These factors strongly suggest that intervals as low as four years, within a varied regime which also includes longer periods between fires, will not disadvantage shrub species.
On the other hand, fire needs to occur before short-lived fire-dependent species and their seedbanks senesce. Themeda is potentially vulnerable: if decline rather than complete local extinction, and ecological importance, are considered, Themeda may be the key fire response species for infrequent fire. In a Victorian grassland this species declined due to self-shading between six and 11 years post-fire, and failed to recover its former abundance when a burn occurred after 11 years (Morgan and Lunt 1999). Themeda in CPW may be slower growing, as CPW shale soils would be less nutrient- rich than the basalt-derived clays which support Victoria’s grasslands, and may therefore take somewhat longer than 11 years to collapse locally. Still, Themeda abundance was much reduced in all low fire frequency sites in the landscape study, suggesting intervals in CPW should not be set too high. Recent research suggests Themeda may play a key role in maintaining ecosystem integrity (Prober et al. 2004), so a precautionary approach to its preservation is warranted. When all these factors are considered, a top threshold of 12 years seems reasonable.
Short-lived shrubs also need a chance to regenerate before soil seedbanks decline. Analysis from a vital characteristics perspective in Chapter 3 led to the recommendation that maximum intervals not exceed 15 years. Factors include observations of the death of significant proportions of local populations of some obligate seeders six to 10 years post-fire, possible decline in productivity of remaining individuals (Auld and Myerscough 1986, MacFarland 1990), and concerns that seedbanks in CPW clay soils may not survive for long periods. The low abundance of obligate seeder shrubs in low fire frequency sites, despite recent fire, suggests the upper threshold should not be set
284 too high, as does the very high vulnerability to low fire frequency evident from the vital attributes analysis. There is ample evidence that CPW shrubs thrive under intervals below 12 years (Chapter 4).
Finally, fire needs to recur with sufficient frequency to maintain a balance between open areas and Bursaria thickets if State 2 structure is to be maintained. Bursaria was a prominent species in all CPW sites, forming some thickets even in high fire frequency areas: it is therefore unlikely to be at risk at this end of the fire frequency spectrum. Keeping this very successful species within bounds is the greater challenge: this is important to ensure populations of ‘open patch’ herbs are maintained (Chapter 8), and that possible competitive effects of Bursaria on obligate seeder shrubs (Chapter 5) do not become too great. The intervals represented in landscape study sites were not sufficiently fine-grained to pin-point the threshold at which Bursaria density begins to rise sharply, although we know it has increased considerably at Mt Annan during a probable 20 year interval (Benson and Howell 2002). Observation at other CPW sites, particularly Prospect and Nurragingy, suggests Bursaria density may build up quite rapidly in some areas, despite a fire recurrence of approximately once a decade. The propensity of Bursaria to dominate may be greatest where nutrients and moisture are most readily available (Crowley and Garnett 1998, Roques et al. 2001, Noble and Grice 2002). Again, intervals between 4 and 12 years would likely achieve a mix of Bursaria and open patches.
Given the variability in juvenile periods in obligate seeder shrubs, and uncertainties about maximum intervals for limiting Bursaria expansion and maintaining Themeda, it is recommended that CPW managers seeking to maintain State 2 characteristics monitor:
• Flowering of obligate seeder shrubs. Fire should normally be excluded until these species have had a couple of good flowering years. In some years, flowering and seed set may be limited by drought (pers. obs. 2002; Auld and Myerscough 1986, Bell et al. 1993). • Bursaria expansion. Fire may need to be applied more frequently if Bursaria is encroaching into previously open areas. • Themeda. Fire may need to be applied more frequently if Themeda clumps are dying.
285 Monitoring these parameters would be relatively simple, and could help tailor fire management to the needs of particular sites, and to variations in plant growth resulting from variations in rainfall.
Experience in low fire frequency sites suggests that State 3 would be adequately maintained by fire exclusion, or a low fire frequency. How often could fire recur without reducing Bursaria density? Research from other localities where shrub encroachment is occurring suggests that once a thicket is established, it has considerable inertia (Scholes and Archer 1997). New successional processes begin to drive the system, shrubs modify soils and microclimate, and seedbanks change (Archer 1989). Monitoring at Mt Annan (Chapter 6) showed an increase, rather than a decline, in Bursaria density after a single fire following a probable 20-year interfire interval. In a well-established Bursaria-dominated woodland, 12 - 15 year intervals might cause little change.
10.3.3 Transitions
Some possible transitions between states are shown in Figure 10.1. Again, these descriptions should be viewed as tentative until experimentally tested.
Transition 1 takes open grassy woodland with shrubs to open grassy woodland with few shrubs. Frequent burning, say every one to three years, should accomplish this transition.
Transition 2 takes open grassy woodland with few shrubs to open grassy woodland with shrubs. A reduction in burning frequency should encourage this transition, although fire should not be completely suppressed. Experimentation with intervals in the 6 - 10 year range is suggested. Six to ten years would allow obligate seeder shrubs time to build up seedbanks, and fuel loads would be sufficiently developed to support a moderately intense fire and thus good germination.
Transition 3 takes grassy woodland with shrubs to Bursaria-dominated woodland. Fire exclusion should support this transition. How quickly Bursaria would spread would almost certainly depend on its initial abundance and distribution (Clarke and Davison 2001), as well as on the local availability of soil nutrients and moisture. Stocking with
286 domestic animals might hasten this process, by removing grass competition and making fire less likely.
What about the transition from State 3 to State 2? This is where State 4 becomes relevant.
Transition 4, which takes State 3 to State 4, is simply a fire in a State 3 woodland. Though simple conceptually, accomplishing this transition may be difficult for managers, as State 3 woodlands are hard to burn (J. Sanders, NPWS, pers. comm 2004). Bursaria does not appear to be as flammable as more sclerophyllous shrubs, and the closed nature of Bursaria thicket may create a more mesic environment than that found in open grassy areas. Similar dynamics have been observed in Australia’s semi-arid rangelands (Noble 1997). Transition 4 planned burns may therefore need to be carried out in conditions outside those appropriate for other vegetation types. In some situations, wildfire may supply this transition. However fragmentation means many remnants are now cut off from wildfire paths (Bond 1997), so this option is unlikely to be viable for many CPW sites.
Transition 4 represents an opportunity to move from State 3 to State 2 via State 4. However management action is required to take advantage of this opportunity. Without action, a burnt Bursaria-dominated woodland will rapidly return to its previous state. This is Transition 5. During the course of the project this process occurred at all three low fire frequency sites: Bursaria resprouts at Orchard Hills and Scheyville reached 2 m in height by 2.5 years post-fire, and Bursaria frequency and density was high in post- fire surveys (Chapter 4). Plant numbers reached 110% of pre-fire totals by two years post-fire in monitoring plots at Mt Annan (Chapter 6).
What might encourage Transition 6, that is a return to grassy woodland with open patches and obligate seeder shrubs? Research from other ecosystems has demonstrated that an increase in fire frequency can be used as a management tool to help regain open grasslands and woodlands in the face of shrub increase (Hodgkinson and Harrington 1985, Roques et al. 2001). Experimentation with 4 or 5 year intervals, but also with intervals between 1 and 3 years in parts of the landscape without obligate seeder shrubs, is suggested. The more open landscape of State 4 would probably carry a fire more easily than State 3; grass growth would provide flammable fuel (Chapter 9) as would dead Bursaria stems remaining from the Transition 4 fire. In semi-arid woodlands,
287 resprouting Eremophila spp. which appear to have similar life-history strategies to Bursaria are vulnerable to two autumn fires a year apart, even though a single fire has little effect (Hodgkinson 1986, cited in Hogkinson 1998).
The primary aim of short intervals would be to reduce Bursaria density and encourage Themeda growth. Themeda clumps and some dead Bursaria plants have been noted in a small patch at Scheyville subject to a two-year interfire interval after a long period of fire exclusion; Themeda is absent from adjacent woodland burnt only in the first fire. This observation suggests long unburnt CPW may be able to demonstrate remarkable resilience when burning is reinstated. However there is no guarantee one or two short intervals will reintroduce Themeda to sites where it is not currently found. Nor is it likely to assist shrubs other than Bursaria to establish and expand; in fact very short intervals might eliminate any remaining obligate seeders. One way to address this problem could be to collect seed of Themeda and local shrubs, particularly obligate seeders, and spread it in patches some time before the second fire (McDougall and Morgan 2005).
10.4 Conservation values and fire management in CPW
Section 10.3 above outlines actions managers could take in CPW remnants, however it does not address the question of which actions should be taken. This section argues for actions I believe would benefit conservation of biological diversity in Cumberland Plain Woodland. As such it reflects my values, which may not be shared universally; values are, by their nature, subjective.
10.4.1 Conservation value of states
All four states described above have high conservation value; no remnant should be devalued because it fits into a particular category. In terms of plant diversity, all three states support many native grass and herb species. The CPW ground layer is resilient in the face of even quite extreme fire regimes (Chapter 8). Ground layer species richness at the 10 m2 scale is similar in States 1, 2 and 4. As State 4 remnants were recently in State 3, presumably this state too hosts similar ground layer diversity, even if some species are not visible above ground.
288
10.4.2 Value of Bursaria spinosa
Although Bursaria spinosa tends to dominate the CPW landscape under particular circumstances, this does not imply that it is a ‘problem plant’ per se. Bursaria is a natural and important component of all CPW remnants. Its thorny nature and tendency to form thickets make it important habitat for birds such as the Striated Thornbill, Eastern Yellow Robin, and White-browed Scrubwren. Many bird species use a mix of shrubby and open patches (H. Recher, unpub data). Bursaria leaves and bark provide food for insects and other grazing animals, including macropods (pers. obs., 2001-4). Intricate Bursaria-insect interactions have been noted (Bellingham 2003, Watson 2004). Unlike exotics such as African Olive and Lantana, the canopy of Bursaria is not completely closed, and it does not shed much litter (pers. obs. while collecting fuel load samples, 2004). Thus it does not exclude ground layer herbaceous species, although as noted in Chapter 8 some species are more likely to be found in open patches.
10.4.3 Value of open grassy woodland
Although maintenance of existing species richness is obviously vital (Benson and Howell 2002), the structural integrity of Western Sydney’s grassy woodlands is also worthy of conservation focus. There is a real risk that the open grassy character of Cumberland Plain Woodland could be completely lost over the next 50 years under the combined influence of fire exclusion and increased atmospheric carbon dioxide, which may also encourage woody plant growth (Berry and Roderick 2002, Bond et al. 2003b).
10.4.4 Management of grassy woodland with shrubs
Resources available for management are often limited, and priorities for conservation must be set. There are strong arguments for giving precedence to State 2 remnants.
Conservation values in State 2 remnants are likely to be relatively high. These remnants have a good shrub complement, and a balance between open and shrubby patches. This may give them particular value as fauna habitat, as habitat requirements vary widely between animal species, and some species use resources from a variety of habitats (Friend 1993, Catling and Burt 1995, York 1999, Tasker and Dickman 2004, Woinarski et al. 2004). Woody weeds are not a major problem in State 2 remnants. Although a
289 number of areas on the Cumberland Plain are in this state, including parts of the large remnants at Prospect and Holsworthy, without conscious management they are at risk.
10.4.5 Management of Bursaria-dominated woodland
State 3 remnants have continuing conservation value, however they do not provide a hospitable environment for either obligate seeder shrubs or open patch ground layer species. Weediness is also an issue. Quite a number of CPW remnants are either currently in State 3, or appear to be heading that way. Fire exclusion, which has been the de facto management strategy for most CPW remnants over the last 15-20 years, is likely to have been the major factor behind this trend, although increased atmospheric carbon dioxide may also have played a role through stimulation of woody plant growth (Berry and Roderick 2002, Bond et al. 2003).
Managers of remnants currently dominated by Bursaria, or in which Bursaria is thickening, need to consider their options carefully. Fire exclusion may be justified in rare instances, for example as protection for cover-dependent fauna. However Bursaria dominance is likely to be associated with a decline in other native shrubs, in ground layer species which favour open patches, and in birds which need open foraging areas (Chapman and Harrington 1997). Weed abundance is likely to continue to increase. Experimentation with Transitions 4 and 6 is suggested. Given the difficulties of burning where Bursaria covers the landscape (see discussion of Transition 4 in Section 10.3.3), managers of remnants with moderate Bursaria density which still retain open patches may be well advised to introduce fire sooner rather than later.
10.4.6 Management of grassy Themeda woodland
State 1 remnants are not common, and those that do exist are not adequately conserved. Although shrub abundance is low, these areas are less weedy than others, have a diverse ground flora, and provide habitat for macropods. While a decrease in fire frequency would be desirable, I believe there is value in conserving these open grassy woodlands, even if arsonists prove intractable.
290 10.4.7 Importance of active management
Elsewhere in Australia, the need for urgent action to preserve fire-maintained habitats is being recognised (eg Crowley and Garnett 1998, Russell-Smith and Stanton 2002). As Lunt (1998a:644) pointed out after cataloguing the conversion of open grassy woodland at Ocean Grove in Victoria to impenetrable Casuarina thicket, “To conserve biodiversity in the future, ecologists and land managers must develop and instil an informed philosophy of active vegetation management, rather than perpetuating a pervasive attitude of passive non-intervention.” These sentiments echo those of non- equilibrium paradigm theorists overseas (Pickett et al. 1992).
10.5 Managing weeds
In this project, frequent fire was associated with significantly lower levels of weediness than infrequent fire, for both shrubs (Chapter 4) and herbs (Chapter 8). The relationship appears to be linear, with very frequently burnt remnants having the lowest abundance of weeds. The dominant tussock grass Themeda australis may play an important role in keeping exotic herbaceous weeds at bay (Prober et al. 2002b, 2004). This species dominated sites with a high or moderate fire frequency, but not those which had been infrequently burnt. These findings have profound implications for management.
The finding of a low weed frequency in frequently burnt areas is not exclusive to this project. For example Yibarbuk et al. (2001) found no weed species whatsoever in traditionally burnt savanna in Arnhemland. However the current project is one of very few Australian studies in which the extent of exotic weed invasion has been compared between sites with different degrees of exposure to fire.
It is possible that a number of exotic shrub species are advantaged, under low fire frequencies, by their ability to recruit all though the interfire interval (T species of Noble and Slayer 1980, see also Kirkpatrick 1986). They may also, however, grow relatively slowly, and thus be vulnerable when young. Fire at intervals shorter than the time taken for these species to reach fire tolerance would prevent their establishment. In CPW, there may be an argument for concentrating on shorter intervals in the moderate 4 - 12 year range, or at least for adding “establishment of woody exotics” to the list of indicators for monitoring in State 2 remnants (Section 10.3.2).
291 Although many exotics may be deterred either directly or indirectly through regular burning, other weed species may be well equipped to take advantage of the “stable invasion window” provided by frequent fire (Morgan 1998d, Setterfield et al. 2005). Concerns that this might apply to two grass species with high invasive potential in temperate areas – Eragrostis curvula (African Love Grass) and Nassella neesiana (Chilean Needle Grass) – have been expressed (Stuwe 1994, Nadolny et al. 2003). Neither of these species is currently a major problem in the areas surveyed for this project, although E. curvula was often found on the edges of frequently burnt remnants, particularly where grass had been mowed, or soil disturbed. Managers are strongly advised to target these species as they regenerate in the weeks after a fire, before they flower.
In addition, although fire appears to have limited weed encroachment into State 1 and 2 remnants, it unfortunately does not follow that frequent fire will eliminate weeds from State 3 remnants. Companion strategies such as hand-weeding and the use of herbicides are almost certain to be needed (Little 2003, Willis et al. 2003). The post-fire environment may present opportunities to target weed species while in an active growth phase, and while they can easily be disentangled from natives. The extent to which fire, and other strategies along with fire, can play a role in reducing weediness in CPW would be an excellent subject for the adaptive management approach (Bradstock et al. 1995, Lunt and Morgan 1999, Whelan and Baker 1999, Gill et al. 2002, Keith et al. 2002b).
10.6 Fire interval domains for Castlereagh woodlands
While the grassy Cumberland Plain Woodland vegetation type was the primary focus of the project, information on regeneration mechanisms and juvenile periods of shrub species in the more shrubby Castlereagh woodlands were also collected. This information was used to inform application of the vital attributes model to Castlereagh woodlands (Chapter 3).
292 10.6.1 Lower thresholds
Petrophile pulchella was identified as the key fire response species amongst Castlereagh woodland taxa for frequent fire. Bradstock and O’Connell (1988) found that this serotinous obligate seeder first flowered at four years post-fire on sandstone, although time to maturity varied between sites. In the current study, plants were observed flowering just under three years post-fire at Castlereagh, although other populations of this post-fire age had not yet flowered. The figure of six years for the juvenile period of this species is an average of 3-9 years (the 9 year figure came from the range in the NSW Database of 4-9 years), and is probably conservative.
Petrophile pulchella occurs in sandy Castlereagh Scribbly Gum sites, but not on the finer-textured Castlereagh Ironbark Forest soils. No other species classified as highly sensitive to frequent fire had a juvenile period above five years.
Shale-Gravel Transition Forest hosts a subset of the shrub species found in Cumberland Plain and Castlereagh woodlands. Short-lived obligate seeder legumes are prominent. There is a rough gradient both in soil texture (clay to sand) and species complement from CPW, though Shale-Gravel Transition Forest and Castlereagh Ironbark Forest, to Castlereagh Scribbly Gum woodland. Keeping this in mind, the following minimum intervals are suggested:
• Castlereagh Scribbly Gum Woodland, 6 years; • Castlereagh Ironbark Forest, 5 years; • Shale-Gravel Transition Forest, 5 years; and, as already discussed, • State 2 Cumberland Plain Woodland, 4 years.
10.6.2 Upper thresholds
Application of the vital attributes model using longevity estimates from the NSW Database led to the recommendation that the maximum interval in Castlereagh woodlands should be no higher than 23 years. In CPW, expansion of Bursaria at the expense of open grassy areas and possibly obligate seeder shrubs led to recommendation of a somewhat lower top threshold than that suggested by vital attributes analysis alone. Are there reasons to think similar issues may also arise in Castlereagh woodlands?
293 No hard data on Castlereagh woodland species able to recruit between fires are available. Only three relatively common ‘T’ species were identified: Bursaria spinosa, Hakea sericea and Styphelia laeta ssp. laeta. Bursaria spinosa did not appear to act in the same manner in Castlereagh woodland as in CPW: no landscapes dominated by this species were encountered in the former vegetation type. Perhaps the existence of other large resprouting shrubs keeps it in check. Hakea sericea was prominent in some long- unburnt woodlands. The inclusion of some intervals just above lower thresholds in variable regimes would ensure this species did not become overwhelming. Styphelia laeta was placed in the T category because observations suggested recruitment between fires. This species is a relatively low-growing obligate seeder shrub which is unlikely to cause problems for other species.
It is possible that a number of other T species occur in Castlereagh woodlands. Persoonia nutans has already been mentioned (Section 3.4.7). Kunzea ambigua is another possibility: this Myrtaceous species dominated wetter depressions in some areas; its dead branches loomed over fields of flowering obligate seeder shrubs 2 years after fire (Figure 10.6). A number of Myrtaceous resprouters, including Melaleuca nodosa and Leptospermum spp. were prominent in areas which had had long intervals between fires. Kunzea, Leptospermum and Melaleuca species have demonstrated encroachment potential elsewhere (Judd 1990 cited in McMahon et al. 1996, Bennett 1994, Crowley and Garnett 1998), as has Casuarina littoralis (Withers and Ashton 1977, Lunt 1998a,b) which is also found in Castlereagh woodlands.
The potential for encroachment by a small number of shrub species in the absence of fire may be greater on the deep clay Castlereagh Ironbark Forest soils than in drier sandy areas (Section 10.8.3). Taking a precautionary approach in relation to possible competition issues, and considering soil productivity gradients, the following maximum intervals are proposed:
• Castlereagh Scribbly Gum Woodland, 20 years; • Castlereagh Ironbark Forest, 18 years; • Shale-Gravel Transition Forest, 15 years; and, as already discussed • State 2 Cumberland Plain Woodland, 12 years.
294
Figure 10.6. Kunzea ambigua skeletons above a field of flowering obligate seeder shrubs, particularly Pimelea linifolia ssp. linifolia and Dillwynia rudis, 23 months post-fire, at Castlereagh.
10.7 Managing fire on the urban fringe
This section addresses the question:
Do fire regimes compatible with bushland conservation in CPW overlap those needed to achieve protection from wildfire?
The answer to this question is basically “yes, to a large extent they do.” If CPW is managed for bushland conservation, fuel loads over much, if not all, of the landscape should remain below the levels considered hazardous from the point of view of property protection.
10.7.1 Intervals for bushland conservation
Fire frequencies for conservation of biodiversity in CPW are outlined in Section 10.3.2. Intervals between 4 and 12 years are predicted to maintain State 2: remnants in this state have high biodiversity values due to a patchy habitat structure, a good complement of obligate seeder shrubs and a Themeda-dominated understorey. Maintenance of the open grassy nature of State 1 remnants could be achieved with intervals between 2 and 4 years, although longer intervals should encourage greater shrub biodiversity.
295 Maintenance of State 3 would not require any fire, however experimentation with intervals between 1 and 5 years may increase habitat diversity, discourage weed invasion, and increase the proportion of Themeda in the understorey. Once this was achieved, State 2 thresholds would be predicted to maintain diversity.
Ideally, then, fire frequency in CPW remnants managed for plant and habitat diversity would involve fires at intervals between 4 and 12 years, with some shorter intervals either to maintain very open grassy habitat, or to open up areas currently occupied by thick Bursaria.
Within these thresholds, variability in both time and space is strongly recommended. Variability in interfire intervals is widely advocated by ecologists (Cowling and Gxaba 1990, Yeaton and Bond 1991, Bradstock et al. 1995, Morrison et al. 1995a, Keith 2002b). This is because even within a single vegetation type, fire-related vital attributes, and thus ability to persist within fire cycles, varies between species (Richardson et al. 1995, Hobbs 2002, Williams et al. 2003). For the full species complement to co-exist, some intervals which are less than ideal for dominant species may be important (Keith and Bradstock 1994, Tozer and Bradstock 2002). Field research indicates that plant species diversity is maximised through diversity in interfire intervals in sandy Sydney environments (Morrison et al. 1995a).
In large CPW remnants, the concept of variability could be expanded to include management practices compatible with maintenance of all three states on site. Given the greater conservation values of State 2 relative to States 1 and 3, it is suggested that while a majority of the landscape should be subject to intervals between 4 and 12 years, shorter intervals could be employed in limited areas, while other restricted parts of the landscape could be managed for denser vegetation through intervals over 12 years. Characteristics of local native fauna, particularly dispersal distances, could inform further thinking about the optimal distribution of states across a landscape.
10.7.2 Intervals for property protection
“The primary motivation for use of prescribed fire is to manipulate fuel structure and quantity to levels which ameliorate intensity of unplanned fires under severe fire weather” (Bradstock 1999:12). The acceptable hazard level in the sclerophyll vegetation of southern Australia is considered to be about 8 to 12 tonnes per hectare
296 (Walker 1981, Simmons and Adams 1986, Fensham 1992, Tolhurst 1996a). In Cumberland Plain Woodland, modelling indicates equilibrium fuel loads of around 9 t/ha. Thus if the aim is to maintain loads below 10, or below 12 t/ha, fuel reduction burning would appear to be unnecessary in this vegetation type – although, as Fensham (1992:315) cautions, figures based on fuel accumulation curves “can only be relied upon to provide management guidelines in a broad sense.”
The point at which a load of 8t/ha is likely to be reached may vary depending on the state of the vegetation. When State 3 remnants were excluded from modelling inputs, it was predicted that this level of fuel would accumulate by 6.1 years post-fire. However this figure dropped to 4.3 years post-fire when three data points from a very infrequently burnt, Bursaria-dominated area were included (Section 9.3.1). The difference almost certainly relates to the rapid build-up of shrub fuel as Bursaria resprouts (Section 9.4.3). Dead Bursaria stems also contribute to the fuel load in State 3 remnants in the months and years following a fire.
Thus intervals of six or fewer years should maintain fuel loads below even the most conservative hazard figure, where CPW is in State 1 or 2. Intervals below four years, however, may be needed in State 3 remnants where Bursaria is thick.
10.7.3 Managing for multiple aims
In State 1 and 2 remnants, then, there is an absolute overlap between fire frequencies compatible with biodiversity conservation and those needed to keep fuel below hazardous levels. A fire frequency of between four and six years should allow obligate seeder shrubs to persist, though perhaps at relatively low levels. Themeda dominance of the understorey would be maintained, Bursaria thickets would survive but be of limited extent, and ground layer forbs would have many opportunities for recruitment. Fuel loads should remain below 8 t/ha.
However better biodiversity outcomes are likely to be achieved if intervals between 6 and 12 years were also included in variable regimes. These intervals would favour shrubs, both Bursaria and obligate seeders. Habitat diversity would increase: Bursaria thickets would be larger, and areas with a time-since-fire of above six years would provide ‘old’ habitat for invertebrates and other fauna with a preference for later times-
297 since-fire (Fox and McKay 1981, Fox 1982, 1983, McFarland 1988b, Friend 1993, Wilson 1996, Woinarski 1999, York 1999).
One option in State 1 and 2 remnants would be to manage areas near built assets on the 4-6 year intervals that would keep fuel levels below 8 t/ha, and areas away from these assets on intervals between 4 and 12 years. Short intervals at remnant edges might have the additional advantage of providing a partial barrier to weed invasion: in this project, very frequently burnt areas were the least weedy (Section 10.2.4). Variable intervals in time and space in areas away from assets would not only have conservation benefits, but would ensure that fuel loads were broken up. Patches of ‘old’ fuel would be interspersed with patches which had been burnt more recently, making suppression in the event of a wildfire more likely to be effective (Raison et al. 1983). As Friend et al. (2003:2) point out, “All fires (and indeed absence of fires) have an ecological dimension, while many ‘ecological burns’ may also provide some asset protection.”
In remnants where the vegetation is in State 3, life and property issues may be somewhat more problematic. While CPW vegetation in this state is probably less flammable than open Themeda-dominated woodland (Section 10.3.3), when a fire does occur fuel loads may be higher, and Bursaria may ladder fire into the canopy (Section 9.4.3). Woody weeds such as olive and privet, which are more likely to be found in long unburnt State 3 than elsewhere (Chapter 4) may enhance fuel loads still further, as these species produce a litter layer that is significantly deeper and heavier than that produced by native species (J. Cooke, unpub. data). In addition, the low fire frequency compatible with maintenance of this state means equilibrium fuel loads will tend to prevail across most of the landscape. As noted above, when a fire does occur, hazardous levels may be achieved relatively rapidly, due to the contribution of dead and resprouting Bursaria.
Here is yet another reason for experimentation with Transition 6. A move from Bursaria-dominated woodland to State 2 woodland with its balance between shrubby and open areas would not only enhance biodiversity values and hopefully slow weed invasion, it would also provide more options for keeping fuel loads across the landscape in a state which would make fire suppression easier if a wildfire were to occur.
In shrubby woodlands on sandstone around Sydney, the conflict between burning frequencies compatible with biodiversity conservation, and those which keep fuel below
298 hazardous levels, is significant. Not only do the two regimes fail to overlap, minimum intervals for biodiversity conservation are years above maximum intervals for keeping fuel loads low, even where 10 or 12 t/ha is used as the benchmark (Morrison et al. 1996, Section 9.3.1). The situation in Western Sydney’s grassy woodlands is very different, and may be reflected in grassy woodlands elsewhere.
10.8 Implications for setting fire frequency thresholds
The findings of this project have implications for the goals and methods of determining fire frequency thresholds for vegetation types, particularly those with a grassy character.
10.8.1 Fire management goals
In NSW, it is widely accepted that the goal of ‘biodiversity conservation’, which fire frequency thresholds are designed to meet (NPWS 2004c), can be translated into the more measurable objective of the avoidance of extinction of all species currently present in a particular unit of interest, such as a vegetation type, a conservation reserve or a remnant (eg Bradstock 1999, Benson and Howell 2002). This definition focuses at the level of individual species, however ‘biodiversity’ also encompasses biotic communities and ecosystems (Noss and Cooperrider 1994, NPWS 2004b). Grassy ecosystems are characterised by their structure as well as by their species complement. The findings of this project suggest that the open grassy structure of some grassy woodland ecosystems may be at risk before individual plant species become locally extinct, although this outcome may follow for some species. It seems somewhat perverse to seek to maintain the species complement of a grassy woodland vegetation type, but to be unconcerned at the loss of its grassy character. In addition, changes in ecosystem structure may affect fauna species; in grassy woodlands, fauna which utilise open patches may no longer find the habitat they need. It is therefore recommended that maintenance of ecosystem structure be considered, along with maintenance of species complements, when fire frequency thresholds are determined.
299 10.8.2 Setting lower thresholds
Vital characteristics data relevant to setting lower thresholds – regeneration modes and juvenile periods – are reasonably easy to obtain. Local observations combined with data from the NSW Flora Fire Response Database (DEC 2002) were able to provide information for the large majority of shrub species in the vegetation types surveyed for this project. In addition, where data were available from both the Database and local observations, agreement was generally good (Chapter 3). Predictions based on vital attributes analysis with respect to the effects of short interfire intervals were borne out in the field (Chapter 4). The findings of this project support the conclusion that:
• The Vital Attributes model provides a sound basis for determining minimum interfire intervals; • Information on regeneration modes and juvenile periods in the NSW Database is accurate and useful for this purpose; • Local observation can contribute valuable input when determining minimum intervals.
10.8.3 Setting upper thresholds
Data relevant to setting upper thresholds – longevity of adults and seeds – are much less readily available. This project’s findings also suggest that the model and assumptions currently being used to set upper thresholds may require some modification.
One issue for upper thresholds concerns assumptions about longevity. In the NSW Flora Fire Response Database (DEC 2002), seeds with hard coats, including legumes, are assumed to live 30 years, while Bradstock and Kenny (2003) use a figure of 40 years. Longevity estimates for Cumberland Plain legumes (lifespan plus seedbank) in the NSW Database clustered around 50 to 60 years (Chapter 3). Subtracting the seedbank estimate of 30 years means adult plants were assumed to live between 20 and 30 years. However local observations of mortality in populations less than 10 years after fire, together with the reduced abundance of obligate seeder legumes in sites whose most recent interfire interval exceeds 20 years, suggests these estimates may be too high.
A second issue concerns the level at which upper thresholds are set vis-a-vis the longevity (y) of the key fire response species. As noted in Chapter 3, in Victoria
300 (Friend et al. 2003) and NSW (NPWS 2004c), the upper threshold is set at y (y includes seedbank longevity, if seeds persist beyond the life-span of adult plants). West Australians Burrows and Abbott (2003:446), however, suggest 0.75y may be more appropriate. Population decline, both above and below ground, may occur over a long period prior to the point of local extinction (Auld 1987). Flowering may peak in the years following the juvenile period: McFarland (1990) found flowering and seeding in south-east Queensland’s wallum heath peaked at four to eight years after a burn, and dropped markedly at 11 years post-fire. A species may therefore still occur in the landscape, but its fecundity might be greatly reduced in later post-fire years (Auld and Myerscough 1986). The need to consider decline rather than local extinction may be particularly important for species which play keystone roles in ecosystem function. Themeda australis may be such a species in some grassy woodland ecosystems, including CPW (Section 10.3.2).
Finally, the potential for one or a small number of species to dominate under extended interfire intervals, and concomitant interspecific interactions, may mean that upper thresholds need to be lower than an assessment based solely on life history characteristics would suggest. Huston (2003) considers competitive exclusion to be the issue for biodiversity conservation under a low disturbance frequency.
The need to consider the effect of dominant shrubs on other species has previously been recognised in Sydney’s heathlands (Keith 2002b, Tozer and Bradstock 2002). When life history characteristics are considered, a feasible fire frequency for the conservation of both dominant obligate seeder shrubs and understorey species appears to be 15-30 years. However at this fire frequency, the dominant species form high-density thickets which reduce the survival and fecundity of some species in the understorey. An understanding of this dynamic has highlighted the need to include in fire regimes some intervals only slightly above the juvenile period of the dominant species, thus reducing overstorey density for a period sufficient to allow understorey species to build up population numbers before again being overshadowed. This need is reflected in fire frequency recommendations in Bradstock et al. (1995), but is not explicitly discussed in the recently-developed fire interval guidelines for NSW, although the importance of variability within thresholds is noted (NPWS 2004c).
301 In Cumberland Plain Woodland, Bursaria spinosa also affects co-occurring species (Chapters 5 and 8). However unlike the heathland dominants, Bursaria resprouts after a fire, and thus will continue to exert competitive pressure by drawing on soil resources, and once its cover is re-established, on light resources. Thus the strategy recommended to provide relief for competitively inferior species in heathlands – inserting one short interval amongst longer ones – is unlikely to work in CPW. The evidence suggests that moderately frequent fires will be needed to allow open patch biota, Bursaria thickets and obligate seeder shrubs to co-exist long-term. A high maximum threshold, particularly if coupled with a recommendation that ecological burning be limited to sites where maximum thresholds have been exceeded, is likely to lead to expansion of Bursaria and increasing capture of resources by this species, to the detriment of other shrubs, open patch forbs, and fauna which forage in open areas.
The potential for Bursaria to dominate in the shrub layer could be predicted once its ability to recruit between fires was recognised. The vital attributes model explicit identifies these species (T and R species), and their propensity to dominate in the absence of disturbance (Noble and Slatyer 1980).
It is therefore recommended that:
• More emphasis be placed on determining the rate of decline with time-since-fire of adult populations and fecundity, in species which recruit only post-fire (I species);
• More emphasis be placed on identifying species able to build up population numbers in the interfire interval, particularly those able to capture large slices of ecosystem resources through this strategy (some T and R species will be more able to capture resources than others). Effects of these species on the next generation of I species also need to be identified.
• These factors be explicitly considered when upper thresholds are set.
• That as an interim rule of thumb, upper thresholds not exceed 0.75y, where y is the longevity (including seedbank) of the key fire response species for infrequent fire.
302 10.9 Relevance of CPW recommendations for grassy woodlands elsewhere
Grassy woodlands occur in a wide range of climatic conditions, and on various substrates. Fire frequencies appropriate for Cumberland Plain Woodland are likely to be most relevant to woodlands growing under similar conditions.
Productivity may hold the key to determining how frequently fire should occur to maintain woodland plant diversity. Huston (1979, 2003) addresses the interaction of productivity and disturbance in mediating species diversity in his dynamic equilibrium model. Where productivity is low, a high disturbance frequency is predicted to reduce diversity, as organisms will be unable to recover between disturbances. Diversity is also predicted to decline in highly productive systems where disturbance frequency is low, due to competitive exclusion. Additional reasons why a high disturbance frequency is likely to be more appropriate in more productive areas can also be advanced.
Productivity in grassy woodlands will increase with:
• Rainfall. At around 800 mm per annum, rainfall on the Cumberland Plain is at the high end of the gradient for Australian temperate grassy woodlands.
• Temperature. Mean temperatures on the Cumberland Plain are lower than those in coastal woodlands to the north of the state, but higher than those in the south and on the Tablelands where altitude contributes to severe frosts (Reseigh et al. 2003).
• Season of rainfall. Where rainfall and warm temperatures coincide, there is a greater potential for plant growth. Rainfall on the Cumberland Plain falls predominantly in summer, as it does over most of northern NSW. The proportion of precipitation falling in winter increases with latitude.
• Soil fertility. The shale-based clays which support Cumberland Plain Woodland would be less productive than the basalt-based soils of the Darling Downs and parts of the Tablelands. However they may be higher in nutrients than the more sandy soils which support grassy woodland elsewhere.
These factors are likely to be associated with extent of grass production, and thus with potential for competitive exclusion of interstitial forbs (Stuwe 1994). Grazing is associated with increased species richness in productive grassy systems, but not in unproductive ones (Proulx and Mazumder 1998); fire, the “large generalist herbivore”
303 (Bond 1997) may operate similarly. The tendency for dominant clump grasses such as Themeda to ‘crash’ due to litter inhibition may also be greater, or occur more rapidly, in more productive environments (Tilman and Wedin 1991). Frequent firing may thus be more appropriate in productive systems than in those whose productivity is limited by poor soils, low rainfall or a short growing season (Huston 2004).
A second reason why shorter interfire intervals may be appropriate in more productive systems is because shrubs may reach life history milestones more rapidly. Juvenile periods of obligate seeder shrubs in the vital attributes study reported in Chapter 3 were often shorter where resources were more readily available. On the New England Tablelands, where the growing season is constrained by severe frosts, shrub juvenile periods can be several years longer than those of the same species in coastal areas (Knox and Clarke 2004). Senescence, and/or overtopping of low growing shrub species, may also occur more rapidly in more productive areas (Specht and Specht 1989).
Encroachment and dominance by one or a small number of shrub species in the absence of fire may be more likely in more productive communities, providing a further rationale for more frequent burning in these areas. Roques et al. (2001) compared their findings of rapid shrub encroachment in fertile, mesic savanna in Swaziland with findings in similar studies, and concluded encroachment occurs more rapidly in more productive areas. Melaleuca viridiflora encroachment on Cape York Peninsula was higher in seasonally inundated grasslands than in nearby, dry eucalypt woodlands (Crowley and Garnett 1998); a similar pattern was found for Melaleuca minutifolia in the Northern Territory (Sharp and Bowman 2004). Woody plant encroachment is most problematic in fertile ‘run on’ areas in semi-arid woodlands (Noble and Grice 2002).
A final argument for more frequent fire in more productive areas is purely pragmatic. Fuel accumulation will be faster in more productive woodlands, and thus the ability to support frequent fire will be higher. In southern Africa, fire frequency follows a rainfall gradient. The mean interfire interval in high rainfall sourveld is three years, while semi- arid sweetveld burns at a mean frequency of eight years. Herbivory plays an increasingly important role in ecosystem dynamics as rainfall decreases. In sourveld, lack of fire is the primary factor in shrub invasion, while in more arid areas grazing is the principal trigger (Bond 1997).
304 Overall, it is suggested that the relevance of the findings of this project to grassy woodlands beyond the Cumberland Plain will depend on the effects of climate and soil fertility on productivity. Related factors for consideration include the resulting potential for competitive exclusion by dominant grasses and/or shrubs, growth rates of shrubs (particularly obligate seeders), and how soon, and often, the vegetation is able to support fire. The potential, need and tolerance for fire will, I predict, decrease with productivity.
10.10 Conclusion
“Living with fire requires cultural shifts. This takes time, requires trial and error and the development of new ways of thinking, new stories and novel approaches.” (Bowman 2003:115-6).
Fire frequency has a profound influence on the grassy woodlands of Western Sydney’s Cumberland Plain. Structure, composition, habitat diversity and fuel loads are all affected. Now isolated in remnants, maintenance of biodiversity in Cumberland Plain Woodland will depend on active fire management. The story of fire in this vegetation type illustrates the importance of disturbance in mediating species co-existence; it is a non-equilibrium paradigm story. Ultimately, conservation of Sydney’s grassy woodlands requires that human beings understand and appreciate fire’s role, and act accordingly.
Similar stories no doubt await telling, and hearing, in grassy woodlands elsewhere.
305 REFERENCES
Adams, R. and Simmons, D. (1996) The impact of fire intensity on litter loads and understorey floristics in an urban fringe dry sclerophyll forest and implications for management. Pp. 21- 35 in Fire and Biodiversity: the Effects and Effectiveness of Fire Management (ed. by J.R. Merrick). Proceedings of conference held in Melbourne on 8-9 October 1994, Commonwealth Department of Environment, Sport and Territories, Canberra.
Andersen, A.N. (1988) Immediate and longer-term effects of fire on seed predation by ants in sclerophyllous vegetation in south-eastern Australia. Australian Journal of Ecology 13, 285-293.
Andersen, A.N., Cook, G.D. and Williams, R.J. (2003) Burning for conservation in Australian tropical savannas: implications from the Kapalga fire experiment. Paper presented at the Third International Wildland Fire Conference and Exhibition, 3-6 October 2003, Sydney, Australia. Published on CD by Third International Wildland Fire Conference.
Archer, S. (1989) Have southern Texas savannas been converted to woodlands in recent history? American Naturalist 134, 545-561.
Archer, S. (1990) Development and stability of grass/woody mosaics in a subtropical savanna parkland, Texas, U.S.A. Journal of Biogeography 17, 453-462.
Archer, S., Scifres, C., Bassham, C.R. and Maggio, R. (1988) Autogenic succession in a subtropical savanna: conversion of grassland to thorn woodland. Ecological Monographs 58, 111-127.
Ashton, D.H. (1976) The vegetation of Mount Piper, Central Victoria: a study of a continuum. Journal of Ecology 64, 463-483.
Ashton, D.H. (1979) Seed harvesting by ants in forests of Eucalyptus regnans F. Muell. in central Victoria. Australian Journal of Ecology 4, 265-277.
Asquith, J. and Messer, J. (1998) Guidelines for Bush Fire Risk Management Planning: a Handbook. National Conservation Council of NSW, Sydney.
Auld, T.D. (1986) Population dynamics of the shrub Acacia suaveolens (Sm.) Willd.: dispersal and the dynamics of the soil seed-bank. Australian Journal of Ecology 11, 235-254.
Auld, T.D. (1987) Population dynamics of the shrub Acacia suaveolens (Sm.) Willd.: survivorship throughout the life cycle, a synthesis. Australian Journal of Ecology 12, 139- 151.
Auld, T.D., Keith, D.A. and Bradstock, R.A. (2000) Patterns in longevity of soil seedbanks in fire-prone communities of south-eastern Australia. Australian Journal of Botany 48, 539- 548.
Auld, T.D. and Myerscough, P.J. (1986) Population dynamics of the shrub Acacia suaveolens (Sm.) Willd.: seed production and predispersal seed predation. Australian Journal of Ecology 11, 219-234.
307 Auld, T.D. and O'Connell, M.A. (1991) Predicting patterns of post-fire germination in 35 eastern Australian Fabaceae. Australian Journal of Ecology 16, 53-70.
Australian National Botanic Gardens (2003) Fire, gardens and fire retardant plants - a bibliography. On line at http://www.anbg.gov.au/bibliography/fire-plants.html. Last amended 21 March 2003. Accessed 28 April 2005.
Baird, A.M. (1977) Regeneration after fire in King's Park, Perth, Western Australia. Journal of the Royal Society of Western Australia 60, 1-22.
Bannerman, S.M. and Hazelton, P.A. (1990) Soil Landscapes of the Penrith 1:100 000 Sheet. Soil Conservation Service of NSW, Sydney.
Barlow, T. (1998) Grassy Guidelines: How to Manage Native Grasslands and Grassy Woodlands on your Property. Trust for Nature, Melbourne.
Baxter, B.J.M., van Staden, J., Granger, J.E. and Brown, N.A.C. (1994) Plant-derived smoke and smoke extracts stimulate seed germination of the fire-climax grass Themeda triandra. Environmental and Experimental Botany 34, 217-223.
Beadle, N.C.W. (1962) Soil phosphate and the delimitation of plant communities in eastern Australia II. Ecology 43, 281-288.
Beadle, N.C.W. (1981) The Vegetation of Australia. Cambridge University Press, Cambridge.
Bear, V. (1997) AABR Lecture Series - Jim Kohen. Aboriginal People of the Sydney Region. Precis of paper presented to the Australian Association of Bush Regenerators (NSW). On line at http://www.zipworld.com.au/~aabr/info/seminars/seminar06.html. Accessed 26 October 2004.
Bell, D.T. (2001) Ecological response syndromes in the flora of southwestern Western Australia: fire resprouters versus reseeders. Botanical Review 67, 417-440.
Bell, D.T. and Koch, J.M. (1980) Post-fire succession in the northern jarrah forest of Western Australia. Australian Journal of Ecology 5, 9-14.
Bell, D.T., Plummer, J.A. and Taylor, S.K. (1993) Seed germination ecology in southwestern Western Australia. Botanical Review 59, 24-73.
Bell, D.T., Vlahos, S. and Watson, L.E. (1987) Stimulation of seed germination of understorey species of the northern Jarrah forest of Western Australia. Australian Journal of Botany 35, 593-599.
Bell, T.L. and Pate, J.S. (1996) Growth and fire response of selected Epacridaceae of south- western Australia. Australian Journal of Botany 44, 509-526.
Bellingham, D. (2003) Elders of the Plains - interview with Roger Oxley. On line at http://plainstalking.deni.net.au/eldersintroxley.html. Accessed 20 July 2004.
Belsky, A.J. (1992) Effects of grazing, competition, disturbance and fire on species composition and diversity in grassland communities. Journal of Vegetation Science 3, 187- 200.
Belsky, A.J., Mwonga, S.M., Amundson, R.G., Duxbury, J.M. and Ali, A.R. (1993) Comparative effects of isolated trees on their undercanopy environments in high- and low- rainfall savannas. Journal of Applied Ecology 30, 143-155.
308 Bennett, L.T. (1994) The expansion of Leptospermum laevigatum on the Yanakie Isthmus, Wilson's Promontory, under changes in the burning and grazing regimes. Australian Journal of Botany 42, 555-564.
Benson, D.H. (1985) Maturation periods for fire-sensitive shrub species in Hawkesbury sandstone vegetation. Cunninghamia 1, 339-349.
Benson, D.H. (1992) The natural vegetation of the Penrith 1:100 000 map sheet. Cunninghamia 2, 541-596.
Benson, D. and Howell, J. (1990) Taken for Granted: The Bushland of Sydney and its Suburbs. Kangaroo Press, Sydney.
Benson, D. and Howell, J. (2002) Cumberland Plain Woodland ecology then and now: interpretations and implications from the work of Robert Brown and others. Cunninghamia 7, 631-650.
Benson, D., Howell, J. and McDougall, L. (1996) Mountain Devil to Mangrove: a Guide to Natural Vegetation in the Hawkesbury-Nepean Catchment. Royal Botanic Gardens, Sydney.
Benson, J.S. and Redpath, P.A. (1997) The nature of pre-European native vegetation in south- eastern Australia: a critique of Ryan, D.G., Ryan, J.R. and Starr, B.J. (1995) The Australian Landscape - Observations of Explorers and Early Settlers. Cunninghamia 5, 285-328.
Benwell, A.S. (1998) Post-fire seedling recruitment in coastal heathland in relation to regeneration strategy and habitat. Australian Journal of Botany 46, 75-101.
Berry, S.L. and Roderick, M.L. (2002) CO2 and land-use effects on Australian vegetation over the last two centuries. Australian Journal of Botany 50, 511-531.
Birk, E.M. and Bridges, R.G. (1989) Recurrent fires and fuel accumulation in even-aged Blackbutt (Eucalyptus pilularis) forests. Forest Ecology and Management 29, 59-79.
Blank, S., Seiter, C. and Bruce, P. (2001) Resampling Stats in Excel (Version 2). Resampling Stats Inc., Arlington, Virginia.
Bond, W.J. (1980) Fire and senescent fynbos in the Swartberg, Southern Cape. South African Forestry Journal 114, 68-71.
Bond, W.J. (1997) Fire. Pp. 421-446 in Vegetation of Southern Africa (ed. by R.M. Cowling, D.M. Richardson and S.M. Pierce). Cambridge University Press, Cambridge.
Bond, W.J. and Archibald, S. (2003) Confronting complexity: fire policy choices in South African savanna parks. International Journal of Wildland Fire 12, 381-389.
Bond, W.J. and Ladd, P.G. (2001) Dynamics of the overstorey and species richness in Australian heathlands. Journal of Mediterranean Ecology 2, 247-257.
Bond, W.J., Midgley, G.F. and Woodward, F.I. (2003) The importance of low atmospheric CO2 and fire in promoting the spread of grasslands and savannas. Global Change Biology 9, 973-982.
Bond, W.J. and van Wilgen, B.W. (1996) Fire and Plants. Chapman and Hall, London.
309 Bond, W.J., Woodward, F.I. and Midgley, G.F. (2005) The global distribution of ecosystems in a world without fire. New Phytologist 165, 525-538.
Bowen, B.J. and Pate, J.S. (2004) Effect of season of burn on shoot recovery and post-fire flowering performance in the resprouter Stirlingia latifolia R. Br. (Proteaceae). Austral Ecology 29, 145-155.
Bowman, D.M.J.S. (1998) Tansley Review No. 101. The impact of Aboriginal landscape burning on the Australian biota. New Phytologist 140, 385-410.
Bowman, D.M.J.S. (2003) Australian landscape burning: a continental and evolutionary perspective. Pp. 107-118 in Fire in Ecosystems of South-west Western Australia (ed. by I. Abbott and N. Burrows). Backhuys Publishers, Leiden, The Netherlands.
Bowman, D.M.J.S., Wilson, B.A. and Hooper, R.J. (1988) Response of Eucalyptus forest and woodland to four fire regimes at Munmarlary, Northern Territory, Australia. Journal of Ecology 76, 215-232.
Bradfield, G.E. (1981) Component analysis of fire patterns in open eucalypt forest. Australian Journal of Ecology 6, 99-109.
Bradstock, R.A. (1990) Demography of woody plants in relation to fire: Banksia serrata Lf. and Isopogon anemonifolius (Salisb.) Knight. Australian Journal of Ecology 15, 117-132.
Bradstock, R.A. (1995) Demography of woody plants in relation to fire: Telopea speciosissima. Proceedings of the Linnean Society of New South Wales 115, 25-33.
Bradstock, R.A. (1999) 'Thresholds' for biodiversity: the National Parks & Wildlife Service approach to planning of fire management for conservation. Pp. 11-18 in Bush Fire Management Conference Proceedings: Protecting the Environment, Land, Life and Property (ed. by F. Sutton, J. Keats, J. Dowling and C. Doig). Proceedings of conference held in Sydney on 27-28 February 1998, Nature Conservation Council of NSW, Sydney.
Bradstock, R.A. and Gill, A.M. (2001) Living with fire and biodiversity at the urban edge: in search of a sustainable solution to the human protection problem in southern Australia. Journal of Mediterranean Ecology 2, 179-195.
Bradstock, R.A., Keith, D.A. and Auld, T.D. (1995) Fire and conservation: imperatives and constraints on managing for diversity. Pp. 323-333 in Conserving Biodiversity: Threats and Solutions (ed. by R.A. Bradstock, T.D. Auld, D.A. Keith, R.T. Kingsford, D. Lunney and D.P. Sivertsen). Surrey Beatty and Sons, Chipping Norton, NSW.
Bradstock, R.A. and Kenny, B.J. (2003) An application of plant functional types to fire management in a conservation reserve in southeastern Australia. Journal of Vegetation Science 14, 345-354.
Bradstock, R.A. and Myerscough, P.J. (1988) The survival and population response to frequent fires of two woody resprouters Banksia serrata and Isopogon anemonifolius. Australian Journal of Botany 36, 415-431.
Bradstock, R.A. and O'Connell, M.A. (1988) Demography of woody plants in relation to fire: Banksia ericifolia L.f. and Petrophile pulchella (Schrad) R.Br. Australian Journal of Ecology 13, 505-518.
310 Bradstock, R.A., Tozer, M.G. and Keith, D.A. (1997) Effects of high frequency fire on floristic composition and abundance in a fire-prone heathland near Sydney. Australian Journal of Botany 45, 641-655.
Bradstock, R.A., Williams, J.E. and Gill, A.M. (eds.) (2002) Flammable Australia: the Fire Regimes and Biodiversity of a Continent. Cambridge University Press, Cambridge.
Bray, J.R. and Curtis, J.T. (1957) An ordination of the upland forest communities of Southern Wisconsin. Ecological Monographs 27, 325-349.
Bureau of Meteorology (1991) Climatic Survey, Sydney, New South Wales. Australian Government Publishing Service, Canberra.
Bureau of Meteorology (2004 ) What is the weather usually like? Climate averages for New South Wales and ACT sites. On line at http://www.bom.gov.au/climate/averages/tables/ca_nsw_names.shtml. Accessed 19 January 2005.
Burrows, N. and Abbott, I. (2003) Fire in south-west Western Australia: synthesis of current knowledge, management implications and new research directions. Pp. 437-452 in Fire in Ecosystems of South-west Western Australia (ed. by I. Abbott and N. Burrows). Backhuys Publishers, Leiden, The Netherlands.
Burrows, N.D., Ward, B. and Robinson, A.D. (1999) The role of indicators in developing appropriate fire regimes. Paper presented at the Australian Bushfire Conference, Albury, July 1999. On line at http://life.csu.edu.au/bushfire99/papers/burrows/index.html. Accessed 3 June 2005.
Burrows, N. and Wardell-Johnson, G. (2003) Fire and plant interactions in forested ecosystems of south-west Western Australia. Pp. 225-268 in Fire in Ecosystems of South-west Western Australia (ed. by I. Abbott and N. Burrows). Backhuys Publishers, Leiden, The Netherlands.
Byth, K. (1982) On robust distance-based intensity estimators. Biometrics 38, 127-135.
Caldwell, M.M. and Richards, J.H. (1989) Hydraulic lift: water efflux from upper roots improves effectiveness of water uptake by deep roots. Oecologia 79, 1-5.
CALM (Department of Conservation and Land Management NSW) (1993 ) Draft Assessment of Crown Land at Scheyville. Sydney District Office, Department of Conservation and Land Management, Blacktown, NSW.
Carpenter, S.R. (1996) Microcosm experiments have limited relevance for community and ecosystem ecology. Ecology 77, 677-680.
Cary, G.J. and Morrison, D.A. (1995) Effects of fire frequency on plant species composition of sandstone communities in the Sydney region: combinations of inter-fire intervals. Australian Journal of Ecology 20, 418-426.
Catling, P.C. and Burt, R.J. (1995) Studies of the ground-dwelling mammals of eucalypt forests in south-eastern New South Wales: the effect of habitat variables on distribution and abundance. Wildlife Research 22, 271-288.
311 Chapman, A. and Harrington, G.N. (1997) Responses by birds to fire regime and vegetation at the wet sclerophyll/tropical rainforest boundary. Pacific Conservation Biology 3, 213-220.
Cheney, P. (1996) The effectiveness of fuel reduction burning for fire management. Pp. 9-16 in Fire and Biodiversity: the Effects and Effectiveness of Fire Management (ed. by J.R. Merrick). Proceedings of conference held in Melbourne on 8-9 October 1994, Commonwealth Department of Environment, Sport and Territories, Canberra.
Cheney, P. (2003) Effectiveness of prescribed burning. Proceedings of the NSW Farmers' Association Bushfire Forum, Sydney, 12 November 2003.
Chilcott, C., Reid, N.C.H. and King, K. (1997) Impact of trees on the diversity of pasture species and soil biota in grazed landscapes on the Northern Tablelands, NSW. Pp. 378-386 in Conservation Outside Nature Reserves (ed. by P. Hale and D. Lamb). Centre for Conservation Biology, University of Queensland, Brisbane.
Christensen, P.E. and Burrows, N.D. (1986) Fire: an old tool with a new use. Pp. 97-105 in Ecology of Biological Invasions: an Australian Perspective (ed. by R.H. Groves and J.J. Burdon). Australian Academy of Science, Canberra.
Christensen, P.E. and Kimber, P.C. (1975) Effect of prescribed burning on the flora and fauna of south-west Australian forests. Proceedings of the Ecological Society of Australia 9, 85- 106.
Clark, S.S. (1988) Effects of hazard-reduction burning on populations of understorey plant species on Hawkesbury sandstone. Australian Journal of Ecology 13, 473-484.
Clark, S.S. and McLoughlin, L.C. (1986) Historical and biological evidence for fire regimes in the Sydney region prior to the arrival of Europeans: implications for future bushland management. Australian Geographer 17, 101-112.
Clarke, K.R. and Gorley, R.N. (2001) PRIMER V5: User Manual/Tutorial. PRIMER-E Ltd., Plymouth.
Clarke, K.R. and Warwick, R.M. (2001) Change in Marine Communities: An Approach to Statistical Analysis and Interpretation. 2nd edition. PRIMER-E Ltd., Plymouth.
Clarke, P.J. (2000) Plant population processes in temperate woodlands of eastern Australia - premises for management. Pp. 248-270 in Temperate Eucalypt Woodlands in Australia: Biology, Conservation, Management and Restoration (ed. by R.J. Hobbs and C.J. Yates). Surrey Beatty and Sons, Chipping Norton, NSW.
Clarke, P.J. (2002) Experiments on tree and shrub establishment in temperate grassy woodlands: seedling survival. Austral Ecology 27, 606-615.
Clarke, P.J. (2003) Composition of grazed and cleared temperate grassy woodlands in eastern Australia: patterns in space and inferences in time. Journal of Vegetation Science 14, 5-14.
Clarke, P.J. and Davison, E.A. (2001) Experiments on the mechanism of tree and shrub establishment in temperate grassy woodlands: seedling emergence. Austral Ecology 26, 400-412.
Clarke, P.J. and Davison, E.A. (2004) Emergence and survival of herbaceous seedlings in temperate grassy woodlands: recruitment limitations and regeneration niche. Austral Ecology 29, 320-331.
312 Clarke, P.J., Davison, E.A. and Fulloon, L. (2000) Germination and dormancy of grassy woodland and forest species: effects of smoke, heat, darkness and cold. Australian Journal of Botany 48, 687-700.
Clarke, P.J. and Knox, K.J.E. (2002) Post-fire responses of shrubs in the tablelands of eastern Australia: do existing models explain habitat differences? Australian Journal of Botany 50, 53-62.
Clarke, S. (2003) The Effect of Fire and Grazing on Cumberland Plain Woodlands. Master of Science (Research) thesis, University of Wollongong.
Clements, F.E. (1916) Plant Succession: an Analysis of the Development of Vegetation. Publication No. 242, Carnegie Institute, Washington D.C.
Connell, J.H. (1978) Diversity in tropical rain forests and coral reefs. Science 199, 1302-1310.
Connell, J.H. and Slatyer, R.O. (1977) Mechanisms of succession in natural communities and their role in community stability and organization. American Naturalist 111, 1119-1144.
Coops, N.C. and Catling, P.C. (2000) Estimating forest habitat complexity in relation to time since fire. Austral Ecology 25, 344-351.
Costello, D.A., Lunt, I.D. and Williams, J.E. (2000) Effects of invasion by the indigenous shrub Acacia sophorae on plant composition of coastal grasslands in south-eastern Australia. Biological Conservation 96, 113-121.
Cowling, R.M. and Gxaba, T. (1990) Effects of a fynbos overstorey shrub on understorey community structure: implications for the maintenance of community-wide species richness. South African Journal of Ecology 1, 1-7.
Cowling, R.M., Lamont, B.B. and Enright, N.J. (1990) Fire and management of south-western Australian banksias. Proceedings of the Ecological Society of Australia 16, 177-183.
Crowley, G.M. and Garnett, S. (1998) Vegetation change in the grasslands and grassy woodlands of east-central Cape York Peninsula, Australia. Pacific Conservation Biology 4, 132-148.
Curtis, D. (1990) Natural regeneration of eucalypts in the New England region. Pp. 7-16 in 'Sowing the Seeds': Direct Seeding and Natural Regeneration Conference . Greening Australia, Canberra.
Dawson, T.E. (1993) Hydraulic lift and water use by plants: implications for water balance, performance and plant-plant interactions. Oecologia 95, 565-574. de Bruin, J. (2004) Time Course of Fuel Accumulation in Cumberland Plain Woodland. Paper prepared for Study Abroad program, University of Western Sydney.
DEC (Department of Environment and Conservation NSW) (2002) NSW Flora Fire Response Database. Version 1.3a. Department of Environment and Conservation, Hurstville, NSW.
Denham, A.J. and Whelan, R.J. (2000) Reproductive ecology and breeding system of Lomatia silaifolia (Proteaceae) following a fire. Australian Journal of Botany 48 , 261-269.
Diamond, J. (1986) Overview: laboratory experiments, field experiments, and natural experiments. Pp. 3-22 in Community Ecology (ed. by J. Diamond and T.J. Case). Harper and Row, New York.
313 Dixon, K.W., Roche, S. and Pate, J.S. (1995) The promotive effect of smoke derived from burnt native vegetation on seed germination of Western Australian plants. Oecologia 101, 185-192.
Dixon, P.M. (2001) Nearest neighbor methods. Department of Statistics, Iowa State University. On line at www.stat.iastate.edu/preprint/articles/2001-19.pdf. Accessed 30 January 2003.
Engle, D.M., Stritzke, J.F. and Claypool, P.L. (1987) Herbage standing crop around eastern redcedar trees. Journal of Range Management 40, 237-239.
Enright, N.J. and Goldblum, D. (1999) Demography of a non-sprouting and resprouting Hakea species (Proteaceae) in fire-prone Eucalyptus woodlands of southeastern Australia in relation to stand age, drought and disease. Plant Ecology 144, 71-82.
Enright, N.J., Marsula, R., Lamont, B.B. and Wissel, C. (1998) The ecological significance of canopy seed storage in fire-prone environments: a model for non-sprouting shrubs. Journal of Ecology 86, 946-959.
Facelli, J.M. and Temby, A.M. (2002) Multiple effects of shrubs on annual plant communities in arid lands of South Australia. Austral Ecology 27, 422-432.
Faith, D.P., Minchin, P.R. and Belbin, L. (1987) Compositional dissimilarity as a robust measure of ecological distance: a theoretical model and computer simulations. Vegetatio 69, 57-68.
Fensham, R.J. (1990) Interactive effects of fire frequency and site factors in tropical Eucalyptus forest. Australian Journal of Ecology 15, 255-266.
Fensham, R.J. (1992) The management implications of fine fuel dynamics in bushlands surrounding Hobart, Tasmania. Journal of Environmental Management 36, 301-320.
Fensham, R.J. (1998) The grassy vegetation of the Darling Downs, south-eastern Queensland, Australia. Floristics and grazing effects. Biological Conservation 84, 301-310.
Fensham, R.J. and Butler, D.W. (2004) Spatial pattern of dry rainforest colonizing unburnt Eucalyptus savanna. Austral Ecology 29, 121-128.
Fensham, R.J. and Fairfax, R.J. (1996) The disappearing grassy balds of the Bunya Mountains, south-eastern Queensland. Australian Journal of Botany 44, 543-558.
Flematti, G.R., Ghisalberti, E.L., Dixon, K.W. and Trengove, R.D. (2004) A compound from smoke that promotes seed germination. Science 305, 977.
Florence, R.G. (1996) Ecology and Silviculture of Eucalypt Forests. CSIRO Publishing, Collingwood.
Floyd, A.G. (1966) Effect of fire upon weed seeds in the wet sclerophyll forests of Northern New South Wales. Australian Journal of Botany 14, 243-256.
Floyd, A.G. (1976) Effect of burning on regeneration from seeds in wet sclerophyll forest. Australian Forestry 39, 210-220.
Fox, B.J. (1982) Fire and mammalian secondary succession in an Australian coastal heath. Ecology 63, 1332-1341.
314 Fox, B.J. (1983) Mammal species diversity in Australian heathlands: the importance of pyric succession and habitat diversity. Pp. 473-489 in Mediterranean-Type Ecosystems: the Role of Nutrients ( ed. by F.J. Kruger, D.T. Mitchell and J.U.M. Jarvis). Springer-Verlag, Berlin.
Fox, B.J., Fox, M.D. and McKay, G.M. (1979) Litter accumulation after fire in a eucalypt forest. Australian Journal of Botany 27, 157-165.
Fox, B.J. and McKay, G.M. (1981) Small mammal responses to pyric successional changes in eucalypt forest. Australian Journal of Ecology 6, 29-41.
Fox, M.D. and Fox, B.J. (1986) The effect of fire frequency on the structure and floristic composition of a woodland understorey. Australian Journal of Ecology 11, 77-85.
French, K., Callaghan, B. and Hill, S. (2000) Classifying endangered vegetation communities: a case study of Cumberland Plain Woodlands. Pacific Conservation Biology 6, 120-129.
French, K. and Westoby, M. (1996) Vertebrate-dispersed species in a fire-prone environment. Australian Journal of Ecology 21, 379-385.
Friend, G.R. (1993) Impact of fire on small vertebrates in mallee woodlands and heathlands of temperate Australia: a review. Biological Conservation 65, 99-114.
Friend, G.R., Leonard, M., Troy, S., Tolhurst, K.G. and Wouters, M. (2003) Fire and biodiversity management in Victoria - integrating the science, planning and implementation processes. Paper presented at the Third International Wildland Fire Conference and Exhibition, 3-6 October 2003, Sydney, Australia. Published on CD by Third International Wildland Fire Conference.
Gibbs, L., Reid, N. and Whalley, R.D.B. (1999) Relationships between tree cover and grass dominance in a grazed temperate stringybark (Eucalyptus laevopinea) open-forest. Australian Journal of Botany 47, 49-60.
Gill, A.M. (1981) Adaptive responses of Australian vascular plant species to fires. Pp. 243- 272 in Fire and the Australian Biota (ed. by A.M. Gill, R.H. Groves and I.R. Noble). Australian Academy of Science, Canberra.
Gill, A.M. (1997) Eucalypts and fires: interdependent or independent? Pp. 151-167 in Eucalypt Ecology: Individuals to Ecosystems (ed. by J. Williams and J. Woinarski). Cambridge University Press, Cambridge.
Gill, A.M. and Bradstock, R.A. (1992) A national register for the fire responses of plant species. Cunninghamia 2, 653-660.
Gill, A.M. and Bradstock, R. (1995) Extinction of biota by fires. Pp. 309-322 in Conserving Biodiversity: Threats and Solutions (ed. by R.A. Bradstock, T.D. Auld, D.A. Keith, R.T. Kingsford, D. Lunney and D.P. Sivertsen). Surrey Beatty and Sons, Chipping Norton.
Gill, A.M., Bradstock, R.A. and Williams, J.E. (2002) Fire regimes and biodiversity: legacy and vision. Pp. 429-446 in Flammable Australia: the Fire Regimes and Biodiversity of a Continent (ed. by R.A. Bradstock, J.E. Williams and A.M. Gill). Cambridge University Press, Cambridge.
Gill, A.M., Christian, K., Moore, P. and Forrester, R. (1987) Bushfire incidence, fire hazard and fuel reduction burning. Australian Journal of Ecology 12, 299-306.
315 Gill, A.M. and Nicholls, A.O. (1989) Monitoring fire-prone flora in reserves for nature conservation. Pp. 137-151 in WA Department of Conservation and Land Management Occasional Paper 1/89. CALM, Perth.
Gillison, A.N. (1994) Woodlands. Pp. 227-255 in Australian Vegetation (2nd edition) (ed. by R.H. Groves). Cambridge University Press, Cambridge.
Gleadow, R.M. and Ashton, D.H. (1981) Invasion by Pittosporum undulatum of the forests of central Victoria. I. Invasion patterns and plant morphology. Australian Journal of Botany 29, 705-720.
Gott, B. (1991) Victorian Koorie Plants: Some Plants Used by Victorian Koories for Food, Fibre, Medicines and Implements. Yangennanock Women's Group, Aboriginal Keeping Place, Hamilton, Victoria.
Gott, B. (1999) Fire as an Aboriginal management tool in south-eastern Australia. Paper presented at the Australian Bushfire Conference, Albury, July 1999. On line at http://www.csu.edu.au/special/bushfire99/papers/gott/. Accessed 3 June 2005.
Gott, B. (2004) The role of indigenous fire regimes in guiding bushfire management today. Pp. 123-128 in Bushfire in a Changing Environment: New Directions in Management. Proceedings of conference held in Sydney, 24-25 June 2004. Nature Conservation Council of NSW, Sydney.
Gould, J.S., McCaw, W.L. and Cheney, N.P. (2004) Project Vesta - findings, extension and validation for south-eastern Australia eucalypt forests. Paper presented at the Bushfire 2004 Conference, Adelaide, May 2004. http://www.environment.sa.gov.au/biodiversity/bushfire_proceedings.html.
Grant, C.D. and MacGregor, C.M. (2001) Topsoil seed banks in grazed and ungrazed eucalypt woodlands at Newholme, Armidale, New South Wales, Australia. New Zealand Journal of Botany 39, 471-481.
Green, D.G. (1989) Simulated effects of fire, dispersal and spatial pattern on competition within forest mosaics. Vegetatio 82, 139-153.
Grice, A.C. (1997) Post-fire regrowth and survival of the invasive tropical shrubs Cryptostegia grandiflora and Ziziphus mauritiana. Australian Journal of Ecology 22, 49-55.
Griffiths, T. (2002) How many trees make a forest? Cultural debates about vegetation change in Australia. Australian Journal of Botany 50 , 375-389.
Hagon, M.W. (1976) Germination and dormancy of Themeda australis, Danthonia spp., Stipa bigeniculata and Bothriochloa macra. Australian Journal of Botany 24, 319-327.
Hammill, K.A., Bradstock, R.A. and Allaway, W.G. (1998) Post-fire seed dispersal and species re-establishment in proteaceous heath. Australian Journal of Botany 46, 407-419.
Harden, G.J. (ed.) (1992) Flora of New South Wales, Volume 3. NSW University Press, Sydney.
Harden, G.J. (ed.) (1993) Flora of New South Wales, Volume 4. NSW University Press, Sydney.
Harden, G.J. (ed.) (2000) Flora of New South Wales, Volume 1. Revised edition. NSW University Press, Sydney.
316 Harden, G.J. (ed.) (2002) Flora of New South Wales, Volume 2. Revised edition. NSW University Press, Sydney.
Harrington, G.N. (1991) Effects of soil moisture on shrub seedling survival in a semi-arid grassland. Ecology 72, 1138-1149.
Harrington, G.N. and Johns, G.G. (1990) Herbaceous biomass in a Eucalyptus savanna woodland after removing trees and/or shrubs. Journal of Applied Ecology 27, 775-787.
Harrington, G.N. and Sanderson, K.D. (1994) Recent contraction of wet sclerophyll forest in the wet tropics of Queensland due to invasion by rainforest. Pacific Conservation Biology 1, 319-327.
Harrold, A. (1979) Heathland regeneration after fire at Noosa. Queensland Naturalist 22, 88- 96.
Henderson, M.K. and Keith, D.A. (2002) Correlation of burning and grazing indicators with composition of woody understorey flora of dells in a temperate eucalypt forest. Austral Ecology 27, 121-131.
Henry, N.B. and Florence, R.G. (1966) Establishment and development of regeneration in Spotted Gum - Ironbark forests. Australian Forestry 30, 304-316.
Hester, A.J. and Hobbs, R.J. (1992) Influence of fire and soil nutrients on native and non- native annuals at remnant vegetation edges in the Western Australian wheatbelt. Journal of Vegetation Science 3, 101-108.
Hill, S. (2000) The Regeneration Capacity of a Remnant of Cumberland Plain Woodlands in Holsworthy Military Area. Master of Science (Honours) thesis, University of Wollongong.
Hill, S.J. and French, K. (2003) Response of the soil seed-bank of Cumberland Plain Woodland to heating. Austral Ecology 28, 14-22.
Hill, S.J. and French, K. (2004) Potential impacts of fire and grazing in an endangered ecological community: plant composition and shrub and eucalypt regeneration in Cumberland Plain Woodland. Australian Journal of Botany 52, 23-29.
Hitchmough, J.D., Curtain, H., Hammersley, L. and Kellow, J. (1996) Effect of gap width and turf type on the establishment of the Australian forb Bulbine bulbosa. Restoration Ecology 4, 25-32.
Hobbs, R.J. (2002) Fire regimes and their effects in Australian temperate woodlands. Pp. 305- 326 in Flammable Australia: the Fire Regimes and Biodiversity of a Continent (ed. by R.A. Bradstock, J.E. Williams and A.M. Gill). Cambridge University Press, Cambridge.
Hobbs, R.J. and Atkins, L. (1988) Spatial variability of experimental fires in south-west Western Australia. Australian Journal of Ecology 13, 295-299.
Hobbs, R.J. and Atkins, L. (1991) Interactions between annuals and woody perennials in a Western Australian nature reserve. Journal of Vegetation Science 2, 643-654.
Hobbs, R.J. and Huenneke, L.F. (1992) Disturbance, diversity, and invasion: implications for conservation. Conservation Biology 6, 324-337.
Hobbs, R.J. and Mooney, H.A. (1986) Community changes following shrub invasion of grassland. Oecologia 70, 508-513.
317 Hodgkinson, K.C. (1979) The shrubs of poplar box (Eucalyptus populnea) lands and their biology. Australian Rangeland Journal 1, 280-293.
Hodgkinson, K.C. (1998) Sprouting success of shrubs after fire: height-dependent relationships for different strategies. Oecologia 115, 64-72.
Hodgkinson, K.C. and Harrington, G.N. (1985) The case for prescribed burning to control shrubs in eastern semi-arid woodlands. Australian Rangeland Journal 7, 64-74.
Holmes, P.M. and Newton, R.J. (2004) Patterns of seed persistence in South African fynbos. Plant Ecology 172, 143-158.
House, A. (1995) Fire ecology research in Queensland native forests - current status and new directions. Pp. 86-97 in Working Papers - Sixth Queensland Fire Research Workshop, 8-10 March 1995, Bargara, Queensland . Land Use Study Centre, University of Southern Queensland, Toowoomba.
Hunter, J.T. (1998) Granite outcrop vegetation of Wilsons Promontory. The Victorian Naturalist 115, 322-325.
Hunter, J.T. (2003) Persistence on inselbergs: the role of obligate seeders and resprouters. Journal of Biogeography 30, 497-510.
Huston, M. (1979) A general hypothesis of species diversity. American Naturalist 113, 81- 101.
Huston, M. (2003) Understanding the effects of fire and other mortality-causing disturbances on species diversity. Pp. 37-70 in Fire in Ecosystems of South-west Western Australia (ed. by I. Abbott and N. Burrows). Backhuys Publishers, Leiden, The Netherlands.
Huston, M.A. (2004) Management strategies for plant invasions: manipulating productivity, disturbance, and competition. Diversity and Distributions 10, 167-178.
Jackson, J. and Ash, A.J. (1998) Tree-grass relationships in open eucalypt woodlands of northeastern Australia: influence of trees on pasture productivity, forage quality and species distribution. Agroforestry Systems 40, 159-176.
James, T., McDougall, L. and Benson, D. (1999) Rare Bushland Plants of Western Sydney. Royal Botanic Gardens, Sydney.
Jeltsch, F., Milton, S.J., Dean, W.R.J., van Rooyen, N. and Moloney, K.A. (1998) Modelling the impact of small-scale heterogeneities on tree-grass coexistence in semi-arid savannas. Journal of Ecology 86, 780-793.
Johnson, M.L. and Faunt, L.M. (1992) Parameter estimation by least-squares methods. Methods in Enzymology 210, 1-37.
Kartzoff, M. (1969) Nature and a City: the Native Vegetation of the Sydney Area. Edwards and Shaw, Sydney.
Keeley, J.E. and Keeley, S.C. (1984) Postfire recovery of California coastal sage scrub. American Midland Naturalist 111, 105-117.
Keith, D. (1996) Fire-driven extinction of plant populations: a synthesis of theory and review of evidence from Australian vegetation. Proceedings of the Linnean Society of New South Wales 116, 37-78.
318 Keith, D. (2004) Ocean Shores to Desert Dunes: the Native Vegetation of New South Wales and the ACT. Department of Environment and Conservation (NSW), Hurstville, NSW.
Keith, D.A. and Bradstock, R.A. (1994) Fire and competition in Australian heath: a conceptual model and field investigations. Journal of Vegetation Science 5, 347-354.
Keith, D.A., McCaw, W.L. and Whelan, R.J. (2002a) Fire regimes in Australian heathlands and their effects on plants and animals. Pp. 199-237 in Flammable Australia: the Fire Regimes and Biodiversity of a Continent (ed. by R.A. Bradstock, J.E. Williams and A.M. Gill). Cambridge University Press, Cambridge.
Keith, D.A., Williams, J.E. and Woinarski, J.C.Z. (2002b) Fire management and biodiversity conservation: key approaches and principles. Pp. 401-425 in Flammable Australia: The Fire Regimes and Biodiversity of a Continent (ed. by R.A. Bradstock, J.E. Williams and A.M. Gill). Cambridge University Press, Cambridge.
Kershaw, K.A. (1973) Quantitative and Dynamic Plant Ecology. Edward Arnold, London.
Kington, D. (1997) Where Have the Pretty Face Gone? Brisbane Forest Park, Brisbane.
Kirkpatrick, J.B. (1986) The viability of bush in cities - ten years of change in an urban grassy woodland. Australian Journal of Botany 34, 691-708.
Kirkpatrick, J., McDougall, K. and Hyde, M. (1995) Australia's Most Threatened Ecosystem: the Southeastern Lowland Native Grasslands. Surrey Beatty and Sons, Chipping Norton, NSW.
Kitchin, M.B. (2001) Fire Ecology and Fire Management for the Conservation of Plant Species and Vegetation Communities in a National Park in Northern NSW, Australia. PhD thesis, University of New England.
Knox, K.J.E. and Clarke, P.J. (2004) Fire response syndromes of shrubs in grassy woodlands in the New England Tableland Bioregion. Cunninghamia 8, 348-353.
Kohen, J. (1993) The Darug and Their Neighbours: the Traditional Aboriginal Owners of the Sydney Region. Darug Link in association with Blacktown and District Historical Society, Sydney.
Kohen, J.L. and Downing, A.J. (1992) Aboriginal use of plants on the western Cumberland Plain. Sydney Basin Naturalist 1, 1-8.
Krebs, C. (1999) Ecological Methodology (2nd edition). Benjamin Cummings, Menlo Park, California.
Lamb, D., Ash, D. and Landsberg, J. (1992) The effect of fire on understorey development and nitrogen cycling in Eucalyptus maculata forests in south east Queensland . Pp. 496-506 in Fire Research in Rural Queensland: Selected Papers from the Queensland Fire Research Workshop Series 1980-1989 (ed. by B.R. Roberts). Land Use Study Centre, University of Southern Queensland, Toowoomba.
Lamont, B.B., Olesen, J.M. and Briffa, P.J. (1998) Seed production, pollinator attractants and breeding system in relation to fire response - are there reproductive syndromes among co- occurring proteaceous shrubs? Australian Journal of Botany 46, 377-385.
319 Leigh, J.H. and Holgate, M.D. (1979) The responses of the understorey of forests and woodlands of the Southern Tablelands to grazing and burning. Australian Journal of Ecology 4, 25-45.
Leigh, J.H., Wood, D.H., Slee, A.V. and Holgate, M.D. (1991) The effects of burning and simulated grazing on productivity, forage quality, mortality and flowering of eight subalpine herbs in Kosciusko National Park. Australian Journal of Botany 39, 97-118.
Liedloff, A.C., Coughenour, M.B., Ludwig, J.A. and Dyer, R. (2001) Modelling the trade-off between fire and grazing in a tropical savanna landscape, northern Australia. Environment International 27, 173-180.
Little, D. (2003) Bringing the Bush Back to Western Sydney. Best Practice Guidelines for Bush Regeneration on the Cumberland Plain. Department of Infrastructure, Planning and Natural Resources, Parramatta.
Lloret, F., Verdu, M., Flores-Hernandez, N. and Valiente-Banuet, A. (1999) Fire and resprouting in Mediterranean ecosystems: insights from an external biogeographical region, the Mexical shrubland. American Journal of Botany 86, 1655-1661.
Lodge, G.M. and Whalley, R.D.B. (1981) Establishment of warm- and cool-season native perennial grasses on the north-west slopes of New South Wales. I. Dormancy and germination. Australian Journal of Botany 29, 111-119.
Ludwig, F., Dawson, T.E., Kroon, H., Berendse, F. and Prins, H.H.T. (2003) Hydraulic lift in Acacia tortilis trees on an East African savanna. Oecologia 134, 293-300.
Ludwig, F., Dawson, T.E., Prins, H.H.T., Berendse, F. and de Kroon, H. (2004) Below-ground competition between trees and grasses may overwhelm the facilitative effects of hydraulic lift. Ecology Letters 7, 623-631.
Ludwig, J.A. and Reynolds, J.F. (1988) Statistical Ecology: A Primer on Methods and Computing. John Wiley & Sons, New York.
Lunt, I.D. (1990) Impact of an autumn fire on a long-grazed Themeda triandra (Kangaroo Grass) grassland: implications for management of invaded, remnant vegetations. Victorian Naturalist 107, 45-51.
Lunt, I.D. (1994) Variation in flower production of nine grassland species with time since fire, and implications for grassland management and restoration. Pacific Conservation Biology 1, 359-366.
Lunt, I.D. (1995a) European management of remnant grassy forests and woodlands in south- eastern Australia - past, present and future? Victorian Naturalist 112, 239-249.
Lunt, I.D. (1995b) Seed longevity of six native forbs in a closed Themeda triandra grassland. Australian Journal of Botany 43, 439-449.
Lunt, I.D. (1997a) Effects of long-term vegetation management on remnant grassy forests and anthropogenic native grasslands in south-eastern Australia. Biological Conservation 81, 287-297.
Lunt, I.D. (1997b) A multivariate growth-form analysis of grassland and forest forbs in south- eastern Australia. Australian Journal of Botany 45, 691-705.
320 Lunt, I.D. (1997c) Germinable soil seed banks of anthropogenic native grasslands and grassy forest remnants in temperate south-eastern Australia. Plant Ecology 130, 21-34.
Lunt, I.D. (1997d) Tree densities last century on the lowland Gippsland plain, Victoria. Australian Geographical Studies 35, 342-348.
Lunt, I.D. (1998a) Two hundred years of land use and vegetation change in a remnant coastal woodland in southern Australia. Australian Journal of Botany 46, 629-647.
Lunt, I.D. (1998b) Allocasuarina (Casuarinaceae) invasion of an unburnt coastal woodland at Ocean Grove, Victoria: structural changes, 1971-1996. Australian Journal of Botany 46, 649-656.
Lunt, I.D. and Morgan, J.W. (1999) Vegetation changes after 10 years of grazing exclusion and intermittent burning in a Themeda triandra (Poaceae) grassland reserve in south-eastern Australia. Australian Journal of Botany 47, 537-552.
Lunt, I.D. and Morgan, J.W. (2002) The role of fire regimes in temperate lowland grasslands of southeastern Australia. Pp. 177-196 in Flammable Australia: the Fire Regimes and Biodiversity of a Continent (ed. by R.A. Bradstock, J.E. Williams and A.M. Gill). Cambridge University Press, Cambridge.
Madany, M.H. and West, N.E. (1983) Livestock grazing-fire regime interactions within montane forests of Zion National Park, Utah. Ecology 64, 661-667.
Magcale-Macandog, D.B. and Whalley, R.D.B. (1994) Factors affecting the distribution and abundance of Microlaena stipoides (Labill.) R.Br. on the Northern Tablelands of New South Wales. Rangeland Journal 16, 26-38.
Manly, B.F.J. (1991) Randomization and Monte Carlo Methods in Biology. Chapman and Hall, London.
Martin, T.G. and Green, J.L. (2002) Wildlife and core conservation areas. Pp. 111-142 in Managing and Conserving Grassy Woodlands (ed. by S. McIntyre, J.G. McIvor and K.M. Heard). CSIRO Publishing, Collingwood, Victoria.
McAlpine, R.S. (1995) Testing the effect of fuel consumption on fire spread rate. International Journal of Wildland Fire 5, 143-152.
McCaw, W.L. (1988) Regeneration of Acacia and Kennedia from soil stored seed following an autumn fire in jarrah (Eucalyptus marginata) forest. Journal of the Royal Society of Western Australia 71, 1-6.
McCaw, W.L. (1991) Measurement of fuel quantity and structure for bushfire research and management. Pp. 147-155 in Conference on Bushfire Modelling and Fire Danger Rating Systems: Proceedings (ed. by N.P. Cheney and A.M. Gill). Proceedings of conference held in Canberra on 11-12 July 1988, CSIRO, Canberra.
McDougall, K.L. and Morgan, J.W. (2005) Establishment of native grassland vegetation at Organ Pipes National Park near Melbourne, Victoria: vegetation changes from 1989 to 2003. Ecological Management and Restoration 6, 34-42.
McFarland, D.C. (1988a) Fire and the vegetation composition and structure of subtropical heathlands in south-eastern Queensland. Australian Journal of Botany 36, 533-546.
321 McFarland, D.C. (1988b) The composition, microhabitat use and response to fire of the avifauna of subtropical heathlands in Cooloola National Park, Queensland. Emu 88, 249- 257.
McFarland, D.C. (1990) Flower and seed phenology of some plants in the subtropical heathlands of Cooloola National Park, Queensland, Australia. Australian Journal of Botany 38, 501-509.
McIntyre, S. (2002) Trees. Pp. 79-110 in Managing and Conserving Grassy Woodlands (ed. by S. McIntyre, J.G. McIvor and K.M. Heard). CSIRO Publishing, Collingwood, Victoria.
McIntyre, S., Lavorel, S. and Tremont, R.M. (1995) Plant life-history attributes: their relationship to disturbance response in herbaceous vegetation. Journal of Ecology 83, 31- 44.
McIntyre, S. and Martin, T.G. (2001) Biophysical and human influences on plant species richness in grasslands: comparing variegated landscapes in subtropical and temperate regions. Austral Ecology 26, 233-245.
McIntyre, S., McIvor, J.G. and Heard, K.M. (eds.) (2002) Managing and Conserving Grassy Woodlands. CSIRO Publishing, Collingwood, Victoria.
McMahon, A.R.G., Carr, G.W., Bedggood, S.E., Hill, R.J. and Pritchard, A.M. (1996) Prescribed fire and control of coast wattle (Acacia sophorae (Labill.) R.Br.) invasion in coastal heath south-west Victoria. Pp. 87-96 in Fire and Biodiversity: the Effects and Effectiveness of Fire Management (ed. by J.R. Merrick). Proceedings of conference held in Melbourne on 8-9 October 1994, Commonwealth Department of the Environment, Sport and Territories, Canberra.
Milberg, P. and Lamont, B.B. (1995) Fire enhances weed invasion of roadside vegetation in southwestern Australia. Biological Conservation 73, 45-49.
Moles, A.T., Warton, D.I. and Westoby, M. (2003) Seed size and survival in the soil in arid Australia. Austral Ecology 28, 575-585.
Moles, A.T. and Westoby, M. (2004) Seed mass and seedling establishment after fire in Ku- ring-gai Chase National Park, Sydney, Australia. Austral Ecology 29, 383-390.
Moreira, M.Z., Scholz, F.G., Bucci, S.J., Sternberg, L.S., Goldstein, G., Meinzer, F.C. and Franco, A.C. (2003) Hydraulic lift in a neotropical savanna. Functional Ecology 17, 573- 581.
Morgan, J.W. (1996) Secondary juvenile period and community recovery following late-spring burning of a Kangaroo Grass Themeda triandra grassland. Victorian Naturalist 113, 47-57.
Morgan, J.W. (1998a) Importance of canopy gaps for recruitment of some forbs in Themeda triandra-dominated grasslands of south-eastern Australia. Australian Journal of Botany 46, 609-627.
Morgan, J.W. (1998b) Composition and seasonal flux of the soil seed bank of species-rich Themeda triandra grasslands in relation to burning history. Journal of Vegetation Science 9, 145-156.
Morgan, J.W. (1998c) Comparative germination responses of 28 temperate grassland species. Australian Journal of Botany 46, 209-219.
322 Morgan, J.W. (1998d) Patterns of invasion of an urban remnant of a species-rich grassland in southeastern Australia by non-native plant species. Journal of Vegetation Science 9, 181- 190.
Morgan, J.W. (1999) Defining grassland fire events and the response of perennial plants to annual fire in temperate grasslands of south-eastern Australia. Plant Ecology 144, 127-144.
Morgan, J.W. (2001) Seedling recruitment patterns over 4 years in an Australian perennial grassland community with different fire histories. Journal of Ecology 89, 908-919.
Morgan, J.W. and Lunt, I.D. (1999) Effects of time-since-fire on the tussock dynamics of a dominant grass (Themeda triandra) in a temperate Australian grassland. Biological Conservation 88, 379-386.
Morris, E.C. (2000) Germination response of seven east Australian Grevillea species (Proteaceae) to smoke, heat exposure and scarification. Australian Journal of Botany 48, 179-189.
Morrison, D.A. (1995) Some effects of low-intensity fires on populations of co-occurring small trees in the Sydney region. Proceedings of the Linnean Society of NSW 115, 109-119.
Morrison, D.A., Buckney, R.T., Bewick, B.J. and Cary, G.J. (1996) Conservation conflicts over burning bush in south-eastern Australia. Biological Conservation 76, 167-175.
Morrison, D.A., Cary, G.J., Pengelly, S.M., Ross, D.G., Mullins, B.J., Thomas, C.R. and Anderson, T.S. (1995a) Effects of fire frequency on plant species composition of sandstone communities in the Sydney region: inter-fire interval and time-since-fire. Australian Journal of Ecology 20, 239-247.
Morrison, D.A., LeBrocque, A.F. and Clarke, P.J. (1995b) An assessment of some improved techniques for estimating the abundance (frequency) of sedentary organisms. Vegetatio 120, 131-145.
Morrison, D.A. and Renwick, J.A. (2000) Effects of variation in fire intensity on regeneration of co-occurring species of small trees in the Sydney region. Australian Journal of Botany 48, 71-79.
Nadolny, C., Ranclaud, H., Whalley, W., McLeish, T., Wheeler, J., Morsley, R., Croft, P., McCormick, L., Ede, A., Hawes, W., Watson, C. and Austin, J. (2003) Grassy Vegetation in North-Western NSW and Guidelines for its Management for Conservation. Armidale Tree Group, Armidale, NSW.
Neumann, F.G. and Tolhurst, K.G. (1991) Effects of fuel reduction burning on epigeal arthropods and earthworms in dry sclerophyll eucalypt forest of west-central Victoria. Australian Journal of Ecology 16, 315-330.
Nieuwenhuis, A. (1987) The effect of fire frequency on the sclerophyll vegetation of the West Head, New South Wales. Australian Journal of Ecology 12, 373-385.
Noble, I.R. (1980) Interactions between tussock grass (Poa spp.) and Eucalyptus pauciflora seedings near treeline in south-eastern Australia. Oecologia 45, 350-353.
Noble, I.R. (1984) Mortality of lignotuberous seedlings of Eucalyptus species after an intense fire in montane forest. Australian Journal of Ecology 9, 47-50.
323 Noble, I.R. and Gitay, H. (1996) A functional classification for predicting the dynamics of landscapes. Journal of Vegetation Science 7, 329-336.
Noble, I.R. and Slatyer, R.O. (1980) The use of vital attributes to predict successional changes in plant communities subject to recurrent disturbances. Vegetatio 43, 5-21.
Noble, J.C. (1982) The significance of fire in the biology and evolutionary ecology of mallee Eucalyptus populations. Pp. 153-159 in Evolution of the Flora and Fauna of Arid Australia (ed. by W.R. Barker and P.J.M. Greenslade). Peacock Publications, Adelaide.
Noble, J.C. (1997) The Delicate and Noxious Scrub: CSIRO Research into Native Tree and Shrub Proliferation in the Semi-arid Woodlands of Eastern Australia. CSIRO Division of Wildlife and Ecology, Canberra.
Noble, J.C. and Grice, A.C. (2002) Fire regimes in semi-arid and tropical pastoral lands: managing biological diversity and ecosystem function. Pp. 373-400 in Flammable Australia: the Fire Regimes and Biodiversity of a Continent (ed. by R.A. Bradstock, J.E. Williams and A.M. Gill). Cambridge University Press, Cambridge.
Noss, R.F. and Cooperrider, A.Y. (1994) Saving Nature's Legacy: Protecting and Restoring Biodiversity. Island Press, Covelo, California.
NPWS (National Parks and Wildlife Service) (1997) Urban Bushland Biodiversity Survey. NSW National Parks and Wildlife Service, Hurstville, NSW.
NPWS (National Parks and Wildlife Service) (1999) Castlereagh, Agnes Banks and Windsor Downs Nature Reserves Plan of Management. NSW National Parks and Wildlife Service, Sydney.
NPWS (National Parks and Wildlife Service) (2000) The Native Vegetation of the Cumberland Plain, Western Sydney - Technical Report. NSW National Parks and Wildlife Service, Hurstville, NSW.
NPWS (National Parks and Wildlife Service) (2004a) Cumberland Plain Vegetation Mapping Project. On line at http://www.nationalparks.nsw.gov.au/npws.nsf/Content/Cumberland+ Plain+Vegetation+Mapping+Project. Last amended 4 February 2004. Accessed throughout 2004.
NPWS (National Parks and Wildlife Service) (2004b) Conserving biodiversity in NSW. On line at http://www.nationalparks.nsw.gov.au/npws.nsf/Content/Conserving+biodiversity+in +NSW. Last amended 4 February 2004. Accessed 9 November 2004.
NPWS (National Parks and Wildlife Service) (2004c) Fire Interval Guidelines for Broad Vegetation Types. NSW National Parks and Wildlife Service, Hurstville, NSW.
O'Connell, A.M. (1991) Fuel quantity: the time dimension for forest litter. Pp. 109-118 in Conference on Bushfire Modelling and Fire Danger Rating Systems: Proceedings (ed. by N.P. Cheney and A.M. Gill). Proceedings of conference held in Canberra on 11-12 July 1988, CSIRO, Canberra.
Odgers, B.M. (1996) Fire, buried germinable seed banks and grass species establishment in an urban eucalypt forest reserve. Australian Journal of Botany 44, 413-419.
Odgers, B.M. (1999) Seasonal variation in buried germinable seed banks of grass species in an urban eucalypt forest reserve. Australian Journal of Botany 47, 623-638.
324 Outhred, R.K. (1984) Semi-quantitative sampling in vegetation survey. Pp. 87-100 in Survey Methods for Nature Conservation ( ed. by K. Myers, C.R. Margules and I. Musto). CSIRO, Canberra.
Peacock, R.W. (1900) Our western lands. Their deterioration and possible improvement. Agricultural Gazette of New South Wales 11, 652-657.
Pickett, S.T.A., Parker, V.T. and Fiedler, P.L. (1992) The new paradigm in ecology: implications for conservation biology above the species level. Pp. 65-88 in Conservation Biology: the Theory and Practice of Nature Conservation Preservation and Management (ed. by P.L. Fiedler and S.K. Jain). Chapman and Hall, New York.
Pieper, R.D. (1990) Overstory-understory relations in pinyon-juniper woodlands in New Mexico. Journal of Range Management 43, 413-415.
Posamentier, H.G., Clark, S.S., Hain, D.L. and Recher, H.F. (1981) Succession following wildfire in coastal heathland (Nadgee Nature Reserve N.S.W.). Australian Journal of Ecology 6, 165-175.
Prober, S.M. (1996) Conservation of the grassy White Box woodlands: rangewide floristic variation and implications for reserve design. Australian Journal of Botany 44, 57-77.
Prober, S.M., Lunt, I.D. and Thiele, K.R. (2002a) Determining reference conditions for management and restoration of temperate grassy woodlands: relationships among trees, topsoils and understorey flora in little-grazed remnants. Australian Journal of Botany 50, 687-697.
Prober, S.M., Lunt, I.D. and Thiele, K.R. (2004) Restoring ecological function in woodlands - is Themeda australis a key to restoring nitrogen cycles and resistance to annual exotics in grassy understoreys? Pp. 90-91 in Abstracts Booklet (ed. by V. Sander, J. Potter and J.M. Facelli). Proceedings of conference held in Adelaide on 7 - 10 December 2004, Ecological Society of Australia, Adelaide.
Prober, S.M. and Thiele, K.R. (1995) Conservation of the grassy White Box woodlands: relative contributions of size and disturbance to floristic composition and diversity of remnants. Australian Journal of Botany 43, 349-366.
Prober, S.M. and Thiele, K.R. (2004) Floristic patterns along an east-west gradient in grassy box woodlands of Central New South Wales. Cunninghamia 8, 306-325.
Prober, S.M., Thiele, K.R. and Lunt, I.D. (2002b) Identifying ecological barriers to restoration in temperate grassy woodlands: soil changes associated with different degradation states. Australian Journal of Botany 50, 699-712.
Proulx, M. and Mazumder, A. (1998) Reversal of grazing impact on plant species richness in nutrient-poor vs. nutrient-rich ecosystems. Ecology 79, 2581-2592.
Raison, R.J., Woods, P.V. and Khanna, P.K. (1983) Dynamics of fine fuels in recurrently burnt eucalypt forests. Australian Forestry 46, 294-302.
Rawling, J. (1994) Does Sydney have ground flora? Victorian Naturalist 111, 169-171.
Read, T.R., Bellairs, S.M., Mulligan, D.R. and Lamb, D. (2000) Smoke and heat effects on soil seed bank germination for the re-establishment of a native forest community in New South Wales. Austral Ecology 25, 48-57.
325 Recher, H.F., Hutchings, P.A. and Rosen, S. (1993) The biota of the Hawkesbury-Nepean catchment: reconstruction and restoration. Australian Zoologist 29, 3-41.
Reid, N., Gross, C., Jarman, P. and Van der Lee, G. (1996) Fire Management Criteria and Guidelines for Conservation of the Natural and Cultural Heritage of Guy Fawkes River National Park. Department of Ecosystem Management, University of New England, Armidale.
Rice, B. and Westoby, M. (1983) Species richness in vascular vegetation of the West Head, New South Wales. Australian Journal of Ecology 8, 163-168.
Richards, J.H. and Caldwell, M.M. (1987) Hydraulic lift: substantial nocturnal water transport between soil layers by Artemisia tridentata roots. Oecologia 73, 496-489.
Richardson, D.M., Cowling, R.M., Lamont, B.B. and van Hensbergen, H.J. (1995) Coexistence of Banksia species in southwestern Australia: the role of regional and local processes. Journal of Vegetation Science 6, 329-342.
Robertson, D. (1985) Interrelationships between Kangaroos, Fire and Vegetation Dynamics at Gellibrand Hill Park, Victoria. PhD thesis, University of Melbourne.
Robertson, G., Matthes, M. and Smith, M. (1996) Conservation Research Statement and Species Recovery Plan for Persoonia nutans R.Br. NSW National Parks and Wildlife Service, Hurstville, NSW.
Robinson, L. (1994) Field Guide to the Native Plants of Sydney. Second edition. Kangaroo Press, East Roseville, NSW.
Robinson, L. (2003) Field Guide to the Native Plants of Sydney. Third edition. Kangaroo Press, East Roseville, NSW.
Roche, S., Dixon, K.W. and Pate, J.S. (1997) Seed ageing and smoke: partner cues in the amelioration of seed dormancy in selected Australian native species. Australian Journal of Botany 45, 783-815.
Roche, S., Dixon, K.W. and Pate, J.S. (1998) For everything a season: smoke-induced seed germination and seedling recruitment in a Western Australian Banksia woodland. Australian Journal of Ecology 23, 111-120.
Roques, K.G., O'Connor, T.G. and Watkinson, A.R. (2001) Dynamics of shrub encroachment in an African savanna: relative influences of fire, herbivory, rainfall and density dependence. Journal of Applied Ecology 38, 268-280.
Russell, M.J. and Roberts, B.R. (1996) Effects of four low-intensity burns over 14 years on the floristics of a Blackbutt (Eucalyptus pilularis) forest in southern Queensland. Australian Journal of Botany 44, 315-329.
Russell-Smith, J. (2002) Pre-contact Aboriginal, and contemporary fire regimes of the savanna landscapes of northern Australia: patterns, changes and ecological processes. In Australian Fire Regimes: Contemporary Patterns (April 1998 - March 2000) and Changes Since European Settlement (ed. by J. Russell-Smith, R. Craig, A.M. Gill, R. Smith and J. Williams). Department of the Environment and Heritage, Canberra.
326 Russell-Smith, J. and Stanton, P. (2002) Fire regimes and fire management of rainforest communities across northern Australia. Pp. 329-350 in Flammable Australia: the Fire Regimes and Biodiversity of a Continent (ed. by R.A. Bradstock, J.E. Williams and A.M. Gill). Cambridge University Press, Cambridge.
Russell-Smith, J., Whitehead, P.J., Cook, G.D. and Hoare, J.L. (2003) Response of Eucalyptus-dominated savanna to frequent fires: lessons from Munmarlary, 1973-1996. Ecological Monographs 73, 349-375.
Ruyle, G. and Frost, B. (1993) Monitoring Rangeland Browse Vegetation. Pp. 27-31 in Arizona Ranchers' Management Guide (ed. by R. Gum, G. Ruyle and R. Rice). Arizona Cooperative Extension, Tuscon, Arizona.
Rymer, P.D. (1999) The Reproductive Biology and Population Genetics of the Rare and Threatened Plant, Dillwynia tenuifolia (Fabaceae). Honours thesis, University of Western Sydney.
San Jose, J.J. and Farinas, M.R. (1991) Temporal changes in the structure of a Trachypogon savanna protected for 25 years. Acta Oecologica 12, 237-247.
Sandercoe, C.S. (1989) A review of fire research in Queensland heathlands. Paper presented at the Bushfires and Shrublands Conference, Canberra, 8-10 May 1989.
Scanlan, J.C. and Burrows, W.H. (1990) Woody overstorey impact on herbaceous understorey in Eucalyptus spp. communities in central Queensland. Australian Journal of Ecology 15, 191-197.
Scholes, R.J. and Archer, S.R. (1997) Tree-grass interactions in savannas. Annual Review of Ecology and Systematics 28, 517-544.
Semple, W.S. and Koen, T.B. (2003) Effect of pasture type on regeneration of eucalypts in the woodland zone of south-eastern Australia. Cunninghamia 8, 76-84.
Setterfield, S.A., Douglas, M.M., Hutley, L.B. and Welch, M.A. (2005) Effects of canopy cover and ground disturbance on establishment of an invasive grass in an Australia savanna. Biotropica 37, 25-31.
Sharp, B.R. and Bowman, D.M.J.S. (2004) Net woody vegetation increase confined to seasonally inundated lowlands in an Australian tropical savanna, Victoria River District, Northern Territory. Austral Ecology 29, 667-683.
Shea, S.R., McCormick, J. and Portlock, C.C. (1979) The effect of fires on regeneration of leguminous species in the northern Jarrah (Eucalyptus marginata Sm) forest of Western Australia. Australian Journal of Ecology 4, 195-205.
Siddiqi, M.Y., Carolin, R.C. and Myerscough, P.J. (1976) Studies in the ecology of coastal heath in New South Wales. III. Regrowth of vegetation after fire. Proceedings of the Linnean Society of New South Wales 101, 53-63.
Simmons, D. and Adams, R. (1986) Fuel dynamics in an urban fringe dry sclerophyll forest in Victoria. Australian Forestry 49, 149-154.
Simmons, D. and Adams, R. (1999) Fuel loads and characteristics in some structurally diverse vegetation types in Victoria. Paper presented at the Australian Bushfire Conference, Albury, July 1999. On line at http://life.csu.edu.au/bushfire99/papers/simmons/index.html. Accessed 6 June 2005.
327 Sivertsen, D. (1993) Conservation of remnant vegetation in the Box and Ironbark lands of New South Wales. Victorian Naturalist 110, 24-29.
Sivertsen, D. and Clarke, P.J. (2000) Temperate woodlands in New South Wales: a brief overview of distribution, composition and conservation. Pp. 6-16 in Temperate Eucalypt Woodlands in Australia: Biology, Conservation, Management and Restoration (ed. by R.J. Hobbs and C.J. Yates). Surrey Beatty and Sons, Chipping Norton, NSW.
Skarpe, C. (1992) Dynamics of savanna ecosystems. Journal of Vegetation Science 3, 293- 300.
Sokal, R.R. and Rohlf, F.J. (1995) Biometry: the Principles and Practice of Statistics in Biological Research. Third edition. W.H. Freeman, New York.
Specht, R.L. (1970) Vegetation. Pp. 44-67 in The Australian Environment (ed. by G.W. Leeper). Melbourne University Press, Carlton, Victoria.
Specht, R.L., Rayson, P. and Jackman, M.E. (1958) Dark Island heath (Ninety-Mile Plain, South Australia). VI. Pyric succession: changes in composition, coverage, dry weight, and mineral nutrient status. Australian Journal of Botany 6, 59-88.
Specht, R.L. and Specht, A. (1989) Species richness of sclerophyll (heathy) plant communities in Australia - the influence of overstorey cover. Australian Journal of Botany 37, 337-350.
Stenhouse, R.N. (2004) Fragmentation and internal disturbance of native vegetation reserves in the Perth metropolitan area, Western Australia. Landscape and Urban Planning 68, 389- 401.
Stewart, B. (1999) Effects of Long-term Repeated Prescribed Burning on Soil-stored Seed Banks in a Coastal Blackbutt Forest. Honours thesis, University of Wollongong.
Stohlgren, T.J., Falkner, M.B. and Schell, L.D. (1995) A modified-Whittaker nested vegetation sampling method. Vegetatio 117, 113-121.
Stuwe, J. (1994) The role of fire in ground flora ecology. Victorian Naturalist 111, 93-5.
Stuwe, J. and Parsons, R.F. (1977) Themeda australis grasslands on the Basalt Plains, Victoria: floristics and management effects. Australian Journal of Ecology 2, 467-476.
Tasker, E.M. and Dickman, C.R. (2004) Small mammal community composition in relation to cattle grazing and associated burning in eucalypt forests of the Northern Tablelands of New South Wales. Pp. 721-740 in The Conservation of Australia's Forest Fauna (2nd edition) (ed. by D. Lunney). Royal Zoological Society of New South Wales, Mosman, NSW.
Taylor, M. (1992) Fire in subtropical forest management. Pp. 172-184 in Fire Research in Rural Queensland: Selected Papers from the Queensland Fire Research Workshop Series 1980-1989 (ed. by B.R. Roberts). Land Use Study Centre, University of Southern Queensland, Toowoomba.
Thomas, J. (1994) Effects of Hazard Reduction Burning on a Grassy Woodland Remnant in Western Sydney. Master of Science thesis, University of New South Wales.
Thomas, P.B., Morris, E.C. and Auld, T.D. (2003) Interactive effects of heat shock and smoke on germination of nine species forming soil seed banks within the Sydney region. Austral Ecology 28, 674-683.
328 Tilman, D. and Wedin, D. (1991) Oscillations and chaos in the dynamics of a perennial grass. Nature 353, 653-655.
Tolhurst, K. (1996a) Effects of fuel reduction burning on fuel loads in a dry sclerophyll forest. Pp. 17-20 in Fire and Biodiversity: the Effects and Effectiveness of Fire Management (ed. by J.R. Merrick). Proceedings of conference held in Melbourne on 8-9 October 1994, Commonwealth Department of Environment, Sport and Territories, Canberra.
Tolhurst, K. (1996b) Effects of fuel reduction burning on flora in a dry sclerophyll forest. Pp. 97-107 in Fire and Biodiversity: the Effects and Effectiveness of Fire Management (ed. by J.R. Merrick). Proceedings of conference held in Melbourne on 8-9 October 1994, Commonwealth Department of the Environment, Sport and Territories, Canberra.
Tongway, D.J. and Ludwig, J.A. (1990) Vegetation and soil patterning in semi-arid mulga lands of eastern Australia. Australian Journal of Ecology 15, 23-34.
Tozer, M. (2003) The native vegetation of the Cumberland Plain, western Sydney: systematic classification and field identification of communities. Cunninghamia 8, 1-75.
Tozer, M.G. and Bradstock, R.A. (2002) Fire-mediated effects of overstorey on plant species diversity and abundance in an eastern Australian heath. Plant Ecology 164, 213-223.
Trapnell, C.G. (1959) Ecological results of woodland burning experiments in northern Rhodesia. Journal of Ecology 47, 129-168.
Tremont, R.M. (1994) Life-history attributes of plants in grazed and ungrazed grasslands on the Northern Tablelands of New South Wales. Australian Journal of Botany 42, 511-530.
Tremont, R.M. and McIntyre, S. (1994) Natural grassy vegetation and native forbs in temperate Australia: structure, dynamics and life histories. Australian Journal of Botany 42, 641-658.
Turner, J. and Lambert, M.J. (2002) Litterfall and forest floor dynamics in Eucalyptus pilularis forests. Austral Ecology 27, 192-199.
Turner, M.G., Gardner, R.H. and O'Neill, R.V. (2001) Landscape Ecology in Theory and Practice: Pattern and Process. Springer-Verlag, New York.
Uys, R.G., Bond, W.J. and Everson, T.M. (2004) The effect of different fire regimes on plant diversity in southern African grasslands. Biological Conservation 118, 489-499. van Langevelde, F., van de Vijver, C.A.D.M., Kumar, L., van de Koppel, J., de Ridder, N., van Andel, J., Skidmore, A.K., Hearne, J.W., Stroosnijder, L., Bond, W.J., Prins, H.H.T. and Rietkerk, M. (2003) Effects of fire and herbivory on the stability of savanna ecosystems. Ecology 84, 337-350. van Vegten, J.A. (1983) Thornbush invasion in a savanna ecosystem in eastern Botswana. Vegetatio 56, 3-7.
Vaughton, G. (1998) Soil seed bank dynamics in the rare obligate seeding shrub, Grevillea barklyana (Proteaceae). Australian Journal of Ecology 23, 375-384.
Vigilante, T. and Bowman, D.M.J.S. (2004) Effects of fire history on the structure and floristic composition of woody vegetation around Kalumburu, North Kimberley, Australia: a landscape-scale natural experiment. Australian Journal of Botany 52, 381-404.
329 Vlahos, S. and Bell, D.T. (1986) Soil seed-bank components of the northern jarrah forest of Western Australia. Australian Journal of Ecology 11, 171-179.
Vlok, J.H.J. and Yeaton, R.I. (2000) Competitive interactions between overstorey proteas and sprouting understorey species in South African mountain fynbos. Diversity and Distributions 6, 273-281.
Walker, J. (1981) Fuel dynamics in Australian vegetation. Pp. 101-127 in Fire and the Australian Biota (ed. by A.M. Gill, R.H. Groves and I.R. Noble). Australian Academy of Science, Canberra.
Walker, J., Robertson, J.A., Penridge, L.K. and Sharpe, P.J.H. (1986) Herbage response to tree thinning in a Eucalyptus crebra woodland. Australian Journal of Ecology 11, 135-140.
Warcup, J.H. (1980) Effect of heat treatment of forest soil on germination of buried seed. Australian Journal of Botany 28, 567-571.
Wark, M.C., White, M.D., Robertson, D.J. and Marriott, P.F. (1987) Regeneration of heath and heath woodland in the north-eastern Otway Ranges following the wildfire of February 1983. Proceedings of the Royal Society of Victoria 99, 51-88.
Warren, D. (2003) Seedy stories of intrigue and mystery from the wilds of Western Sydney. Corridors of Green: Newsletter of Greening Western Sydney 35 (Spring), 6-7.
Watson, L. and Dallwitz, M.J. (1992 onwards) Grass Genera of the World. On line at http://biodiversity.uno.edu/delta/. Accessed 15 March 2005.
Watson, P. (2001a) The Role and Use of Fire for Biodiversity Conservation in Southeast Queensland: Fire Management Guidelines Derived from Ecological Research. SEQ Fire and Biodiversity Consortium, Brisbane.
Watson, P. (2001b) The Southeast Queensland Fire and Biodiversity Consortium: a case study in interagency cooperation and community education. Pp. 205-211 in Bushfire 2001 (ed. by G. Pearce and L. Lester). Proceedings of conference held in Christchurch, New Zealand on 3-6 July 2001, New Zealand Forst Research Institute, Rotorua, New Zealand.
Watson, P. and Wardell-Johnson, G. (2004) Fire frequency and time-since-fire effects on the open-forest and woodland flora of Girraween National Park, south-east Queensland, Australia. Austral Ecology 29, 225-236.
Watson, Phil. (2001) So what's interesting about the Sweet Bursaria? Indigenotes 14, 2-3.
Wellington, A.B. and Noble, I.R. (1985) Post-fire recruitment and mortality in a population of the mallee Eucalyptus incrassata in semi-arid, south-eastern Australia. Journal of Ecology 73, 645-656.
Weltzin, J.F. and Coughenour, M.B. (1990) Savanna tree influence on understory vegetation and soil nutrients in northwestern Kenya. Journal of Vegetation Science 1, 325-334.
Westfall, R.H., Everson, C.S. and Everson, T.M. (1983) The vegetation of the protected plots at Thabamhlope Research Station. South African Journal of Botany 2, 15-25.
Westoby, M., Walker, B. and Noy-Meir, I. (1989) Opportunistic management for rangelands not at equilibrium. Journal of Range Management 42, 266-274.
330 Whelan, R.J. and Baker, J.R. (1999) Fire in Australia: coping with variation in ecological effects of fire. Pp. 71-79 in Bush Fire Management Conference Proceedings: Protecting the Environment, Land, Life and Property (ed. by F. Sutton, J. Keats, J. Dowling and C. Doig). Proceedings of conference held in Sydney on 27-28 February 1998, Nature Conservation Council of NSW, Sydney.
Whelan, R.J., De Jong, N.H. and von der Burg, S. (1998) Variation in bradyspory and seedling recruitment without fire among populations of Banksia serrata (Proteaceae). Australian Journal of Ecology 23, 121-128.
Williams, J.E. and Gill, A.M. (1995) The Impact of Fire Regimes on Native Forests in Eastern New South Wales. NSW National Parks and Wildlife Service, Sydney.
Williams, J.E., Whelan, R.J. and Gill, A.M. (1994) Fire and environmental heterogeneity in southern temperate forest ecosystems: implications for management. Australian Journal of Botany 42, 125-137.
Williams, P.R., Congdon, R.A., Grice, A.C. and Clarke, P.J. (2003) Effect of fire regime on plant abundance in a tropical eucalypt savanna of north-eastern Australia. Austral Ecology 28, 327-338.
Williams, R.J. (2000) Vegetation dynamics. P.6 in Fire and Biodiversity. CSIRO's Kapalga Fire Experiment, Kakadu National Park. Workshop Notes. CSIRO, Darwin.
Williams, R.J., Cook, G.D., Gill, A.M. and Moore, P.H.R. (1999) Fire regime, fire intensity and tree survival in a tropical savanna in northern Australia. Australian Journal of Ecology 24, 50-59.
Williams, R.J., Griffiths, A.D. and Allan, G.E. (2002) Fire regimes and biodiversity in the savannas of northern Australia. Pp. 281-304 in Flammable Australia: the Fire Regimes and Biodiversity of a Continent (ed. by R.A. Bradstock, J.E. Williams and A.M. Gill). Cambridge University Press, Cambridge.
Williams, W.M. (1996) Genetic resources of temperate native and low-input grasses in New Zealand and Australian collections. New Zealand Journal of Agricultural Research 39, 513-526.
Willis, A.J. and Groves, R.H. (1991) Temperature and light effects on the germination of seven native forbs. Australian Journal of Botany 39, 219-228.
Willis, A.J., McKay, R., Vranjic, J.A., Kilby, M.J. and Groves, R.H. (2003) Comparative seed ecology of the endangered shrub, Pimelea spicata and a threatening weed, Bridal Creeper: smoke, heat and other fire-related germination cues. Ecological Management and Restoration 4, 55-65.
Wilson, B.A. (1996) Fire effects on vertebrate fauna and implications for fire management and conservation. Pp. 131-147 in Fire and Biodiversity: the Effects and Effectiveness of Fire Management (ed. by J.R. Merrick). Proceedings of conference held in Melbourne on 8-9 October 1994, Commonwealth Department of Environment, Sport and Territories, Canberra.
Wimbush, D.J. and Costin, A.B. (1979) Trends in vegetation at Kosciusko. I. Grazing trials in the subalpine zone, 1957-1971. Australian Journal of Botany 27, 741-487.
331 Withers, J. and Ashton, D.H. (1977) Studies on the status of unburnt Eucalyptus woodland at Ocean Grove, Victoria. I. The structure and regeneration. Australian Journal of Botany 25, 623-637.
Woinarski, J.C.Z. (1999) Fire and Australian birds: a review. Pp. 55-180 in Australia's Biodiversity - Responses to Fire. Plants, Birds and Invertebrates (ed. by A.M. Gill, J.C.Z. Woinarski and A. York). Department of the Environment and Heritage, Canberra.
Woinarski, J.C.Z., Risler, J. and Kean, L. (2004) Response of vegetation and vertebrate fauna to 23 years of fire exclusion in a tropical Eucalyptus open forest, Northern Territory, Australia. Austral Ecology 29, 156-176.
Wood, G.B., Turner, B.J. and Brack, C.L. (1999) Code of Forest Mensuration Practice: a Guide to Good Tree Measurement Practice in Australia and New Zealand. On line at http://www.anu.edu.au/Forestry/mensuration/rwg2/code. Accessed 8 November 2002.
Wood, P.D. (2001) What are the Germination Signals of the Cumberland Plain Woodland Soil Seed Bank? Honours thesis, University of Western Sydney.
Woods, P.V., Raison, R.J. and Khanna, P.K. (1983) Effects of prescribed burning on forest- floor microclimate and on subsequent rates of litter decomposition in a Eucalyptus pauciflora forest. Proceedings of the Ecological Society of Australia 12, 174-175.
Yates, C.J. and Hobbs, R.J. (2000) Temperate eucalypt woodlands in Australia - an overview. Pp. 1-5 in Temperate Eucalypt Woodlands in Australia: Biology, Conservation, Management and Restoration (ed. by R.J. Hobbs and C.J. Yates). Surrey Beatty and Sons, Chipping Norton, NSW.
Yeaton, R.I. and Bond, W.J. (1991) Competition between two shrub species: dispersal differences and fire promote coexistence. American Naturalist 138, 328-341.
Yibarbuk, D., Whitehead, P.J., Russell-Smith, J. , Jackson, D., Godjuwa, C., Fisher, A., Cooke, P., Choquenot, D. and Bowman, D.M.J.S. (2001) Fire ecology and Aboriginal land management in central Arnhem Land, northern Australia: a tradition of ecosystem management. Journal of Biogeography 28, 325-343.
York, A. (1999) Long-term effects of repeated prescribed burning on forest invertebrates: management implications for the conservation of biodiversity. Pp. 181-266 in Australia's Biodiversity - Responses to Fire. Plants, Birds and Invertebrates (ed. by A.M. Gill, J.C.Z. Woinarski and A. York). Department of the Environment and Heritage, Canberra.
Zammit, C. and Westoby, M. (1987) Population structure and reproductive status of two Banksia shrubs at various times after fire. Vegetatio 70, 11-20.
Zar, J.H. (1999) Biostatistical Analysis. 4th edition. Prentice Hall, Upper Saddle River, New Jersey.
332 APPENDICES
1. Climate data for Cumberland Plain and comparison weather stations
2. Rainfall during the project
3. Vital attributes, plant functional types and sensitivity to disturbance regimes
4. Regeneration modes of Cumberland Plain shrub species
5. Juvenile periods of Cumberland Plain shrub species
6. Comparison of Cumberland Plain data and data in NSW Flora Fire Response Database
7. Fire-related attributes of two Cumberland Plain vegetation types
8. Species in landscape study sites
9. Summary of landscape study findings
10. Grazing-intolerant species
11. Dry weight, litter depth, bare ground and understorey cover, by data point
12. Fuel components by data point
333 Appendix 1 Climate data for Cumberland Plain and comparison weather stations
Table A1.1. Location, elevation, and years of data available for eight weather stations on the Cumberland Plain. Stations in Sydney City (Observatory Hill) and the Blue Mountains (Katoomba) are included for comparison (italics). Source: Bureau of Meteorology 2004.
Location Latitude Longitude Elevation Years of data: Years of data: (°S) (°E) (m asl) temperature rainfall Richmond 33.62 150.75 20 60 123 Seven Hills 33.77 150.93 50 18 51 Orchard Hills 33.80 150.71 93 14 32 Prospect 33.82 150.91 61 38 118 Parramatta 33.82 151.00 15 40 100 Bankstown 33.92 150.99 7 34 27 Liverpool 33.93 150.91 20 38 39 Campbelltown 34.08 150.82 75 21 25 Sydney City 33.86 151.21 39 145 146 Katoomba 33.71 150.30 1030 70 118
Table A1.2. Monthly and yearly mean rainfall (mm) at eight Cumberland Plain and two comparison locations detailed in Table A1.1 (stations not on the Cumberland Plain in italics). The two months with the highest average rainfall in each location are highlighted in blue, those with the lowest average rainfall, in pink. Source: Bureau of Meteorology 2004.
Location Average rainfall (mm) Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Year Richmond 96 93 89 66 60 60 47 44 43 57 72 75 801 Seven Hills 106 111 108 75 75 77 45 58 45 71 82 71 923 Orchard Hills 103 109 95 70 67 46 40 43 38 61 76 68 817 Prospect 95 94 97 75 74 74 59 51 47 59 72 76 871 Parramatta 89 96 99 91 80 82 80 55 51 63 63 72 921 Bankstown 94 107 118 94 67 81 43 52 42 64 82 73 917 Liverpool 98 95 101 85 69 71 40 57 45 62 78 66 866 Campbelltown 91 79 101 64 60 82 34 50 41 74 84 71 830 Sydney City 103 117 131 127 123 128 98 82 69 77 83 78 1217 Katoomba 158 173 168 122 104 116 85 82 72 90 104 124 1399
334 Table A1.3. Monthly and yearly mean daily maximum temperatures (°C) at eight Cumberland Plain and two comparison locations detailed in Table A1.1 (stations not on the Cumberland Plain in italics). The month with the highest average temperature in each location is highlighted in pink, that with the lowest, in blue. Source: Bureau of Meteorology 2004.
Locations Mean daily maximum temperature (°C) Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Year Richmond 29.4 28.9 27.1 23.8 20.3 17.4 17.3 18.9 22.1 25.0 27.1 29.1 23.9 Seven Hills 28.3 27.7 27.0 24.1 20.0 17.5 17.4 18.6 21.5 23.8 26.3 28.4 23.3 Orchard Hills 28.3 27.8 26.5 23.7 20.2 17.3 17.1 18.9 21.8 23.9 25.8 28.5 23.5 Prospect 28.1 27.8 26.1 23.6 20.2 17.3 16.7 18.5 21.1 23.5 25.1 27.3 22.9 Parramatta 28.1 27.7 26.4 23.4 20.3 17.5 17.0 18.7 21.4 23.7 25.8 27.6 23.1 Bankstown 27.9 27.6 26.2 23.6 20.3 17.7 17.1 18.8 21.4 23.6 25.0 27.3 23.1 Liverpool 28.2 27.9 26.4 23.9 20.5 17.8 17.3 18.9 21.5 23.7 25.3 27.5 23.2 Campbelltown 28.2 28.4 26.8 24.0 20.4 17.6 17.2 18.6 21.3 23.5 25.8 27.9 23.5 Sydney City 25.8 25.7 24.7 22.4 19.3 16.9 16.2 17.7 19.9 22.0 23.6 25.1 21.6 Katoomba 23.1 22.4 20.1 16.5 12.9 9.8 9.2 11.0 14.3 17.5 20.1 22.4 16.6
Table A1.4. Monthly and yearly mean daily minimum temperatures (°C) at eight Cumberland Plain and two comparison locations detailed in Table A1.1 (stations not on the Cumberland Plain in italics. The month (or months) with the highest average temperature in each location is highlighted in pink, that with the lowest, in blue. Source: Bureau of Meteorology 2004.
Location Mean daily minimum temperature (°C) Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Year Richmond 16.8 16.8 15.0 11.3 7.3 4.7 3.2 4.4 7.1 10.5 13.1 15.5 10.5 Seven Hills 16.5 17.0 15.7 12.3 8.3 6.3 4.5 5.9 7.9 11.0 12.9 15.4 11.0 Orchard Hills 16.9 17.4 16.1 13.1 9.9 6.8 5.3 6.0 8.8 11.1 13.2 15.5 11.7 Prospect 17.5 17.7 16.1 13.1 10.1 7.4 6.1 6.8 9.3 12.0 14.1 16.4 12.3 Parramatta 16.7 16.6 14.9 11.5 8.2 5.7 4.5 5.4 7.7 10.8 13.3 15.6 10.9 Bankstown 18.0 18.0 16.2 12.7 9.8 6.7 5.0 6.0 8.6 11.8 14.1 16.6 12.0 Liverpool 17.6 17.7 15.9 12.6 9.3 6.3 4.7 5.9 8.3 11.5 13.8 16.2 11.7 Campbelltown 16.7 16.9 15.0 11.0 7.5 5.2 3.1 4.6 7.0 10.4 12.6 15.1 10.7 Sydney City 18.6 18.7 17.5 14.7 11.5 9.2 8.0 8.9 11.0 13.5 15.5 17.5 13.7 Katoomba 12.6 12.8 11.4 8.6 6.1 3.6 2.5 3.2 5.3 7.6 9.7 11.6 7.9
335 Appendix 2 Rainfall during the project
Data on monthly rainfall for the years 2001 to 2004 were sourced from the Bureau of Meteorology in January 2005, for eight locations on the Cumberland Plain, and two comparison locations. Recent data was available for some of the stations for which long-term climate data was reported in Appendix 1. However in some cases 2001-4 data came from nearby stations, as old stations were not longer operational. No data was available for Liverpool from January to August 2001, or for Orchard Hills for November and December 2004.
To characterise rainfall patterns during the course of the project, rainfall figures for each year (2001-2004) were compared with long-term average yearly rainfall from the same, or a nearby, station. Monthly rainfall figures were grouped into seasons (Autumn, March-May; Winter, June-August; Spring, September-November; Summer, December- February), and compared with average seasonal rainfall, for each location (Table A2.1).
Table A2.1. Percentage of mean yearly rainfall experienced at eight Cumberland Plain and two comparison suburbs/towns in 2001, 2002, 2003 and 2004 (locations not on the Cumberland Plain in italics). Column 2 - same: recent data was collected at the same site as the long-term data reported in Table A1.2; nearby: recent data was collected from the same local area as the long-term data, but not from exactly the same site. -, missing data. Source: Bureau of Meteorology, pers. comm. 2005.
Suburb/town Sites for Percent long-term average rainfall data falling in: collection 2001 2002 2003 2004 Richmond nearby 93 73 87 87 Seven Hills same 94 67 84 71 Orchard Hills same 89 62 80 - Prospect same 97 69 86 81 Parramatta nearby 100 84 99 77 Bankstown same 90 66 87 77 Liverpool nearby - 75 80 82 Campbelltown nearby 59 63 70 76 Observatory Hill same 112 71 99 82 Katoomba same 89 73 86 69
336 Appendix 3 Vital attributes, plant functional types and sensitivity to disturbance regimes
Noble and Slatyer (1980) initially formulated the typology presented below. The scheme was further elaborated by Noble and Gitay (1996), and was recently applied to fire management in NSW by Bradstock and Kenny (2003). The NSW Flora Fire Response Database (DEC 2002) uses categories based on Noble and Slatyer’s work, and the notes accompanying the Database define vital attributes, plant functional types and sensitivities to disturbance. While the descriptions in each of these sources are generally consistent, some anomalies do exist. This appendix draws on all the above documents, plus, to some extent, on my own understanding.
Noble and Slatyer’s first set of vital attributes address the method of persistence of plant species through disturbance. Four propagule-based methods likely to be ‘observed in nature’ are identified. (Text in brackets explains the genesis of the letters Noble and Slatyer use to symbolise each category.) • D – species in this category have propagules which are capable of long-distance dispersal at any time and thus are always available (dispersed propagules) • S – propagules are available at all life stages (including juvenile) and persist beyond the death of mature individuals. Propagule stores are not exhausted by a single disturbance event (long-lived propagule store) • G – propagules are available during the adult life stage, and persist beyond the death of mature individuals. Propagule stores are exhausted through germination by a single disturbance event (germinates) • C – propagules are available during the adult life stage only, and do not survive once adults die. This category includes many species which store short-lived propagules in the canopy (canopy)
Noble and Slatyer (1980) also identify a number of vegetative mechanisms of persistence: • V – adults and juveniles resprout, but adults lose all reproductively mature tissue and must regrow for a time before again becoming reproductive (vegetative) • U – adults and juveniles resprout, and adults rapidly regain their capacity to reproduce. This category includes species which flower rapidly post-fire (unaffected by disturbance) • W – adults resprout (and are reproductively mature) but juveniles are killed (w, a letter to follow u and v).
Some species may possess two mechanisms for persistence. After considering the possible combinations, Noble and Slatyer (1980) identified three additional patterns: • Δ (delta) – DU, DW – resprouting species with reproductive tissue whose adults are capable of long-distance dispersal of propagules
337 • Σ (sigma) – SU, SW, GU – resprouting species with reproductive tissue and stored propagules • Γ (gamma) – GW – resprouting species with reproductive tissue which have a propagule store exhausted by a single disturbance event. There are thus ten methods of persistence which are likely to be encountered in nature, according to Noble and Slatyer (1980).
The NSW Database (DEC 2002) does not use Roman letters to designate vital attribute categories for individual species. Rather, resprouting plants which also reproduce via propagules are given a resprouting code (U, V or W), plus a propagule code (D, S, G or C). This convention has been used in the current project.
In reality, almost all resprouting species in the Database are coded either U or V. Although W species (rapidly reproductive resprouters whose juveniles are killed by disturbance) undoubtedly exist, in practice virtually no information on the relative timing of fire tolerance and sexual maturity is available.
Noble and Slatyer’s second set of vital attributes address the question of when, in relation to disturbance, a species can establish and grow to maturity. An initial post- disturbance phase where there is little competition for resources is identified, as is a second phase during which competition for resources progressively increases. Three likely patterns are identified: • I – species able to establish only immediately after disturbance. In obligate seeders, this pattern results in single-age cohorts, with all individuals dating from just after the last disturbance (species intolerant of competition) • T – species in this category can establish and grow at any time after disturbance, and thus exhibit a mixed-age structure (species tolerant of competition) • R – species unable to establish immediately after disturbance as they require some conditions associated with an older community for recruitment. Almost all species in the NSW Database have been given an establishment code of either I or T. Logically, R species could be expected to be uncommon in regularly- disturbed communities.
Combining these two sets of vital attributes gives 30 possible species types, which Noble and Slatyer (1980) group into 14 categories according to their behaviour in post- disturbance replacement sequences. Noble and Slatyer call each of these unique behaviour patterns a “group”, while Bradstock and Kenny (2002), drawing on Noble and Gitay (1996) use the term ‘Plant Functional Type’ or PFT. Noble and Slatyer (1980) divide Group 10 into two subgroups, on the basis of differences in invasive properties. • Group 1 – DT, DVT, ST, VST, VGT, VCT, VT. These species always have either adults or propagules available and can establish at any time. They are therefore not sensitive to any disturbance regime.
338 • Group 2 – GT, CT. Species in this group are obligate seeders whose propagules cannot persist through two disturbance events. They will therefore become locally extinct if a second fire occurs before juveniles reach maturity. They can, however, establish at any time between disturbances, and thus are not vulnerable to long intervals between fires. • Group 3 – DI, VDI – Species in this group have propagules which can reach any site at any time. Noble and Slatyer therefore regard them as not sensitive to any disturbance regime. Bradstock and Kenny (2003), however, consider them vulnerable to persistent frequent disturbance. They are also regarded as partially sensitive to infrequent disturbance by DEC (2002), presumably because opportunities for establishment would be limited. • Group 4 – SI, VSI, VGI – These species can only establish after disturbance. Propagules persist after adults die, however if a disturbance does not occur while propagules remain viable, these species will become locally extinct. They are thus vulnerable to intervals between disturbances longer than life span plus propagule longevity. Obligate seeder species in this group are also vulnerable to multiple cycles of frequent disturbance within the juvenile period, as although propagules stores are not exhausted by a single disturbance event, they will be exhausted eventually(Bradstock and Kenny 2003). Resprouters too may decline if fires continue to occur at intervals shorter than primary and secondary juvenile periods. • Group 5 – GI – These obligate seeder species are vulnerable to long intervals between disturbances in the same way as Group 4 species. They differ in also being liable to local extinction in the face of one interval during the juvenile period, as propagule stores are exhausted by a single disturbance event. • Group 6 – CI – These obligate seeders, which include those that store short-lived propagules in the canopy, are also vulnerable in the face of both frequent and infrequent disturbance. Like Group 5 species, they will become locally extinct if exposed to an interval between disturbances of less than the juvenile period. As propagules do not persist after adults die, and establishment can only occur immediately after disturbance, they will also become locally extinct if the interval between fires exceeds their lifespan. • Group 7 – VI, VCI – These resprouter species can only establish new plants immediately after fire and do not have a mechanism for persistence after adults die. Thus an interval greater than the lifespan of individuals will cause local extinction. Resprouting provides protection from frequent fire, although repeated short intervals may prevent recruitment and thus lead to a decline in populations, as these species need time to again become reproductively mature after a disturbance (Bradstock and Kenny 2003). • Group 8 – DR, VDR, SR, VSR, VGR – Although these species can establish only some years after disturbance, they always have propagules available. Noble and Slatyer (1980) therefore consider them not sensitive to any disturbance regime. Bradstock and Kenny (2003), however, consider them sensitive to persistent frequent disturbance, presumably because frequently disturbed sites may rarely reach the stage of post-fire regeneration in which these species can establish. • Group 9 – GR, CR, VR, VCR – Although these combinations of vital attributes are theoretically possible, they are unlikely to be observed in nature as these species
339 would become locally extinct at the first disturbance event, according to Noble and Slatyer (1980), although VR and VCR species could resprout. • Group 10a – UDT, WDT, UST, WST, UGT, WGT, UCT, WCT, UT, WT. These species resprout after disturbance and can establish any time after it. They are therefore not vulnerable to any disturbance regime. Noble and Gitay (1996) call them ‘superspecies’. • Group 10b – UDR, WDR, USR, WSR, UGR, WGR, UCR, WCR, UR, WR. These species resprout after disturbance and can establish new plants once the immediate post-fire period is past. Noble and Slatyer (1980) do not consider them vulnerable to any disturbance regime. Bradstock and Kenny (2003) and DEC (2002), however, list them as vulnerable to persistent frequent disturbance, again presumably because frequently disturbed sites may not reach the stage of post-fire regeneration necessary for new plants to establish. • Group 11 – UDI, WDI – Noble and Slatyer (1980) do not consider these resprouting species with long-distance dispersal of propagules sensitive to any disturbance regime. Bradstock and Kenny (2003) and DEC (2002) consider them sensitive to persistent frequent disturbance. These species would have at least the same protection against frequent disturbance as DI species, but with the additional protection of resprouting. • Group12 – USI, WSI, UGI – Although these species resprout and have long-lived propagules, disturbance frequencies longer than lifespan of adults plus propagules will lead to local extinction, as it does for Group 4 species. Again, Bradstock and Kenny (2003) and DEC (2002) consider them vulnerable to persistent frequent disturbance, presumably because propagule stores may be exhausted and adult plants may eventually die if frequently disturbed. • Group 13 – WGI – These species resprout, and have propagules which persist after adult plants die. However as they can only establish in the immediate post-fire period, eventually the entire population will age and die without replacement. Intervals between disturbance events greater than the lifespan of adults plus propagules will therefore lead to local extinction. These species are also considered vulnerable to frequent disturbance by all authors, because propagule stores would be exhausted by the first disturbance. However resprouting would be protective in the same manner as it is for group 7 species; persistent frequent disturbance, however, could lead to decline. I have therefore rated them sensitive to persistent frequent disturbance. • Group 14 – UI, WI, UCI, WCI – These plants resprout and reproduce either vegetatively or through short-lived propagules which germinate readily, but reproduction can only take place in the immediate post-fire period. If disturbance does not occur before plants age and die, then they will become locally extinct. Bradstock and Kenny (2002) also consider them vulnerable to persistent frequent disturbance, although Noble and Slatyer (1980) do not. This may apply to those UCI shrub species which, although flowering relatively rapidly post-fire, take some time for seedlings to become fire-tolerant.
Bradstock and Kenny (2003) point out that although species is groups 4, 5, 12 and 13 are theoretically vulnerable to infrequent fire, longevity of propagules is rarely known.
340 Noble and Slatyer’s third set of vital attributes concerns the timing of life stages of a species. They are: • m – the time taken for a species to reach reproductive maturity • l – the lifespan of the species in an undisturbed environment. For T and R species this is effectively infinite, since they form self-maintaining populations. • e – the time taken for all propagules to be lost from the community.
Sensitivities to disturbance regimes are summarised in the table below. Sensitivity categories are: 1. regime leads to local decline or extinction 2. persistent regime likely to lead to local decline or extinction 3. regime unlikely to lead to local decline or extinction
Table A3.1. Sensitivity to frequent and infrequent disturbance of species in 14 plant functional types.
Plant Vital attributes Sensitivity to Sensitivity to functional frequent infrequent type disturbance disturbance 1 DT, VDT, ST, VST, VGT, VCT, VT 3 3 2 GT, CT 1 (
341 Appendix 4 Regeneration modes of Cumberland Plain shrub species
Table A4.1. Regeneration modes of Cumberland Plain shrub species. S, obligate seeder; R, resprouter; r, usually killed by fire but sometimes resprouts; s, usually resprouts but sometimes killed; pf, post-fire; NFRR, National Fire Response Register; NR, Nature Reserve. Reliability ratings (column 3): 1, tested by following the fate of individual plants through time, at more than one site; 2, tested by following the fate of individual plants through time, at a single site; 3. post-burn observation in three or more sites; 4, post-burn observation in two sites; 5, post-burn observation in one site. See Figure 3.1 for site locations.
Regeneration Relia- Evidence Taxon mode bility (dates of observations in brackets) Acacia brownii R 3 Resprouting observed at various sites, including: • Nutt Road 7 mths pf (7/02) • Taylor Road 8 mths pf (9/04) • Llandilo 9 mths pf (8/02) • Castlereagh NR 10 mths pf (9/03) Acacia bynoeana R 3 Resprouting observed at: Taylor Road 12 mths pf (12/04) – some seedlings also present
342 • • Castlereagh NR 14 mths pf (1/04) • Devlin Road 2.8 yrs pf (9/04) Acacia decurrens S 3 Dead adults and live seedlings observed at: • Lansdowne 16 mths pf (4/03); • Holsworthy 12 mths pf (12/03). Individuals planted at Hoxton Park did not recover after fire (pers. comm. David Warren 2001). Large adults subject to low intensity fires have been observed to survive where leaf scorch is less than 100% eg at Landsdowne. Acacia elongata R 3 Resprouting observed at various sites, including: • Devlin Road 3 mths pf (2/02) • Shanes Park 6 mths pf (6/02) • Llandilo 9 mths pf (8/02) • Castlereagh NR 14 mths pf (1/04) Post-fire seedlings co-occurred with resprouting plants at Shanes Park (6/02). Senescence and death of adult plants observed in long unburnt areas at Shanes Park (9/04, 11/04). Regeneration Relia- Evidence Taxon mode bility (dates of observations in brackets) Acacia falcata S 3 Dead adults and live seedlings observed at many sites including: • Shanes Park 5 mths pf (4/03) • Windsor Downs 10 mths pf (8/03) • Holsworthy 13 mths pf (1/04) • Lansdowne 17 mths pf (4/03) Senescence and death of adult plants observed 7 yrs post-fire at Windsor Downs (9/04), 10 yrs post-fire at Scheyville (9/03) and in long-unburnt areas at Shanes Park (9/04, 11/04). Acacia implexa R 3 Resprouting from root suckers observed at: • Mt Annan 4 mths pf (1/03); • Orchard Hills 15 mths pf (3/03); • Prospect 11 mths pf (8/03). Acacia parramattensis R 3 Resprouting from root suckers observed at many sites including: • Hoxton Park (D. Warren pers. comm. 2001); • Nurragingy 4 mths pf (4/02); • Windsor Downs 3 mths pf (4/03). 343 Acacia ulicifolia Sr 3 • Dead adults and seedlings observed approx 4 mths pf at Mirambeena (1/02) • Seedlings but no adults observed at Castlereagh NR 23 mths pf (10/04) • Seedlings but no adults observed at Post Office Road 23 mths pf (10/04) • Two resprouting plants observed at Taylor Road 8 mths pf (9/04) Allocasuarina littoralis S 3 Dead adults and no resprouts observed at: • Windsor Downs 10 mths pf (8/03) • Mulgoa 2.7 yr pf (9/03) Dead adults and many seedlings observed at Castlereagh NR 23 mths pf (10/04) – however some plants had survived where leaf scorch was not complete. Astroloma humifusum R 5 Resprouting observed at Holsworthy 13 mths pf (1/04). Baeckea diosmifolia R 3 Resprouting observed at: • Taylor Road 8 mths pf (9/04) • Castlereagh NR 10 mths pf (9/03) • Nutt Road 11 mths pf (11/04) • Post Office Road 23 mths pf (9/04) Banksia oblongifolia R 3 Resprouting observed at various sites, including: • Nutt Road 2 mths pf (2/02) • Devlin Road 3 mths pf (2/02) • Taylor Road 2 mths pf (3/04) Regeneration Relia- Evidence Taxon mode bility (dates of observations in brackets) Banksia spinulosa var. spinulosa R 3 Resprouting observed at various sites, including: • Nutt Road 2 mths pf (2/02) • Devlin Road 3 mths pf (2/02) • Taylor Road 8 mths pf (9/04) • Llandilo 9 mths pf (8/02) Billardiera scandens var. R 3 Resprouting observed at: scandens • Shanes Park 11 mths pf (8/03) • Holsworthy 13 mths pf (1/04) • Llandilo 22 mths pf (9/03) Boronia polygalifolia R 4 Resprouting observed at: • Windsor Downs 10 mths pf (8/03) • Shanes Park 11 mths pf (8/03) Bossiaea obcordata R 3 Resprouting observed at: • Llandilo 9 mths pf (8/02) • Windsor Downs 10 mths pf (8/03) • Taylor Road 12 mths pf (9/04) • Castlereagh NR 14 mths pf (1/04) 344 Bossiaea prostrata R 3 Resprouting observed at numerous sites, including: • Nurragingy 4 mths pf (4/02) • Scheyville 6 mths pf (5/02) • Ropes Creek 9 mths pf (2/03) • Holsworthy 12 mths pf (12/03) Bossiaea rhombifolia R 3 Resprouting observed at: • Llandilo 9 mths pf (8/02) • Devlin Road 9 mths pf (8/02) • Castlereagh NR 10 mths pf (9/03) • Taylor Road 12 mths pf (9/04) Brachyloma daphnoides R 3 Resprouting observed at: • Castlereagh NR 14 mths pf (1/04) • Llandilo 22 mths pf (9/03) • Castlereagh NR 23 mths pf (10/04) • Post Office Road 24 mths pf (10/04) Breynia oblongifolia R 4 Small numbers of plants observed resprouting at: • Prospect 13 mths pf (10/03) • Holsworthy 13 mths pf (1/04) Regeneration Relia- Evidence Taxon mode bility (dates of observations in brackets) Bursaria spinosa R 2 • Vigorous resprouting observed at many sites across the Cumberland Plain, including nine sites surveyed for landscape study (Chapter 5) • Experimental evidence from Mt Annan (Chapter 8) Callistemon linearis R 4 Resprouting observed at: • Tadmore Road 8 mths pf (9/04) • Devlin Road 23 mths pf (9/04) Callistemon pinifolius R 3 Resprouting observed at: • Tadmore Road 8 mths pf (9/04) • Nutt Road 21 mths pf (9/03) • Devlin Road 23 mths pf (9/04) • Castlereagh NR 23 mths pf (10/04) Calotis lappulacea R 4 Resprouting observed at: • Scheyville 4 mths pf (3/02) • Lansdowne 15 mths pf (3/03) 345 Chorizema parviflorum R 3 Resprouting observed at various sites, including: • Shanes Park 6 mths pf (7/02) • Plumpton 8 mths pf (8/03) • Ropes Creek 9 mths pf (7/03) • Holsworthy 12 mths pf (12/03) Clematis glycinoides R 5 Resprouting observed at Prospect 12 mths pf (10/3). Clerodendrum tormentosum R 5 Resprouting observed at Mt Annan 15 mths pf (12/03). Conospermum taxifolium R 4 Resprouting observed at: • Taylor Road 2 mths pf (3/04) • Nutt Road 11 mths pf (1/04) Cryptandra amara var. amara S 3 Seedlings but no adults observed at: • Taylor Road 8 mths pf (9/04) • Shanes Park 18 mths pf (7/03) • Castlereagh NR 23 mths pf (10/04) Seedlings of this species appeared later than those of other seeder species. Cryptandra spinescens R 5 Resprouting observed at Shanes Park 11 mths pf (8/03) Daviesia genistifolia R 5 Numerous plants (over 100) observed resprouting at Prospect over a wide area 13 mths pf (10/03). Seedlings often co-occurred with resprouting plants. Regeneration Relia- Evidence Taxon mode bility (dates of observations in brackets) Daviesia squarrosa R 3 Resprouting observed at: • Nutt Road 7 mths pf (8/02) • Plumpton 8 mths pf (8/03) • Tadmore Road 8 mths pf (8/04) • Devlin Road 9 mths pf (8/02) Daviesia ulicifolia Sr 3 Dead adults and large numbers of live seedlings observed at many sites, including: • Nurragingy 6 mths pf (6/02) • Shanes Park 6 mths pf (6/02) • Taylor Road 12 mths pf (9/04) • Lansdowne 18 mths pf (4/03) • Post Office Road 24 mths pf (10/04) This a consistent observation across Cumberland Plain Woodland, Shale-Gravel Transition Forest and some Castlereagh Woodland sites, however this species was also observed resprouting in other Castlereagh Woodland sites, eg Llandilo 8mths pf (8/02); Castlereagh NR 14 mths pf (1/04). 346 Dillwynia rudis (previously D. S 3 Many seedlings but no adults observed at various sites, including: sericea) • Nutt Road 7 mths pf (7/02) • Devlin Road 8 mths pf (7/02) • Post Office Road 23 mths pf (10/04) • Castlereagh NR 23 mths pf (10/04) Dillwynia sieberi S 3 Dead adults and live seedlings observed at many sites including: • Nurragingy 6 mths pf (6/02); • Orchard Hills 7 mths pf (7/02); • Lansdowne 18 mths pf (4/03). Dillwynia tenuifolia S 3 Many seedlings but no adults observed at:: • Shanes Park 6 mths pf (6/02) • Nutt Road 8 mths pf (8/02) • Tadmore Road 8 mths pf (9/04) • Windsor Downs 10 mths pf (8/03) Senescence and death of adult plants observed 10 yrs post-fire at Scheyville (9/03) and in long-unburnt areas at Shanes Park (9/04, 11/04). Dodonaea falcata S 4 Many seedlings but no adults observed at: • Shanes Park 5 mths pf (6/02) • Castlereagh NR 23 mths pf (10/04) Dodonaea triquetra S 5 Seedlings but no adults observed at Castlereagh NR 23 mths pf (10/04) Regeneration Relia- Evidence Taxon mode bility (dates of observations in brackets) Dodonaea viscosa ssp. cuneata Sr 4 Dead adults and many seedlings observed at: • Mt Annan 14 mths pf (10/03); • Orchard Hills 15 mths pf (3/03); • Prospect 12 mths pf (10/03). However a small number of plants were observed resprouting 11 mths pf at Shanes Park (11/02), and at Prospect 11 mths pf (8/03). Eremophila debilis R 3 Resprouting observed at many sites including: • Bligh Park 3 mths pf (11/04) • Scheyville 3 mths pf (1/02); • Mulgoa 3 mths pf (3/02); • Lansdowne 16 mths pf (3/03). Exocarpos cupressiformis R 3 Resprouting observed at many sites including: • Plumpton 2 yrs pf (10/02); • Windsor Downs 10 mths pf (8/03);
347 • Holsworthy 12 mths pf (12/03) • Scheyville 21 mths pf (8/03). Exocarpos strictus R 3 Resprouting observed at: • Holsworthy 12 mths pf (10/02); • Castlereagh NR 23 mths pf (10/04) • Mulgoa 2.7 yrs pf (9/04) Gompholobium glabratum S 5 Seedlings but no adults observed at Post Office Road 23 mths pf (9/04). Gompholobium inconspicuum S 5 Seedlings but no adults observed at Post Office Road 24 mths pf (10/04) Gompholobium minus R 3 Resprouting observed at many Castlereagh Woodland sites, including: • Castlereagh NR 10 mths pf (9/03) • Taylor Road 12 mths pf (9/04) • Holsworthy 13 mths pf (1/04) • Llandilo 22 mths pf (9/03) Gompholobium pinnatum S 4 Seedlings but no adults observed at: • Castlereagh NR 23 mths pf (10/04) • Post Office Road 24 mths pf (10/04) Grevillea juniperina ssp, S 3 • Many seedlings but no adults observed at Plumpton 8 mths pf (8/03), juniperina • Dead adults and many seedlings observed at Shanes Park 11 mth pf (8/03). • Many seedlings observed at Castlereagh NR 23 mths pf (10/04) Regeneration Relia- Evidence Taxon mode bility (dates of observations in brackets) Grevillea mucronulata R 3 Resprouting observed at various sites including: • Nutt Road 7 mths pf (7/02) • Llandilo 9 mths pf (8/02) • Windsor Downs 10 mths pf (8/03) • Castlereagh NR 14 mths pf (1/04) Hakea laevipes (previously H. R 3 Resprouting observed at various sites including: dactyloides) • Devlin Road 3 mths pf (2/02) • Taylor Road 8 mths pf (8/04) • Llandilo 9 mths pf (8/02) • Castlereagh NR 10 mths pf (9/03) Hakea sericea S 3 Many dead adults and numerous seedlings observed in various sites including: • Devlin Road 8 mths pf (7/02) • Taylor Road 8 mths pf (9/04) • Llandilo 9 mths pf (8/02) • Windsor Downs 10 mths pf (8/03) 348 Hardenbergia violacea R 3 Resprouting observed at many CPW sites including: • Mulgoa 6 mths pf (6/02) • Lansdowne 4 mths pf (3/03) • Plumpton 8 mths pf (8/03) • Prospect 12 mths pf (9/03) Hibbertia aspera R 4 Resprouting observed at: • Holsworthy 12 mths pf (12/03) • Windsor Downs 23 mths pf (9/04) Seedlings were also found at Holsworthy. Hibbertia diffusa R 3 Many plants observed resprouting at various sites including: • Shanes Park 5 mths pf (6/02) • Plumpton 8 mths pf (8/03) • Ropes Creek 9 mths pf (7/03) • Orchard Hills 15 mths pf (4/03) Hibbertia pedunculata R 5 Resprouting observed at Holsworthy 13 mths pf (1/04) Regeneration Relia- Evidence Taxon mode bility (dates of observations in brackets) Hovea linearis Rs 3 Resprouting observed at: • Nutt Road 9 mths pf (9/02) • Taylor Road 12 mths pf (9/04) • Castlereagh NR 23 mths pf (10/04) • Post Office Road 23 mths pf (9/04) Seedlings of this species were also observed in various sites, generally with resprouting plants but sometimes without (eg Taylor Road, 8 mths pf, 9/04) Indigofera australis Rs 3 Resprouting observed at various sites including: • Holsworthy 12 mths pf (1/02); • Shanes Park 6 mths pf (6/02); • Mt Druitt 6 mths pf (8/03). However there was also evidence of the death of some adult plants at Shanes Park (6/02), and post-fire seedlings often co-occurred with resprouting plants. Isopogon anemonifolius R 3 Resprouting observed at various Castlereagh woodland sites including: • Nutt Road 7 mths pf (7/02) • Taylor Road 8 mths pf (9/04) 349 • Llandilo 9 mths pf (8/02) • Castlereagh NR 10 mths pf (9/03) Jacksonia scoparia R 3 Resprouting observed at: • Windsor Downs 10 mths pf (8/03) • Shanes Park 10 mths pf (8/03) • Holsworthy 13 mths pf (1/04) Kunzea ambigua Sr 4 Many large dead adults, but also a few resprouting plants, observed at: • Devlin Road 23 mths and 2.7 yr pf (9/04) • Nutt Road 2.8 yrs pf (11/04) Resprouting observed at Shanes Park 11 mths pf (8/03) Kunzea capitata R 3 Resprouting observed at: • Nutt Road 8 mths pf (8/02) • Taylor Road 8 mths pf (9/04) • Devlin Road 9 mths pf (8/02) • Windsor Downs 10 mths pf (8/03) Leptospermum parvifolium R 3 Resprouting observed at many Castlereagh woodland sites including: • Devlin Road 10 mths pf (9/02) • Nutt Road 21 mths pf (9/03) • Llandilo 22 mths pf (9/03) Regeneration Relia- Evidence Taxon mode bility (dates of observations in brackets) Leptospermum polygalifolium R 3 Resprouting observed at: ssp. polygalifolium • Windsor Downs 10 mths pf (8/03) • Post Office Road 23 mths pf (9/04) • Devlin Road 23 mths pf (9/04) Leptospermum trinervium R 3 Resprouting observed at: • Nutt Road 7 mths pf (7/02) • Llandilo 8 mths pf (7/02) • Taylor Road 8 mths pf (8/04) • Castlereagh NR 10 mths pf (9/03) Leucopogon virgatus R 5 Resprouting observed at Castlereagh NR 23 mths pf (10/04) Lissanthe strigosa R 3 Resprouting observed at many sites, including: • Shanes Park 5 mths pf (6/02) • Plumpton 8 mths pf (8/03) • Tadmore Road 8 mths pf (8/03) • Ropes Creek 9 mths pf (7/03) Macrozamia spiralis R 3 Resprouting observed at: 350 • Llandilo 3 mths pf (2/02) • Shanes Park 11 mths pf (8/03) • Holsworthy 13 mths pf (1/04) • Castlereagh NR 23 mths pf (10/04) Maytenus silvestris R 5 Resprouting observed at Holsworthy 13 mths pf (1/04) Melaleuca erubescens R 3 Resprouting observed at: • Shanes Park 2 mths pf (11/04) • Windsor Downs 10 mths pf (8/03) • Castlereagh NR 10 mths pf (9/03) Melaleuca nodosa R 3 Resprouting observed at many sites, including: • Taylor Road 8 mths pf (9/04) • Llandilo 9 mths pf (8/02) • Devlin Road 9 mths pf (8/02) • Windsor Downs 10 mths pf (8/03) Melaleuca thymifolia R 3 Resprouting observed at: • Tadmore Road 8 mths pf (9/04) • Taylor Road 8 mths pf (9/04) • Holsworthy 13 mths pf (1/04) • Post Office Road 23 mths pf (9/04) Regeneration Relia- Evidence Taxon mode bility (dates of observations in brackets) Micromyrtus minutiflora S 5 Seedlings but no adults observed at: • Post Office Road 24 mths pf (10/04) Mirbelia rubiifolia S 4 Seedlings but no adults observed at: • Castlereagh NR 23 mths pf (10/04) • Post Office Road 24 mths pf (10/04) Monotoca scoparia R 3 Resprouting observed at: • Llandilo 9 mths pf (8/02) • Nutt Road 9 mths pf (10/02) • Taylor Road 12 mths pf (9/04) • Castlereagh NR 23 mths pf (10/04) Olearia viscidula R 5 Resprouting observed at Holsworthy 13 mths pf (1/04) Ozothamnus diosmifolius Sr 3 • Dead adults and seedlings observed at Shanes Park 5 mths pf (6/02). • Seedlings but no adults observed 12 mths pf at Holsworthy (1/04). • Didn't recover from burn at Hoxton Park 30 mths after germination (David Warren pers.
351 comm. 2001). However a few individuals observed resprouting at Windsor Downs 10 mths pf (8/03). Many seedlings and 1 resprouting plant observed at Llandilo 22 mths pf (9/03). Senescence and death of adult plants observed 7 yrs post-fire at Windsor Downs (9/04), 10 yrs post-fire at Scheyville (9/03) and in long-unburnt areas at Shanes Park (9/04, 11/04). Persoonia laurina R 3 Resprouting observed at: • Nutt Road 7 mths pf (7/02) • Llandilo 9 mths pf (8/02) • Taylor Road 12 mths pf (9/04) • Castlereagh NR 23 mths pf (10/04) Persoonia linearis R 3 Resprouting observed at: • Shanes Park 11 mths pf (8/03); • Holsworthy 13 mths pf (1/04); • Castlereagh NR 23 mths pf (10/04). Persoonia nutans S 4 Scattered seedlings but no adults observed at: • Castlereagh NR 23 mths pf (10/04) • Post Office Road 23 mths pf (9/04) NPWS Threatened Species information describes this species as an obligate seeder (NPWS 2003) Regeneration Relia- Evidence Taxon mode bility (dates of observations in brackets) Petrophile pulchella S 3 • Dead adults observed at Nutt Road 7 mths pf (7/02) • Dead adults observed at Taylor Road 8 mths pf (9/04) • Seedlings but no adults observed at Nutt Road 2.7 yrs pf (8/02) • Dead adults and seedlings observed at Devlin Road 2.8 yrs pf (9/04) Philotheca salsolifolia R 3 Resprouting observed at: • Nutt Road 7 mths pf (7/02) • Devlin Road 8 mths pf (7/02) • Taylor Road 12 mths pf (9/04) • Castlereagh NR 23 mths pf (10/04) Phyllanthus hirtellus R 3 Resprouting observed at: • Nutt Road 7 mths pf (7/02) • Tadmore Road 8 mths pf (9/04) • Taylor Road 12 mths pf (9/04) • Windsor Downs 10 mths pf (8/03) Pimelea curviflora var. S 5 Many seedlings but no adults observed at Prospect 12 mths pf (10/03). subglabrata Pimelea linifolia ssp. linifolia S 3 Many seedlings but no adults observed at:
352 • Devlin Road 8 mths pf (7/02) • Taylor Road 8 mths and 12 mths pf (8/04 – 2 different fires) • Windsor Downs 10 mths pf (8/03) • Holsworthy 13 mths pf (1/04) This finding is consistent across Castlereagh Woodland sites, although this species has been observed resprouting in vegetation types not included in this study. Pimelea spicata R 4 Resprouting observed at: • Mt Annan 4 mths pf (1/03) • Prospect 12 mths pf (9/03) Post-fire seedlings co-occurred with resprouting plants at both sites. Platysace ericoides R 4 Resprouting observed at: • Nutt Road 7 mths pf (7/02) • Devlin Road 8 mths pf (7/02) Podolobium scandens R 3 Resprouting observed at: • Shanes Park 6 mths pf (5/02) • Ropes Creek 9 mths pf (7/03) • Plumpton 8 mths pf (8/03) • Windsor Downs 10 mths pf (8/03) Regeneration Relia- Evidence Taxon mode bility (dates of observations in brackets) Pomax umbellata S 3 Seedlings but no adults observed at: • Holsworthy 13 mths pf (1/04) • Castlereagh NR 23 mths pf (10/04) • Post Office Road 24 mths pf (10/04) Seedlings observed in long unburnt area at Shanes Park (11/04). Ready establishment in absence of fire also observed on sandstone. Prostanthera scutellarioides S 5 Seedlings but no adults observed 23 mths pf at Castlereagh NR (10/04) Pultenaea elliptica R 3 Resprouting observed at various CW sites including: • Nutt Road 7 mths pf (7/02) • Devlin Road 8 mths pf (6/02) • Tadmore Road 8 mths pf (9/04) • Castlereagh NR 14 mths pf (1/04) Pultenaea microphylla S 4 • Many seedlings but no adults observed at Plumpton 8 mths pf (8/03); • Very numerous seedlings, and no adults, over a wide area at Prospect 11 mths pf (8/03) However adult plants were found surviving in frequently burnt areas at Shanes Park, possibly in unburnt patches, or perhaps resprouting after low intensity fire. 353 Pultenaea parviflora S 3 • Many seedlings but no adults observed at Shanes Park 5 mths pf (6/02) • Many seedlings but no adults observed at Castlereagh NR 23 mths pf (10/04) • Dead adults and seedlings observed at Windsor Downs 10 mths pf (8/03). Senescence and death of adult plants observed in long-unburnt areas at Shanes Park (9/04, 11/04). Pultenaea villosa S 3 • Dead adults and many seedlings observed at Rickards Road 9 mths pf (10/04) • Dead adults and many seedlings observed at Holsworthy 13 mths pf (1/04) • Many seedlings and no adults observed at Post Office Road 24 mths pf (10/04) Rubus parvifolius R 5 Large robust plant observed 12 mths pf at Prospect. Solanum cinereum R 2 80% of tagged plants resprouted after a moderate intensity planned burn at Mt Annan in September 2002 (pers. comm. Jocelyn Howell 2003). Post-fire seedlings of this species were also relatively common at Mt Annan (12/03). Styphelia laeta ssp. laeta S 5 Dead adults observed at Windsor Downs 10 mths pf (8/03) Neither resprouts nor seedlings were encountered in recently-burnt areas. May establish in later post-fire years, through animal dispersal.
Appendix 5 Juvenile periods of Cumberland Plain shrub species
Table A5.1. Juvenile periods of Cumberland Plain shrub species. S, obligate seeder; R, resprouter; r, usually killed by fire but sometimes resprouts; s, usually resprouts but sometimes killed; pf, post-fire; NR, Nature Reserve; uk, unknown. Reliability ratings (columns 4 and 7): 1, tested by following the fate of individual plants through time, at more than one site; 2, tested by following the fate of individual plants through time, at a single site; 3. post-burn observation in three or more sites; 4, post-burn observation in two sites; 5, post-burn observation in one site. Columns 5 and 8: dates of observations in brackets. See Figure 3.1 for site locations.
Regener- Primary juvenile period Secondary juvenile period ation Taxon mode Years Relia- Evidence Years Relia- Evidence (App 4.2) bility bility Acacia brownii R uk 1 - 2 3 Budding observed: • 10 mths pf at Windsor Downs (8/03) Flowering observed: • 12 mths pf at Taylor Road (9/04) • 20 mths pf at Devlin Road (7/03)
354 • 20 mths pf at Shanes Park (9/03) • 21 mths pf at Nutt Road (9/03) Acacia bynoeana R uk 2 - 3 4 Flowering observed: • 14 mths pf at Castlereagh NR (1/04) • 2.2 yrs pf at Nutt Road (6/02) Acacia decurrens S 3 - 7 4 • Some plants flowering 2.7 yrs pf at na na na Orchard Hills (9/04) • Plants 3.0 yrs pf at Lansdowne were 1.8m tall, but bore no fruit (10/04) • Flowering approx 7 yrs pf at Plumpton (8/04) Acacia elongata R > 3 2 Three seedlings monitored at Shanes 2 3 Flowering observed: Park had not flowered by 2.8 yrs pf • 21 mths pf at Shanes Park (8/03) • 22 mths pf at Llandilo (9/03) • 23 mths pf at Devlin Road (9/04) Fruiting observed: • 21 mths pf at Nutt Road (9/03) Regener- Primary juvenile period Secondary juvenile period ation Taxon mode Years Relia- Evidence Years Relia- Evidence (App 4.2) bility bility Acacia falcata S 2 - 3 2 One of eight seedlings monitored at na na na Shanes Park flowered at 2.4 yrs pf, as did some other plants in the area (6/04) Fruits observed: • 23 mths pf at Castlereagh NR (10/04) • 23 mths pf at Post Office Road (9/04) • 2.7 yrs pf at Mulgoa (9/04) • 2.7 yrs pf at Orchard Hills (9/04) • 2.8 yrs pf at Devlin Road (9/04) Acacia parramattensis R uk 3 5 Budding observed 22 mths pf at Lansdowne (10/04) 355 Allocasuarina littoralis S 5 4 Fruiting observed: na na na • 4.3 yrs pf at Windsor Downs (3/02) • 4.7 yrs pf at Castlereagh NR (10/04) Baeckea diosmifolia R uk 1 - 2 3 Flowering observed: • 11 mths pf at Nutt Road (11/04) • 17 mths pf at Devlin Road (4/03) • 23 mths pf at Post Office Road (9/04) Banksia oblongifolia R uk 3 - 4 3 • Flowering observed 2.3 yrs pf at Devlin Road • A few plants with cones observed 2.8 yrs pf at Nutt Road (but many without) (11/04) • Burnt cones observed post-fire at Taylor Road after a 4-yr interfire interval (3/04) Regener- Primary juvenile period Secondary juvenile period ation Taxon mode Years Relia- Evidence Years Relia- Evidence (App 4.2) bility bility Banksia spinulosa var. R uk Two seedlings were tagged as part of 2 - 3 3 Flowering observed: spinulosa Castlereagh monitoring, however they • 23 mths pf at Castlereagh NR had both died by 14 mths pf. (10/04) • 2.5 yrs pf at Nutt Road (6/02) • 2.7 yrs pf at Tadmore Road (9/04) • 2.8 yrs pf at Devlin Road (9/04) No flowers 21 mths pf at Nutt Road (9/03), nor 22 mths pf at Llandilo (9/03). Billardiera scandens R uk 2 5 Flowering observed 20 mths pf at var. scandens Shanes Park (9/03) Boronia polygalifolia R uk 1 4 Flowering observed: • 10 mths pf at Windsor Downs (8/03) • 11 mths pf at Shanes Park (8/03) 356 Bossiaea obcordata R uk 2 5 Flowering observed 23 mths pf at Castlereagh NR (10/04) Bossiaea prostrata R uk 1 - 2 3 • Buds observed 10 mths pf at Windsor Downs (8/03) • Buds observed 11 mths pf at Shanes Park (8/03) • Flowers and fruits observed 22 mths pf at Lansdowne (10/03) Bossiaea rhombifolia R uk 3 5 Flowering observed 2.8 yrs pf at Devlin Road (9/04) Brachyloma R uk 2 3 Flowering observed: daphnoides • 22 mths pf at Llandilo (9/03) • 23 mths pf at Castlereagh NR (10/04) • 24 mths pf at Post Office Road (10/04) Regener- Primary juvenile period Secondary juvenile period ation Taxon mode Years Relia- Evidence Years Relia- Evidence (App 4.2) bility bility Bursaria spinosa R > 2 2 Post-fire seedlings tagged at Mt Annan 2 - 3 3 Flowering observed: were no more than 15cm high 2.1 yrs • 16 mths pf at Mt Annan (1/04). pf, and had not flowered. Fruiting observed: • 23 mths pf at Windsor Downs (9/04) • 2.7 yrs pf at Orchard Hills (9/04) Callistemon linearis R uk 2 - 3 4 • Budding observed 23 mths pf at Devlin Road (9/04) • Flowering observed 2.8 yrs pf at Nutt Road (11/04) Callistemon pinifolius R uk 1 - 3 3 Budding observed: • 8 mths pf at Taylor Road (9/04) • 21 mths pf at Nutt Road (9/03) 23 mths pf at Castlereagh NR 357 • (10/04) • 23 mths pf at Devlin Road and Post Office Road (9/04) Flowering observed 2.8 yrs pf at Nutt Road (11/04) Calotis lappulacea R uk 1 - 2 4 Flowering observed: • 4 mths pf at Scheyville (3/02) • 15 mths pf at Lansdowne (3/03) Chorizema parviflorum R uk 1 3 Flowering observed: • 3 mths pf at Shanes Park (2/02) • 8 mths pf at Plumpton (8/03) • 9 mths pf at Ropes Creek (7/03) Conospermum R uk 1 5 Flowering observed 11 mths pf at Nutt taxifolium Road (11/04) Regener- Primary juvenile period Secondary juvenile period ation Taxon mode Years Relia- Evidence Years Relia- Evidence (App 4.2) bility bility Cryptandra amara S 3 1 • All five seedlings monitored at na na na Shanes Park had flowered by 2.7 yrs pf. • The single seedling monitored at Castlereagh had well-developed buds by 2.6 yrs pf, and a nearby plant was flowering (6/04) Cryptandra spinescens R uk 1 5 Flowering observed 11 mths pf at Shanes Park (8/03) Daviesia genistifolia R uk 1 5 Many plants observed flowering 12 mths post-fire at Prospect (9/03). Daviesia squarrosa R uk 1 - 2 3 • Flowering observed 8 mths pf at Tadmore Road (9/04) • Flowering observed 21 mths pf at 358 Nutt Road (9/03) • Flowering observed 23 mths pf at Devlin Road (9/04) • Flowering and fruiting observed 23 mths pf at Devlin Road (9/04) Regener- Primary juvenile period Secondary juvenile period ation Taxon mode Years Relia- Evidence Years Relia- Evidence (App 4.2) bility bility Daviesia ulicifolia Sr 2 - 4 2 • One of 13 seedlings monitored at 2 3 Flowering on resprouts observed: Shanes Park had flowered and • 22 mths pf at Llandilo (9/03) fruited by 2.8 yrs pf. • 23 mths pf at Devlin Road (9/04) • Fruits observed 21 mths pf at • 23 mths pf at Castlereagh NR Mulgoa East (10/03) (10/04) • Flowering and fruiting observed 22 mths pf at Lansdowne (10/04) • Many seedlings flowering 23 mths pf at Windsor Downs (9/04) • Many seedlings fruiting 24 mths pf at Post Office Road (10/04) • Many plants flowering 2.7 yr pf at Mulgoa (9/04) 359 • However no flowers 3.1 yrs pf at Mt Druitt, nor 2.9 yrs pf at Plumpton (8/03) Dillwynia rudis S 2 - 4 2 None of the 20 seedlings monitored at na na na (previously D. sericea) Castlereagh had flowered by 3 yrs pf. However a few plants in the adjacent area flowered at 22 mths pf (9/03). Flowering also observed: • 23 mths pf at Devlin Road (9/04) • 24 mths pf at Post Office Road (10/04) – profuse flowering in moist area • 2.7 yrs pf at Nutt Road (8/02) • 2.8 yrs pf at Devlin Road (9/04) • 3.6 yrs pf at Castlereagh NR (9/03) – flowering profuse Regener- Primary juvenile period Secondary juvenile period ation Taxon mode Years Relia- Evidence Years Relia- Evidence (App 4.2) bility bility Dillwynia sieberi S 2 - 3 3 • Flowering and fruiting observed 22 na na na mths pf at Lansdowne (10/04) • Buds observed 2.7 yrs pf at Nurragingy (8/04) • Many plants flowering or budding 2.7 yrs pf at Orchard Hills (9/04) • Buds observed at Plumpton 2.9 yrs pf (8/03) Dillwynia tenuifolia S 2 - 3 2 • One of 15 seedlings monitored at na na na Shanes Park had flowered by 2 yrs pf, as had some other plants in the same area. All 15 monitored seedlings had flowered by 2.8 yrs 360 pf. Flowering also observed: • 21 mths pf at Nutt Road, in an area which had had 4 fires in 10 years (9/02) • 23 mths pf at Castlereagh NR (10/04) • 23 mths pf at Devlin Road (9/04) • 24 mths pf at Post Office Road (10/04) – flowering profuse Dodonaea falcata S 2 2 • Seedlings adjacent to study site at na na na Shanes Park were monitored at three-monthly intervals. Fruiting first observed at 18 mths pf (7/03). Profuse flowering observed 2.7 yrs pf (9/04) • Many plants fruiting 23 mths pf at Castlereagh NR (10/04) Regener- Primary juvenile period Secondary juvenile period ation Taxon mode Years Relia- Evidence Years Relia- Evidence (App 4.2) bility bility Dodonaea viscosa Sr 3 3 • Some plants fruiting 2.7 yrs pf at 1 5 Fruits observed on resprouts 12 mths pf ssp. cuneata Orchard Hills (9/04) at Shanes Park (12/02). • Some plants fruiting 2.7 yrs pf at Mulgoa (9/04) • Takes 3 years pf to set seed of any quantity. Good crops in years 4-6 post-fire (pers. comm. David Warren, Greening Australia, 2002) Eremophila debilis R uk 1 3 Observed flowering: • 2 mths pf at Bligh Park (11/04) • 3 mths pf at Scheyville (1/02) • 3 mths pf at Mulgoa (3/02) 361 Exocarpos strictus R uk 3 5 Flowering observed 2.7 yrs pf at Mulgoa (9/04). Gompholobium S 2 5 Flowering and fruiting observed 23 na na na glabratum mths pf at Post Office Road (9/04) Gompholobium S 2 5 Seedlings observed flowering and na na na inconspicuum fruiting 24 mths pf at Post Office Road (10/04) Gompholobium minus R uk 1 - 2 3 • Buds observed 10 mths pf at Castlereagh NR (9/03) • Fruiting observed 22 mths pf at Llandilo (9/03) • Fruiting observed 23 mths pf at Castlereagh NR (10/04) • Profuse flowering 2.8 yrs pf at Nutt Road (11/04) Gompholobium S 2 - 4 4 Flowering observed: na na na pinnatum • 24 mths pf at Post Office Road (10/04) • 3.6 yrs pf at Castlereagh NR (9/03) Regener- Primary juvenile period Secondary juvenile period ation Taxon mode Years Relia- Evidence Years Relia- Evidence (App 4.2) bility bility Grevillea juniperina S 4 - 5 4 • 2 seedlings monitored at Shanes na na na ssp, juniperina Park had not flowered by 2.8 yrs pf., nor had other plants in the area • Flowering observed 3.1 yrs pf at Mt Druitt (8/03) • Many flowers 4.7 yrs pf at Castlereagh NR (10/04) Grevillea mucronulata R uk 2 3 Flowering observed: • 21 mths pf at Nutt Road (9/03) • 22 mths pf at Llandilo (9/03) • 24 mths pf at Shanes Park (9/04) Hakea laevipes R uk 2 - 3 4 Flowering observed: (previously H. • 23 mths pf at Castlereagh NR 362 dactyloides) (10/04) • 2.8 yrs pf at Nutt Road (11/04) – profuse flowering Regener- Primary juvenile period Secondary juvenile period ation Taxon mode Years Relia- Evidence Years Relia- Evidence (App 4.2) bility bility Hakea sericea S 2 – 4 3 None of the 12 seedlings monitored at na na na Castlereagh had flowered by 3 yrs pf • Small dead plants in a recently- burnt area on Nutt Road with a 2- year inter-fire interval had fruits (8/02) • Small fruits observed 23 mths pf at Devlin Road (9/04) – 1 plant • Flowering observed 23 mths pf at Post Office Road (9/04) – 1 plant • Flowering observed 2.7 yrs pf at Nutt Road (9/02) • Flowering observed 2.8 yrs pf at 363 Tadmore Road (9/04) • Flowering and some fruiting observed 2.8 yrs pf at Devlin Road (9/04) • Many plants with fruits 3.6 yrs pf at Castlereagh NR (9/03); • Many plants with fruits 4.0 yrs pf at Windsor Downs (12/01) Hardenbergia violacea R uk Observed flowering <18 mths after 1 3 Flowering observed: germination on sandstone (8/03). • 6 mths pf at Mt Druitt (8/03) • 8 mths pf at Plumpton (8/03) • 9 mths pf at Ropes Creek (7/03) • 11 mths pf at Shanes Park (8/03) Hibbertia aspera R uk 1 5 Flowering observed 12 mths pf at Holsworthy (12/03). Hibbertia diffusa R uk 1 3 Flowering observed: • 8 mths pf at Plumpton (8/03) • 9 mths pf at Ropes Creek (7/03) • 10 mths pf at Windsor Downs (8/03) Regener- Primary juvenile period Secondary juvenile period ation Taxon mode Years Relia- Evidence Years Relia- Evidence (App 4.2) bility bility Hibbertia pedunculata R uk 2 5 Flowering observed 13 mths pf at Holsworthy (1/04) Hovea linearis R uk 1 - 2 3 Flowering observed: • 9 mths pf at Nutt Road (9/02) • 23 mths pf at Post Office Road (9/04) • 23 mths pf at Castlereagh NR (10/04) Fruiting observed 12 mths pf at Taylor Road (9/04) Indigofera australis R 2 - 3 2 • Two of four seedlings monitored at 2 4 Flowering on resprouts observed: Shanes Park had flowered by 2 yrs • 20 mths pf at Nurragingy (8/03) pf, as had some plants in the • 21 mths pf at Holsworthy (8/02). adjacent area. • Small single-stemmed plants 364 budding 2.7 yrs pf at Orchard Hills (9/04). Isopogon R uk 1 - 2 3 Flowering observed: anemonifolius • 11 mths pf at Devlin Road (10/02) • 12 mths pf at Taylor Road (10/04) - buds • 22 mths pf at Llandilo (9/03) Fruiting observed: • 14 mths pf at Castlereagh NR (1/04) • 23 mths pf at Post Office Road (9/04) Cones with seeds observed on burnt, dead shoots at Nutt Road after a 2-year interfire interval (7/02). Kunzea ambigua Sr uk 3 5 Flowering on resprouts observed 2.8 yrs pf at Nutt Road (11/04) Kunzea capitata R uk 1 4 • Flowering observed11 mths pf at Devlin Road (10/02) • Budding observed 10 mths pf at Windsor Downs (8/03) Regener- Primary juvenile period Secondary juvenile period ation Taxon mode Years Relia- Evidence Years Relia- Evidence (App 4.2) bility bility Leptospermum R uk 1 - 2 3 Flowering observed: parvifolium • 10 mths pf at Devlin Road (9/02) • 21 mths pf at Nutt Road (9/03) • 22 mths pf at Llandilo (9/03) • 23 mths pf at Post Office Road (9/04) Leptospermum R uk 2 - 3 3 • Fruiting observed 23 mths pf at polygalifolium ssp. Devlin Road (9/04) polygalifolium • Flowering observed 24 mths pf at Post Office Road (10/04) • Flowering observed 2.8 yrs pf at Nutt Road (11/04) Leptospermum R uk 3 3 Flowering observed: trinervium • 2.1 yrs pf at Devlin Road (12/03) • 2.7 yrs pf at Tadmore Road (9/04) - buds 365 • 2.8 yrs pf at Nutt Road (11/04) Leucopogon virgatus R uk 2 5 Flowering observed 23 mths pf at Castlereagh NR (10/04) Lissanthe strigosa R uk 1 3 • Flowering observed 5 mths pf at Shanes Park (6/02) • Budding observed 8 mths pf at Tadmore Road (9/04) • Flowering observed 10 mths pf at Windsor Downs (8/03) • Flowering observed 12 mths pf at Taylor Road (9/04) • Fruiting observed 12 mths pf at Holsworthy (1/02). Macrozamia spiralis R uk 2 - 3 4 • Developing cones observed 23 mths pf at Castlereagh NR (10/04) • Fresh, open cone observed 2.2 yrs pf at Shanes Park (6/04) Regener- Primary juvenile period Secondary juvenile period ation Taxon mode Years Relia- Evidence Years Relia- Evidence (App 4.2) bility bility Melaleuca erubescens R uk 3 5 Fruiting observed 2.8 yrs pf at Devlin Road (9/04) Melaleuca nodosa R uk 1 - 2 3 • Flowering observed 11 mths pf at Devlin Road (10/02) • Flowering observed 23 mths pf at Castlereagh NR (10/04) - buds • Fruiting observed 23 mths pf at Devlin Road (9/04) • Flowering observed 24 mths pf at Post Office Road (10/04) Melaleuca thymifolia R uk 2 - 3 3 • Flowering observed 13 mths pf at Holsworthy (1/04) • Fruiting observed 23 mths pf at Devlin Road (9/04)
366 • Fruiting observed 2.7 yrs pf at Tadmore Road (9/04) • Flowering observed 2.8 yrs pf at Nutt Road (11/04) Micromyrtus minutiflora S 2 - 5 4 • Flowering observed 24 mths pf at na na na Post Office Road (10/04) • Profuse flowering observed 4.7 yrs pf at Castlereagh NR (10/04) Mirbelia rubiifolia S 2 - 3 4 Flowering observed: na na na • 24 mths pf at Post Office Road (10/04) • 2.8 yrs pf at Devlin Road Monotoca scoparia R uk 1 - 2 3 Flowering observed: • 9 mths pf at Nutt Road (10/02) • 9 mths pf at Llandilo (8/02) - buds • 23 mths pf at Castlereagh NR (10/04) • 23 mths pf at Post Office Road (9/04) – small fruits Regener- Primary juvenile period Secondary juvenile period ation Taxon mode Years Relia- Evidence Years Relia- Evidence (App 4.2) bility bility Ozothamnus Sr 2 - 3 2 • Five of 12 seedlings monitored at 2 5 Resprouting plant observed flowering 23 diosmifolius Shanes Park had flowered by 2.8 mths pf at Windsor Downs. yrs pf, as had other plants in the area Seedlings also observed flowering: • 23 mths pf at Windsor Downs (6/02) – large buds • 23 mths pf at Castlereagh NR (10/04) • 2.7 yrs pf at Orchard Hills (9/04) • 2.7 yrs pf at Tadmore Road (9/04) Persoonia linearis R uk 2 5 Fruiting observed 23 mths pf at Castlereagh NR 367 Persoonia nutans S 5 5 • Fruiting observed 4.8 yrs pf at na na na Windsor Downs (10/02) • No fruits on seedlings 2.8 yrs pf at Devlin Road (9/04) Petrophile pulchella S 3+ 5 • Many seedlings flowering 2.8 yrs pf na na na at Nutt Road (11/04) • However no sign of flowers 2.7 yrs pf at Nutt Road (8/02), or 2.8 yrs pf at Devlin Road (9/02) • Cones found on dead plants at Nutt Road 7 mths pf, after a 2-yr interfire interval, however no seedlings found (7/02). Philotheca salsolifolia R uk 2 4 Flowering observed: • 14 mths pf at Devlin Road (1/03) • 23 mths pf at Castlereagh NR (10/04) – buds Regener- Primary juvenile period Secondary juvenile period ation Taxon mode Years Relia- Evidence Years Relia- Evidence (App 4.2) bility bility Phyllanthus hirtellus R uk 1 - 2 3 • Flowering observed 12 mths pf at Taylor Road (9/04) • Fruiting observed 18 mths pf at Shanes Park (7/03) • Flowering observed 20 mths pf at Nutt Road (9/03) • Flowering observed 23 mths pf at Castlereagh NR (10/04) Pimelea curviflora var. S 1 5 Some seedlings observed flowering at na na na subglabrata Prospect 12 mths pf (10/03).
Pimelea linifolia ssp. S 1 - 2 2 Three of 26 tagged seedlings flowered na na na linifolia at 22 mths pf, as did some nearby plants. An additional 7 seedlings had 368 flowered by three years pf. Flowering also observed: • 12 mths pf at Taylor Road (9/04) – a few plants • 21 mths pf at Nutt Road (9/03) • 22 mths pf at Llandilo (9/03) • 23 mths pf at Devlin Road and Post Office Road (9/04) - profuse • 23 mths pf at Castlereagh NR (10/04) Pimelea spicata R uk 1 4 Flowering observed: • 7 mths pf at Mt Annan (4/03); • 12 mths pf at Prospect (9/03). Platysace ericoides R > 3 2 None of the 4 seedlings monitored at 2 5 Flowering observed 14 mths pf at Devlin Castlereagh had flowered by 3.0 yrs Road (1/03). pf. These seedlings emerged later than other species. Regener- Primary juvenile period Secondary juvenile period ation Taxon mode Years Relia- Evidence Years Relia- Evidence (App 4.2) bility bility Podolobium scandens R uk 1 3 Flowering observed: • 8 mths pf at Plumpton (8/03) • 10 mths pf at Windsor Downs (8/03) • 11 mths pf at Shanes Park (8/03). Pomax umbellata S 2 3 Seedlings observed fruiting: na na na • 13 mths pf at Holsworthy (1/04) • 18 mths pf at Shanes Park (7/03) • 20 mths pf at Plumpton (8/04) • 23 mths pf at Windsor Downs (9/04). Prostanthera S 2 5 Flowering observed 23 mths pf at na na na scutellarioides Castlereagh NR Pultenaea elliptica R uk 3 5 Flowering observed 2.8 yrs pf at Nutt Road (11/04) Pultenaea microphylla Sr 3 - 4 4 Flowering observed: 2 5 Observed flowering 19 mths pf at 369 • 2.9 yrs pf at Plumpton (8/03) Shanes Park, probably on resprouting • 3.1 yrs pf at Mt Druitt (8/03). shoots after low intensity fire (7/03). Pultenaea parviflora S 2 - 3 2 • Two of 36 seedlings monitored at na na na Shanes Park had flowered by 2 yrs pf, as had some plants in the adjacent area. Another 30 plants had flowered by 2.8 yrs pf. • Flowering also observed 23 mths pf at Castlereagh NR (10/04) Pultenaea villosa S 2 - 3 4 Budding observed: na na na • 21 mths pf at Holsworthy (8/02) • 24 mths pf at Post Office Road (10/04) – single plant, others without buds Solanum cinereum R uk 2 5 Observed flowering and fruiting 14 mths pf at Mt Annan.
Appendix 6 Comparison of Cumberland Plain data and data in NSW Flora Fire Response Database
With the exception of the final column, data in this appendix has been copied either from the NSW Flora Fire Response Database (accessed August 2004), or from Appendices 4 and 5 which detail observations from the Cumberland Plain (CP).
Table A6.1. Comparison of Cumberland Plain data and data in NSW Flora Fire Response Database. Regeneration mode (columns 2 and 3): S, obligate seeder; R, resprouter; r, usually killed by fire but sometimes resprouts; s, usually resprouts but sometimes killed; SR, S and R modes equally reported; nl, not listed. Colour codes for CP observations (column 3): black, regeneration mode from CP observation corresponds with that in NSW Database (where both a primary and a secondary response is given in the NSW Database – Sr or Rs – only the primary mode is considered); red, regeneration mode from CP observation differs from that in NSW Database; pink, NSW Database response is coded SR, response from CP
370 observation is more definitive; blue, species not listed in NSW Database. CP rel., reliability ratings for Cumberland Plain data (columns 4, 7 and 10): 1, tested by following the fate of individual plants through time, at more than one site; 2, tested by following the fate of individual plants through time, at a single site; 3. post-burn observation in three or more sites; 4, post-burn observation in two sites; 5, post-burn observation in one site. Colour codes for CP juvenile period observations (columns 6 and 9): black, juvenile period from CP observation falls within one year of that given in NSW Database; red, juvenile period from CP observation differs from that in NSW Database by more than 1 year; pink, NSW Database lists juvenile period as “< x”, CP information is more specific; blue – no information on juvenile period in NSW Database. Vital attributes (columns 11 and 12): see Appendix 3 for explanation of initials, and Section 3.2.5 for process used to assign vital attributes to Cumberland Plain species. Colour codes for vital attributes groups from NSW Database (these are taken directly from the Database): blue, based on assumed category; red, based on other species in genus; pink, conflicting data, most sensitive category used; green, conflicting data, resprouter category used.
Taxon name Regeneration mode Primary juvenvile period Secondary juvenvile Vital attributes (years) period (years) from from CP CP from from CP CP from from CP from for CP NSW obs'n rel. NSW obs'n rel. NSW CP rel. NSW Dbase Dbase Dbase obs'n Dbase Acacia brownii Rs R 3 1 - 2 3 VSI USI Acacia bynoeana nl R 3 2 - 3 4 VSI Acacia decurrens S S 3 3 - 7 4 SI SI Acacia elongata Sr R 3 > 3 2 2 3 SI USI Acacia falcata S S 3 2 2 - 3 2 SI SI Acacia implexa Sr R 3 1 - 5 SI VSI Acacia parramattensis S R 3 5 3 5 SI VSI Acacia ulicifolia Sr Sr 3 2 - 3 SI SI Allocasuarina littoralis Rs S 3 3 - 6 5 4 3 VCT CT Astroloma humifusum R R 5 2 V VGI Baeckea diosmifolia R R 3 1 1 - 2 3 UI UGI 371 Banksia oblongifolia R R 3 5 - 10 2 - 5 3 - 4 3 VCT VCT Banksia spinulosa var. spinulosa R R 3 3 - 8 3 2 - 3 3 VC VCI Billardiera scandens var. scandens R R 3 2 5 V UGI Boronia polygalifolia Rs R 4 1 4 VGI UGI Bossiaea obcordata Rs R 3 < 2 2 5 USI USI Bossiaea prostrata R R 3 < 2 1 - 2 3 USI USI Bossiaea rhombifolia SR R 3 3 5 SI VSI Brachyloma daphnoides R R 3 1 - 2 2 3 UGI UGI Breynia oblongifolia Rs R 4 1 - 5 1 - 5 VGT VGT Bursaria spinosa R R 2 3 > 2 2 1 - 2 2 - 3 3 U VGT Callistemon linearis R R 4 3 2 - 3 4 VC VCI Callistemon pinifolius R R 3 1 - 3 3 VC VCI Calotis lappulaceae nl R 4 1 - 2 4 UGI Chorizema parviflorum Rs R 3 < 2 1 3 USI USI Clematis glycinoides R R 5 3 - 5 VI VDI Clerodendrum tomentosum R R 5 V VGI Taxon name Regeneration mode Primary juvenvile period Secondary juvenvile Vital attributes (years) period (years) from from CP CP from from CP CP from from CP from for CP NSW obs'n rel. NSW obs'n rel. NSW CP rel. NSW Dbase Dbase Dbase obs'n Dbase Conospermum taxifolium SR R 4 4 1 5 GI UGI Cryptandra amara var. amara R S 3 1 - 2 3 1 SI SI Cryptandra spinescens nl R 5 1 5 UGI Daviesia genistifolia S R 5 1 5 SI USI Daviesia squarrosa SR R 3 1 - 2 3 SI USI Daviesia ulicifolia Rs Sr 3 2 - 4 2 2 3 VSI SI Dillwynia rudis (perviously D. sericea) Sr S 3 2 - 4 2 2 SI SI Dillwynia sieberi S S 3 2 - 3 3 SI SI Dillwynia tenuifolia S S 3 3 - 4 2 - 3 2 SI SI Dodonaea falcata R S 4 2 2 VSI SI Dodonaea triquetra S S 5 3 - 6 SI SI
372 Dodonaea viscosa ssp. cuneata R Sr 4 3 3 1 5 WSI SI Eremophila debilis R R 3 1 3 V UGI Exocarpos cupressiformis Rs R 3 > 3 V VGI Exocarpus strictus R R 3 3 5 V VGI Gompholobium glabratum Sr S 5 3 2 5 SI SI Gompholobium inconspicuum S S 5 2 5 SI SI Gompholobium minus Sr R 3 1 - 2 3 SI USI Gompholobium pinnatum Rs S 4 2 - 4 4 VSI SI Grevillea juniperina ssp. juniperina S S 3 4 - 5 4 G GT Grevillea mucronulata Sr R 3 2 3 GI UGI Hakea laevipes (previously H. R R 3 1 2 - 3 4 UCI VCI dactyloides) Hakea sericea S S 3 3 - 4 2 - 4 3 CT CT Hardenbergia violacea Rs R 3 1 1 1 3 USI USI Hibbertia aspera SR R 4 1 1 5 G UGI Hibbertia diffusa SR R 3 0.5 1 3 G UGI Hibbertia pedunculata R R 5 2 5 VGI UGI Taxon name Regeneration mode Primary juvenvile period Secondary juvenvile Vital attributes (years) period (years) from from CP CP from from CP CP from from CP from for CP NSW obs'n rel. NSW obs'n rel. NSW CP rel. NSW Dbase Dbase Dbase obs'n Dbase Hovea linearis Rs Rs 3 2 2 - 3 1 - 2 3 VSI USI Indigofera australis Rs Rs 3 2 2 - 3 2 2 2 4 VSI USI Isopogon anemonifolius Rs R 3 10 2 1 - 2 3 VCI UCI Jacksonia scoparia R R 3 5 < 2 USI VSI Kunzea ambigua S Sr 4 2 - 4 3 5 GI GI Kunzea capitata Rs R 3 < 2 1 4 UGI UGI Leptospermum parvifolium Rs R 3 < 4 1 - 2 3 UCI UCI Leptospermum polygalifolium ssp. Rs R 3 < 3 2 - 3 3 VCI VCI polygalifolium Leptospermum trinervium R R 3 0.5 3 3 UCI VCI Leucopogon virgatus R R 5 2 2 5 VGI UGI Lissanthe strigosa R R 3 1 1 3 U UGI
373 Macrozamia spiralis R R 3 10 - 20 2 - 3 4 VC VCI Maytenus silvestris R R 5 V VGI Melaleuca erubescens nl R 3 3 5 VCI Melaleuca nodosa R R 3 < 4 1 - 2 3 VCI UCI Melaleuca thymifolia R R 3 0.5 2 - 3 3 UCI VCI Micromyrtus minutiflora nl S 5 2 - 5 4 GI Mirbelia rubiifolia Sr S 4 1 - 3 2 - 3 4 SI SI Monotoca scoparia Rs R 3 2 1 - 2 3 VGI UGI Olearia viscidula R R 5 V VDI Ozothamnus diosmifolius SR Sr 3 2 - 3 2 2 5 G GI Persoonia laurina R R 3 VG VGI Persoonia linearis R R 3 2 5 VG VGI Persoonia nutans S S 4 5 5 GI GI Petrophile pulchella Sr S 3 4 - 9 3 + 5 CI CI Philotheca salsolifolia Rs R 3 5 2 4 VGI UGI Phyllanthus hirtellus Rs R 3 1 1 - 2 3 UCI UCI Taxon name Regeneration mode Primary juvenvile period Secondary juvenvile Vital attributes (years) period (years) from from CP CP from from CP CP from from CP from for CP NSW obs'n rel. NSW obs'n rel. NSW CP rel. NSW Dbase Dbase Dbase obs'n Dbase Pimelea curviflora var. subglabrata nl S 5 1 5 GI Pimelea linifolia ssp. linifolia SR S 3 1 - 2 2 GI GI Pimelea spicata R R 4 2 0.5 1 4 U UGI Platysace ericoides Sr R 4 > 3 2 1 2 5 GI UGI Podolobium scandens R R 3 < 2 1 3 VSI USI Pomax umbellata S S 3 < 1 2 3 GI GI Prostanthera scutellarioides S S 5 3 2 5 SI SI Pultenaea elliptica Rs R 3 2 3 5 VSI VSI Pultenaea microphylla Sr S 4 3 - 4 4 2 5 SI SI Pultenaea parviflora S S 3 3 - 4 2 - 3 2 SI SI Pultenaea villosa S S 3 2 - 3 4 SI SI Rubus parvifolius R R 5 VD VDI Solanum cinereum nl R 2 2 5 UGI
374 Styphelia laeta ssp. laeta SR S 5 G GT
Appendix 7 Fire-related attributes of two Cumberland Plain vegetation types
This appendix details the fire-related attributes of shrub species in two broad Cumberland Plain vegetation types: Cumberland Plain woodlands and Castlereagh woodlands. Species in each table are arranged in two groups. The first group includes all species listed by James et al. (1999) as either ‘very common’, ‘common’ or ‘frequent’ in the relevant vegetation type. The second group includes species listed ‘occasional’, plus species designated ‘rare’ which are also listed under the NSW Threatened Species Conservation (TSC) Act 1995. Within groups, species are arranged by Plant Functional Type (PFT).
Regeneration mode (column 2): S, obligate seeder; R, resprouter; r, usually killed by fire but sometimes resprouts; s, usually resprouts but sometimes killed; SR, S and R modes equally reported. Colour codes: black, from local observations; pink, from NSW Database; green, no information available, decision rules in Section 3.2.6 applied.
Vital attributes (column 3): see Appendix 3 for explanation of initials. Colour codes: black, from Appendix 6 (species with local information); pink, from NSW Database; green; no information available, decision rules in section 3.2.6 applied.
375 Plant functional types (PFT, column 4): see Appendix 3 for explanation of these groups.
Sensitivity to frequent fire (column 5), sensitivity to infrequent fire (column 6): 1, regime leads to local decline or extinction; 2, persistent regime likely to lead to local decline or extinction; 3, regime unlikely to lead to local decline or extinction.
Juvenile periods (columns 7 and 8). Colour codes: black, from local observations; pink, from NSW Database.
Lifespan + seedbank (column 9): Figures in this column are taken from the NSW Database. Values in blue include an estimate for seedbank longevity: 10 years for persistent seedbank without hard seedcoat, 30 years for persistent seedbank with hard seedcoat.
Frequency (column 10): Frequency in relevant vegetation type, from James et al. (1999). 5, very common; 4, common; 3, frequent; 2, occasional; 1, rare.
Conservation status (column 11). Listing under the NSW Threatened Species Conservation (TSC) Act 1995: E, endangered; V, vulnerable. Table A7.1. Fire-related attributes of Cumberland Plain woodlands shrubs.
Taxon name Regen Vital PFT Sens. Sens. to Primary Seconary Lifespan + Family Freq. Cons'n mode attributes to f.f. i.f. juvenile juvenile seedbank status period (yrs) period (yrs) More common species: Bursaria spinosa R VGT 1 3 3 > 2 2 - 3 70 Pittosporaceae 5 Hakea sericea S CT 2 1 3 2 - 4 > 25 Proteaceae 3 Acacia decurrens S SI 4 2 1 3 - 7 60 Mimosaceae 3 Acacia falcata S SI 4 2 1 2 - 3 50 Mimosaceae 4 Acacia parramattensis R VSI 4 2 1 5 3 > 55 Mimosaceae 4 Astroloma humifusum R VGI 4 2 1 2 Epacridaceae 3 Daviesia ulicifolia Sr SI 4 2 1 2 - 4 2 55 Fabaceae 4 Dillwynia sieberi S SI 4 2 1 2 - 3 55 Fabaceae 4 Dodonaea triquetra S SI 4 2 1 3 - 6 > 55 Sapindaceae 3 Dodonaea viscosa ssp. Sr SI 4 2 1 3 1 Sapindaceae 3 cuneata 376 Exocarpos cupressiformis R VGI 4 2 1 > 3 150 Santalaceae 4 Persoonia linearis R VGI 4 2 1 2 Proteaceae 3 Leucopogon juniperinus Sr GI 5 1 1 Epacridaceae 3 Opercularia varia Sr GI 5 1 1 1 - 2 35 Rubiaceae 3 Ozothamnus diosmifolius Sr GI 5 1 1 2 - 3 2 Asteraceae 4 Pomax umbellata S GI 5 1 1 2 35 Rubiaceae 3 Bossiaea prostrata R USI 12 2 1 1 - 2 Fabaceae 3 Chorizema parviflorum R USI 12 2 1 1 Fabaceae 3 Eremophila debilis R UGI 12 2 1 1 Myoporaceae 3 Hardenbergia violacea R USI 12 2 1 1 1 60 Fabaceae 5 Hibbertia diffusa R UGI 12 2 1 1 Dilleniaceae 3 Indigofera australis Rs USI 12 2 1 2 - 3 2 30 Fabaceae 4 Lissanthe strigosa R UGI 12 2 1 1 60 Epacridaceae 3
Less common species: Breynia oblongifolia R VGT 1 3 3 1 - 5 1 - 5 > 40 Euphorbiaceae 2 Pittosporum revolutum Rs VDT 1 3 3 1 - 5 2 - 3 > 30 Pittosporaceae 2 Taxon name Regen Vital PFT Sens. Sens. to Primary Seconary Lifespan + Family Freq. Cons'n mode attributes to f.f. i.f. juvenile juvenile seedbank status period (yrs) period (yrs) Polyscias sambucifolia Rs VDT 1 3 3 110 Araliaceae 2 Grevillea juniperina ssp. S GT 2 1 3 4 - 5 Proteaceae 2 V juniperina Styphelia laeta ssp. laeta S GT 2 1 3 Epacridaceae 2 Cassinia aculeata S DI 3 2 2 40 Asteraceae 2 Olearia viscidula R VDI 3 2 2 Asteraceae 2 Rapanea variabilis R VDI 3 2 2 < 3 100 Myrsinaceae 2 Acacia binervia S SI 4 2 1 2 130 Mimosaceae 2 Acacia floribunda SR SI 4 2 1 3 - 4 2.5 90 Mimosaceae 2 Acacia implexa R VSI 4 2 1 1 - 5 80 Mimosaceae 2 Acacia longifolia var. Sr SI 4 2 1 2 - 5 90 Mimosaceae 2 longifolia Acacia pubescens Rs VSI 4 2 1 3 - 5 80 Mimosaceae 2 V Acacia stricta S SI 4 2 1 4 Mimosaceae 2 Bossiaea buxifolia S SI 4 2 1 Fabaceae 2 377 Clerodendrum tomentosum R VGI 4 2 1 Verbenaceae 2 Dillwynia tenuifolia S SI 4 2 1 2 - 3 60 Fabaceae 2 V Dodonaea viscosa ssp. S SI 4 2 1 Sapindaceae 2 angustifolia Gompholobium glabratum S SI 4 2 1 2 55 Fabaceae 2 Jacksonia scoparia R VSI 4 2 1 5 < 2 Fabaceae 2 Maytenus silvestris R VGI 4 2 1 > 30 Celastraceae 2 Pultenaea microphylla Sr SI 4 2 1 3 - 4 2 Fabaceae 2 Pultenaea parviflora S SI 4 2 1 2 - 3 55 Fabaceae 2 E Pultenaea villosa S SI 4 2 1 2 - 3 Fabaceae 2 Atriplex semibaccata S GI 5 1 1 Chenopodiaceae 2 Goodenia ovata Sr GI 5 1 1 1 - 3 40 Goodeniaceae 2 Kunzea ambigua Sr GI 5 1 1 2 - 4 3 50 Myrtaceae 2 Pimelea linifolia ssp. linifolia S GI 5 1 1 1 - 2 > 32 Thymelaeaceae 2 Zieria smithii Sr GI 5 1 1 Rutaceae 2 Cassinia arcuata S CI 6 1 1 2 - 3 20 Asteraceae 2 Olearia microphylla S CI 6 1 1 Asteraceae 2 Taxon name Regen Vital PFT Sens. Sens. to Primary Seconary Lifespan + Family Freq. Cons'n mode attributes to f.f. i.f. juvenile juvenile seedbank status period (yrs) period (yrs) Leptospermum trinervium R VCI 7 2 1 3 80 Myrtaceae 2 Melaleuca erubescens R VCI 7 2 1 3 Myrtaceae 2 Melaleuca nodosa R VCI 7 2 1 1 - 2 Myrtaceae 2 Phyllanthus gunnii R VI 7 2 1 1 - 2 Euphorbiaceae 2 Notelaea longifolia R UDT 10a 3 3 1 - 5 < 2 > 30 Oleaceae 2 Boronia polygalifolia R UGI 12 2 1 1 Rutaceae 2 Daviesia genistifolia R USI 12 2 1 1 Fabaceae 2 Grevillea mucronulata R UGI 12 2 1 2 35 Proteaceae 2 Hibbertia aspera R UGI 12 2 1 1 > 32 Dilleniaceae 2 Hibbertia empetrifolia R UGI 12 2 1 1 70 Dilleniaceae 2 Pimelea spicata R UGI 12 2 1 2 1 Thymelaeaceae 2 E Podolobium scandens R USI 12 2 1 1 Fabaceae 2 Phyllanthus hirtellus R UCI 14 2 1 1 - 2 Euphorbiaceae 2
378 Table A7.2. Fire-related attributes of Castlereagh woodlands shrubs.
Taxon name Regen Vital PFT Sens. Sens. Primary Seconary Lifespan + Family Freq. Cons'n mode attributes to f.f. to i.f. juvenile juvenile seedbank status period period (yrs) (yrs) More common species: Bursaria spinosa R VGT 1 3 3 > 2 2 - 3 70 Pittosporaceae 4 Hakea sericea S CT 2 1 3 2 - 4 > 25 Proteaceae 5 Styphelia laeta ssp. laeta S GT 2 1 3 Epacridaceae 3 Acacia falcata S SI 4 2 1 2 - 3 50 Mimosaceae 4 Acacia parramattensis R VSI 4 2 1 5 3 > 55 Mimosaceae 3 Daviesia ulicifolia Sr SI 4 2 1 2 - 4 2 55 Fabaceae 4 Dillwynia tenuifolia S SI 4 2 1 2 - 3 60 Fabaceae 4 V Exocarpos cupressiformis R VGI 4 2 1 > 3 150 Santalaceae 3 Pultenaea parviflora S SI 4 2 1 2 - 3 55 Fabaceae 3 E Pultenaea villosa S SI 4 2 1 2 - 3 Fabaceae 3 379 Opercularia varia Sr GI 5 1 1 1 - 2 35 Rubiaceae 3 Ozothamnus diosmifolius Sr GI 5 1 1 2 - 3 2 Asteraceae 3 Pimelea linifolia ssp. linifolia S GI 5 1 1 1 - 2 > 32 Thymelaeaceae 3 Pomax umbellata S GI 5 1 1 2 35 Rubiaceae 3 Olearia microphylla S CI 6 1 1 Asteraceae 3 Banksia spinulosa var. R VCI 7 2 1 3 - 8 2 - 3 100 Proteaceae 3 spinulosa Callistemon pinifolius R VCI 7 2 1 1 - 3 Myrtaceae 3 Hakea laevipes (previously H. R VCI 7 2 1 2 - 3 80 Proteaceae 3 dactyloides) Leptospermum trinervium R VCI 7 2 1 3 80 Myrtaceae 3 Macrozamia spiralis R VCI 7 2 1 10 - 20 2 - 3 > 60 Zamiaceae 4 Melaleuca erubescens R VCI 7 2 1 3 Myrtaceae 3 Melaleuca nodosa R VCI 7 2 1 1 - 2 Myrtaceae 5 Melaleuca thymifolia R VCI 7 2 1 2 - 3 80 Myrtaceae 3 Acacia brownii R USI 12 2 1 1 - 2 Mimosaceae 3 Acacia elongata R USI 12 2 1 > 3 2 Mimosaceae 4 Taxon name Regen Vital PFT Sens. Sens. Primary Seconary Lifespan + Family Freq. Cons'n mode attributes to f.f. to i.f. juvenile juvenile seedbank status period period (yrs) (yrs) Boronia polygalifolia R UGI 12 2 1 1 Rutaceae 3 Brachyloma daphnoides R UGI 12 2 1 2 110 Epacridaceae 3 Daviesia squarrosa R USI 12 2 1 1 - 2 Fabaceae 3 Gompholobium minus R USI 12 2 1 1 - 2 55 Fabaceae 3 Grevillea mucronulata R UGI 12 2 1 2 35 Proteaceae 4 Hardenbergia violacea R USI 12 2 1 1 1 60 Fabaceae 4 Hibbertia diffusa R UGI 12 2 1 1 Dilleniaceae 3 Hovea linearis Rs USI 12 2 1 2 1 - 2 Fabaceae 3 Kunzea capitata R UGI 12 2 1 1 90 Myrtaceae 3 Lissanthe strigosa R UGI 12 2 1 1 60 Epacridaceae 4 Leptospermum parvifolium R UCI 14 2 1 1 - 2 80 Myrtaceae 3 Phyllanthus hirtellus R UCI 14 2 1 1 - 2 Euphorbiaceae 3
380 Less common species: Banksia oblongifolia R VCT 1 3 3 5 - 10 3 - 4 80 Proteaceae 2 Grevillea juniperina ssp. S GT 2 1 3 4 - 5 Proteaceae 2 V juniperina Persoonia lanceolata S DI 3 2 2 3 - 7 70 Proteaceae 2 Acacia bynoeana R VSI 4 2 1 2 - 3 Mimosaceae 1 V Acacia decurrens S SI 4 2 1 3 - 7 60 Mimosaceae 2 Acacia parvipinnula S SI 4 2 1 Mimosaceae 2 Acacia pubescens Rs VSI 4 2 1 3 - 5 80 Mimosaceae 2 V Acacia ulicifolia Sr SI 4 2 1 2 - 3 > 55 Mimosaceae 2 Astroloma humifusum R VGI 4 2 1 2 Epacridaceae 2 Bossiaea buxifolia S SI 4 2 1 Fabaceae 2 Bossiaea rhombifolia R VSI 4 2 1 3 90 Fabaceae 2 Cryptandra amara var. amara S SI 4 2 1 3 > 50 Rhamnaceae 2 Daviesia acicularis S SI 4 2 1 Fabaceae 2 Dillwynia floribunda S SI 4 2 1 2 - 6 60 Fabaceae 2 Dillwynia parvifolia S SI 4 2 1 Fabaceae 2 Dillwynia rudis (previously D. S SI 4 2 1 2 - 4 > 52 Fabaceae 2 Taxon name Regen Vital PFT Sens. Sens. Primary Seconary Lifespan + Family Freq. Cons'n mode attributes to f.f. to i.f. juvenile juvenile seedbank status period period (yrs) (yrs) sericea) Dillwynia sieberi S SI 4 2 1 2 - 3 55 Fabaceae 2 Dodonaea falcata S SI 4 2 1 2 Sapindaceae 2 Dodonaea multijuga S SI 4 2 1 Sapindaceae 2 Gompholobium glabratum S SI 4 2 1 2 55 Fabaceae 2 Gompholobium pinnatum S SI 4 2 1 2 - 4 Fabaceae 2 Hibbertia acicularis Rs VGI 4 2 1 Dilleniaceae 2 Hibbertia riparia R VGI 4 2 1 2 90 Dilleniaceae 2 Jacksonia scoparia R VSI 4 2 1 5 < 2 Fabaceae 2 Mirbelia rubiifolia S SI 4 2 1 2 - 3 55 Fabaceae 2 Mirbelia speciosa S SI 4 2 1 55 Fabaceae 2 Persoonia laurina R VGI 4 2 1 Proteaceae 2 Persoonia linearis R VGI 4 2 1 2 Proteaceae 2 Prostanthera scutellarioides S SI 4 2 1 2 > 55 Lamiaceae 2
381 Pultenaea elliptica R VSI 4 2 1 3 110 Fabaceae 2 Pultenaea retusa S SI 4 2 1 40 Fabaceae 2 Sphaerolobium vimineum Sr SI 4 2 1 1 - 5 2 55 Fabaceae 2 Baeckea ramosissima SR GI 5 1 1 3 Myrtaceae 2 Comesperma ericinum Sr GI 5 1 1 2 Polygalaceae 2 Cryptandra propinqua S GI 5 1 1 Rhamnaceae 2 Kunzea ambigua Sr GI 5 1 1 2 - 4 3 50 Myrtaceae 2 Leucopogon ericoides Sr GI 5 1 1 > 3 50 Epacridaceae 2 Micromyrtus minutiflora S GI 5 1 1 2 - 5 Myrtaceae 2 V Persoonia nutans S GI 5 1 1 5 Proteaceae 1 E Petrophile pulchella S CI 6 1 1 3 - 9 60 Proteaceae 2 Allocasuarina glareicola R VCI 7 2 1 > 3 100 Casuarinaceae 1 E Leptospermum polygalifolium R VCI 7 2 1 2 - 3 80 Myrtaceae 2 ssp. polygalifolium Xanthorrhoea minor R UR 10b 2 3 1 Xanthorrhoeaceae 2 Bossiaea obcordata R USI 12 2 1 2 130 Fabaceae 2 Bossiaea prostrata R USI 12 2 1 1 - 2 Fabaceae 2 Taxon name Regen Vital PFT Sens. Sens. Primary Seconary Lifespan + Family Freq. Cons'n mode attributes to f.f. to i.f. juvenile juvenile seedbank status period period (yrs) (yrs) Chorizema parviflorum R USI 12 2 1 1 Fabaceae 2 Cryptandra spinescens R UGI 12 2 1 1 Rhamnaceae 2 Hibbertia aspera R UGI 12 2 1 1 > 32 Dilleniaceae 2 Hibbertia pedunculata R UGI 12 2 1 2 Dilleniaceae 2 Indigofera australis Rs USI 12 2 1 2 - 3 2 30 Fabaceae 2 Leucopogon virgatus R UGI 12 2 1 2 Epacridaceae 2 Monotoca scoparia R UGI 12 2 1 1 - 2 50 Epacridaceae 2 Philotheca salsolifolia R UGI 12 2 1 5 2 50 Rutaceae 2 Platysace ericoides R UGI 12 2 1 > 3 2 > 32 Apiaceae 2 Isopogon anemonifolius R UCI 14 2 1 10 1 - 2 80 Proteaceae 2 Xanthorrhoea media R UCI 14 2 1 0.5 Xanthorrhoeaceae 2
382 Appendix 8 Species in landscape study sites
Table A8.1. List of native shrub species found in landscape study sites, showing family, abundance in Cumberland Plain woodlands, frequency in study sites, percent frequency in 2 x 2 m subplots, conservation status (*, listed under NSW Threatened Species Conservation Act 1995), life form and post-fire regeneration mode.
Species Family Estimated abundance Number of Average Regional Life form Post-fire in CP woodlands sites with frequency conservation status regeneration mode (James et al. 1999) species in subplots (James et al. 1999) (n = 9) Acacia decurrens Mimosaceae frequent 2 6.0 conserved shrub obligate seeder Acacia falcata Mimosaceae common 5 0.7 conserved shrub obligate seeder Acacia implexa Mimosaceae occasional 3 2.7 conserved shrub resprouter Acacia parramattensis Mimosaceae common 6 2.9 conserved shrub resprouter Boronia polygalifolia Rutaceae occasional 1 0.1 vulnerable subshrub resprouter Bossiaea prostrata Fabaceae frequent 7 3.7 conserved subshrub resprouter Breynia oblongifolia Euphorbiaceae occasional 1 0.1 conserved shrub resprouter Bursaria spinosa Pittosporaceae very common 9 50.3 conserved shrub resprouter Calotis lappulacea Asteraceae occasional 2 0.2 vulnerable subshrub resprouter Chorizema parviflorum Fabaceae frequent 6 2.9 vulnerable subshrub resprouter Clematis glycinoides Ranunculaceae occasional 2 0.1 conserved subshrub resprouter Clerodendrum tormentosum Verbenaceae occasional 1 0.0 conserved shrub resprouter Daviesia genistifolia Fabaceae occasional 1 1.1 vulnerable shrub resprouter Daviesia squarrosa Fabaceae not listed 1 0.1 conserved shrub resprouter 383 Daviesia ulicifolia Fabaceae common 6 8.6 conserved shrub obligate seeder Dillwynia sieberi Fabaceae common 6 8.4 conserved shrub obligate seeder Dodenaea viscosa ssp. cuneata Sapindaceae frequent 5 2.1 vulnerable shrub obligate seeder Eremophila debilis Myoporaceae frequent 6 2.8 vulnerable subshrub resprouter Exocarpos cupressiformis Santalaceae common 3 0.2 conserved shrub resprouter Exocarpos strictus Santalaceae rare 1 0.1 conserved shrub resprouter Grevillea juniperina ssp. juniperina Proteaceae occasional 2 2.7 vulnerable* shrub obligate seeder Hardenbergia violacea Fabaceae very common 8 18.0 conserved subshrub resprouter Hibbertia aspera Dilleniaceae not listed 1 0.3 conserved shrub resprouter Hibbertia diffusa Dilleniaceae frequent 6 1.0 conserved subshrub resprouter Indigofera australis Fabaceae common 6 2.9 conserved shrub resprouter Lissanthe strigosa Epacridaceae frequent 4 0.4 conserved shrub resprouter Ozothamnus diosmifolius Asteraceae common 3 0.1 conserved shrub obligate seeder Persoonia linearis Proteaceae frequent 1 0.1 conserved shrub resprouter Pimelea curviflora var. subglabrata Thymelaeaceae rare 2 0.3 vulnerable subshrub obligate seeder Pimelea spicata Thymelaeaceae occasional 2 0.8 vulnerable* subshrub resprouter Platysace ericoides Apiaceae not listed 1 0.0 conserved subshrub resprouter Podolobium scandens Fabaceae occasional 1 0.1 conserved subshrub resprouter Pomax umbellata Rubiaceae frequent 2 0.2 conserved subshrub obligate seeder Pultenaea microphylla Fabaceae occasional 4 11.9 vulnerable shrub obligate seeder Pultenaea parviflora Fabaceae occasional 1 0.0 conserved* shrub obligate seeder Rubus parvifolius Rosaceae not listed 1 0.0 vulnerable subshrub resprouter Solanum cinereum Solanaceae rare 1 1.1 vulnerable shrub resprouter Table A8.2. Frequency of native shrub species in nine sites, three in each of three fire frequency categories. Taxa are listed in order of average frequency across all study sites.
Species Percentage of subplots with cover of species High fire frequency Moderate fire frequency Low fire frequency All sites Ropes Creek Shanes Park Holsworthy Prospect Plumpton Park Lansdowne Mt Annan Orchard Hills Scheyville Bursaria spinosa 28.730.828.038.2 27.0 42.872.890.294.250.3 Hardenbergia violacea 2.8 8.8 31.5 89.5 0.5 27.3 0.7 1.2 18.0 Pultenaea microphylla 1.2 14.5 65.2 26.5 11.9 Daviesia ulicifolia 3.2 19.7 19.0 3.5 31.8 0.5 8.6 Dillwynia sieberi 3.8 6.8 36.3 11.5 17.2 0.2 8.4 Acacia decurrens 30.0 24.2 6.0 Bossiaea prostrata 1.7 2.0 12.7 1.3 1.8 13.5 0.2 3.7 Indigofera australis 0.8 2.0 20.0 2.8 0.5 0.2 2.9 Chorizema parviflorum 4.5 0.2 15.7 3.7 0.8 1.3 2.9 Acacia parramattensis 0.3 1.3 3.7 13.2 2.2 5.3 2.9 Eremophila debilis 2.0 0.5 1.5 7.8 3.0 10.3 2.8 Grevillea juniperina spp. juniperina 0.8 23.7 2.7 Acacia implexa 16.7 6.5 1.2 2.7 Dodenaea viscosa ssp. cuneata 0.3 1.5 0.8 12.8 3.2 2.1 Daviesia genistifolia 10.2 1.1 Solanum cinereum 9.5 1.1
384 Hibbertia diffusa 0.7 5.7 0.3 0.3 1.2 0.5 1.0 Pimelea spicata 3.3 3.8 0.8 Acacia falcata 1.8 0.3 3.0 0.5 0.5 0.7 Lissanthe strigosa 0.2 2.0 0.2 1.7 0.4 Hibbertia aspera 3.0 0.3 Pimelea curviflora var. subglabrata 0.3 2.7 0.3 Pomax umbellata 1.0 1.2 0.2 Calotis lappulacea 0.7 1.0 0.2 Exocarpos cupressiformis 0.5 0.7 0.3 0.2 Daviesia squarrosa 1.0 0.1 Exocarpos strictus 1.0 0.1 Boronia polygalifolia 0.8 0.1 Breynia oblongifolia 0.8 0.1 Podolobium scandens 0.8 0.1 Clematis glycinoides 0.2 0.5 0.1 Ozothamnus diosmifolius 0.2 0.2 0.3 0.1 Persoonia linearis 0.5 0.1 Platysace ericoides 0.3 0.0 Pultenaea parviflora 0.3 0.0 Clerodendrum tormentosum 0.2 0.0 Rubus parvifolius 0.2 0.0
Table A8.3. List of exotic shrub species found in the landscape study, showing frequency in nine sites, three in each of three fire frequency categories.
Species Frequency (v, <1%, vv, 1-10%, vvv, >10%) High fire frequency Moderate fire frequency Low fire frequency Ropes Creek Shanes Park Holsworthy Prospect Plumpton Park Lansdowne Mt Annan Orchard Hills Scheyville Araujia sericiflora vvvv Lantana camara v Lycium ferocissimum vvv Ligustrum sinense v Morus abla v Olea europaea ssp. africana vvvv Ochna serrulata vvv Protasparagus aethiopicus vv Polygala virgata vv Sida rhombifolia vv vv vvv vv vvv
385 Table A8.4. List of tree species found in the landscape study, showing abundance in nine sites, three in each of three fire frequency categories.
Species Number of adult trees along all transects (7200m2 per site) High fire frequency Moderate fire frequency Low fire frequency Ropes Creek Shanes Park Holsworthy Prospect Plumpton Park Lansdowne Mt Annan Orchard Hills Scheyville Eucalyptus moluccana 53 265 79 62 200 56 43 107 197 Eucalyptus tereticornis 2765241 42 2841309 Eucalyptus crebra 1285 17 9 38 Melaleuca decora 76 8 2 Eucalyptus fibrosa 142211 11 Eucalyptus eugenioides 7 2 12 37 Angophora floribunda 111 Allocasuarina torulosa 1 Melaleuca styphelioides 1 Appendix 9 Summary of landscape study findings
Table A9.1. Mean (S.E.) of landscape study variables for nine sites. Variables are defined in Section 4.2.3. Frequencies and indices are scaled from 0 to 100.
Variable Mean (SE) High fire frequency Moderate fire frequency Low fire frequency Ropes Creek Shanes Park Holsworthy Prospect Plumpton Park Lansdowne Mt Annan Orchard Hills Scheyville Bursaria frequency 28.7 (3.1) 30.8 (6.4) 28.0 (2.6) 38.2 (18.0) 27.0 (1.4) 42.8 (1.3) 72.8 (4.6) 90.2 (6.0) 94.2 (3.6) Bursaria dominance index 18.6 (2.8) 19.0 (3.6) 13.4 (0.6) 29.2 (14.2) 15.8 (2.1) 31.6 (3.5) 60.1 (2.9) 63.3 (6.5) 77.1 (4.0) Bursaria density (plants over 80cm/ha) 131 (31) 95 (50) 52 (5) 364 (234) 153 (23) 299 (62) 575 (188) 2009 (841) 3033 (571)
Other native shrub species richness (/200m2) 1.9 (0.5) 4.7 (1.6) 6.6 (0.2) 6.8 (1.5) 7.0 (0.5) 6.1 (0.6) 3.3 (0.4) 2.8 (0.3) 1.5 (0.1) Other native shrub frequency 10.3 (0.4) 30.3 (10.3) 66.8 (6.1) 99.3 (0.4) 67.0 (6.7) 67.3 (7.4) 34.7 (8.1) 26.3 (7.1) 16.3 (2.1) Subshrub frequency 8.8 (1.3) 16.2 (10.7) 42.0 (10.0) 90.2 (3.9) 7.7 (2.0) 36.8 (9.2) 11.7 (8.9) 5.8 (3.2) 10.5 (5.4) Larger shrub frequency 1.7 (0.9) 22.5 (5.8) 47.7 (2.6) 88.3 (4.2) 64.5 (6.8) 49.0 (7.7) 25.8 (10.9) 21.5 (10.0) 6.0 (3.3) Obligate seeder frequency 1.5 (0.8) 19.8 (4.9) 43.8 (3.3) 72.0 (6.0) 55.3 (5.7) 49.0 (7.7) 12.8 (10.1) 19.3 (10.4) 0.7 (0.7) Resprouter frequency 9.0 (1.2) 18.3 (11.9) 45.0 (8.1) 96.2 (2.6) 23.7 (4.8) 36.8 (9.2) 25.7 (5.8) 9.2 (2.9) 15.7 (2.7) Exotic shrub frequency 0.3 (0.2) 0.0 (0.0) 0.0 (0.0) 1.3 (1.3) 6.0 (5.5) 8.0 (4.3) 52.5 (7.4) 6.2 (1.9) 31.8 (17.7) 386
Adult trees/ha 221 (31) 450 (111) 340 (16) 149 (33) 392 (19) 172 (3) 129 (44) 190 (7) 339 (41) Adult eucalypts/ha 114 (18) 438 (113) 339 (18) 147 (32) 392 (19) 168 (1) 129 (44) 190 (7) 339 (41) Juvenile trees/ha 886 (238) 497 (265) 669 (24) 363 (111) 282 (60) 1231 (70) 206 (89) 594 (157) 378 (141) Juvenile eucalypts/ha 378 (101) 486 (263) 668 (24) 356 (106) 281 (60) 1040 (210) 206 (89) 590 (159) 378 (141) Sapling trees/ha 115 (48) 79 (44) 29 (9) 107 (48) 54 (13) 240 (53) 197 (77) 29 (4) 185 (91) Sapling eucalypts/ha 42 (16) 57 (29) 29 (9) 104 (46) 54 (13) 225 (65) 197 (77) 29 (4) 183 (92)
Basal area of adult trees (m2/ha) 12.3 (0.8) 15.9 (1.4) 14.1 (0.4) 10.2 (1.0) 15.6 (0.5) 10.9 (0.8) 10.4 (2.7) 10.5 (0.4) 15.5 (0.4) Basal area of adult eucalypts (m2/ha) 8.1 (1.7) 15.6 (1.3) 14.1 (0.5) 10.2 (1.0) 15.6 (0.5) 10.8 (0.8) 10.4 (2.7) 10.5 (0.4) 15.5 (0.4) Percent non-adult trees which were saplings 15.3 (4.4) 14.5 (1.6) 5.4 (2.5) 19.4 (4.3) 16.3 (3.8) 15.4 (2.6) 43.6 (5.8) 5.9 (1.7) 31.9 (6.1) Percent non-adult eucalytps which were saplings 10.4 (1.0) 8.9 (2.2) 5.4 (2.5) 19.0 (4.0) 16.4 (3.7) 17.1 (1.2) 43.6 (5.8) 5.9 (1.7) 31.7 (6.3)
Grass cover index 68.4 (9.1) 63.0 (4.6) 75.3 (3.0) 54.1 (6.7) 66.8 (7.1) 78.8 (2.6) 65.8 (2.8) 61.7 (5.5) 57.7 (3.2) Themeda index 57.7 (20.1) 62.8 (7.2) 82.7 (9.4) 69.5 (9.6) 72.5 (10.8) 82.0 (9.0) 21.7 (4.5) 46.2 (12.6) 0.5 (0.3) Microlaena index 13.3 (7.6) 5.2 (0.9) 8.7 (4.9) 14.8 (6.2) 8.7 (3.0) 6.3 (5.8) 28.0 (9.5) 7.3 (1.8) 92.7 (0.9) Aristida index 12.7 (5.6) 23.8 (7.8) 1.0 (1.0) 0.7 (0.7) 7.7 (1.4) 2.8 (1.2) 21.3 (6.7) 24.3 (6.7) 1.2 (0.9)
Table A9.2. Mean (S.E.) of landscape study variables for three fire frequency categories. Variables are defined in Section 4.2.3. Frequencies and indices are scaled from 0 to 100.
Variable Mean (SE) High fire frequency Moderate fire Low fire frequency frequency Bursaria frequency 29.2 (0.9) 36.0 (4.7) 85.7 (6.6) Bursaria dominance index 17.0 (1.8) 25.5 (4.9) 66.8 (5.2) Bursaria density (plants over 80cm/ha) 92 (23) 272 (62) 1872 (713)
Other native shrub species richness (/200m2) 4.4 (1.4) 6.6 (0.3) 2.5 (0.5) Other native shrub frequency 35.8 (16.5) 77.9 (10.7) 25.8 (5.3) Subshrub frequency 22.3 (10.1) 44.9 (24.2) 9.3 (1.8) Larger shrub frequency 23.9 (13.3) 67.3 (11.4) 17.8 (6.0) Obligate seeder frequency 21.7 (12.3) 58.8 (6.9) 10.9 (5.5) Resprouter frequency 24.1 (10.8) 52.2 (22.3) 16.8 (4.8) Exotic shrub frequency 0.1 (0.1) 5.1 (2.0) 30.2 (13.4) 387 Adult trees/ha 337 (66) 238 (77) 219 (62) Adult eucalypts/ha 297 (96) 236 (78) 219 (62) Juvenile trees/ha 684 (113) 625 (304) 393 (113) Juvenile eucalypts/ha 511 (85) 559 (242) 391 (111) Sapling trees/ha 75 (25) 134 (55) 137 (54) Sapling eucalypts/ha 43 (8) 128 (51) 137 (54)
Basal area of adult trees (m2/ha) 14.1 (1.0) 12.2 (1.7) 12.2 (1.7) Basal area of adult eucalypts (m2/ha) 12.6 (2.3) 12.2 (1.7) 12.2 (1.7) Percent non-adult trees which were saplings11.7 (3.2) 17.0 (1.2) 27.1 (11.2) Percent non-adult eucalypts which were saplings8.2 (1.5) 17.5 (0.8) 27.1 (11.1)
Grass cover index 68.9 (3.6) 66.6 (7.1) 61.7 (2.3) Themeda index 67.7 (7.6) 74.7 (3.8) 22.8 (13.2) Microlaena index 9.1 (2.4) 9.9 (2.5) 42.7 (25.7) Aristida index 12.5 (6.6) 3.7 (2.1) 15.6 (7.3)
Appendix 10 Grazing-intolerant species
Species used in the grazing impact index are listed below, along with their source.
Prober and Theile (1995) identified two groups of grazing-sensitive species in the White Box woodlands of the Central Western Slopes region of NSW. Group 1 species declined with any livestock grazing. Group 2 species were tolerant of light, but declined with more extensive, grazing.
P.J. Clarke (2003) also identified two groups of grazing-sensitive species. Both groups were intolerant of intense pastoralism, however their substrates differed: one was found on basalt, the other on metasediments and granite.
Table A10.1. Species found by previous researchers to be in reduced abundance due to grazing. Categories detailed above.
Species Prober and Theile (1995) P.J. Clarke (2003) Ajuga australis basalt Arthropodium milleflorum basalt Asperula conferta Group 2 Brunoniella australis basalt Bulbine bulbosa Group 2 basalt Cheilanthes seiberi Group 2 Convolvulus erubescens Group 2 Cymbonotus lawsonianus Group 2 Dianella longifolia Group 1 Dianella revoluta Group 1 basalt Einadia nutans Group 2 Elymus scaber Group 2 Galium propinquum metasediments and granite Geranium solanderi Group 2 Glycine tabacina Group 1 metasediments and granite Hardenbergia violaceae metasediments and granite Lomandra multiflora Group 2 metasediments and granite Poranthera microphylla metasediments and granite Stackhousia viminea metasediments and granite Themeda australis Group 1 Tricoryne elatior Group 2
388 Appendix 11 Dry weight, litter depth, bare ground and understorey cover, by data point
Table A11.1. Means and standard errors for 14 data points in Cumberland Plain woodland. Variables: weight of oven-dry fuel (tonnes/hectare), residuals in relation to 14-point curve (tonnes/hectare), litter depth (cm), percent bare ground, and percent understorey foliage cover (UFC). *, data point omitted from 11 point curve. -, data point omitted from fire recency analysis. Data Site Time Fire Fire Weight dry Residuals Litter depth Bare ground UFC (percent) point since recency frequency fuel (t/ha) (t/ha) (cm) (percent) number fire category category (yrs) 1 Ropes Creek 0 - frequent 1.02 (0.16) -0.37 0.42 (0.06) 45 (10) 5 (0) 2 Shanes Park 0.5 very recent frequent 2.70 (0.32) -0.27 0.81 (0.26) 20 (5) 5 (0) 3 Orchard Hills 0.6 very recent infrequent 3.43 (0.58) 0.18 0.71 (0.05) 20 (5) 20 (15) 4* Scheyville 0.7 very recent infrequent 5.22 (0.30) 1.70 0.90 (0.11) 13 (3) 25 (0) 5 Ropes Creek 0.8 very recent frequent 3.30 (1.28) -0.47 0.53 (0.27) 20 (5) 8 (3) 389 6 Shanes Park 1.7 recent frequent 3.93 (1.24) -1.64 0.95 (0.20) 15 (0) 5 (0) 7 Orchard Hills 2.4 recent infrequent 5.57 (0.26) -0.97 1.22 (0.02) 15 (0) 18(3) 8* Scheyville 2.5 recent infrequent 9.05 (0.98) 2.39 2.35 (0.05) 5 (0) 33 (3) 9 Plumpton 2.9 recent frequent 6.59 (0.78) -0.47 1.82 (0.47) 10 (5) 8 (3) 10 Windsor Downs 5.8 - frequent 8.35 (0.90) -0.20 2.29 (0.09) 5 (0) 15 (0) 11 Pitt Town 9.8 not recent frequent 9.26 (2.18) 0.26 4.21 (0.26) 5 (0) 8 (3) 12 Mt Annan 21.6 not recent infrequent 7.74 (0.01) -1.34 1.61 (0.08) 10 (0) 33 (3) 13 Richmond 22 not recent infrequent 8.73 (0.79) -0.35 1.83 (0.16) 5 (0) 10 (0) 14* Scheyville 50 not recent infrequent 10.65 (1.35) 1.57 2.02 (0.07) 5 (0) 40 (5)
Appendix 12 Fuel components by data point
A. Composition of fine fuel in frequently burnt sites
12
10
8
6
4 Fine fuel load (t/ha) load Fine fuel 2
0 0 0.5 0.8 1.7 2.9 5.8 9.8 Time since fire (years)
B. Composition of fine fuel in infrequently burnt sites
12
10
8
6
4 Fine fuel load (t/ha) load Fine fuel 2
0 0.6 0.7 2.4 2.5 21.6 22 50 Time since fire (years)
Figure A12.1. Mean dry weight of four components of CPW fuels, for A. seven data points from frequently burnt areas, and B. seven data points from infrequently burnt areas. Green, material from tree species; pink, material from shrub species; cream, material from grasses and other herbaceous species; brown, the ‘comminuted fraction’.
390 A. Points from frequent burnt areas
100%
80%
60%
40%
20% Percent total fine load fuel total Percent
0% 0 0.5 0.8 1.7 2.9 5.8 9.8 Time since fire (years)
B. Points from infrequently burnt areas
100%
80%
60%
40%
20% Percent total fine load fuel total Percent
0% 0.6 0.7 2.4 2.5 21.6 22 50 Time since fire (years)
Figure A12.2. Percent contribution of four components to total dry weight of fine fuel, for A. seven data points from frequently burnt areas, and B. seven data points from infrequently burnt areas. Green, material from tree species; pink, material from shrub species; cream, material from grasses and other herbaceous species; brown, the ‘comminuted fraction’.
391